Climate Change Impacts Assessment on Snow Hydrology and Irrigation in Turkey
Abdullah Gokhan Yilmaz
Submitted in total fulfilment of the requirements for the degree of Doctor of Philosophy
Faculty of Engineering and Industrial Sciences Swinburne University of Technology 2010 Abstract
Climate is changing in a way that can not be explained by natural variability. One of the most affected areas by global warming is hydrology and water resources. Regions where majority of runoff consists of snow melt, which is an important water resource to many aspects of hydrology including water supply, flood control and erosion, are more sensitive to climate change, particularly temperature increases. Euphrates River located in the mountainous Eastern Anatolia in Turkey, is one of the major rivers within Middle Eastern countries. Euphrates Basin is the largest basin of Turkey and it has 17% of entire country’s water potential. Snow is the main water source of Euphrates Basin, particularly for headwater of Euphrates Basin, which is also called Karasu Basin. Despite the significance of Euphrates Basin, there are very limited hydrological modelling studies for this basin. Moreover, impacts of climate change on hydrology of Euphrates Basin have not received sufficient attention. One of the most snow dominated subbasin of Euphrates Basin, Karasu Basin, was selected as study area and climate change effects were investigated for this subbasin in this study. Firstly trend analysis of hydro-meteorological data was performed based on non-parametric trend tests for better understanding of historical trends and investigating climate change finger prints in Karasu Basin. Statistically significant warming trends were detected in particular since 1992. Then, historical flows were simulated by using two conventional hydrological models. In addition, Artificial Neural Networks based hydrological model was developed to simulate runoffs in Karasu Basin. Then, future climate data over period of 2070-2100 was preceded for Karasu Basin and projected outcomes of climate models were used in calibrated hydrological models to predict future streamflows of Karasu Basin for the purpose of having an idea on future water availability in the basin. Largest decreases in streamflows were detected for summer season due to larger temperature increases and precipitation decreases. Runoff decreases were modelled for all seasons except autumn season. Based on available data, impacts of climate change on water sectors in Karasu and Euphrates basin were discussed.
ii Agriculture sector is the largest water consumer with 70% over the world and it has a direct relation with climate. Thus, it is among the most sensitive sectors to climate change. Although this study does not include any modelling study on agriculture and climate change relation, climate change impacts on agriculture and irrigation were explained in detail based on studies in the literature for different parts of the world and Turkey.
iii Acknowledgment
First and foremost, I would like to show my deepest gratitude and appreciation to my supervisor, Dr. Monzur Imteaz for his excellent supervision and friendship throughout my candidature. Without his constant support, it would be very challenging to complete my study. I also thank my co-supervisor Dr. Shirley Gato-Trinidad for her support during my study. I am grateful to Turkish State Meteorological Service in particular to Ismail Demir for providing historical and future climate data for my study. I also would like to thank Atilla Gurbuz from General Directorate of Electrical Power Resources Survey and Development Administration for supporting my study by providing hydrological data.
I would like to show my gratitude to my father, Halil Ibrahim Yilmaz, for his encouragement and excellent mentoring during my entire life and my PhD. I owe my sincere gratitude to my wife, Meleknur Yilmaz, for being such a supportive & thoughtful partner during my study and being shine of my life. I also would like to thank Associate Prof Ashish Sharma and water research group at the University of New South Wales for helping me a lot during initial years of my study.
Lastly, I present my regards and blessings to all of those who supported me in any respect during the completion of the study.
iv Declaration
I hereby declare that this thesis contains no material which has been accepted for the award of any other degree or diploma, except where due reference is made in the text of the thesis. To the best of my knowledge, this thesis contains no material previously published or written by another person except where due reference is made in the text of the thesis. Parts of the work described in Chapter two, three, four, five, six and seven have previously appeared or currently under review in the following conference or journal papers:
Yilmaz, AG, Imteaz, MA, Gato-Trinidad, S & Hossain, I 2010,’Climate Change Finger Prints in Mountainous Upper Euphrates Basin’, International Journal of Environmental Science and Engineering, vol.3 (1), pp. 13-21.
Yilmaz, AG & Imteaz, MA 2010, ‘Climate Change Impact on Runoffs in the Head Region of Euphrates Basin’, Journal of the International Association of Hydrological Sciences, Under review.
Yilmaz, AG & Imteaz, MA 2010, ‘Simulation of Snow Runoffs in Upper Euphrates Basin by Using Two Lumped Conceptual Hydrological Models’, Journal of Hydrologic Engineering, Under review.
Yilmaz, AG & Imteaz, MA 2010, ‘Climate Change in Snow Dominated Upper Euphrates Basin for the last three decades’, River Basin Management 2011 , 25 May – 27 May 2011, Riverside, California, USA, Preliminary acceptance.
Yilmaz, AG & Imteaz, MA 2010, ‘Development of a Hydrologic Model Using Artificial Intelligence for Upper Euphrates Basin in Turkey’, 4th International Conference on Water Resources and Arid Environments 2010 (ICWRAE 4), Riyadh, Saudi Arabia.
v Yilmaz, AG & Bakir, H 2008, ‘Climate Change Effects on Snow Coverage and Snow Melting’, Second National Snow Hydrology Congress, Erzurum, Turkey.
Name: Abdullah Gokhan Yilmaz
Signed:
Date:
vi Table of Contents
Abstract…………………………………………………………………………………………………………………………………….ii
Acknowledgment……………………………………………………………………………………………………………………..iv
Declaration……………………………………………………………………………………………………………………………….v
Table of Contents……………………………………………………………………………………………………………………vii
List of Figures………………………………………………………………………………………………………………………….xi
List of Tables………………………………………………………………………………………………………………………….xvi
Chapter 1: Introduction...... 1 1.1. Hydrological Significance of Snow...... 1 1.2. Climate Change Impacts on Snow Hydrology ...... 2 1.3. Problem Definition...... 3 1.4. Steps of Study...... 6 Chapter 2: Climate Change...... 8 2.1. Greenhouse Effect...... 10 2.2. Human Impact on Climate Change ...... 12 2.3. Observed Changes of Water Related Climate Variables...... 15 2.3.1. Temperature Changes...... 15 2.3.2. Precipitation Changes ...... 16 2.3.3. Sea Level Changes...... 19 2.3.4. Evapotranspiration Changes...... 21 2.3.5. Runoff and River Discharge Changes ...... 21 2.3.6. Extreme Weather Events Changes...... 22 2.3.7. Changes in Cryosphere Components...... 22 2.4. Future Climate Projections in Global Scale ...... 29 2.4.1. Temperature Projections ...... 30 2.4.2. Precipitation Projections...... 31 2.4.3. Snow and Land Ice Projections ...... 34 2.4.4. Sea Level Projections ...... 35 2.4.5. Evapotranspiration Projections ...... 36 2.4.6. Soil Moisture Projections...... 36 2.4.7. Runoff and River Discharge Projections...... 37
vii 2.5. Climate Change Impacts on Hydrology and Water Resources ...... 40 2.5.1. Observed Impacts ...... 40 2.5.1.1. Change in surface water and groundwater systems………………………………………40 2.5.1.2. Change in Water Quality……………………………………………………………………………….40 2.5.1.3. Change in Flood Events…………………………………………………………………………………41 2.5.1.4. Change in Drought Events…………………………………………………………………………….42 2.5.2. Climate Change Impacts on Future Water Availability and Demand……………………..48 2.5.2.1. Climate Change Impact on Ground Water…………………………………………………….48 2.5.2.2. Climate Change Impact on Floods…………………………………………………………………50 2.5.2.3. Climate Change Impact on Droughts…………………………………………………………….52 2.5.2.4. Climate Change Impact on Water Quality……………………………………………………..54 2.5.2.5. Climate Change Impact on Water Erosion and Sedimentation……………………..55 2.5.2.6. Climate Change Impact on Future Freshwater Availability……………………………56 2.5.2.7. Climate Change Impact on Future Freshwater Demand………………………………..56 2.5.2.8. Climate Change Impact on Future Water Stress……………………………………………57 2.5.2.9. Climate Change Impact on Water Costs and Other Socio-economic Aspects of Freshwater……………………………………………………………………………………………….58 2.5.2.10. Vulnerable Freshwater Areas to Climate Change………………………………………..62 2.5.2.11. Uncertainties in Impact Projection Studies………………………………………………….63 2.5.3. Water Issues Based Adaptation to Climate Change ...... 63 2.6. Climate Change Impacts on Ecosystems and Biodiversity...... 65 2.7. Climate Change Impacts on Agriculture and Irrigation...... 65 2.7.1. Reducing Impacts of Climate Change on Agriculture Sector ...... 76 2.7.2. Climate Change and Irrigation ...... 79 2.7.3. Climate Change Impacts on Agriculture of Turkish Republic...... 126 Chapter 3: Study Area ...... 139 3.1. Turkey...... 139 3.1.1. Turkey’s Climate...... 142 3.1.2. Greenhouse Gases Emission Sectors in Turkey ...... 145 3.1.2.1. Energy………………………………………………………………………………………………………………145 3.1.2.2. Transportation………………………………………………………………………………………………….147 3.1.2.3. Industry……………………………………………………………………………………………………………148 3.1.2.4. Residential Impacts…………………………………………………………………………………………..149 3.1.2.5. Solid Waste……………………………………………………………………………………………………….150
viii 3.1.2.6. Agriculture…………………………………………………………………………………………………..150 3.1.2.7. Livestock Sector…………………………………………………………………………………………..150 3.1.2.8. Forestry…………………………………………………………………………………………………….…151 3.1.3. Historical Changes in Climate Variables in Turkey………………………………………………..151 3.1.3.1. Changes in Temperature………………………………………………………………………….….152 3.1.3.2. Changes in Precipitation………………………………………………………………………………163 3.1.3.3. Changes in Streamflow……………………………………………………………………..…………167 3.1.4. Future Climate Projections for Turkey...... 168 3.1.5. Exposure Potential of Turkey to Climate Change Impacts and Adaptation Need .. 175 3.2. Karasu Basin ...... 178 Chapter 4: Literature Review ...... 183 Chapter 5: Methodology ...... 207 5.1. Trend Analysis Methods ...... 207 5.1.1. Mann-Kendal Non-parametric Trend Analysis……………………………………………………..207 5.1.2. Spearman’s Rho Trend Analysis...... 208 5.2. Hydrological Models ...... 209 5.2.1. HEC-HMS Model...... 213 5.2.2. LBRM...... 214 5.2.3. Artificial Neural Network (ANN) Model ...... 217 5.2.3.1. Introduction to Artificial Neural Networks…………………………………………………..217 5.2.3.2. Training…………………………………………………………………………………………………….…219 5.2.3.3. Popular ANNs Architectures and Training Algorithms in Hydrological Studies…………………………………………………………………………………………………220 5.3. Climate Models ...... 221 5.3.1. Global Climate Model – ECHAM5 ...... 228 5.3.2. Regional Climate Model – RegCM3 ...... 229 5.4. Interface between Climate Model and Hydrological Model ...... 230 Chapter 6: Results and Discussion ...... 233 6.1. Historical Hydro-meteorological Data Trend Analysis Results ...... 233 6.1.1. Precipitation Trends...... 233 6.1.2. Temperature Trends ...... 243 6.1.3. Flow Trends...... 268 6.2. Historical Runoff Simulations ...... 270 6.2.1. HEC-HMS Results ...... 270
ix 6.2.2. LBRM Results...... 273 6.2.3. ANN Results ...... 277 6.2.3.1. Data Set Selection and Standardization……………………………………………………….277 6.2.3.2. Network Type and Training Algorithm Selection………………………………………….280 6.2.3.3. Annual ANN model………………………………………………………………………………….…..280 6.2.3.4. Seasonal ANN Model……………………………………………………………………………………284 6.3. Future Climate and Flow Projections ...... 286 6.3.1.Temperature and Precipitation Projections in Karasu Basin ...... 286 6.3.2.Streamflow Projections in Karasu Basin ...... 288 6.3.3. Influences of Streamflow Decrease in Karasu Basin...... 290 6.4. Uncertainties of Study ...... 293 Chapter 7: Conclusion and Recommendations ...... 296 References ...... 305
x List of Figures
FIGURE PAGE
Chapter 2
Fig.2. 1: Change in Earth temperature due to natural forces and human effects ...... 10 Fig.2. 2: Schematic illustration of greenhouse effect...... 11 Fig.2. 3: Increase in greenhouse gases amount in atmosphere...... 13 Fig.2. 4: Global mean temperature trend ...... 16 Fig.2. 5: Mean annual precipitation anomalies (%) over land from 1900 to 2005 relative their 1961–1990 means...... 17 Fig.2. 6: Changes in temperature, sea level and snow cover area...... 20 Fig.2. 7: Surface air temperature (A and E), Northern Hemisphere seasonally frozen ground extent (B), NH snow cover extent for March–April (C), and global glacier mass balance (D) anomaly time series...... 23 Fig.2. 8: Average snow covered area in March-April from Brown (2000) ...... 25 Fig.2. 9: Summary of observed changes in cryosphere components...... 28 Fig.2. 10: Average projected global surface temperature changes in 2020-2029 and 2090-2099 relative to 1980-1999 based on B1, A1B and A2 scenarios ...... 31 Fig.2. 11: Global precipitation, soil moisture, runoff and evaporation changes for the scenario SRES A1B for the period 2080–2099 relative to 1980–1999 ...... 32 Fig.2. 12: Changes in extremes based on nine global coupled climate models in 2080–2099 relative to 1980–1999 for the A1B scenario ...... 34 Fig.2. 13: Annual runoff changes at global scale for the period 2090–2099, relative to 1980–1999 based on study of Milly, Dunne & Vecchia (2005) ...... 38 Fig.2. 14: Average annual runoff change by the 2050s based on different climate models ...... 39
xi Fig.2. 15: Drought analysis based on monthly Palmer Drought Severity Index (PDSI) for the period of 1900-2002...... 43 Fig.2. 16: Turkey annual total precipitation anomalies (mm)...... 46 Fig.2. 17: Aegean, Central Anatolia and Marmara regions cumulative precipitations...... 47 Fig.2. 18: Climate change impact on long-term average annual diffuse groundwater recharge ...... 49 Fig.2. 19: Change of 100-year droughts future recurrence...... 53 Fig.2. 20: Map of some regions where are sensitive to climate change impacts on freshwater availability...... 62 Fig.2. 21: Greenhouse emission sources...... 67 Fig.2. 22: Graphical illustration of possible advantages and disadvantages of climate change on agriculture...... 70 Fig.2. 23: Change in Australian agricultural production owing to climate change by 2030 and 2050...... 72 Fig.2. 24: Growing season length increase in Turkey...... 128 Fig.2. 25: Regional rainfall in 2007 in Turkey ...... 134 Fig.2. 26: Yield changes in 2007 in Turkey...... 135 Fig.2. 27: Regional yield percentage changes of wheat, corn, sun flower and cotton crops due to 2007 drought...... 136 Fig.2. 28: Production and price percentage change of wheat, corn, rice, lentil, sun flower, tomato, beef and milk...... 137 Fig.2. 29: Production change of wheat, barley, corn, cotton and sun flower owing to drought conditions...... 138 Chapter 3
Fig.3. 1: Location map of Turkey...... 139 Fig.3. 2: Geographical regions in Turkey...... 141 Fig.3. 3: Climate zones of Turkey...... 142 Fig.3. 4: Distribution of mean annual temperature in Turkey...... 144 Fig.3. 5: Distribution of mean annual rainfall in Turkey...... 144 Fig.3. 6: Trend of energy use in Turkey...... 146
xii Fig.3. 7: The distribution of energy consumption by fuels in 1990 and 2004...... 147 Fig.3. 8: Road network in Turkey...... 148 Fig.3. 9: Building heating requirement zones in Turkey (Unit: kwh/m2)...... 149 Fig.3. 10: Temperature trend in Turkey between 1941 and 2007...... 153 Fig.3. 11: Average air temperature trend in Turkey between 1950 and 2004...... 154 Fig.3. 12: Local relative temperature differences and their low pass filtered signals……………………………………………………………………………………………………. 155 Fig.3. 13: Mann-Kendal trend analysis results a) Maximum temperature b) Minimum temperature c) Average temperature...... 157 Fig.3. 14: Changing trend in annual average temperatures (1952-2006)...... 159 Fig.3. 15: Changing trend in annual maximum temperatures (1952-2006)...... 160 Fig.3. 16: Changing trend in annual minimum temperatures (1952-2006)...... 161 Fig.3. 17: Seasonal annual average temperature trends...... 162 Fig.3. 18: Turkey annual average precipitation (1941-2007)...... 163 Fig.3. 19: Long term geographical precipitation distribution of Turkey...... 164 Fig.3. 20: 1951-2004 seasonal precipitation trends a) Winter b) Spring c)Summer d)Autumn...... 165 Fig.3. 21: Regional precipitation trend analysis for 30 year slices...... 166 Fig.3. 22: Streamflow trend analysis in Turkey...... 168 Fig.3. 23: Expected seasonal temperature change in 2070-2100 relative to 1961-1990 (a) Winter season (b) Spring Season (c) Summer season (d) Autumn season...... 169 Fig.3. 24: Expected seasonal precipitation change in 2070-2100 relative to 1961-1990 (a) Winter season (b) Spring Season (c) Summer season (d) Autumn season...... 170 Fig.3. 25: Mean temperature difference map of Turkey between 1960-1990 and 2070-2100 periods...... 172 Fig.3. 26: Mean temperature difference graph of Turkey between 1960-1990 and 2070-2100 periods...... 172 Fig.3. 27: Average annual precipitation difference in mm/year between 1960-1990 and 2071-2100 periods...... 173
xiii Fig.3. 28: Annual precipitation change rate in percentage between reference and future climate...... 174 Fig.3. 29: Winter season snow depth change (mm)...... 175 Fig.3. 30: Location map of Karasu Basin...... 180 Fig.3. 31: Karasu Basin elevation map…………………………………………………………….… 180 Fig. 3.32: Meteorology and flow stations in Karasu Basin...... 181 Chapter 5
Fig. 5.1: Tank cascade schematic...... 216 Fig. 5.2: Feed-forward network architecture...... 218 Fig. 5.3: Schematic diagram of single node...... 219 Fig. 5.4: Schematic of the climate system...... 221 Fig. 5.5: Graphical illustration of climate model...... 222 Fig. 5.6: Global mean near-surface temperatures over the 20th century...... 223 Fig. 5.7: Schematic illustration of global climate model...... 224 Fig. 5.8: Improvement of GCMs resolutions at IPCC assessment reports...... 226 Fig. 5.9: Schematic illustration of hydrological impact study steps...... 231 Chapter 6
Fig. 6.1: Total annual precipitation graphs of meteorology stations in Karasu Basin...... 234 Fig. 6.2: Winter season precipitation trends at meteorological stations...... 236 Fig. 6.3: Summer season precipitation trends at meteorological stations...... 238 Fig. 6.4: Spring season precipitation trends at meteorological stations...... 240 Fig. 6.5: Autumn season precipitation trends at meteorological stations...... 242 Fig. 6.6: Annual average, minimum and maximum temperatures of Erzurum station……………………………………………………………………………………………. 244 Fig. 6.7: Annual average, minimum and maximum temperatures of Erzincan station………………………………………………………………………………………….. 246 Fig. 6.8: Annual average, minimum and maximum temperatures of Tercan station...... 248 Fig. 6.9: Seasonal average temperature trends in Erzurum...... 250 Fig. 6.10: Seasonal minimum temperature trends in Erzurum...... 252
xiv Fig. 6.11: Seasonal maximum temperature trends in Erzurum...... 254 Fig. 6.12: Seasonal average temperature trends in Erzincan...... 256 Fig. 6.13: Seasonal minimum temperature trends in Erzincan...... 258 Fig. 6.14: Seasonal maximum temperature trends in Erzincan...... 259 Fig. 6.15: Seasonal average temperature trends in Tercan...... 261 Fig. 6.16: Seasonal minimum temperature trends in Tercan...... 263 Fig. 6.17: Seasonal maximum temperature trends in Tercan...... 265 Fig. 6.18: Summer season maximum temperature trends...... 267 Fig. 6.19: Time series graph of annual average runoff in Karasu Basin...... 269 Fig. 6.20: HEC-HMS comparison hydrograph for calibration run at outlet point of Karasu Basin...... 271 Fig. 6.21: HEC-HMS comparison hydrograph for validation run at outlet point of Karasu Basin...... 272 Fig. 6.22: LBRM comparison hydrograph for calibration run at outlet point of Karasu Basin...... 273 Fig. 6.23: LBRM comparison hydrograph for validation run at outlet point of Karasu Basin...... 274 Fig. 6.24: Precipitation time series from January 1995 to December 2004...... 275 Fig. 6.25: Observed flow time series for modelling period...... 276 Fig. 6.26: Flow time series graphs for training and test periods...... 278 Fig. 6.27: Runoff graph in Karasu Basin in 1995...... 279 Fig. 6.28: Flow scatter diagrams of training and test phases in annual analysis. 282 Fig. 6.29: a) Observed-Modelled flow graph at Karasu Basin outlet point for training phase b) Observed-Modelled flow graph at Karasu Basin outlet point for test phase.. 283 Fig. 6.30: Seasonal flow scatter diagrams of training and test phases in seasonal analysis…………………………………………………………………………………………. 285 Fig. 6.31: a) Seasonal observed-modelled flow graphs for training phase b) Seasonal observed-modelled flow graphs for test phase...... 286 Fig. 6.32: HEC-HMS and LBRM monthly average flow estimations for 2070-2100...... 289
xv List of Tables
TABLE PAGE
Chapter 2
Table 2.1: Monthly snow covered area trend in Northern Hemisphere (106 km2 per decade)...... 25 Table 2.2: IPCC scenarios population and economy conditions ...... 29
Table 2.3: IPCC scenarios CO2 emission predictions...... 30 Table 2.4: Observed changes in runoff/streamflow, lake levels, floods and droughts...... 45 Table 2.5: Possible influences of changes in extreme precipitations due to climate change based on climate projections to the mid- to late 21st century...... 51 Table 2.6: Population growth and climate change effects on the population living in water stressed catchments...... 58 Table 2.7: Summary of supply-side and demand-side adaptation options...... 63 Table 2.8: Some adaptation examples...... 64 Table 2.9: Climate change impact on agricultural productivity at 2050...... 71 Table 2.10: Regional impacts of climate change on wheat production at 10 regions over Australia...... 72 Table 2.11: Grain yield estimations of paddy rice for possible climate change scenarios...... 133 Chapter 3
Table 3.1: Primary energy resources of Turkey...... 146 Table 3.2: Manufacturing industry indicators...... 149 Chapter 6
Table 6.1: Mann-Kendal & Spearman’s Rho trend analysis results for annual precipitation………………………………………………………………………………………………………. 235 Table 6.2: Winter season precipitation trend analysis results at meteorological stations.. 237
xvi Table 6.3: Summer season precipitation trend analysis results at meteorological stations………………………………………………………………………………………………….. 239 Table6.4: Spring season precipitation trend analysis results at meteorological stations.....241 Table 6.5: Autumn season precipitation trend analysis results at meteorological stations.243 Table 6.6: Average, maximum and minimum temperature trends of Erzurum Station..... 245 Table 6.7: Average, maximum and minimum temperature trends of Erzincan station...... 247 Table 6.8: Average, maximum and minimum temperature trends of Tercan station...... 249 Table 6.9: Trends analysis results of seasonal average temperature in Erzurum...... 251 Table 6.10: Trends analysis results of seasonal minimum temperature in Erzurum...... 253 Table 6.11: Trends analysis results of seasonal maximum temperature in Erzurum...... 255 Table 6.12: Trends analysis results of seasonal average temperature in Erzincan...... 257 Table 6.13: Trends analysis results of seasonal maximum temperature in Erzincan...... 260 Table 6.14: Trends analysis results of seasonal average temperatures in Tercan...... 262 Table 6.15: Trends analysis results of seasonal minimum temperature in Tercan...... 264 Table 6.16: Trends analysis results of seasonal maximum temperature in Tercan...... 266 Table 6.17: Significance levels of maximum temperature increasing trends in summer season...... 268 Table 6.18: Flow trend analysis in Karasu Basin...... 269 Table 6.19: Peak flow dates...... 270 Table 6.20: Calibration run performance in HEC-HMS...... 272 Table 6.21: Validation run performance in HEC-HMS...... 272 Table 6.22: Calibration run performance in LBRM...... 273 Table 6.23: Validation run performance in LBRM...... 274 Table 6.24: Basic statistical characteristics of used data...... 277 Table 6.25: Summary of different annual ANN runoff models...... 281 Table 6.26: Summary of different annual ANN runoff models...... 284 Table 6.27: Modelled changes in precipitation and air temperature in Karasu Basin over 2070-2100...... 288 Table 6.28: Annual surface water potential estimation of hydrological models...... 293
xvii Chapter 1: Introduction
1.1. Hydrological Significance of Snow
Cryosphere, which consists of snow, river and lake ice, sea ice, glaciers and ice caps, ice shelves and ice sheets, and frozen ground, is the second largest part of climate system. The cryosphere relation to climate change and variability is based on physical characteristics such as surface reflectivity. The cryosphere has 75% of world’s fresh water and all components of cryosphere have strong relation with climate change (Lemke et al. 2007).
Snow has positive feedbacks related to albedo, moisture storage and latent heat in the climate system. Albedo positive feedback is stronger than the other feedbacks of snow. Because of the high albedo of snow, snow has a significant impact on the surface energy budget and Earth radiative balance. The depth, age of a snow cover, solar radiation, vegetation and cloudiness are the effective factors on snow albedo and its feedback. Besides direct snow-albedo influence, snow can also affect climate indirectly such as summer soil moisture (Lemke et al. 2007).
Snow is sensitive to climate change so it is a good and reliable indicator of climate change. Thus, snow is utilized as a helpful indicator for both testing and monitoring climate change (Changchun 2007).
The snowpack in the mountains stores water naturally. Schlenker, Hanemann and Fischer (2007) explained that natural snowpack water storage almost equals to the water stored in state’s major reservoirs in California. Moreover, snowmelt runoff is major water source at snow dominated basins. It is used to provide sufficient water to vital sectors including agriculture, industry, hydropower generation and urban. Barnett et al. (2005) reported that according to a year 2000 population map, snowmelt- dominated regions contains approximately one-sixth of the world’s population. They also implied that snowmelt-dominated regions include much of the industrialized world, accounting for approximately one-quarter of the global gross domestic product.
1 Therefore, it is significant to express climate change and snow hydrology relation and it is imperative to investigate how changing climate affects hydrology at snow dominated basins.
1.2. Climate Change Impacts on Snow Hydrology
One of the most affected areas by global warming is hydrology and water resources. Climate change results increase in temperature and it changes precipitation patterns over many regions in the world. Change in temperature and precipitation leads to alteration of hydrological cycles resulting changes in streamflow regimes. As a result of global warming, changes in globally averaged water vapour concentrations, evaporation, precipitation, humidity and wind speed have been observed and it is expected to be observed with increasing intensity in following years (Beldring et al. 2008). However, each variable is defined in terms of its frequency distribution and its distribution over space and time, hence, each variable will be affected differently. Mentioned changes in climate variables lead significant changes on hydrological cycle. For example, any change in temperature and precipitation will directly influence runoff quantity and quality; moreover, it will also affect evapotranspiration. Therefore, any change in hydrological cycle and water resources will result significant effects on irrigation, hydro-electrical power generation and water supply.
Regions where majority of runoff consists of snow melt, which is an important water resource to many aspects of hydrology including water supply, flood control and erosion, are more sensitive to climate change, particularly changes in temperature. Temperature is the decisive variable for precipitation type and snow melt timing (Vicuna & Dracup 2007). Therefore, it is necessary to establish climate change- temperature-snow hydrology relations and to assess climate change impacts in global and regional scales.
Due to increasing temperatures, probability of rainfall is more than snowfall especially in beginning and end of the snow season particularly in places where temperatures are near freezing. It leads to reduced snowpack and increased rainfall. Snowpack is a very 2 important water resource especially for summer season. It is possible to observe snow covered area globally by continuous satellite measurements. Based on these observations, it is revealed that in many locations in Northern Hemisphere spring snow cover has declined by around 2% per decade since 1966 (IPCC 2007).
There are two basic impacts of climate change on snow hydrology. One of them is decrease of snowpack which leads problems to supply sufficient water for water users. The second one is shifts in snow melting time or early snow melting which may result floods in winter & spring and droughts in summer seasons.
Changes in snowpack amount and runoff timing (early runoffs) will high possibly result substantial modifications to the hydrologic cycle such as increased winter and spring flooding; changes in lake, stream, and wetland ecology; reduced warm season flows (Cayan et al. 2007) and decrease in water storages. For instance, forest fire frequency and intensity could increase as a result of lower summer soil moisture (Westerling et al. 2006). Furthermore, water supplies for sectors including agriculture, hydroelectric power generation and recreational use could be intensively influenced (Rauscher et al. 2008; Purkey et al. 2008; Markoff & Cullen 2008; Vicuna et al. 2008, Hayhoe et al. 2004). These significant conclusions require precautions such as additional reservoirs and/or extended reservoir capacity. Moreover, these changes to the hydrological cycle will lead social and economic changes/problems which require improving new and efficient management strategies for water and land use management in the future (Rauscher et al. 2008).
1.3. Problem Identification
Euphrates Basin is the largest basin of Turkey. Moreover, it is an international basin and its water is also used by Syria and Iraq. Largest dams of Turkey are located in Euphrates Basin. Ataturk, Keban and Karakaya dams are the most significant ones amongst these. Ataturk is the largest dam with a volume of 84.5x106 m3. Keban and Karakaya dam volumes are 16.7x106 m3, 2x106 m3 respectively (SHW 2010). These
3 dams were constructed for irrigation, water supply and power generation purposes. Moreover, there are very large irrigation districts in Euphrates Basin.
Despite the importance of Euphrates Basin for Turkey, Syria and Iraq, there are very limited streamflow simulation and prediction studies for basin. Moreover, any modelling study which investigates climate change impact on hydrology in Euphrates Basin could not find in literature. Most of the water in Euphrates Basin (60%-70% of all streamflow) is sourced from snow melt. Irrigation and hydro-power generation activities are very intense in Euphrates Basin. Any change in snowmelt runoff amount or seasonality can produce important problems for these sectors. Possible reduction in available water can also intensify competition among water sectors and water allocation can be a problematic issue in the future. Moreover, more dams and water structures are planned to be built in Euphrates Basin in near future. It means more water use for irrigation and hydropower generation. It also means more severe debates between Turkey, Syria and Iraq.
Climate change will very likely result substantial consequences in Euphrates Basin water availability. Climate change impacts on snow hydrology are explained above. Due to changing runoff amount and seasonality, Euphrates Basin can face difficulties to provide sufficient water to significant water sectors. Thus, climate change may affect Euphrates Basin socio-economically. It is a fact that developing countries will be affected socio-economically more than developed countries from global warming.
Karasu Basin is headwater of Euphrates Basin and it is a mountainous part of Euphrates Basin. Furthermore, it is one of the most snow dominated sub-basins in Euphrates Basin. Focus of this study is investigating climate change influences on snow hydrology in Turkey. Thus, one of the most snow dominated part of Turkey’s largest and one of the most important basins was selected as study area.
The main purpose of this study is to generate accurate predictions on future water availability of Karasu Basin. In other words, purpose of study is to understand how
4 climate change will affect streamflows of Karasu Basin in future. It is believed that despite not to perform any modelling for middle and lower Euphrates Basin, streamflow predictions on Upper Euphrates Basin helps to produce realistic forecasts on Euphrates Basin future water availability. Furthermore, influence of changing climate on water sectors was discussed in dependence to available data. In addition, some more information about climate change influence on agriculture and irrigation was presented under climate change chapter due to sensitivity of this sector to climate change because of being the largest consumer of available water. Finally, objectives of this study can be presented as follows:
Investigation of climate change finger prints on Karasu Basin’s hydrology by performing non-parametric trend analysis for hydro-meteorological data in Karasu Basin
Simulation of historical streamflows of Karasu Basin by using suitable hydrological models
Assessing future climate data for Karasu Basin and predicting future climate of Karasu Basin
Projecting future runoffs of Karasu Basin by using future climate outputs of climate models in calibrated hydrological models
Generating realistic predictions for future water availability of Karasu Basin
Discussing effects of future water availability on water sectors in Karasu Basin and Euphrates Basin
Producing a review on climate change impacts and agriculture relation by investigating relevant papers in literature for the purpose of generating initial steps to further modelling studies in Turkey.
5 1.4. Steps of Study
When studies on climate change impacts on hydrology and water resources have been investigated, it is realized that there are three major steps; (1) determining changes in climate variables by performing historical data analysis, (2) generating future climate data set by using climate models, (3) investigating climate change impacts on hydrology and water resources by using hydrological models.
The first step of climate change investigation on hydrology over a region is historical climate and streamflow data trend analysis. It is important to investigate whether climate change finger prints have already been observed and what type of trends the climate variables are following. Moreover, it is very beneficial to compare Global Climate Models’ future projections with historical data trends.
Secondly, outputs of Global Climate Models (GCMs) are used to predict future climate variables including temperature and precipitation in study basin. Climate models are mathematical representation of the major features of earth’s climate, explained as computer codes based on physical laws such as conservation of mass, energy and momentum coupled with observations. Although GCMs are the best tool to predict future climate currently, spatial resolution of GCMs are too coarse to use their outputs in hydrological impact studies. Thus, GCM outputs must be downscaled to finer spatial resolutions by using stochastic or dynamic approaches. In this study, data which is generated by downscaling ECHAM5 GCM outputs dynamically by using Regional Climate Model (RCM) named RegCM3, was used to predict future streamflows of study basin.
The next step in hydrological impact study is selecting suitable hydrological models to project basins’ future streamflows by using future climate data as an input to hydrological models. In this study, two lumped conceptual models which are using temperature index approach to calculate snow melt process were selected to model future streamflows. The names of the models are the Hydrologic Engineering Center- Hydrologic Modeling System (HEC-HMS) and Large Basin Runoff Model (LBRM). The 6 main reason for selecting these models was relevant data availability for the study area.
Moreover, Artificial Neural Networks (ANNs) based hydrological model was developed to simulate snow melt runoffs in study basin. Although ANN model showed sufficient performance to simulate runoffs in study basin; based on available future climate data, only HEC-HMS and LBRM were used to predict future streamflows of Karasu basin. Available water in catchment is used by different sectors including agriculture, industry, hydropower generation and urban. Agriculture is the largest consumer of available water resources with globally around 70% (Fischer et al. 2007). Thus, climate change on agriculture & irrigation and adaptation policies of agriculture sector are discussed separately in this study.
In summary, there are five steps of this study. Firstly, climate variables and flow data trend analysis were performed to detect trends in Karasu Basin. Then, outputs of RCM were preceded and assessed for Karasu Basin. Then, snow melt runoffs were simulated by using three models including HEC-HMS, LBRM and ANN models. As a next step, Karasu Basin’s future streamflows were predicted by using future climate data as an input to calibrated hydrological models. Finally, climate change impact on future water availability and influences of climate change on different sectors, especially agriculture, were discussed in this study.
7 Chapter 2: Climate Change
Climate system consists of the atmosphere, land surface, snow and ice, ocean & other water bodies and living things. It is a complicated and interactive system. Climate is generally described as ‘average weather’; it is defined based on the mean and variability of temperature, precipitation and wind over a period of time, ranging from months to millions of years. Changes in external factors also called forcing result effects on the internal dynamics of climate which leads evolve of the climate system. External forcing can be either natural phenomena such as solar radiations or anthropogenic changes in atmospheric composition (IPCC 2007).
Observations demonstrate that weather components such as temperature, rainfall have been changing with the progression of time. Climate change is defined as the statistics of these changes in weather over time. Climate change is sometimes confused with global warming but there is a difference between these two terms. While global warming absolutely focus on the Earth’s average surface temperature and described as the gradual increase of Earth’s surface temperature, climate change contains broader area and investigate changes in other climate parameters with temperature such as rainfall.
Particularly since industry revolution in the world, very significant changes have been observed in temperature, precipitation, snow and ice covered area. These changes have very important impacts on sea level rise, hydrological cycles-quantity and quality of water, agriculture, forests, ecosystems and health.
Global warming and climate change can occur both naturally and due to anthropogenic reasons. Climate has changed on all ages during Earth’s history. However, particularly after industrial era, it has started showing unusual changing trends. CO2 concentration in atmosphere has reached values which are more than the past half-million years. Due to greenhouse effect, global temperatures currently are “warmer than they have ever
8 been during at least the past five centuries” (IPCC 2007, p. 114). Amount of greenhouse gases particularly CO2 concentration in atmosphere has been increasing due to human activities. There are supporting evidences to state this idea. Firstly, the increase in atmospheric CO2 concentration closely follows the rise in emissions related to fossil fuel burning. Secondly, the inter-hemispheric gradient in atmospheric CO2 concentration is growing correspondingly with CO2 emissions. There is more land and population in Northern Hemisphere means higher human activities and CO2 emission, which is reflected in the CO2 growth in the Northern hemisphere compared to the Southern Hemisphere (McNeil 2009).
One of the important points while investigating climate change is not to confuse global and local changes. Local changes can be different from global changes. For instance, for a specific part of the world, local temperatures can decrease while global temperatures are increasing. Moreover, it is significant to separate time scales in climate studies. Climate changes over millions of years can be much larger than centennial time scale climate changes and have different results (IPCC 2007).
It is shown in Fig. 2.1 that the main reason for current climate change is human activities. In Fig. 2.1, blue band demonstrates how climate would change owing to natural forces only based on climate models. Red band shows how climate would change with the combination of natural and human forces. Black line illustrates actual observations of global surface temperature (IPCC 2007).
9 Fig.2.1: Change in Earth temperature due to natural forces and human effects (IPCC 2007)
Fig. 2.1 demonstrates the importance of human activities on climate change. Thus, it is substantial to comprehend how human activities influence climate of the earth.
2.1. Greenhouse Effect
Solar radiation has a great importance for climate system due to its effect on the total energy of climate system. Around 30% of the incoming solar radiation to the top of the atmosphere is reflected back to spaces by mostly clouds and small particles named aerosols in the atmosphere. Light coloured areas in earth surface such as snow, ice and deserts reflect the remaining one third of incoming solar radiation. The most important reason of aerosols in the atmosphere is volcanic eruptions. Normally rain clears aerosols from atmosphere but if aerosols are located far above the highest cloud, these aerosols typically impact the climate for about a year or two before being cleared by precipitation. Main volcanic eruptions may result cooling in global surface temperature around half a degree for months sometimes years (IPCC 2007). 10 Approximately 240 w/m2 of incoming energy is absorbed by the earth’s surface and atmosphere. For the purpose of balance the incoming energy, the earth emits same amount of energy back to space as long wave radiation. Earth’s surface would have a temperature around -19 °C to emit 240 w/m2 energy, however it is around 14 °C. Reason of 33 °C increase in surface temperature is existence of greenhouse gases which is a partial blanket for the longwave radiation coming from the surface. This effect is designated as the natural greenhouse effect. Carbon dioxide, methane, nitrous oxide, CFC-12, HCFC-22, perfluoromethane, sulphur hexa-fluoride and water vapour are the main greenhouse gases; however, carbon dioxide and water vapour are the most significant greenhouse gases. Clouds show similar influence of greenhouse gases by reflecting back incoming long wave radiation from earth surface, on the other hand they also reflect incoming solar radiation so overall its cooling effect is stronger than warming effect during day time (IPCC 2007). Schematic illustration of greenhouse effect is shown in Fig. 2.2.
Fig.2.2: Schematic illustration of greenhouse effect (IPPC 2007)
11 2.2. Human Impact on Climate Change
Human activities contribute to climate change by increasing amount of greenhouse gases, which is a natural blanket warming the Earth, aerosols and cloudiness in the atmosphere. Activities such as burning coal, oil and natural gas for energy supply, transportation activities, residential and commercial buildings, industry, agriculture particularly agricultural fertilizers, deforestation, land use change and waste disposal intensify the greenhouse effect. Among these activities, burning fossil fuels is the most significant contribution to climate change by human. Amount of carbon dioxide which is the most important greenhouse has increased around 35% in the industrial era owing to human activities (IPCC 2007).
Change in amount of aerosols and greenhouse gases alter incoming solar radiation and out-going infrared (thermal) radiation. It means change in Earth energy balance that can result cooling or warming of Earth. However, overall impact of human on Earth’s temperature is warming since the start of industrial era. Climate change occurred by human impact during industrial era extremely exceeds the climate change due to natural processes such as volcanic eruptions and solar changes (IPCC 2007).
Remarkable increases in all major greenhouse gases have been observed in the industrial era (since around 1750). Increase in greenhouse gases over the last 2000 years is shown in Fig. 2.3.
12 Fig.2.3: Increase in greenhouse gases amount in atmosphere (IPCC 2007)
Main human activities resulting carbon dioxide release are fossil fuel use in transpor- tation, building heating & cooling and the manufacture of cement & other goods.
Moreover, deforestation by human is another factor of increasing CO2 amounts in the atmosphere. Fossil fuel burning and cement manufacturing are reasons for more than
%75 of human-induced CO2 releases while land use change, particularly deforestation, is the reason of remaining CO2 emission. It is not possible to explain observed -1 atmospheric CO2 increase, which is from 3.2 to 4.1 GtC yr over the last 25 years, with natural carbon cycle (IPCC 2007). Human activities finger prints on the increase in CO2 concentration can be seen clearly by investigating character of CO2 in the atmosphere. “The ratio of heavy to light carbon atoms has changed in a way that can be attributed to addition of fossil fuel carbon” (IPCC 2007, p. 115). Moreover, while CO2 concentration has increased in the atmosphere, the ratio of oxygen to nitrogen has decreased, explained with oxygen decrease due to fossil fuel burning (IPCC 2007).
13 Methane (CH4) is another important greenhouse gas and human-caused methane exceeded natural emissions over the last 25 years. Main reason of increase in methane is naturally wetlands, termites, oceans, vegetation and CH4 hydrates. Human activities resulting increase in methane in atmosphere are energy production from coal and natural gas, waste disposal in landfills, raising ruminant animals and agricultural activities (IPCC 2007).
Nitrous oxide (N2O) is also released by human activities such as fertilizer usage in agriculture, biomass burning, raising cattle, some industrial activities and fossil fuel burning. For nitrous oxide, emissions generated by human activities and natural emissions to atmosphere are very similar. Halocarbon gas concentrations, which did not exist in the atmosphere before industrial era, have increased due to human activities particularly by refrigeration agents, which are using halocarbon gases extensively, also owing to some other industrial processes. Nevertheless, the amount of chlorofluorocarbon gases is declining based on international regulations designed to protect the ozone layer (IPCC 2007).
Water vapour is the most abundant greenhouse gas in the atmosphere. However, human activities do not have a direct influence on water vapour amount in the atmosphere. But, indirectly human activities impact water vapour concentration by changing climate. Finally, aerosols are small particles existing in the atmosphere with different size, concentration and chemical composition. While some aerosols are released to atmosphere directly, some of them consist of emitted compounds. Human activities such as fossil fuel and biomass burning increase “aerosols containing sulphur compounds, organic compounds and black carbon” (IPCC 2007, p. 100). Furthermore, industrial activities like surface mining grow up dust amount in the atmosphere (IPCC 2007).
The concentration of CO2 is currently “379 parts per million (ppm) while methane is greater than 1,774 parts per billion (ppb)” (IPCC 2007, p.116).
14 Both very likely much higher than any time in at least 650 kyr (during which CO2 remained between 180 and 300 ppm and methane between 320 and 790 ppb).
The recent rate of change is dramatic and unprecedented; increases in CO2
never exceeded 30 ppm in 1 kyr – yet now CO2 has risen by 30 ppm in just the last 17 years”. (IPCC 2007, p.116).
2.3. Observed Changes of Water Related Climate Variables
2.3.1. Temperature Changes
Based on instrumental observations, Earth’s surface temperature has been increasing globally. According to IPCC fourth assessment report, warming in the last century has to be investigated in two phases: from 1910s to 1940s and from 1970s to present. IPCC (2007) stated that Earth’s surface temperature ascended 0.35°C in first phase while it increased more strongly, 0.55°C, in the second phase. Another significant issue which has to be noted is 11 of the 12 warmest years on record have observed in the past 12 years. Furthermore, above the surface troposphere (up to about 10 km) has been warmed more than surface based on observations since 1950s. Warming of oceans, rising sea levels, glaciers melting and decrease in snow covered area in Northern Hemisphere can be shown as a proof of global warming (IPCC 2007).
Surface temperatures have risen by about 0.74 °C between 1906 and 2005. After an increase of 0.35 °C from the 1910s to 1940s, a slight cooling of 0.1 °C was observed and then an increasing jump of 0.55 °C has been detected until the end of 2006 (IPCC 2007). Graphical illustration of temperature trend is shown in Fig. 4.
15 Fig.2.4: Global mean temperature trend (IPCC 2007)
Warming over land is larger than over oceans especially since the 1970s. Warming has been observed slightly larger in the winter hemisphere. Also additional warming has been reported in urban areas due to urban heat island effect (IPCC 2007).
2.3.2. Precipitation Changes
Changes have been occurring in the amount, intensity, frequency and type of precipitation based on long term observations. According to observations from 1900 to 2005, eastern North and South America, northern Europe, northern and central Asia have been wetter while Sahel, southern Africa, the Mediterranean and southern Asia have been getting drier. Due to increase in air temperature, more rain is falling than snow in northern regions. Moreover, increases in heavy precipitation events have been observed widely even in locations where total amount of precipitation has decreased. Due to increase in heavy precipitation and heat waves events, increase in drought and flood events has also occurred (IPCC 2007).
16 Trenberth et al. (2007) presented global precipitation changes based on large number of observations provided by Global Historical Climatology Network (GHCN), also the Precipitation Reconstruction over Land (PREC/L: Chen, Xie & Janowiak 2002), the Global Precipitation Climatology Project (GPCP: Adler et al. 2003), the Global Precipitation Climatology Centre (GPCC: Beck, Grieser & Rudolf 2005) and the Climatic Research Unit (CRU: Mitchell & Jones 2005). They reported general precipitation increase over the 20th century between 30°N and 85°N, “but notable decreases have occurred in the past 30–40 years from 10°S to 30°N” (Trenberth et al 2007, p. 265). It is shown in Fig. 2.5.
Fig.2.5: Mean annual precipitation anomalies (%) over land from 1900 to 2005 relative their 1961–1990 means (Trenberth et al. 2007)
They also stated similar precipitation changes over the ocean based on salinity decreases in the North Atlantic and south of 25°S. Trenberth et al. (2007) reported remarkable precipitation increase from 10°N to 30°N during 1900-1950 period, but then decreases after 1970. They did not present any strong hemispheric-scale trends
17 over Southern Hemisphere extra-tropical land masses. Bates et al. (2008) explained that: Across South America, increasingly wet conditions have been observed over the Amazon Basin and south-eastern South America, including Patagonia, while negative trends in annual precipitation have been observed over Chile and parts of the western coast of the continent. Variations over Amazonia, Central America and western North America are suggestive of latitudinal changes in monsoon features. The largest negative trends since 1901 in annual precipitation are observed over western Africa and the Sahel, although there were downward trends in many other parts of Africa, and in south Asia. Since 1979, precipitation has increased in the Sahel region and in other parts of tropical Africa, related in part to variations associated with teleconnection patterns. Over much of north-western India the 1901–2005 period shows increases of more than 20% per century, but the same area shows a strong decrease in annual precipitation since 1979. North-western Australia shows areas with moderate to strong increases in annual precipitation over both periods. Conditions have become wetter over northwest Australia, but there has been a marked downward trend in the far south-west, characterised by a downward shift around 1975.
Widespread increases in heavy precipitation events (e.g., above the 95th percentile) have been observed, even in places where total amounts have decreased. These increases are associated with increased atmospheric water vapour and are consistent with observed warming. However, rainfall statistics are dominated by interannual to decadal-scale variations, and trend estimates are spatially incoherent (Peterson et al. 2002; Griffiths et al. 2003; Herath and Ratnayake 2004). Moreover, only a few regions have data series of sufficient quality and length to assess trends in extremes reliably. Statistically significant increases in the occurrence of heavy precipitation have been observed across Europe and North America (Klein Tank and Können 2003; Kunkel et al. 2003; Groisman et al. 2004; Haylock and Goodess 2004). Seasonality of changes
18 varies with location: increases are strongest in the warm season in the USA, while in Europe changes were most notable in the cool season (Groisman et al. 2004; Haylock and Goodess 2004) (Bates et al. 2008, p.16).
IPCC (2007) stated that increased temperature due to climate change resulted increased water vapour. It is expressed that over the 20th century atmospheric water vapour increased around 5% over oceans. It resulted increasing number of intensive rainfall and snowfall events. Basic theory utters that independently from increase or decrease in total amount of annual precipitation, warmer climate due to increased water vapour results more severe precipitation events. Although basic theory on intense precipitation is independent from total amount of annual precipitation, it is significant to declare that intense precipitation events are stronger in the cases of increase in total annual precipitation. Warmer climate leads to droughts when it is not raining whereas it causes floods when it rains. Hence, warmer climate enhances risk of both floods and droughts. Aerosol pollution acts as a mask hindering incoming solar radiation to Earth surface. So, it decreases evaporation and water vapour amount of atmosphere. Thus, even increased intensity is observed in precipitations, duration and frequency of events may decrease (IPCC 2007).
Increase in extreme precipitations is a good indicator of human induced warming by contributing greenhouse gases. Thus, human impact may be detected easier in extreme precipitation than mean. The reason of this is the control of extreme precipitation by the availability of water vapour, while mean precipitation is controlled by the ability of the atmosphere to radiate long-wave energy to space, and the latter is restricted by increasing greenhouse gases (Bates et al. 2008). In summary, upward trend of heavy precipitation events is more likely to occur owing to anthropogenic contribution.
2.3.3. Sea Level Changes
Sea level increased in the 20th century and it shows rising trend currently. It is foreseen that it will increase more intensively in this century. Primary reasons of sea level rise
19 are stated as thermal expansion of the oceans because water expands as it gets warm, and the loss of land ice owing to increased melting in IPCC reports (IPCC 2007). Summary graph of changes in temperature, sea level and snow cover is shown in Fig. 2.6.
Fig.2.6: Changes in temperature, sea level and snow cover area (IPCC 2007)
If time duration between mid-19th century and mid-20th century was divided into three parts and investigated separately, it is seen that sea level increase is getting higher towards current date. The average sea level rise was 1.7 ± 0.5 mm/ yr for the 20th century, it was 1.8 ± 0.5 mm/yr for 1961–2003, and finally 3.1 ± 0.7 mm/yr sea level rise was detected for 1993–2003 (Bates et al. 2008; Bindoff et al. 2007).
Increases in sea level are in agreement with warming, and modelling studies state that overall it is high possibility that human induced climate change contributed a lot sea- level rise during the latter half of the 20th century; however, it is not possible to quantify the anthropogenic contribution to sea level rise due to the observational uncertainties and lack of studies on sea level rise (Bates et al. 2008; Bindoff et al. 2007). 20 2.3.4. Evapotranspiration Changes
Changes in evapotranspiration depend on moisture supply, energy availability and surface wind. Moreover, the direct effects of atmospheric CO2 enrichment on plant physiology influence actual evapotranspiration. Direct measurement of actual evapotranspiration over global land areas is very limited. It is very difficult to perform trend analysis by using global analysis products due to large errors. There is limited literature which investigates observed trends of actual and potential evapotranspiration (Trenberth et al. 2007; Bates et al. 2008).
A decreasing trend in pan evaporation was detected in US, India, Australia, China and Thailand. However, pan evaporation does not represent actual evaporation and it may be affected by many external factors. There are increases in actual evapotranspiration during the second half of the 20th century in US and Russia (Trenberth et al. 2007; Bates et al. 2008).
2.3.5. Runoff and River Discharge Changes
There are many studies which examined potential trends of river discharges during 20th century from basin scale to global scale in literature. Some studies reported statistically significant trends and some studies stated significant links with trends in temperature or precipitation. For example, high latitudes and large parts of the US experienced an increase in runoffs; on the other hand, studies reported significant runoff decreases at parts of West Africa, southern Europe and southern South America. Bates et al. (2008) stated that Labat et al. (2004) reported a “4% increase in global total runoff per 1°C rise in temperature during the 20th century”. There is widespread evidence that river flows timing in snow dominated basins has been substantially changed. Due to higher temperatures, large portion of precipitation falls as rain rather than snow. Moreover, snow melt season starts earlier (Rosenzweig et al. 2007; Bates et al. 2008).
21 2.3.6. Extreme Weather Events Changes
Due to climate change, number of heat waves, and floods increased since the 1950s. The more regions have been affected by droughts. Based on observations and reports, since the 1970s, intensity and frequencies of tropical storm and hurricane have increased significantly (IPCC 2007).
It would not be scientific or reasonable to clarify all extreme events such as heat waves or floods depending on climate change. Because extreme events are under impacts of many factors and many extreme events can be observed even in an unchanging climate. Nonetheless, there is a fact that climate change has clearly increased the likelihood of extreme events.
2.3.7. Changes in Cryosphere Components
Climate change impact on snow hydrology is main purpose of this study. Thus, it is important to present historical changes on cryosphere components. Due to importance of cryosphere for human life (more than one sixth of world’s population lives in snow melt dominated basins), it is necessary to analyse changes in crysophere components of the climate system. Fig. 2.7 demonstrates cryosphere trends. It is shown in Fig. 2.7 that there are significant declines in ice storages in many components.
22 Fig.2.7: Surface air temperature (A and E), Northern Hemisphere seasonally frozen ground extent (B), NH snow cover extent for March–April (C), and global glacier mass balance (D) anomaly time series (IPCC 2007)
23 Lemke et al. (2007) expressed the changes in cryosphere components in detail. They firstly explained globally available snow data. They stated that daily snow depth of snow and new snowfall have been performed by many countries such as Switzerland, US, the former Soviet Union and Finland, dating to the late 1880s. Snow water equivalent (SWE) and snow depth measurements have been made widespread since 1950s in the mountains of western North America and Europe, and a few snow station measured snow data in Australia since 1960. Snow data are influenced by observation station location, measurement techniques and land cover and they are not uniformly distributed (Lemke et al. 2007).
Satellite-derived data have been used to evaluate snow covered area since 1966. However, for Southern Hemisphere, snow covered area observations started in 2000 by using the advantage of Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data. More currently, space-borne passive microwave sensors enable to perform observations not only for snow cover but also snow depth and SWE. Micro wave sensors are not affected by cloud cover and winter darkness (Lemke et al. 2007).
According to observations between 1966 and 2004, mean annual snow cover in Northern Hemisphere is 23.9 × 106 km2, excluding the Greenland Ice Sheet. Snow covered area interannual variability is largest in autumn or summer (Lemke et al. 2007).
Snow covered area has decreased in spring and summer but not significantly in winter in spite of winter warming since 1920s, in particular since 1970s. Due to changes in climate, a maximum snow covered area date shifted a month earlier from February to January, a statistically significant decrease in annual average snow covered area, a shift to approximately two weeks earlier snow melt in 1972-2000 period were observed (Dye 2002). Brown (2000) reported significant decrease of 2.7 ± 1.5 × 106 km2 or 7.5 ± 3.5% in Northern Hemisphere snow covered area in March and April months (Lemke et al. 2007). Explained changes in snow covered area are shown in Fig. 2.8 and Table 2.1.
24 Fig.2.8: Average snow covered area in March-April from Brown (2000) (Lemke et al. 2007)
Table 2.1: Monthly snow covered area trend in Northern Hemisphere (106 km2 per decade) (Lemke et al. 2007)
Moreover, regional (North America, Europe and Eurasia, South America, Australia and New Zealand) snow data changes were explained as follows based on report of Lemke et al. (2007). Snow covered area in North America has increased between 1915 and 2004 due to precipitation increases in North America. In the second half of 20th
25 century, snow covered area change is more important, particularly in spring season. Moreover, shifts by around eight days earlier in snow melting time since mid-1960s were reported by Stone et al. (2002) in northern Alaska (Lemke et al. 2007). Furthermore, annual measurements of mountain snow water equivalent near 1 April demonstrated that there is declining trend in 1 April snow water equivalent since 1950 at about 75% of locations observed in western North America (Mote et al. 2005). Moreover, streamflow observations indicated shifts to two weeks earlier in the date of maximum mountain SWE (Stewart, Cayan & Dettinger 2005) (Lemke et al. 2007).
Snow covered area decreases were observed in Europe and Eurasia and they are characterized by large elevation variations. Recently, reductions have been observed in the mountains of Switzerland and Slovakia (Scherrer, Appenzeller & Laternser 2004; Vojtek, Fasko & St’astny 2003). However, no change in snow covered area was documented in Bulgaria according to data between 1931 and 2000 (Petkova, Koleva & Alexandrov 2004). Declines are more important at lower elevations where are more sensitive to temperature changes (Lemke et al. 2007). About 1 day yr-1 snow covered area reductions in central Europe was reported by Falarz (2002). Since the late 1970s shorter snow season trends were observed in Finland, Russia and in the Tibetan Plateau (Hyvärinen 2003; Ye & Ellison 2003; Lemke et al. 2007).
Lemke et al. (2007) expressed that except Antarctica, very less amount of land area is snow covered in Southern Hemisphere. Moreover, snow data are limited in Southern Hemisphere and quality of available data in Southern Hemisphere is much lower than data in Northern Hemisphere.
Prieto et al. (2001) stated long-term increasing trend in the number of snow days in the eastern side of the central Andes region (33°S) based on data from 1885 to 1996. Later winter (August–September) snow depth in the most mountainous south-eastern part of Australia have demonstrated significant decreases up to %40 since 1962. The main reason of significant decrease was explained as spring season warming (Hennessy
26 et al. 2003; Nicholls 2005). Furthermore, no trend of snow covered area was reported in New Zealand (Lemke et al. 2007).
Bates et al. (2008) expressed based on Lemke et al. (2007) that:
Degradation of permafrost and seasonally frozen ground is leading to changes in land surface characteristics and drainage systems. Seasonally frozen ground includes both seasonal soil freeze–thaw in non-permafrost regions and the active layer over permafrost that thaws in summer and freezes in winter. The estimated maximum extent of seasonally frozen ground in non-permafrost areas has decreased by about 7% in the Northern Hemisphere from 1901 to 2002, with a decrease of up to 15% in spring. Its maximum depth has decreased by about 0.3 m in Eurasia since the mid-20th century in response to winter warming and increases in snow depth. Over the period 1956 to 1990, the active layer measured at 31 stations in Russia exhibited a statistically significant deepening of about 21 cm. Records from other regions are too short for trend analyses. Temperature at the top of the permafrost layer has increased by up to 3°C since the 1980s in the Arctic. Permafrost warming and degradation of frozen ground appear to be the result of increased summer air temperatures and changes in the depth and duration of snow cover (Bates et al. 2008, p. 19).
Moreover, remarkable amount of mass loss has been noted on the majority of glaciers and ice caps in global scale with increasing rates. Bates et al. (2008) expressed the loss based on Lemke et al. (2007) as follows: ”From 1960/61 to 1989/90 the loss was 136 ± 57 Gt/yr (0.37 ± 0.16 mm/yr sea-level equivalent, SLE), and between 1990/91 and 2003/04 it was 280 ± 79 Gt/yr (0.77 ± 0.22 mm/yr SLE)” (Bates et al. 2008, p. 19).
Summary of observed changes in cryosphere components are indicated graphically in Fig. 2.9.
27 Fig.2.9: Summary of observed changes in cryosphere components (Lemke et al. 2007)
So as to mitigate climate change impact or for adapting climate change, performing reliable future climate projections are as significant as (maybe more) understanding what happened in the past. Thus, in the following section IPCC future climate projections based on the Global Climate Models (GCMs), which have high reliability to forecast future climate, were presented. Then, study was specialized into Turkey scale.
28 2.4. Future Climate Projections in Global Scale
Climate models, which are the representation of climate components, interactions of these components and feedbacks, are used to forecast future climate. Different emission scenarios are used in models which are utilized to obtain future climate projections. These scenarios were prepared by IPCC and presented by special report on emission scenarios (SRES). While greenhouse gases emissions were projected in scenarios, population increase, energy usage, economies, technologic development and changes in agriculture and land use were considered. Based on mentioned criteria, four main group of scenario were generated, A1, A2, B1 and B2. Then, forty different scenarios were generated by using these four main scenarios (Ministry of Environment and Forestry 2008).
Demographic, socio-economic and technologic development projections which are used in scenarios preparation are shown in Table 2.2.
Table 2.2: IPCC scenarios population and economy conditions (IPCC 2000)
In addition, according to four main scenarios, projected annual change of CO2 amount which are released to atmosphere is shown in Table 2.3.
29 Table 2.3: IPCC scenarios CO2 emission predictions (IPCC 2000)
Scenario Fossil Fuel Induced CO2 Land Induced CO2 Cumulative CO2 (Billion ton C/year) (Billion ton C/year) (Billionton C/year) 1990 2050 2100 1990 2050 2100 1990-2100 A1F1 6 23.1 30.3 1.1 0.8 -2.1 2.189 A1B 6 16 13.1 1.1 0.4 0.4 1.499 A1T 6 12.3 4.3 1.1 0.0 0 1.068 A2 6 16.5 28.9 1.1 0.9 0.2 1.862 B1 6 11.7 5.2 1.1 -0.4 -1 983 B2 6 11.2 13.8 1.1 -0.2 -0.5 1.164
2.4.1. Temperature Projections
In IPCC fourth assessment report (2007), an increase of 0.2°C/ten years in next 20 years is reported for the global average Earth surface temperature based on studies by using most advanced climate models and a range of SRES scenarios. Moreover, even greenhouse gases emissions and aerosol release could stabilize at current level; global average Earth surface temperature will increase 0.1°C/ten years. An increase in Earth average surface temperature in 2090-2099 relative to 1980-1999 is expected average 1.8°C (1.1°C-2.9°C) based on the most optimistic scenario (B1), when it is expected average 4°C (2.4°C -6.4°C)according to the most pessimistic scenario (A1F1) (Demir et al. 2008).
30 Fig.2.10: Average projected global surface temperature changes in 2020-2029 and 2090-2099 relative to 1980-1999 based on B1, A1B and A2 scenarios (IPCC 2007)
2.4.2. Precipitation Projections
Climate model projections indicated increases in globally averaged water vapour, evaporation and precipitation over the 21st century. The climate models show generally precipitation increase in the areas of regional tropical precipitation maxima (such as the monsoon regimes, and the tropical Pacific in particular) and at high latitudes. On the other hand, they show general decreases in the sub-tropics (Bates et al. 2008).
Climate models are in agreement on precipitation increases at high latitudes in both winter and summer seasons. Precipitation increases of the tropical oceans and in some of the monsoon regimes are remarkable. Except eastern Asia precipitation increases, summer precipitation decreases were projected in mid-latitude (Meehl et al. 2007; Bates et al. 2008). 31 Fig. 2.11 shows the global distribution of the 2080–2099 change in annual mean precipitation for the SRES A1B scenario together with quantities from a 15-model ensemble. High latitude precipitation increases exceeds 20% mostly as well as in eastern Africa, the northern part of central Asia and the equatorial Pacific Ocean. Significant precipitation reduction up to 20% was projected in the Mediterranean and Caribbean regions and on the sub-tropical western coasts of each continent (Meehl et al. 2007; Bates et al. 2008).
Fig.2.11: Global precipitation, soil moisture, runoff and evaporation changes for the scenario SRES A1B for the period 2080–2099 relative to 1980–1999 (Meehl et al. 2007)
Precipitation increase over lands is overall 5% while over oceans is projected by 4%. The net precipitation change over land was predicted by climate models as 24% of the global mean increase in precipitation (Meehl et al. 2007). Furthermore, it is very likely 32 that extreme precipitation events will be observed more frequent. Especially in tropical and high-latitude areas, increasing precipitation intensity was projected. Moreover, there is a drying tendency in mid-continental areas during summer, suggesting very likely intensive droughts in these regions. In addition, extreme precipitation increases in most tropical and mid- and high-latitude areas are more than mean precipitation (Meehl et al. 2007).
Climate model projections indicate increases in both precipitation intensity and number of consecutive dry days in many regions over the 21st century. It is shown in Fig. 2.12. Precipitation intensity increases almost everywhere, however, it is more significant at mid and high latitudes where mean precipitation also increases (Meehl et al. 2007; Bates et al. 2008).
33 Fig.2.12: Changes in extremes based on nine global coupled climate models in 2080–2099 relative to 1980–1999 for the A1B scenario (Meehl et al. 2007)
2.4.3. Snow and Land Ice Projections
Due to increase in temperatures, climate models projected decline in snow cover, lose mass of glaciers and ice caps as a result of larger increase in summer melting than increase in winter snowfall. Furthermore, widespread increases in thaw depth over much of the permafrost regions are predicted to be observed due to global warming. Snow cover is under control of two climate variables: temperature and precipitation. Due to dominant impact of temperature which will increase based on future
34 projections, decreases in snow cover over 21st century was projected. However, climate models projected some increases at higher altitudes. For instance, Bates et al. (2008) explained based on Mehhl et al. (2007) that “Climate models used in the Arctic Climate Impact Assessment (ACIA) predict a 9–17% decline in the annual mean Northern Hemisphere snow coverage under the B2 scenario by the end of the century” (Bates et al. 2008, p. 28). Climate models projected delay in snow accumulation season and earlier beginning in melting season. It results shorter snow seasons and decrease in snow covered area during the snow season (Meehl et al. 2007).
According to models which are forced by a range of IPCC climate scenarios, permafrost area in the Northern Hemisphere will possibly show reduction by 20– 35% by the mid- 21st century (Anisimov et al. 2007). Anisimov et al. (2007) stated that active layer depths are projected to be within 10–15% of their present values over most of the permafrost area in the next three decades. Moreover they reported that an increase by 15–25% on average was projected in the depth of seasonal thawing. Moreover, 50% or more was projected for the northernmost locations. Finally, it will high possibly rise by 30–50% or more over all permafrost areas by 2080 (Anisimov et al. 2007; Bates et al. 2008).
Based on simulations at 11 different glaciers, 60% volume loss was projected in glaciers by 2050 (Schneeberger et al. 2003). Bradley et al. (2004) stated disappearance of many glaciers because of an increase in the equilibrium-line altitude based on seven GCM simulations at 2 × atmospheric CO2 conditions (Meehl et al. 2007; Bates et al. 2008). Bates et al. (2008) expressed by referring Meehl et al. (2007) and Lemke et al. (2007) that projections for 21st century demonstrated “glacier and ice cap shrinkage of 0.07– 0.17 m sea-level equivalent (SLE) out of today’s estimated glacier and ice cap mass of 0.15–0.37 m SLE” (Bates et al. 2008, p. 28).
2.4.4. Sea Level Projections
Sea level rise is expected 18-38 cm based on the most optimistic scenario (B1) and 26- 59 cm according to the most pessimistic one (A1F1) for 2090-2099 relative to 1980-
35 1999 (IPCC 2007).Thermal expansion is the largest impacts on sea level rise, moreover, glaciers, ice caps and the Greenland ice sheet were also projected to contribute positively to sea level. It is projected that sea level rise during 21st century will be very sensitive to geographical variability (Bates et al. 2008).
2.4.5. Evapotranspiration Projections
Potential evaporation is expected to increase for almost everywhere over the world. This increase can be explained by increase in the water-holding capacity of the atmosphere due to higher temperatures. However, relative humidity is not predicted to alter remarkably. Thus, water vapour deficit in the atmosphere increases and it results increases in evaporation rate (Trenberth et al. 2003). Furthermore, actual evaporation over open water is predicted to rise. Changes in evapotranspiration over land are depending on changes in precipitation and radiative forcing (Bates et al. 2008; Kundzewicz et al. 2007).
Carbon dioxide increase in the atmosphere has two potential influences on evapotranspiration from vegetation. First of these, higher CO2 concentrations can decline transpiration because the stomata of leaves. On the other hand, higher CO2 concentrations can increase plant growth which results increased leaf area leading increased transpiration. Effects of increased CO2 on plant transpiration show alteration “in terms of plant types and in response to other influences such as the availability of nutrients and the effects of changes in temperature and water availability” (Bates et al. 2008, p. 29).
2.4.6. Soil Moisture Projections
Models have a capability to assess upper few metres of the land surface’s moisture, so evaluation of future soil moisture is very difficult. Projections demonstrate soil moisture decreases in the subtropics and the Mediterranean region and increases East Africa, central Asia and some other regions with increased precipitation. Soil moisture
36 decreases were also projected at high latitudes, where snow cover will be strongly affected by warming (Meehl et al. 2007).
2.4.7. Runoff and River Discharge Projections
Future changes in river flows, lake and wetland levels owing to climate change are basically related with changes in the volume and timing of precipitation and precipitation type (snowfall or rainfall). Moreover, evaporation changes have an impact on river flows. Climate change impact studies on runoffs in the literature are generally focused on Europe, North America and Australasia. Large majority of studies use different types of hydrological models driven by future climate projections which are generated by climate models. However, a few global scale studies, which investigated climate change influences on river basin flows, used runoff both simulated directly by climate models and hydrological models. Study findings demonstrated runoff increases in high latitudes and the wet tropics, and decreases in mid-latitudes and some parts of the dry tropics (Bates et al. 2008).
Fig. 2.11 (c) indicated mean runoff changes under A1B greenhouse emission scenario. It is seen in Fig. 2.11 (c) that southern Europe will experience significant reductions in runoff. On the other hand, increased runoffs were projected in south-east Asia and in high latitudes. Up to 20% runoff changes relative to 1980–1999 values were projected. Mentioned changes “range from 1 to 5 mm/day in wetter regions to below 0.2 mm/day in deserts” (Bates et al. 2008, p. 29). Furthermore, Meehl et al. (2007) expressed that high-latitude river flows are projected to increase, “while those from major rivers in the Middle East, Europe and Central America tend to decrease” (Meehl et al. 2007, p. 769). However, runoff change magnitudes show large variations among climate models, even, some climate models project increases while others predict decreases at some regions such as southern Asia (Bates et al. 2008).
Fig. 2.13 which is presented by Bates et al. (2008) based on Milly, Dunne and Vecchia (2005) and Kundzewicz et al. (2007) demonstrates annual runoff change for 2090– 2099 relative to 1980–1999 based on 12 climate models under A1B scenario. In Fig. 37 2.13, hatching was used to imply areas where models are in agreement and whitening are used to mark areas where models are not in agreement. Fig. 2.13 gives a general idea about annual runoff changes however; it is not useful to interpret smaller temporal and spatial scales (Bates et al. 2008).
Fig.2.13: Annual runoff changes at global scale for the period 2090–2099, relative to 1980–1999 based on study of Milly, Dunne and Vecchia (2005) (Bates et al. 2008)
Moreover, Fig. 2.14 indicates the climate change impacts on long-term average annual river runoff by the 2050s worldwide under the A2 emissions scenario (Kundzewicz et al. 2007).
38 Fig.2.14: Average annual runoff change by the 2050s based on different climate models (Arnell 2003; Kundzewicz et al. 2007)
Changes in seasonality of river flows are expected due to warming especially at snow dominated basins where majority of precipitation falls as snow. Decreases in spring flows are projected owing to decrease in snowpack and earlier snowmelt. Moreover, increasing winter flows were projected. Seasonality changes were reported for around the Baltic, Russia, the Himalayas, and western, central and eastern North America. The
39 impacts are expected larger at lower elevations. Moreover, a month shift to earlier dates in peak flow was presented for many regions over the world by the middle of the 21st century. In rainfall dominated basins, runoff changes are dominantly based on precipitation changes than temperature changes (Bates et al. 2008; Kundzewicz et al. 2007).
2.5. Climate Change Impacts on Hydrology and Water Resources
2.5.1. Observed Impacts
2.5.1.1. Change in surface water and groundwater systems
Changes in annual runoff explained above. Moreover, detailed review of studies which investigated climate change influence on runoff was presented in literature review.
Groundwater flow in shallow aquifers is one of the important components of the hydrological cycle and is impacted by climate variability and human interventions (Rosenzweig et al. 2007). Declining trends have been observed in groundwater levels of many aquifers over the world in last few decades (Kundzewicz et al. 2007). However, current declining trend are not due to climate change, it is owing to groundwater pumping surpassing groundwater recharge rates. However, in some regions such as south-western Australia, climate change may be a reason of decreasing groundwater levels as well (Government of Western Australia 2003). Climate based groundwater recharge changes have not been detected because of lack of data and very slow respond of groundwater systems to changing recharge conditions (Rosenzweig et al. 2007; Bates et al. 2008).
2.5.1.2. Change in Water Quality
Owing to warming of lakes and rivers, freshwater ecosystems have demonstrated alterations in species composition, organism abundance, productivity and phenological
40 shifts (Rosenzweig et al. 2007). However, there is no evidence for stating climate related influences on water quality parameters (Bates et al. 2008).
2.5.1.3. Change in Flood Events
Many climate related or non-climatic processes affect floods including river floods, flash floods, urban floods, sewer floods, glacial lake outburst floods and coastal floods. Long-lasting and intensive precipitation, snowmelt, dam break, reduced conveyance due to ice jams or landslides can be shown as main reasons of floods (Bates et al. 2008; Kundzewicz et al. 2007). Bates et al. (2008) stated based on Kundzewicz et al. (2007) that flood events “depend on precipitation intensity, volume, timing, precipitation type (rain or snow), rivers and their drainage basins’ antecedent conditions including existence of snow and ice, soil character and status, wetness, rate and timing of snow and ice melt, urbanisation, existence of dykes, dams and reservoirs” (Bates et al. 2008, p.37). Constructing into flood plains and lack of flood management plans increase the floods’ damages. The observed increase in precipitation intensity and changes in other climate variables due to climate change is a proof of climate change influences on the intensity and frequency of floods (Kundzewicz et al. 2007). Bates et al. (2008) reported increase in frequency of heavy precipitation events over most areas during the late 20th century based on IPCC Fourth Assessment Report and they underlined the high level possibility of human contributing role on it.
In global scale, number of inland flood catastrophes during the last decade (1996- 2005) almost doubled when it is compared with inland flood catastrophes per decade between 1950 and 1980 (Kundzewicz et al. 2007). Kron and Berz (2007) reported five times larger economic losses owing to increasing number of flood catastrophes (Kundzewicz et al. 2007).
Although climate change has an impact on flood events, dominant drivers of the upward trend of flood damage are socio-economic factors including economic growth, increases in population and land-use change. Kundzewicz et al. (2007) implied in their
41 study that “Floods have been the most reported natural disaster events in Africa, Asia and Europe, and have affected more people across the globe (140 million/yr on average) than all other natural disasters (WDR, 2003, 2004)” (Kundzewicz et al. 2007, p. 178). For example, in Bangladesh flooding event in 1998, about 70% of the country’s area exposure inundation (Mirza 2003; Clarke & King 2004) (Kundzewicz et al. 2007; Bates et al. 2008).
Both warming and increased water vapour resulted increase in the frequency of heavy precipitation events (Trenberth et al. 2007). Milly et al. (2002) detected an obvious increase in the frequency of ‘large’ floods, whose return period is larger than 100 years, across much of the globe from the analysis of data from large river basins. Moreover, Kundzewicz et al. (2005) studied 195 catchments over world and identified increases in 27 locations and decreases in 31 locations and no trend in the remaining 137 (Rosenzweig et al. 2007).
2.5.1.4. Change in Drought Events
Drought can be defined in four different ways: meteorological drought, hydrological drought, agricultural drought and environmental drought. Meteorological drought refers to precipitation well below average. Hydrological drought is explained by low river flows and low water levels in rivers, lakes and groundwater. Agricultural drought corresponds to low soil moisture and finally environmental drought is a combination of the above. Droughts have very substantial socio-economic consequences and intensity can vary based on the interaction between natural conditions and human factors “such as changes in land use, land cover, and the demand for and use of water” (Bates et al. 2008, p. 38).
Frequency of droughts has increased, especially in the tropics and sub-tropics, since the 1970s. Reduced land precipitation and rising temperatures which increase evapotranspiration and decrease soil moisture, are significant factors which have contributed to droughts (Dai, Trenberth & Qian 2004; Kundzewicz et al. 2007; Bates et al. 2008). 42 Dai et al. (2004) used Palmer Drought Severity Index (PDSI) approach to detect drought trend worldwide and they identified a large drying trend over Northern Hemisphere land since the mid-1950s, with widespread drying over much of Eurasia, northern Africa, Canada and Alaska (Trenberth et al. 2007). Results of study are shown in Fig. 2.15.
Fig.2.15: Drought analysis based on monthly Palmer Drought Severity Index (PDSI) for the period of 1900-2002 (Dai, Trenberth & Qian 2004; Trenberth et al. 2007)
In Fig. 2.15, the lower panel indicates how the sign and strength of PDSI has altered since 1900. In lower plot, positive and negative values correspond to drier or wetter
43 conditions respectively. In upper map, the red & orange areas are drier and the blue & green areas are wetter than average. Moreover, the smooth black curve refers decadal variations in Fig. 2.15. Widespread increasing African drought is remarkable; especially it is dramatic in the Sahel. On the other hand, eastern North and South America and northern Eurasia have been getting wetter notably (Dai, Trenberth & Qian 2004; Trenberth et al. 2007; Bates et al. 2008).
Dai, Trenberth and Qian (2004) detected a large drying trend over Northern Hemisphere land since the mid-1950s, in particular, over much of Eurasia, northern Africa, Canada and Alaska. “In the Southern Hemisphere, land surfaces were wet in the 1970s and relatively dry in the 1960s and 1990s” (Bates et al. 2008, p. 38). There is an overall drying trend in the Southern Hemisphere (Trenberth et al. 2007).
Trenberth et al. (2007) explained the main reason based on Dai, Trenberth and Qian (2004) behind drying trends as decreases in land precipitation in recent decades. In addition, large surface warming during the 20-30 years have accelerated drying. Globally, very dry areas with PDSI less than -3 more than twice larger and it reached from ~12% to 30% since the 1970s (Trenberth et al. 2007).
Droughts cause significant problems for rain-fed agricultural production and water supply for domestic, industrial and agricultural purposes. For instance, some semi-arid and sub-humid regions, such as Australia, western US, southern Canada and the Sahel have substantial socio-economical problems due to intense and multi-annual droughts (Kundzewicz et al. 2007).
Schär et al. (2004) reported a link between 2003 heatwave in Europe and global warming. Up to 300 mm decreases were observed in annual precipitation during this drought. This drought resulted approximately 30% reduction in gross primary production of terrestrial ecosystems over Europe (Ciais et al. 2005). Many major rivers such as Po, Rhine, Loire and Danube experienced low levels, leading significant
44 problems especially for water sectors (Beniston and Diaz 2004; Zebisch et al. 2005; Bates et al. 2008).
Table 2.4 is a good summary table which presents observed changes in runoff/streamflow, lake levels, floods and droughts over the world.
Table 2.4: Observed changes in runoff/streamflow, lake levels, floods and droughts (Rosenzweig et al. 2007)
In regional scale, Turkey is in semi-arid climate zone and spatial precipitation pattern shows large difference because of topography. Thus, drought risk is a continuous threat for Turkey. Droughts were observed in Turkey in 1955-1961,1970-1977,1982- 1986, 1989-1994 and 1999-2006 (except 2001 and 2003) periods. It was particularly significant for 1973–74, 1988–89, 1993–94 and 2001–02 agriculture years (Environment and Forestry Ministry 2008). Drought seasons are demonstrated in Fig. 2.16.
45 Fig.2.16: Turkey annual total precipitation anomalies (mm) (Environment and Forestry Ministry 2008)
Southern parts of Turkey have been experiencing drought events more frequently and intensive than northern parts. In particular, middle Anatolia, western Anatolia, eastern Mediterranean and South eastern Anatolia are driest parts of Turkey. On the other hand, Black Sea (Northern Anatolia) is the wettest part of Turkey.
Turkish State Meteorology Service (TSMS) (2007) stated that Marmara, Aegean and Middle Anatolia geographical regions are the most drought affected regions in Turkey. The seven months (April to October) cumulative precipitation graphs of these three regions are shown in Fig. 2.17.
46 Fig.2.17: Aegean, Central Anatolia and Marmara regions cumulative precipitations (TSMS 2007) 47 Drought in Turkey can not be explained by just global warming. Inefficient usage of water resources, population growth, agricultural land increase in proportion to agricultural demand and urbanization are the primary reasons of droughts (TSMS 2007).
2.5.2. Climate Change Impacts on Future Water Availability and Demand
2.5.2.1. Climate Change Impact on Ground Water
Climate change has an influence on groundwater recharge rates and depth of groundwater table. Nonetheless, due to lack of available data in both developed and developing countries groundwater studies are limited. Moreover, there is less number of studies which investigate effects of climate change on groundwater and groundwater-surface water relation (Kundzewicz et al. 2007; Bates et al. 2008).
Due to very strong interaction between groundwater and surface water, climate change impacts on surface water flow regimes are expected to influence groundwater. Bates et al. (2008) explained based on Kundzewicz et al. (2007) that enhanced precipitation variability may reduce groundwater recharge in humid areas. On the other hand, increased precipitation variability may rise groundwater recharge in arid and semi-arid areas, “because only high-intensity rainfalls are able to infiltrate fast enough before evaporating” (Bates et al. 2008, p. 40).
Fig. 2.18 shows findings of global scale hydrological study which investigates plausible climate change impacts on groundwater. Based on study of Döll and Flörke (2005) which is presented by Kundzewicz et al. (2007), increases in groundwater recharge by 2% were detected and it is less than total runoff change which is predicted by 9% until the 2050s when it is globally averaged for the ECHAM4 climate change response under SRES A2 scenario.
48 Fig.2.18: Climate change impact on long-term average annual diffuse groundwater recharge (Döll & Flörke 2005; Kundzewicz et al. 2007)
All simulations, above mentioned study, based on ECHAM4 and HadCM3 GCMs with the SRES A2 and B2 emission scenarios showed decrease in groundwater recharge by the 2050s by more than 70% in north-eastern Brazil, south-western Africa and the southern rim of the Mediterranean Sea. Nonetheless, this study did not consider increases in daily precipitation variability. Thus, estimations can be higher than real conditions (Kundzewicz et al. 2007).
49 For the locations where water table depth rises and groundwater recharge decreases, wetlands dependent on aquifers are under risk. Moreover, “the base flow runoff in rivers during dry seasons is reduced” (Bates et al. 2008, p. 40). Locations, where simulations projected groundwater recharge increases by more than 30% by the 2050s, involve the Sahel, the Near East, northern China, Siberia and the western US. Increased groundwater recharge may lead problems in residential locations and agricultural areas by soil salinisation and waterlogged soils in regions in which water tables are already high (Kundzewicz et al. 2007; Bates et al. 2008).
2.5.2.2. Climate Change Impact on Floods
As stated before in this study, heavy precipitation events are expected to be more frequent in the 21st century. Increasing frequency of extreme precipitation would very likely enlarge risk of flash flooding and urban flooding (Meehl et al. 2007; Kundzewicz et al. 2007). Some potential effects of floods due to heavy precipitation events are shown in Table 2.5 by Bates et al. (2008) based on IPCC (2007).
50 Table 2.5: Possible influences of changes in extreme precipitations due to climate change based on climate projections to the mid- to late 21st century (IPCC 2007, Bates et al. 2008)
Review of some papers regarding climate change and extreme precipitation events & floods were presented by Bates et al. (2008) based on Kundzewicz et al. (2007). Palmer and Raisanen (2002) reported a remarkable increase in the risk of a “very wet winter over much of central and northern Europe” and they explained this condition with an increase in intense precipitation associated with mid-latitude storms (Bates et al. 2008, p. 41). Milly et al. (2002) expressed that in 15 out of 16 large basins worldwide, “the control 100-year peak volumes of monthly river flow are projected to be exceeded more frequently for a CO2-quadrupling” (Bates et al. 2008, p. 41). In some areas, 100-year flood events are predicted to occur very frequently, even every 2-5
51 years for some cases. Mirza et al. (2003) projected minimum 23–29% increase in the flooded area in Bangladesh based on climate models (Bates et al. 2008).
2.5.2.3. Climate Change Impact on Droughts
It is expected that drought affected regions over world will increase due to global warming. In particular, there is a drying tendency at mid-continental areas during summer, which is a sign of very significant risk of droughts in these regions. Coupled effect of decreased summer precipitation and rising temperatures in southern and central Europe will very likely lead to both reduced summer soil moisture and more frequent and intense droughts (Kundzewicz et al. 2007). Kundzewicz et al. (2007) expressed Fig. 2.19 based on Lehner, Czisch and Vassolo (2005) that “a 100-year drought of today’s magnitude would return, on average, more frequently than every 10 years in parts of Spain and Portugal, western France, theVistula Basin in Poland, and western Turkey” (Kundzewicz et al. 2007, p. 187).
52 Fig.2.19: Change of 100-year droughts future recurrence (Lehner, Czisch & Vassolo 2005) (Kundzewicz et al. 2007)
Some potential effects of increased drought area are shown in Table 2.5. Previously explained climate change impacts on snow hydrology will contribute to drought risk in summer and autumn seasons particularly at snow dominated basins.
Kundzewicz et al. (2007) stated that glacial melt is very important water source by contributing to river flow and water supply for tens of millions of people during the 53 long dry season in Andes. They also explained that many small glaciers such as those in Bolivia, Ecuador and Peru are projected to disappear within the next few decades (Ramirez et al. 2001). Moreover, hundreds of millions of people in China, Pakistan and India will experience sufficient water problems during dry seasons owing to Hindu Kush and Himalayas’ sensitivity to temperature increases (Barnett et al. 2005; Kundzewicz et al. 2007).
2.5.2.4. Climate Change Impact on Water Quality
Increased water temperature and more intensive precipitation events coupled with longer low flow periods are expected to increase water pollution types including sediments, nutrients, dissolved organic carbon, pathogens, pesticides, salt and thermal pollution. Above mentioned changes will influence ecosystems, human health, and the reliability and operating costs of water systems (Kundzewicz et al. 2007).
Higher temperatures are expected to increase thermal stability and change mixing patterns of lakes which result decreased oxygen concentrations and an increased release of phosphorus from the sediments. Thus, water quality of lakes is projected to reduce due to climate change. On the other hand, increase in temperatures can have a positive effect on water quality during winter/spring owing to “earlier ice break-up and consequent higher oxygen levels and reduced winter fish-kill” (Fischlin et al. 2007; Bates et al. 2008, p. 43). Bates et al. (2008) explained that:
The projected increase in precipitation intensity is expected to lead to a deterioration of water quality, as it results in the enhanced transport of pathogens and other dissolved pollutants (e.g., pesticides) to surface waters and groundwater; and in increased erosion, which in turn leads to the mobilisation of adsorbed pollutants such as phosphorus and heavy metals. In addition, more frequent heavy rainfall events will overload the capacity of sewer systems and water and wastewater treatment plants more often. An increased occurrence of low flows will lead to decreased contaminant dilution capacity, and thus higher pollutant concentrations, including pathogens. In 54 areas with overall decreased runoff (e.g., in many semi-arid areas), water quality deterioration will be even worse (Bates et al. 2008, p. 43).
Climate change will high possibly enhance salinisation of shallow groundwater due to increase in evapotranspiration in semi-arid and arid regions. Moreover, due to projected decrease in streamflow in many semi-arid regions, the salinity level of rivers and estuaries will rise (Kundzewicz et al. 2007). For instance, Pittock (2003) reported 13–19% increase in salinity levels in the headwaters of the Murray-Darling Basin in Australia by 2050 (Kundzewicz et al. 2007; Bates et al. 2008).
Increasing sea levels may influence storm-water drainage and sewage disposal negatively in coastal regions. Due to increasing “intrusion of saline water into fresh groundwater in coastal aquifers” proportional to seal level increases, groundwater resources are very likely affected (Kundzewicz et al. 2007, p. 186). Bobba et al. (2000) presented decreases in the thickness of freshwater lenses from 25 m to 10 m and from 36 m to 28 m, respectively for two small and flat coral islands off the coast of India as a respond to a sea-level rise of only 0.1 m (Kundzewicz et al. 2007). Because of strong relation between groundwater recharge and sea-level rise effect, any reduction in groundwater recharge will intensify the effect of sea-level rise. Chen, Grasby and Osadetz (2004) stated that groundwater recharge reduction can cause saltwater intrusion from neighbouring saline aquifers in inland aquifers (Kundzewicz et al. 2007).
2.5.2.5. Climate Change Impact on Water Erosion and Sedimentation
All studies, which investigated climate change and soil erosion relation, stated greater rates of erosion at regions where increase in rainfall intensity is expected. Because of shifts from snow to rain as a result of temperature increases, soil erosion may increase because rainfall is more erosive than snowfall. Permafrost is a non-erodible however the melting of permafrost owing to warming generates an erodible state in soil. Moreover, there are some indirect influences of climate change on erosion which are related to soil and vegetation changes caused by climate change (Kundzewicz et al. 2007; Bates et al. 2008). 55 2.5.2.6. Climate Change Impact on Future Freshwater Availability
Climate based hydrological changes may have both positive and negative influences. For instance, climate change has resulted increasing runoff for some parts of the world and it may be beneficial for water users. However, it increases flood risk at same regions. A wetter trend in parts of southern South America has increased flood events, on the other hand, it contributed yields in the Pampas region of Argentina. Moreover, it helped to new fishing opportunities (Magrin, Travasso & Rodriguez 2005). Furthermore, increased streamflows can be harmful for the regions with a shallow water table. A water table rise in these regions can damage agricultural use and can be harmful for buildings in urban areas. For instance, Kharkina (2004) reported annual damage due to shallow water tables as US$5–6 billion in Russia and it is expected to increase in the future. It is significant to realize that increased runoff is beneficial if it contributes to low flow season runoffs (Kundzewicz et al. 2007; Bates et al. 2008).
2.5.2.7. Climate Change Impact on Future Freshwater Demand
Agriculture is the largest water consumer sector. Increasing temperatures and higher precipitation variability cause increased irrigation water demand. It was projected that net irrigation demand will increase by +2% to +15% in the China and net irrigation demand will change by −6% to +5% in India by applying A2 and B2 scenarios as interpreted by two climate models by 2020 (Döll 2002; Döll, Florke & Vassolo 2003). Estimation of different climate models for worldwide irrigation demand changes ranging from 1 to 3% by the 2020s and 2 to 7% by the 2070s. B2 emission scenario based climate scenario concluded the largest net irrigation demand increase in global scale (Kundzewicz et al. 2007).
Some studies showed that the increase in domestic water demand and industrial water demand owing to climate change will be less than 5% by the 2050s at selected locations (Mote et al. 1999; Downing et al. 2003). However, climate change can increase indirectly water use cooling of thermal power plants which generate electricity to cool buildings. Protopapas, Katchamart and Platonova (2000) reported
56 approximately 2% of daily per capita water use increase in New York City on the days above 25°C (Kundzewicz et al. 2007).
2.5.2.8. Climate Change Impact on Future Water Stress
Different studies in literature give significantly different results about global estimation of the number of people living in areas with water stress. Climate change is only one factor which is affecting future water stress and other factors such as demographic, socio-economic and technological changes are more effective on future water stress at many regions. Estimations of number of people who will experience water stress in the future after the 2050s are very strongly depend on the used SRES scenario. A2 scenario projects significant increase while the increase level is lower under the A1 and B1 scenarios. It is important to note that climate change would decrease water stress at global scale in terms of per capita water availability indicator due to projected increases in runoff in the most populated parts of the world, mainly in eastern and south-eastern Asia. Nonetheless, increases in runoff are projected in high flow season and if the extra water is not stored by using additional reservoirs, runoff increases may not provide benefit for water stress problems. Moreover, seasonal pattern changes and increasing probability of extreme events may offset the influences of annual available freshwater resources rises and demographic changes (Kundzewicz et al. 2007; Bates et al. 2008).
There are very significant non-climatic factors on water stress including income, water- use efficiency, water productivity, and industrial production. Income growth is more effective than population growth on increasing water use and water stress at some regions. Table 2.6 presents the summary of findings from studies on climate change, population and water stress relation done by Alcamo et al. (2007) and Arnell (2004) (Kundzewicz et al. 2007).
57 Table 2.6: Population growth and climate change effects on the population living in water stressed catchments (Kundzewicz et al. 2007, Bates et al. 2008)
In Table 2.6, water stressed basins were defined as basins where per capita renewable water resources of less than 1,000 m3/yr. Projections in above studies are based on emissions scenarios for several climate model runs.
2.5.2.9. Climate Change Impact on Water Costs and Other Socio-economic Aspects of Freshwater
Available water amount for withdrawals is based on runoff, groundwater recharge, aquifer conditions, water quality and water supply infrastructure. Drinking water supply is more related to the level of water supply infrastructure than runoff quantity. Accessing safe drinking water will be challenging in regions where runoff and/or groundwater recharge decreases due to climate change. Moreover, climate change will require additional costs for the water supply sector because of the “changing water levels affecting water supply infrastructure” (Kundzewicz et al. 2007).
It is not accurate approach to just consider climate change impacts on long-term average annual runoff. Changes in seasonal runoff regime and interannual runoff variability can be as substantial as maybe more significant than long term average annual runoff changes. Snow melt dominated basins are good examples for importance of seasonal runoff changes owing to warming. Decreases in snowpack in winter season will very likely result decreasing river flows in summer and autumn
58 seasons. For instance, the Rhine might suffer from a decrease of summer flows by 5– 12% by the 2050s and these decreases will have significant negative impacts on water supply, particularly for thermal power plants (Middelkoop et al. 2001). Moreover, Krysanova and Wechsung (2002) reported actual evapotranspiration increase in the Elbe River Basin by 2050 and Krysanova, Hattermann and Habeck (2005) projected decreases in river flow, groundwater recharge, crop yield and diffuse source pollutions (Kundzewicz et al. 2007).
Studies on future water availability of western China showed that earlier spring snowmelt and reductions in glaciers will likely decrease water availability for irrigated agriculture. Thus, additional water structures including wells and reservoirs which are necessary to guarantee a reliable water supply under climate change are required for China and these investments will result additional costs for water management organizations in China (Kundzewicz et al. 2007). This cost will be low in low water stressed basins while it will be high in high water stressed basins (Kirshen et al. 2005). Thus, climate change and increasing demands will increase water supply costs over the world (Kundzewicz et al. 2007).
Irrigated agriculture is one the most affected sector from future water availability and it is very likely to result substantial socio-economic problems for farmers. For instance, Chen, Gillig and McCarl (2001) projected decreases in the net income of farmers by 16–30% by the 2030s and by 30–45% by the 2090s owing to reductions in irrigation water supply and increases in irrigation water demand in Texas (Kundzewicz et al. 2007).
Desalinated water usage is relatively new water supply technology and if desalinated water has to be used instead of freshwater owing to climate change influences on freshwater availability, climate change will induce additional cost for water supply. The average cost of desalination is currently around US$1.00/m3 for seawater and US$0.60/m3 for brackish water (Zhou & Tol 2005). Moreover, freshwater chlorination
59 cost is approximately US$0.02/m3 (Kundzewicz et al. 2007). However, in developing countries, due to its cost, it is difficult to apply desalination plants.
Bates et al. (2008) expressed that “The impact of climate change on flood damages can be projected, based on modelled changes in the recurrence interval of current 20- or 100-year floods and in conjunction with flood damages from current events as determined from stage discharge relations and detailed property data” (Bates et al. 2008, p. 46). By using above methodology, Schreider, Smith and Jakeman (2000) predicted increases in the average annual direct flood damage by four- to ten-fold under doubled CO2 conditions for three Australian drainage basins. Choi and Fisher (2003) projected change in flood damages for selected US regions by using two climate change scenarios. Projected mean annual precipitation increases in these regions are 13.5% and 21.5%, respectively. They stated increase by more than 140% in the mean and standard deviation of flood damage by using a structural econometric (regression) model based on a time-series of flood damage and input data set consists of population, a wealth indicator and annual precipitation. They also underlined the necessity of strong social infrastructure not to exposure dangerous flood losses. Moreover, Kirshen et al. (2005) investigated the potential flood influences of extreme precipitation changes by utilizing the Canadian Climate Center model and the IS92a scenario for the metro Boston area in the north-eastern US. According to this study, overall cost of flood damage and the number of properties damaged by floods would increase by two times by 2100 without adaptation practices. Kirshen et al. (2005) implied the importance of performing adaptation investment to decrease flood damages (Kundzewicz et al. 2007).
Hall, Sayers and Dawson (2005) conducted similar study to above studies for better understanding of climate change impact on flood damages in England and Wales in the 2080s. They coupled four greenhouse gases emissions scenarios with four socio- economic change scenarios in a SRES-like framework and they projected increases in flood damages for all scenarios without further investments for flood control. They projected £5 billion annual damage in a B1-type world by 2080. On the other hand,
60 according to B2-type world, it is projected as £1.5 billion which is close to current value with £1 billion. However, if results of both scenarios are normalized with respect to gross domestic product, scenarios give approximately same amount of damages (Evans et al. 2004; Hall, Sayers & Dawson 2005; Kundzewicz et al. 2007).
Changes may also have impacts on navigation. For example, increased flood periods in the future would disrupt navigation more frequent, and low flow can extend the loading time of ships (Kundzewicz et al. 2007). For instance, restrictions on loading in the Rhine River is 19 days currently and it may increase to 26–34 days by the 2050s (Middelkoop et al. 2001; Kundzewicz et al. 2007).
Owing to climate change impacts on river discharge, significant changes are expected in hydropower generation sector. Bates et al. (2008) explained these expectations for different parts of the world based on some studies in literature as follows:
Hydropower impacts for Europe have been estimated using a macro-scale hydrological model. The results indicate that by the 2070s the electricity production potential of hydropower plants existing at the end of the 20th century will increase (assuming IS92a emissions) by 15–30% in Scandinavia and northern Russia, where currently between 19% (Finland) and almost 100% (Norway) of electricity is produced by hydropower (Lehner, Czisch & Vassolo 2005). Decreases of 20–50% and more are found for Portugal, Spain, Ukraine and Bulgaria, where currently between 10% (Ukraine, Bulgaria) and 39% of the electricity is produced by hydropower (Lehner, Czisch & Vassolo 2005). For the whole of Europe (with a 20% hydropower fraction), hydropower potential is projected to decrease by 7–12% by the 2070s. In North America, potential reductions in the outflow of the Great Lakes could result in significant economic losses as a result of reduced hydropower generation both at Niagara and on the St. Lawrence River (Lofgren et al. 2002). For a CGCM1 model projection with 2°C global warming, Ontario’s Niagara and St. Lawrence hydropower generation would decline by 25– 35%, resulting in annual losses of
61 Canadian $240–350 million at 2002 prices (Buttle et al. 2004). With the HadCM218 climate model, however, a small gain in hydropower potential (+3%) was found, worth approximately Canadian $25 million per year. Another study that examined a range of climate model scenarios found that a 2°C global warming could reduce hydropower generating capacity on the St. Lawrence River by 1–17% (LOSLR 2006) (Bates et al. 2008, p. 46-47).
2.5.2.10. Vulnerable Freshwater Areas to Climate Change
Above discussed climate change effects on fresh water resources may impact sustainable development and they can produce significant risks for many parts of the globe. Some sensitive regions to climate change induced hydrological alterations are shown in Fig. 2.20.
Fig.2.20: Map of some regions where are sensitive to climate change impacts on freshwater availability (Kundzewicz et al. 2007, Bates et al. 2008)
62 Fig. 2.20 was constituted by Kundzewicz et al. (2007) by using related papers in literature. In Fig. 2.20, case 1 is stated by Bobba et al. (2000), while 2: Barnett et al. (2004), 3: Döll and Flörke (2005), 4: Mirza et al. (2003), 5: Lehner, Czisch and Vassolo (2005), 6: Kistemann et al. (2002), 7: Porter and Semenov (2005).
2.5.2.11. Uncertainties in Impact Projection Studies
The most important uncertainty of climate change impacts studies regarding to water resources is precipitation input. Secondly, uncertainty of greenhouse gas emission is very important on hydrological impact studies. Kay, Bell and Davies (2006) stated that largest uncertainties of flood statistics study in two UK basins are uncertainty from the GCM structure, followed by the emissions scenarios and hydrological modelling (Kundzewicz et al. 2007). More detailed information regarding to uncertainties of hydrological impact studies were presented in the further parts of this study.
2.5.3. Water Issues Based Adaptation to Climate Change
Kundzewicz et al. (2007) summarised some supply-side and demand-side adaptation options in Table 2.7, supply-side adaptation options generally present storage capacity increases or water courses abstractions. On the other hand, demand-side adaptation options are based on the cumulative actions of individuals (Kundzewicz et al. 2007).
Table 2.7: Summary of supply-side and demand-side adaptation options (Kundzewicz et al. 2007)
63 Moreover, Bates et al. (2008) presented Table 2.8 which is a good summary of adaptation policies for different parts of the world. They prepared this table based on IPCC assessment report. In Table 2.8, WGII refers to Working Group II of IPCC fourth assessment report. Moreover, WG III refers to Working Group III of IPCC fourth assessment report.
Table 2.8: Some adaptation examples (Bates et al. 2008)
64 2.6. Climate Change Impacts on Ecosystems and Biodiversity
The key variables which determine distribution, growth, productivity and reproduction of plants and animals are temperature and moisture regimes. The important changes in climate in the coming decades will have effects on moisture availability, timing & volume of streamflow, water levels of wetlands and available mist water in tropic mountain forests (Bates et al. 2008).
In 21st century, projected changes in hydrology, which were performed by IPCC, will influence biodiversity in many regions at every continent. Impacts of changing hydrology on species have already been observed in many areas over the world (Rosenzweig et al. 2007). Root et al. (2003) stated changes in animal and plants due to climate change in 20th century. Approximately 80% of changes are in agreement with observed temperature changes. Moreover, it should be noted that temperature can also indirectly affect species by changing moisture availability (Rosenzweig et al. 2007; Bates et al. 2008).
Some amphibians and other aquatic in Costa Rica, Spain and Australia are expected to extinct due to effects of temperature changes and water stress (Pounds et al. 2006). Drying of wetlands in the Sahel, which are used as stopovers of birds while they migrate to the Northern Hemisphere breeding grounds, will influence migration of birds. Unprecedented levels of plant and animal species in South Africa are expected to face extinction. Among all ecosystems, freshwater aquatic ecosystems are the most vulnerable ecosystems to climate change and they are under great risk. They have the highest proportion of species which are under extinction risk owing to climate change (Millenium Ecosystem Assessment 2005) (Bates et al. 2008; Rosenzweig et al. 2007).
2.7. Climate Change Impacts on Agriculture and Irrigation
Agriculture has a disputable significance in human welfare and it is the main land use over the globe. Currently, approximately 1.2-1.5 billion hectares are under crop cultivation and another 3.5 billion hectares being grazed. Per capita food demand has 65 been rising in parallel to increasing population and agricultural production must continue to meet population growth. Moreover, agriculture is very significant economic, social and cultural activity. Furthermore, it harbours wide range of ecosystems (Howden et al. 2007).
Worldwide and regionally, water is a huge factor for production of food. Close to 80% of global agriculture land is rain fed. In these areas, sufficient precipitation has a critical role on crop productivity due to its key impact to meet soil moisture distribution and evaporative demand (FAO 2003). Bates et al. (2008) expressed based on Easterling et al. (2007) that agricultural productivity is vulnerable to climate change especially in areas such as arid or semi-arid regions in tropics and sub-tropics as well as the Mediterranean type regions in Europe, Australia and South America. Water is very significant for global food production not only as the form of precipitation but also in the form of irrigation and available water resources for irrigation. Approximately 18% of global agricultural land is represented as irrigated land which is producing about 1 billion ton of grain annually or about half of the world’s total supply. This is due to the fact that “irrigated crops yield on average 2-3 times more than rain fed counterparts (FAO 2003)” (Bates et al. 2008, p. 59).
Water is a must for food producing process. Currently, 850 million people in the world are suffering from inadequate food (FAO 2003) due to insufficient agricultural activities and it is related to water deficiency or lack of required irrigation systems at many regions. Water availability for agriculture sector established very strong connection between climate change and agriculture. Because, as it is explained before, climate change has substantial influences on water availability and these effects are expected to be more intensive in coming decades. Over the next few decades, due to socio- economic pressures, there will be a push for increased competition between demand from non-agricultural sectors and irrigation needs, which will ultimately decrease the availability and quality of water resources for food (Bates et al. 2008; Easterling et al. 2007).
66 Due to sensitivity of agriculture to climate variations, climate change can impose agriculture crucially. Global temperature has been rising dramatically especially after 1950s due to human activities and it has been causing changes in many climate variables such as rainfall, humidity. Greenhouse gases emissions caused 0.1 °C warming per decade for several decades (Solomon et al. 2007). Greenhouse gases emission in atmosphere has continued to increase (Howden et al. 2007) and some expected climate change impacts are coming in true faster than its estimated rate.
Agricultural is climate and environment based activity. Thus, despite technological improvements, dependence of agriculture to climate and environment does not change. So, agriculture is among the most affected sectors by climate change. Climate change and agriculture has a bilateral relation. Agriculture sector contributes to climate change, also it is influenced by climate change (Aydinalp & Cresser 2008).
Agriculture sector contributes to climate change by emitting carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O) gases. Contribution of agriculture to greenhouse gases emission is shown graphically in Fig. 2.21.
Forestry
Agriculture 19.40% 17.40% Waste and Wastewater
7.90% 13.50% Energy Supply
Transport 13.10% 2.80% Residential and 25.90% commercial buildings Industry
Fig.2.21: Greenhouse emission sources (IPCC 2007)
67 Main reason of CO2 gas emission regarding to agriculture is deforestation owing to agricultural expansion. Forests are large carbon sinks which has a significant role at carbon cycle. Deforestation due to agricultural activities release substantial amount of carbon to atmosphere which later increases greenhouse gases concentration in atmosphere and contribute climate change. Mentioned activity contributes approximately 30% of total CO2 emissions globally. Moreover, fossil fuels which are used in agricultural activities can be shown among agricultural contribution to climate change (Aydinalp & Cresser 2008).
Methane (CH4) is the most important greenhouse gas released by agriculture sector. Paddy fields are responsible from majority of methane with 91%, then animal husbandry with 7% and finally burning agricultural wastes with 2%. Many parameters including land in cultivation, fertilizer use, water management, density of rice plants and other agricultural practices are effective on methane release of paddy fields. China is the most significant source of methane emission over world (Aydinalp & Cresser 2008).
The most important nitrous dioxide release in agriculture sector is due to nitrogen fertilizer usage, legume cropping and animal waste. Nitrogen fertilizers are used by farmers for the purpose of enhancing crop growth. While some of the fertilizers are being consumed by crop, remaining leaches into surrounding surface & groundwater and joins to atmosphere (Aydinalp & Cresser 2008).
On the other hand, there are some positive effects of agriculture on climate change. Green plants are natural carbon sink which store it so reduce greenhouse gases in the atmosphere. Moreover, lands store carbon and contribute to decreasing carbon concentration in the atmosphere. Furthermore, usage of biofuels or agro fuels instead of fossil fuels is very beneficial to reduce greenhouse gases emission.
Some precautions to reduce greenhouse emission in agriculture sector are better management of rice paddies, some changes in agricultural production type, lessening
68 CO2 emission in agriculture by irrigation scheduling, minimum tillage, solar drying of crops and improved fertilizer management (Aydinalp & Cresser 2008). One should note that the role of forests and vegetation is so substantial in greenhouse gases concentration in the atmosphere due to its sink characteristic for CO2. Vegetation and soils of unmanaged forests has a capability of holding 20-100 times more carbon per unit area than agricultural lands. Thus, forestation is also very beneficial action to reduce greenhouse gases concentration in the atmosphere (Aydinalp & Cresser 2008).
Although agriculture is among the most affected sectors and climate change- agriculture relation is so strong, positive or negative influences of climate change on agricultural sector have not received sufficient attention so far. Increasing temperatures may result “nonlinear and increasingly negative” influences on agriculture. On the other hand, it may be an opportunity for agricultural investments which are adapted climate change earlier (Howden et al. 2007).
Global warming coupled with rainfall amount, intensity and frequency changes will have impacts on global agriculture productivity particularly in tropical regions. Furthermore, increasing number of extreme events such as floods and droughts will directly affect agriculture sector. Moreover, changes in patterns of pests and diseases owing to climate change will influence agriculture. Thus, it is important to develop regional climate change impact studies for agriculture. However, it is not simple because of following reasons: there are uncertainties in regional climate change projections, different crops respond differently to increased levels of atmospheric CO2 and finally, uncertainty coming from ”potential for adaptation of agricultural practices” (Aydinalp & Cresser 2008, p.674).
Aydinalp and Cresser (2008) stated that climate change influence on agriculture will be small or moderate in global scale. However, it can have substantial effects in regional scales. Agricultural impacts of climate change will be intensive particularly in regions where low income populations (developing countries) whose economies are dependent on extremely agricultural systems. The most significant negative
69 agricultural influences owing to climate change are expected in dry land areas at lower latitudes and in arid and semi-arid areas, especially for those dependent on rainfall, non-irrigated agriculture. Mentioned populations which are under risk are located in South, Southeast Asia and Africa. Moreover, marginal agriculture and farmers will be more influenced by climate change. Regions where climate is at the edge or near the climate zone, which is the most suitable climate zone for a specific crop, can be more affected by changes in climate. Because, climate change can lead climate zone alteration and it may influence the crop productivity substantially (Aydinalp & Cresser 2008).
As stated before, there are both positive and negative impacts of climate change on agriculture depending on crop type, geography and climate. Some of them are tried to be explained in previous paragraphs. Graphical illustration of some positive and negative impacts of climate change on agriculture is shown in Fig. 2.22.
Fig.2.22: Graphical illustration of possible advantages and disadvantages of climate change on agriculture (Crimp 2000)
70 Climate change can reduce the yields and it may change geographical limits of crops due to changes in rainfall, temperature, soil moisture and CO2 concentrations. Decrease in streamflows of catchments owing to global warming which is also shown in current study, will directly impact irrigation water which is vital for agriculture in many parts of world. Moreover, climate change will lead changes in soil properties such as “soil organic matters loss, salinization, erosion and soil nutrients leaching” and these changes will affect crop yields and agriculture productivity directly. Furthermore, changes in biodiversity of agricultural lands because of climate change can result negative effects on agriculture. In addition, agriculture in low-lying coastal areas will be affected by climate change induced sea level rises. Flooding can be costly and damaging for agriculture of coastal areas (Aydinalp & Cresser 2008).
It is also possible to give some numerical findings about climate change impacts on agriculture. For instance, Gunasekera et al. (2008) stated estimates of climate change impacts on regional agricultural productivity based on Cline (2007). It is shown in Table 2.9.
Table 2.9: Climate change impact on agricultural productivity at 2050 (Gunasekera et al. 2008; Cline 2007)
Moreover, Gunasekera et al. (2008) reported important declines in agricultural productivity in Australia by 2030 and 2050 which are shown in Fig. 2.23.
71 Fig.2. 23: Change in Australian agricultural production owing to climate change by 2030 and 2050 (Gunasekera et al. 2008)
Moreover, Kingswell (2006) presented study of Howden and Jones (2004) which investigated climate change impacts on wheat production in 10 regions over Australia towards 2070 and summarized their finding in a table which is shown as Table 2.10.
Table 2.10: Regional impacts of climate change on wheat production at 10 regions over Australia (Howden & Jones 2004) (Kingswell 2006)
72 Although climate change has important negative impacts on agriculture, increasing atmospheric CO2 is beneficial for some plants and it increases their growth rate which is enhancing agricultural productivity. Moreover, increasing CO2 concentration helps crop plants to use water more efficient. It is possible to classify plants as C3, C4 and CAM based on the photosynthetic pathways they employ. Among these C3 plants such as potato, rice, and soy bean have benefits from increasing CO2. Furthermore, increasing temperatures provide some agricultural advantages especially for the regions where low average temperatures limit agricultural activities. Temperature increase in these regions can extend growing season available for plants and reduce growing period. In addition, climate change induced precipitation increase will provide more irrigation water and will increase agricultural productivity in some regions (Aydinalp & Cresser 2008).
Moreover, Bates et al. (2008) explained plausible climate change impacts on agriculture based on Easterling et al. (2007) and some more studies as follows:
In general, while moderate warming in high-latitude regions would benefit crop and pasture yields, even slight warming in low-latitude areas, or areas that are seasonally dry, would have a detrimental effect on yields. Modelling results for a range of sites show that, in high-latitude regions, moderate to medium
increases in local temperature (1–3°C), along with associated CO2 increases and rainfall changes, can have small, beneficial impacts on crop yields. However, in low-latitude regions, even moderate temperature increases (1–2°C) are likely to have negative yield impacts for major cereals. Further warming has increasingly negative impacts in all regions.
Regions where agriculture is currently a marginal enterprise, largely due to a combination of poor soils, water scarcity and rural poverty, may suffer increasingly as a result of climate change impacts on water. As a result, even small changes in climate will increase the number of people at risk of hunger, with the impact being particularly great in sub-Saharan Africa.
73 Increases in the frequency of climate extremes may lower crop yields beyond the impacts of mean climate change. Simulation studies since the TAR have considered specific aspects of increased climate variability within climate change scenarios. Rosenzweig et al. (2002) computed that, under scenarios of increased heavy precipitation, production losses due to excessive soil moisture (already significant today) would double in the US to US$3 billion/yr in 2030. In Bangladesh, the risk of crop losses is projected to increase due to higher flood frequency under climate change. Finally, climate change impact studies that incorporate higher rainfall intensity indicate an increased risk of soil erosion; in arid and semi-arid regions, high rainfall intensity may be associated with a higher possibility of salinisation, due to increased loss of water past the crop root zone.
Impacts of climate change on irrigation water requirements may be large. A few new studies have further quantified the impacts of climate change on regional and global irrigation requirements, irrespective of the positive effects of elevated CO2 on crop water-use efficiency. Döll (2002), in considering the direct impacts of climate change on crop evaporative demand, but without any
CO2 effects, estimated an increase in net crop irrigation requirements (i.e., net of transpiration losses) of between 5% and 8% globally by 2070, with larger regional signals (e.g., +15%) in south-east Asia. Fischer et al. (2007), in a study that included positive CO2 effects on crop water-use efficiency, computed increases in global net irrigation requirements of 20% by 2080, with larger impacts in developed versus developing regions, due to both increased evaporative demands and longer growing seasons under climate change. Fischer et al. (2007) and Arnell et al. (2004) also projected increases in water stress (measured as the ratio of irrigation withdrawals to renewable water resources) in the Middle East and south-east Asia. Recent regional studies have likewise underlined critical climate change/water dynamics in key irrigated
74 areas, such as northern Africa (increased irrigation requirements; Abou-Hadid et al. 2003) and China (decreased requirements; Tao et al. 2003).
At the national scale, some integrative studies exist. In the USA, two modelling studies on adaptation of the agricultural sector to climate change (i.e., shifts between irrigated and rainfed production) foresee a decrease in both irrigated areas and withdrawals beyond 2030 under various climate scenarios (Reilly et al. 2003; Thomson et al. 2005). This is related to a declining yield gap between irrigated and rain-fed agriculture caused either by yield reductions of irrigated crops due to higher temperatures, or by yield increases of rain-fed crops due to higher precipitation. These studies did not take into account the increasing variability of daily precipitation and, as such, rain-fed yields are probably overestimated.
For developing countries, a 14% increase in irrigation water withdrawal by 2030 was foreseen in an FAO study that did not consider the impacts of climate change (Bruinsma 2003). However, the four Millennium Ecosystem Assessment scenarios project much smaller increases in irrigation withdrawal at the global scale, as they assume that the area under irrigation will only increase by between 0% and 6% by 2030; and between 0% and 10% by 2050.
The overwhelming water use increases are likely to occur in the domestic and industrial sectors, with withdrawals increasing by between 14% and 83% by 2050 (Millennium Ecosystem Assessment 2005 a,b). This is based on the idea that the value of water will be much higher for domestic and industrial uses, which is particularly true under conditions of water stress.
Locally, irrigated agriculture may face new problems linked to the spatial and temporal distribution of streamflow. For instance, at low latitudes, especially in south-east Asia, early snowmelt may cause spring flooding and lead to a summer irrigation water shortage (Bates et al. 2008, p. 61-62.).
75 2.7.1. Reducing Impacts of Climate Change on Agriculture Sector
The first step to reduce impacts of climate change on agriculture and other sectors should be mitigation of greenhouse gases which is the major source of global warming. Agricultural activities emit significant amount of greenhouses into atmosphere. Some mitigation policies such as emissions trading scheme (Gunasekera et al. 2008) should be encouraged to reduce greenhouse gases emission by agriculture sector. However, there is “high diversity of entities” (Garnaut 2007, p. 5). For instance, there are more than 130000 commercial farms in Australia under different climates and using different agricultural industries. Each agriculture entities have its own unique production system and management approach so it is not simple to apply general uniform policies for agriculture sector (Garnaut 2007). Moreover, majority of greenhouse gases emitted by agriculture sector are the consequence of biological processes that are not highly dependent on human influence. Finally, mitigation policies require initial investment (economical cost) and advantages or benefit of initial mitigation investment will be obtained in future. Early cost & late profit process discourages agriculture sector to implement mitigation policies (Garnaut 2007).
Although mitigation has huge importance to obstruct climate change, it is not easy for governments to actualize theoretical recommendations into practice. Moreover, climate change will continue under any greenhouse emission scenario even greenhouse gases are fixed at current levels. So, agriculture sector must perform accurate actions to achieve adaptation to climate change. Agricultural adaptation connotes effectively managing potential climate risks resulting from climate change. Some key adaptation titles to encounter climate change impacts and taking advantage of changing climate was stated by Aydinalp & Cresser (2008) as follows:
- Changes in crops and crop varieties (especially cultivation of more resistant crops to heat, frosts or drought) - Developing efficient water management and irrigation practices - Improvement in planting schedules and tillage applications - “More efficient use of mineral fertilizers“ 76 - Educating farmers about possible effects of climate change.
In addition to Aydinalp and Cresser (2008), Gunasekera et al. (2008) advised altering pasture rotations and modifying grazing times. Furthermore, additional adaptation opportunities for agriculture sector are stated by Garnaut (2007) as follows:
- Applying changes in harvesting patterns and rotation periods - “Improved moisture management by increased on-farm capture and storage, residue management and weed control” - Improvement in seasonal climate forecasting with the aim of supporting decision making - “Use of financial management tools to manage risk” (Garnaut 2007, p. 4).
In addition to adaptation policies mentioned above, Kingswell (2006) advised to use satellite imagery technology based improved pasture and crop management decision support systems. He also implied to importance of improving accurate climate prediction models which can estimate extent and duration of droughts. Other adaptation recommendations which are explained by Kingswell (2006) are forming of low cost surface sealants on farm dam catchments to generate more streamflow even from small rainfall events, development of low cost desalination plants to use saline groundwater for irrigation purposes, re-design of farm housing, building, machinery and outdoor clothing with the aim of accommodating extreme heat, development of profitable crops or tree species which involve returns as renewable energy or carbon sinks.
Although adaptation to climate change is very significant for agriculture, there are some significant challenges in effectuation of agricultural adaptation. They are explained by Garnaut (2007) as follows:
Need for finding financial support to develop alternative technologies and improved climate data and monitoring
77 Need for cost-benefit analyses of adaptation practices Need for available information relating to climate change impacts and the benefits of adaptation; Challenge regarding to initial investment cost Challenge to educate community leaders and farmers in adaptation and change Challenge for small companies to overcome uncertainty and risk.
Easterling et al. (2007) divided adaptation approaches into two groups, autonomous adaptation and planned adaptation. Bates et al. (2008) defined autonomous adaptation based on Easterling et al. (2007) as “responses that will be implemented by individual farmers, rural communities and/or farmers’ organisations, depending on perceived or real climate change in the coming decades, and without intervention and/or co-ordination by regional and national governments and international agreements” (Bates et al. 2008, p. 63). Moreover, Bates et al. (2008) described planned adaptation as follows: “changes in policies, institutions and dedicated infrastructure which will be needed to facilitate and maximise long-term benefits of adaptation responses to climate change” (Bates et al. 2008, p. 63).
Bates et al. (2008) explained that autonomous adaptation options are largely extended and intensified forms of existing risk management and production activities, so they are already known by farmers and communities. Bates et al. (2008) stated some autonomous adaptation options based on Easterling et al. (2007) item by item: “adoption of varieties/species with increased resistance to heat shock and drought; modification of irrigation techniques, including amount, timing or technology; adoption of water-efficient technologies to ‘harvest’ water, conserve soil moisture (e.g. crop residue retention), and reduce siltation and saltwater intrusion; improved water management to prevent waterlogging, erosion and leaching; modification of crop calendars, i.e., timing or location of cropping activities; implementation of seasonal climate forecasting” (Bates et al. 2008, p. 64).
78 Bates et al. (2008) expressed by using information from Easterling et al. (2007) that developing new infrastructure, policies, and institutions for the purpose of supporting, facilitating, co-ordinating and maximising the benefits of new management and land- use arrangements should be the focus of planned adaptation. Moreover, they linked the success of planned adaptation to following practices: “improved governance, including addressing climate change in development programmes; increasing investment in irrigation infrastructure and efficient water-use technologies; ensuring appropriate transport and storage infrastructure; revising land tenure arrangements (including attention to well defined property rights); and establishing accessible, efficiently functioning markets for products and inputs (including water pricing schemes) and for financial services (including insurance)” (Bates et al. 2008, p. 65).
Large number of studies showed that, in general, semi-arid and arid basins are the most vulnerable regions to water stress. If precipitation decreases occur in arid and semi-arid regions, irrigation water demand would put very strong restriction on other water demands including domestic water demand, industrial water demand and hydropower generation water demand. This offers to focus on policies to reduce agricultural water demand. Improvements in irrigation efficiency, either through market mechanisms or increased regulations and improved governance can be given as example to policies to decrease irrigation water use (Bates et al. 2008).
2.7.2. Climate Change and Irrigation
Producing sufficient food for growing population is a vital challenge of humankind. Water has an indisputable significance for agricultural production. Currently, irrigated area over the world is 270 Mha, approximately 18% of total cultivated land. Agriculture is the largest consumer of available water resources with around 70%, 2630 Gm3/year out of 3815 Gm3/year (Fischer et al. 2007).
Effects of climate change on agriculture can be basically assessed in two distinct ways. First of them is impact of changing climate on crop growth which is explained above.
79 Second one is climate change influence on water availability and supplying sufficient water for irrigation purposes (Schlenker, Hanemann & Fischer 2007).
Climate change has very important consequences on irrigation and drainage over the world. Svendsen and Kunkel (2009) implied that the best defined impact of climate change on irrigation is loss of snowpack and glacier storage of winter precipitation which have been already observed at many parts of the world such as western United States, Australia and India (Barnett et al. 2008). Some more information about this issue can be found in this thesis where climate change and snow hydrology relationship was explained.
The snowpack in the mountains is a significant source of natural storage. Schlenker, Hanemann and Fischer (2007) expressed that water stored in natural snowpack almost matches with the water stored in state’s major reservoirs in California. Climate models project one third of snowpack storage lost owing to increasing winter temperatures by the middle of this century (Schlenker, Hanemann & Fischer 2007). The snowpack losses lead to late season (late spring, summer and early autumn) droughts resulting shift in planting calendars (planting earlier) for the purpose of avoiding late season droughts. Moreover, high spring temperatures may be a reason for a move to earlier planting dates but it requires investigation of runoff and air temperature relation at the basins (Swendsen & Kunkel 2009).
One of the major purposes of constructing water storage reservoirs is providing sufficient water to agriculture sector or in other words irrigation. Reductions in available surface water due to climate change can require new operating plans for reservoirs. According to seasonal changes of streamflows, some new reservoirs or increase of existing reservoirs’ capacity can be needed to store early season runoff with the aim of using it for late season water supply. Artificial recharge of groundwater aquifers by using early season runoff for using groundwater in late season irrigation can be another method to overcome irrigation problems in late summer and autumn seasons when surface water is diminished. Furthermore, recycled urban waste water
80 can be used as irrigation water to deal with climate change induced water availability problem. Israel and Jordan are meeting more than half of their agricultural water demand by using recycled water. Using advanced technology in irrigation systems is one of the best solutions to deal with the irrigation water scarcity due to climate change or any other reasons. Drip irrigation is currently the most popular technology to use irrigation water efficiently (Swendsen & Kunkel 2009).
Following paragraphs review the studies which were done in an effort to investigate climate change impacts on agriculture over the world. Then, climate change and agriculture sector relation was discussed specifically for Turkey.
Mohamed, van Duivenbooden and Abdoussallam (2002) investigated the effects of climate change on agricultural production in the Sahel in Niger. They explained that due to rainfall inter-annual variability, farmers at West African Sahel were affected intensively. They underlined the substantial role of agriculture for West African Sahel. They stated that 80% of the 55 million population of West African Sahel is rural and agriculture is their major income source. They explained that 35% of the countries’ total Gross Domestic Products (GDPs) consist of agriculture and stock-farming sector. Moreover, increasing population with a rate of approximately 3%, poor sanitary facilities form additional problems to agriculture sector. Owing to above mentioned importance of agriculture sector and vulnerability of this sector to climate change, Mohamed, van Duivenbooden and Abdoussallam (2002) investigated current climate variability and future climate change on millet production at three millet-production regions in Niger. They used statistical models with the aim of projecting climate change impacts on future millet production. Mohamed, van Duivenbooden and Abdoussallam (2002) expressed that according to the past 30-years of rainfall and production data, sea surface temperature anomalies, the amount of rainfall in July, August & September, number of rainy days and wind erosion factor are the most important predictors of model. Based on modelling studies, Mohamed, van Duivenbooden and Abdoussallam (2002) projected approximately 13% decrease in millet production due to projected precipitation decrease and temperature increase in their study area.
81 Van Duivenbooden, Abdoussalam and Mohamed (2002) showed examined impacts of current climate variability and potential effects of future climate change on groundnut and cowpea production in Niger for three major agricultural areas including the groundnut basin. Firstly, they explained the significance of groundnut and cowpea production for Niger’s economy. They stated that rainfall in July, August and September is very substantial for groundnut and cowpea production in Niger. Thus, firstly they calculated mean annual and monthly rainfall, beginning, end and length of the rainy season, number of rainy days per month, amount of rainfall per rainy day and the maximum length of dry spell per month for the period of 1951-1998. They detected important changes for the following durations in study regions: Dosso 1951– 1968, 1969–1984 and 1985–1998; Maradi 1951–1970, 1971–1987 and 1988–1998; Zinder 1951–1966, 1967–1984 and 1985–1998. Van Duivenbooden, Abdoussalam and Mohamed (2002) used a statistical modeling approach with the aim of evaluating climate variability and climate change impacts on groundnut and cowpea production. They concluded 11 to 25% reduction in groundnut production while they found maximum 30% decrease in cowpea yield by 2025. Finally, they presented different strategies to counter projected production losses.
Rosenszweig et al. (2004) investigated climate change impacts on different parts of the world by improving links between climate change scenarios and hydrologic, agricultural, and planning models to study water availability for agriculture under climate change conditions. They implemented their approach for major agricultural regions in Argentina, Brazil, China, Hungary, Romania and the US. They stated that among study regions Northeastern China is the one which has the greatest water availability problem for agriculture both in current climate and future climate projections. Climate change will not affect runoffs in the Danube Basin substantially. They explained that existing water supply problems may become more intensive in Northern Argentina and it may require further investments to overcome future water stress. They projected no water problem for agriculture sector in Southeastern Brazil. Moreover, they had optimistic results for most of the regions studied in US. Rosenszweig et al. (2004) implied that only Brazil can tolerate agricultural land
82 expansion under changing climate. Finally, they explained that increasing water demand for urban and agricultural purposes may require improvement in crop cultivars, irrigation and drainage technology, and water management.
Gbetibouo and Hassan (2005) examined climate change influences on South Africa’s field crops by implementing Ricardian model. They performed analysis based on regression of farm net revenue on climate, soil and other socioeconomic variables. They implemented their methodology for seven field crops including maize, wheat, sorghum, sugarcane, groundnut, sunflower and soybean. Gbetibouo and Hassan (2005) reported that crop yields are more sensitive to marginal temperature increases than precipitation changes. They stated positive effects of temperature increases on net revenue. However, they expressed that reduction in precipitation will impact net revenue negatively. They implied the significance of season and location and added that adaptation strategies should not be uniform due to spatial variability of climate change influences. Gbetibouo and Hassan (2005) concluded that climate change scenarios may result different shifts in farming practices and patterns in different regions over South Africa.
Olesen (2005) explored climate change impacts on agriculture in Denmark. He stated based on previous climate change studies in Denmark that annual mean increase will increase around 1 to 4°C by the end of the 21st century depending on socio-economic development. In addition, he expressed winter rainfall increase up to 20%. He explained that agricultural productivity in Denmark will increase according to future climate projections and increasing CO2 concentrations. Olesen (2005) stated increase in grain production at both arable and dairy farms. He expressed that there are large number of uncertainties in agricultural impacts studies. He stated that major uncertainty is coming from climate change scenario and models which were used to simulate climate change impacts. In addition, uncertainty of secondary climate change factors including soil fertility, weeds, pests and diseases are alo significant due to significance of these factors on crop production.
83 Fischer et al. (2005) accomplished a comprehensive evaluation study for the purpose of investigating climate change impacts on agro-ecosystems over this century at global scale. They used integrated ecological-economic modelling methodology by using climate scenarios, agro-ecological zoning information, socio-economic drivers. They implemented their simulations by using the FAO/IIASA agro-ecological zone model and IIASAs global food system model. With the aim of achieving future simulations, they used future climate data projected by five different general circulation models, under four different socio-economic scenarios from the intergovernmental panel on climate change. Fischer et al. (2005) utilized dynamic crop models including DSSAT, EPIC, TEM for computing crop growth, water dynamics and harvest yield. Fischer et al. (2005) presented their approach and findings of study as follows:
First, impacts of different scenarios of climate change on bio-physical soil and crop growth determinants of yield are evaluated on a 5’ × 5’ latitude/longitude global grid; second, the extent of potential agricultural land and related potential crop production is computed. The detailed bio-physical results are then fed into an economic analysis, to assess how climate impacts may interact with alternative development pathways, and key trends expected over this century for food demand and production, and trade, as well as key composite indices such as risk of hunger and malnutrition, are computed. This modelling approach connects the relevant bio-physical and socioeconomic variables within a unified and coherent framework to produce a global assessment of food production and security under climate change. The results from the study suggest that critical impact asymmetries due to both climate and socio- economic structures may deepen current production and consumption gaps between developed and developing world; it is suggested that adaptation of agricultural techniques will be central to limit potential damages under climate change (Fischer et al. 2005, p. 2067).
Erda et al. (2005) used data from regional climate model named PRECIS, which is developed by the UK’s Hadley Centre, to generate China’s future climate projections.
84 By using PRECIS model, they predicted future climate for three different periods, 2010- 2019, 2040-2049 and 2070-2079 under A2 and B2 grennhouse gases emission scenarios. They explained that according to PRECIS model, temperature increases between 3 and 4° were projected for China by the end of twenty-first century. In addition, they reported precipitation increases ranging from 3.3 to 12.9%. They used regional crop models driven by PRECIS climate outputs to determine future yield of important three crops of China including rice, maize and wheat. According to modelling studies, Erda et al. (2005) reported that yields of rice, maize and wheat will decrease up to 37% in the next 20–80 years without carbon dioxide (CO2) fertilization. Moreover, they explained that “More complete reporting of free-air carbon enrichment experiments than was possible in the Intergovernmental Panel on Climate
Change’s Third Assessment Report confirms that CO2 enrichment under field conditions consistently increases biomass and yields in the range of 5–15%, with CO2 concentration elevated to 550 ppm levels of CO2 that are elevated to more than 450 ppm will probably cause some deleterious effects in grain quality” (Erda et al. 2005, p. 2149).
Ewert et al. (2005) developed a model to investigate climate change impacts on crop productivity over Europe. Summary of their study was given by them as follows:
The future of agricultural land use in Europe is unknown but is likely to be influenced by the productivity of crops. Changes in crop productivity are difficult to predict but can be explored by scenarios that represent alternative economic and environmental pathways of future development. We developed a simple static approach to estimate future changes in the productivity of food crops in Europe (EU15 member countries, Norway and Switzerland) as part of a larger approach of land use change assessment for four scenarios of the IPCC Special Report on Emission Scenarios (SRES) representing alternative future developments of the world that may be global or regional, economic or environmental. Estimations were performed for wheat (Triticum aestivum) as a reference crop for the time period from 2000 until 2080 with particular
85 emphasis on the time slices 2020, 2050 and 2080. Productivity changes were
modelled depending on changes in climatic conditions, atmospheric CO2 concentration and technology development. Regional yield statistics were related to an environmental stratification (EnS) with 84 environmental strata for Europe to estimate productivity changes depending on climate change as projected by the global climate model HadCM3. A simple empirical relationship
was used to estimate crop productivity as affected by increasing CO2 concentration simulated by the global environment model IMAGE 2.2. Technology was modelled to affect potential yield and the gap between actual and potential yield.
We estimated increases in crop productivity that ranged between 25 and 163% depending on the time slice and scenario compared to the baseline year (2000). The increases were the smallest for the regional environmental scenario and the largest for the global economic scenario. Technology development was identified as the most important driver but relationships that determine technology development remain unclear and deserve further attention. Estimated productivity changes beyond 2020 were consistent with changes in the world-wide demand for food crops projected by IMAGE. However, estimated increases in productivity exceeded expected demand changes in Europe for most scenarios, which is consistent with the observed present oversupply in Europe. The developed scenarios enable exploration of future land use changes within the IPCC SRES scenario framework (Ewert et al. 2005, p. 101-102).
Mutke, Gordo and Gil (2005) explored the response of Mediterranean Stone pine cone production to climate change. They stated that cone yields provide higher profit for the owners of these pine forests than any other forest product. Moreover, they noted that large annual variation of cone yields is a substantial point for forest management. Thus, Mutke, Gordo and Gil (2005) performed analysis on historical weather and yield registers over 41 years in the Northern Inland Plateau of Spain which is one of the
86 world’s major Stone pine areas. They concluded their findings as follows: “(1) Significant relationships found between rainfall and temperatures at certain key periods during the 4-year cone development period allowed for a multiple linear regression model for the log-transformed annual cone yield to be set up, (2) The observed trend of cone-yield reduction from 180 to less than 100 kg ha-1 in the last 40 years was slightly overestimated by the predicted effects of the covariables that show significant tendencies to a warmer and drier climate” (Mutke, Gordo & Gil 2005, p. 263).
Abraha and Savage (2006) evaluated potential impacts of climate change on grain yield of maize for midlands of KwaZulu-Natal in South Africa. They investigated climate change assessment on maize yield by using generated weather data, modified weather data under future climate change scenarios with a normal planting date, 15 days earlier and 15 days later by using CropSyst. They implemented above mentioned analysis for Cedara where is defined as summer rainfall location by them and located in the midlands of KwaZulu-Natal in South Africa. They generated baseline climate scenario by using 30 years observed data in stochastic weather generator named ClimGen. Abraha and Savage (2006) computed maize grain yields by using observed and generated weather data series with different planting dates and they compared the results. Simulated grain yields did not show large variance between generated weather data and observed weather data series. Then they modified baseline weather data in an effort to use them for generating climatic scenarios. They explained that “the climate changes corresponded to a doubling of carbon dioxide concentration to 700 µl l-1 without air temperature and water regime changes, and a doubling of carbon dioxide concentration accompanied by mean daily air temperature and precipitation increases of 2 °C and 10%, 2 °C and 20%, 4 °C and 10%, and 4 °C and 20%, respectively” (Abraha & Savage 2006, p. 150). They assumed that increase in the daily mean minimum air temperature will be three times larger than increase in daily mean maximum temperature. Moreover, they adjusted crop parameters regarding to radiation use and biomass transpiration efficiencies for maize in CropSyst with an aim of considering physiological changes owing to increased carbon dioxide concentration.
87 Abraha and Savage (2006) explained that maize grain yields are much more sensitive to temperature changes than precipitation under increased carbon dioxide concentration regimes.
Gay et al. (2006) conducted a study on relation between coffee production and climatic and economic variables in Veracruz, Mexico, with the aim of project potential impacts of climate change on agriculture. They developed an econometric model and validate it by means of statistical analysis. They used their model to project coffee production under different climate scenarios. Gay et al. (2006) generated future climate projections for using them in their model by assuming that the observed trends of climate variables will be same until the year 2020. They found based on their modelling study that temperature is the most significant climate factor affecting coffee production. It is very sensitive to seasonal temperature changes. Gay et al. (2006) stated that modelling study offered reduction in coffee production by 34% relative to current production. They explained that results showed that coffee production might not be economically viable for producers by 2020. They considered different economic variables including the state and international coffee prices, a producer price index for raw materials for coffee benefit, the national and the US coffee stocks, they expressed that the state real minimum wage is the most significant economic variable influencing coffee production. They made definition of real minimum wage by “a proxy for the price of labor employed for coffee production” (Gay et al. 2006, p. 259). They investigated different approaches on real minimum wage evolvement for the year 2020 due to its importance for coffee production.
Fuhrer et al. (2006) made a study on climate change risk to agriculture and forestry in Switzerland. They explained that some of the intensive weather events have become more frequent over Europe for the next 50 to 100 years because of climate change. They presented two main purposes of their study and their findings as follows:
The paper aims to (i) describe observed trends and scenarios for summer heat waves, wind storms and heavy precipitation, based on results from simulations
88 with global circulation models, regional climate models, and other downscaling procedures, and (ii) discuss potential impacts on agricultural systems and forests in Switzerland. Trends and scenarios project more frequent heavy precipitation during winter corresponding, for example, to a three-fold increase in the exceedance of today’s 15-year extreme values by the end of the 21st century. This increases the risk of large-scale flooding and loss of topsoil due to erosion. In contrast, constraints in agricultural practice due to waterlogged soils may become less in a warmer climate. In summer, the most remarkable trend is a decrease in the frequency of wet days, and shorter return times of heat waves and droughts. This increases the risk of losses of crop yield and forage quality. In forests, the more frequent occurrence of dry years may accelerate the replacement of sensitive tree species and reduce carbon stocks, and the projected slight increase in the frequency of extreme storms by the end of the century could increase the risk of windthrow. Some possible measures to maintain goods and services of agricultural and forest ecosystems are mentioned, but it is suggested that more frequent extremes may have more severe consequences than progressive changes in means. In order to effectively decrease the risk for social and economic impacts, long-term adaptive strategies in agriculture and silviculture, investments for prevention and new insurance concepts seem necessary (Fuhrer et al. 2006, p. 79-80).
Ludwig and Asseng (2006) investigated effects of climate change on wheat production in Western Australia. They expressed that CO2 concentrations and temperature will probably increase and winter rainfall will decrease in south-west Australia. They implied that for achieving efficient agricultural adaptation, climate change impacts on agricultural production should be understood clearly. Thus, Ludwig and Asseng (2006) improved different future climate scenarios and investigated response of wheat yield and grain protein to changing climate. They assumed that temperature increases by 2,
4 and 6 °C and elevated CO2 between 525 and 700 ppm. They explored climate change influences by using simulation performed with the Agricultural Production Systems Simulator (APSIM-Nwheat). Ludwig and Asseng (2006) simulated fifty years of yield
89 and grain protein concentrations for three soil types at different locations spreading along north–south transect within the wheatbelt of south-west Australia. They stated that increase in temperature, elevated CO2 and rainfall changes conclude nonlinear and different effects for different parts of the region. They explained that higher CO2 concentration positively affected yield especially at drier sites. Increased temperatures had a positive effect for the cooler and wetter southern part of the region. They expressed that different soil types were affected distinctively by changing climate. For instance, they explained that heavier clay soils are most vulnerable to reduced rainfall, on the other hand, sandy soils were more sensitive to increased temperatures. Finally, they reported that elevated CO2 decreased protein concentration and reduced rainfall increased protein levels at all sites over the region. However, they expressed that increased temperature effect on protein concentration is uncertain, it may both increase or reduce protein concentrations.
Benhin (2006) investigated climate change impacts on South Africa agriculture. He discussed the impacts and adaptations in his study. He summarized his paper as follows: Statistical evidence suggests that South Africa has been getting hotter over the past four decades, with average yearly temperatures increasing by 0.13°C a decade between 1960 and 2003, with relatively higher levels for the fall, winter and summer periods. There has also been an increase in the number of warmer days and a decrease in the number of cooler days. Moreover, the average rainfall in the country is very low, estimated at 450 mm per year – well below the world’s average of 860mm per year – while evaporation is comparatively high. In addition, surface and underground water are very limited, with more than 50% of the available water resources being used for only 10% of the country’s agricultural activities. Climate change, which may make temperatures climb and reduce the rains and change their timing, may therefore put more pressure on the country’s scarce water resources, with implications for agriculture, employment and food security. Not only South Africa but also the
90 sub-region will be affected, given that more than half of the region’s staple, maize, is produced in South Africa.
This study attempts to assess the economic impact of the expected adverse changes in the climate on crop farming in the country. It estimates a revised Ricardian model for South Africa, using farm household crop farming data from selected districts in the nine provinces, long-term climate data, major soil types in the country, runoff in the districts, and adaptation related variables such as irrigation, livestock ownership, access to output markets and access to public and other extension services.
The analysis shows that climate change affects irrigated farms and dryland farms differently. Irrigated farms are cushioned against climate effects because they have alternatives to rain water. There are also some differences between the ways large- and small-scale farms are affected, but such differences are blurred by the influence of irrigation or dryland farming. The results also show that climate variables, especially for precipitation, have a non-linear relationship with crop net revenues in South Africa. Certain soil types, such as vertisols and xerosols, may be harmful to crop farming and therefore aggravate the harmful effects of climate change, while other types, such as acrisols and arenosols, may help reduce them. Runoff will also benefit crop farming, but when it is excessive it can be harmful.
In general, adaptations such as irrigation may help reduce the harmful effects of climate change, but if not properly implemented they may aggravate them. Of relevance here is public extension service, which was found to rather negatively affect crop net revenues, suggesting that the information provided by this service may not be very relevant to farmers, even though it can be an important tool for controlling the harmful effects of climate change if properly managed.
91 One significant finding is that there are seasonal differences in the climate effects, and these differences must not be overshadowed by looking only at the mean annual effects. Increased temperatures will be harmful in the summer farming season but beneficial in the winter one. The overall annual effects will therefore depend on the relative magnitudes of the positive and negative effects. This means that advantage should be taken of the positive effects, while controlling or reducing the negative effects. If this can be done, temperature changes should be beneficial rather than harmful to the country. Some of the adaptation strategies identified in the study could help achieve this. Changes in precipitation will also have different seasonal effects. Again there is the need to fine-tune policy to take advantage of the relative benefits. The analysis also shows that the effects of changes in both temperature and precipitation may be different for the different farming systems in the country – irrigated, dryland, large-scale and small-scale farms. The effects would also be different at the provincial levels. This finding is important for knowing how and where to direct the relevant policies for controlling the effects of climate change.
Using selected climate scenarios, the study also predicts that crop net revenues are expected to fall by as much as 90% by 2100 and that small-scale farmers will be the most severely affected. However, if proper adaptations are made these losses are expected to be reduced. Analysis of farmers’ perceptions of a change in the climate shows that most farmers across the country are aware of it. They cite increased temperature and reduced volume and altered timing of rain as signs of the change, and they are using various strategies to cope with it. Policy makers must be aware of these strategies, assess their effectiveness and find ways of improving them so as to limit their harmful effects and enhance the benefits that may be had from climate change. In general, climate change is expected to be harmful to crop farming in South Africa.
92 However, there are expected to be gains and losses specific to each farming system and each province. If policy makers and farmers are able to identify where the gains and losses are, and direct the appropriate policies and adaptation strategies to these areas, the expected overall negative effect may be reduced, and it is even possible that the agriculture sector in South Africa may reap benefits from climate change (Benhin 2006, p. 9-10).
Tao et al. (2006) explored the climate change impacts on field crops’ yield in China. They explained that warming trend started being significant in China since 1980s and they added that it will be more intensive in the future. They explained that despite the many studies on agricultural vulnerability to climate change, it is not possible to make a statement that crop production has been affected by observed climate change impacts. Tao et al. (2006) collected crop and climate data over China for the period of 1981-200 and they investigated if significant trends occurred in climate variables at different regions. Moreover, they examined if changes in climate variables affect on staple crop (rice, wheat, and maize) production. Tao et al. (2006) stated that they found significant warming trends for many regions over China. In addition, they revealed that significant warming resulted shifts in crop phenology and affected crop yield for two decades. Moreover, they expressed the spatial variability of effects over China and they stated that effects were not spatially uniform. Tao et al. (2006) implied that further investigations on coupled effects of temperature and CO2 concentration on physiological processes and mechanisms governing crop growth and production are necessary.
Isik and Devadoss (2006) developed a framework to investigate the potential impacts of climate change on agriculture. They underlined that in particular their study experiences an econometric model to predict “stochastic production functions and quantify the impacts of temperature and precipitation on the mean, variance, and covariance for wheat, barley, potato, and sugar beet yields in Idaho” (Isik & Devadoss 2006, p. 842-843). They stated that after estimating production functions, they used them to draw inferences for investigating climate change impacts on Idaho agriculture.
93 They implemented econometric model by using the historical climate and yield data. They stated that model results demonstrated that climate change impacts show difference across different crops. They explained that impacts of trend are positive on both the mean yield and variance of wheat, barley, potato, and sugar beet yields. Isik and Devadoss (2006) projected that climate change for Idaho (increase in both temperature and precipitation) will result increase in most crop yields studied in their study. Moreover, they expressed that climate change will plausibly affect the yield variability and covariance of crop yields substantially. Isik and Devadoss (2006) stated that due to changes in climate, the variance of yields will likely decrease for wheat, barley, and sugar beets, but the potato yield variability is projected to increase slightly. Moreover, they expressed that covariance of wheat and potato yields and covariance of barley and potato yields are projected to decrease remarkably. On the other hand, they reported marginal increase in the covariance of wheat and barley yields. Isik and Devadoss (2006) stated that changes in the variance and covariance of crops influence land allocations of farmers’ agricultural decisions on land allocation. They explained that farmers will high possibly increase the acreage of crops whose mean yield increases and/or variability decreases in response to the projected climate change. According to this statement, they expressed that while potato production decrease, production of wheat and barley is expected to increase in Idaho.
Alcamo et al. (2007) assessed the climate change impacts on food production in Russia. They implied that while previous studies focus was just changing climate influences on Russian agriculture and water resources, they also considered effect of changing frequency and spatial heterogeneity of extreme climate events. During their study, Alcamo et al. (2007) used combination of models including Global Assessment of Security (GLASS) model containing the Global Agro-Ecological Zones (GAEZ) crop production model and the Water-Global Assessment and Prognosis (WaterGAP 2) water resources model. They stated that under normal climate conditions, food production shortfalls (defined as a year which potential production of most crops below 50% of its average production) happened approximately 1-3 years each decade. However, Alcamo et al. (2007) expressed that this frequency may double at major agricultural regions in the 2020s and triple in the 2070s. They reported nearly 50 94 million people living in regions that experience one or more production shortfalls each decade. They warned that this number can increase to 82–139 million in the 2070s. Alcamo et al. (2007) announced water availability increase in Russia due to climate change. However, they stated that increasing number of intensive runoff events may be observed in much of central Russia. On the other hand, more frequent low runoff events in South Russia whereas already dry crop growing regions may be observed. Alcamo et al. (2007) expressed that increasing frequency of extreme climate events will be an important threat to Russia’s food production and water resources.
Naylor et al. (2007) stated that El Nino events have an impact on Indonesia’s main rice- growing regions by prolonging the hungry season and increasing the risk of annual rice deficits due to delayed rainfall and reduced rice planting. Naylor et al. (2009) purposed tom explore potential impacts of El Nino events and natural variability on Indonesian rice agriculture in 2050 under future climate conditions. They specifically focused on two rice cultivation area, Java and Bali. They selected a “30-day delay in monsoon onset as a threshold beyond which significant impact on the country’s rice economy is likely to occur” (Naylor et al. 2007, p. 7752). They used empirical downscaling model to obtain suitable future climate data set for their study. They found a remarkable increase in the probability of a 30-day delay in monsoon onset in 2050 in response to changes in the mean climate. Naylor et al. (2007) reported an increase in precipitation later in the crop year (April–June) by around 10%. On the other hand, Naylor et al. (2007) projected a significant decrease (up to 75% at the tail) in precipitation later in the dry season (July–September). In terms of above mentioned summary results, they recommended to develop and implement adaptation strategies including increased investments in water storage, drought-tolerant crops, crop diversification, and early warning systems.
Yao et al. (2007) evaluated climate change effects on irrigated rice yield in the main rice area of China. They explained the importance of rice for China by stating that rice cropping area correspond to 29.1% of the total crop cultivation area and more importantly, rice production accounts for 43.7% of total national grain production and
95 22.8% and 36.9% of the total world cropping area and production respectively. Yao et al. (2007) stated that rice ranks first among cereal crop production in China. In summary, they stated that rice production of China is very important for food safety both for China and the world. Moreover, Yao et al. (2007) warned about constraints including population growth, sharp decreases in cultivated land, lack of water resources, pollution and frequent natural disasters. They selected eight rice stations which are located in the Middle and South China to investigate climate change effects on rice production in China. They used B2 greenhouse gases emission scenario. Yao et al. (2007) used HadAM3H global climate model and they downscaled its outputs by using PRECIS regional climate model. They selected 1961-1990 as present climate and 2071-2100 as future climate. They expressed the spatial resolution of their data which is very important for regional impacts studies as 50 km. Yao et al. (2007) used present and future climate model in Crop Estimation through Resource and Environment Synthesis (CERES)-rice model in an effort of determining yield changes in response to climate change. Steps and findings of study were summarized by Yao et al. (2007) as follows: First, Crop Estimation through Resource and Environment Synthesis (CERES)- rice model is validated using farm experiment data in selected stations. The simulated results represent satisfactorily the trend of flowering duration and yields. The deviation of simulation within ±10% of observed flowering duration and ±15% of observed yield. Second, the errors of the outputs of RCM due to the difference of topography between station point and grid point is corrected. The corrected output of the RCM used for simulating rice flowering duration
and yield is more reliable than the not corrected. Without CO2 direct effect on crop, the results from the assessment explore that B2 climate change scenario would have a negative impact on rice yield at most rice stations and have little impacts at Fuzhou and Kunming. To find the change of inter-annual rice yield, a preliminary assessment is made based on comparative cumulative probability at low and high yield and the coefficient variable of yield between the B2
scenario and baseline. Without the CO2 direct effect on rice yield, the result indicates that frequency for low yield would increase and it reverses for high
96 yield, and the variance for rice yield would increase. It is concluded that high frequency at low yield and high variances of rice yield could pose a threat to
rice yield at most selected stations in the main rice areas of China. With the CO2 direct effect on rice yield, rice yield increase in all selected stations (Yao et al. 2007, p. 395-396).
De Silva et al. (2007) explored climate change impacts and paddy irrigation requirement in Sri Lanka. They expressed that paddy is a predominant crop in Sri Lanka. They stated that nearly 800 000 farmers are dependent on paddy and paddy rice agriculture corresponds to 30% of the Sri Lanka land area. In addition, they reported that total cultivation area of paddy rice (in wet season and dry season) exceeded 852 000 ha in 2002. They explained that around 72% of paddy production is done during the wet season in dry regions where water resources are already limited. They explained that 96% of water withdrawals is used for paddy rice agriculture in the drier regions. However, an increase in water demand was projected in response to population increase, expansion of irrigated agricultural land and industrialisation. De Silva et al. (2007) selected UK Hadley Centre for Climate Prediction and Research Model (HadCM3) to produce future climate projections for 2050s. They used A2 and B2 SRES scenarios. They developed a model based on CROPWAT model to simulate paddy irrigation requirement. They explained climate change effects on reference evapotranspiration and rainfall as follows:
Average rainfall during the wet season is predicted to decrease by 17% in the A2 scenario and by 9% in the B2 scenario, even though the average annual rainfall is predicted to increase by 14% (A2) and 5% (B2). The greatest reductions in wet season rainfall (16% in A2, 12% in B2) are predicted in Batticaloa. By contrast, wet season rainfall increases by 10% (A2) and 12% (B2) in 2050 in Hambantota. The average wet season temperature (the average of minimum and maximum air temperature) increases by 1.6 °C (A2) and 1.3 °C (B2) and the average reference evapotranspiration increases by 2% (A2) and 1% (B2). In Batticaloa, during the wet season, the average temperature increases
97 by 1.4 °C (A2) and 1°C (B2) and the average reference evapotranspiration increases by 1.3% (A2) and 1.1% (B2). The reductions in wet season rainfall combined with higher temperatures will lead to higher irrigation requirements (De Silva et al. 2007, p. 23).
Paddy irrigation requirements are predicted to increase on average across Sri Lanka by 23 and 13%, for the A2 and B2 scenarios, respectively. The highest proportional increase is predicted to occur in Batticaloa, by 45% (A2) and 15% (B2). Only around Hambantota, a comparatively dry area is the paddy irrigation requirement predicted to decrease, by 2% (A2) and 4% (B2). Even though there is a slight increase in rainfall in November, larger reductions in rainfall occur during January and February, particularly in the A2 scenario (+13, -77 and -65%, respectively). This earlier end to the seasonal rains appears to occur across most of Sri Lanka. The reduction in rainfall and the increase in potential evapotranspiration during January and February might have serious implication for paddy production, favouring earlier planting and/or shorter duration varieties (De Silva et al. 2007, p. 24).
Moreover, De Silva et al. (2007) discussed the adaptation strategies for Sri Lanka paddy rice agriculture. Firstly, they stated that despite the positive impact of climate change for Hambantota region regarding to paddy rice agriculture, most regions have been affected negatively by climate change in Sri Lanka. They expressed that precipitation in wet season will decrease. Especially, January and February will be influenced more due to earlier end of rains. De Silva et al. (2007) explained that decrease in rainfall will cause higher paddy irrigation requirement and it will result lower water availability which is increasing water stress. De Silva et al. (2007) suggested following adaptation strategies: increasing water use efficiency, water harvesting and/or reducing cropped areas. Moreover, they recommended farmers to take into account earlier planting and shorter duration varieties to avoid the impacts of less rainfall in January and February.
98 Luo et al. (2007) conducted a study on risk analysis of potential effects of climate change on South Australian wheat production. They examined a model based risk assessment on wheat production in eight locations over South Australia according to climate change scenarios by 2080. They investigated the vulnerability of wheat production based on climate change scenarios. They applied risk analysis to wheat production in terms of critical yield threshold. According to their study, Luo et al. (2007) reported increasing risk. They warned that drier locations such as Minnipa, Orroroo and Wanbi will experience intensive risk and they will have difficulty to reach economically viable wheat production.
Anwar et al. (2007) studied on influences of climate change on south-eastern Australia rain-fed wheat. They generated optimistic, average and pessimistic future daily climate projections by using the Australian Commonwealth Scientific and Industrial Research Organisation’s (CSIRO’s) global atmosphere models. Anwar et al. (2007) reported significant downward trend in wheat yield of south-eastern Australia based on three climate scenarios with and without elevated CO2 concentration. Anwar et al. (2007) revealed wheat yield reductions around 29% in terms of low, mid and high global warming scenarios. When elevated CO2 effect is taken into account, median wheat yield decrease was presented as 25% by Anwar et al. (2007). Elevated CO2 effect on wheat yield is positive, however it is only 4%. Finally, Anwar et al. (2007) concluded with importance of advances in agronomy and breeding to counteract changing climate impacts on wheat yield in south-eastern Australia.
Fischer et al. (2007) examined climate change impacts on irrigation requirement and influence of mitigation for the period of 1990-2080. They summarised their study as follows:
Potential changes in global and regional agricultural water demand for irrigation were investigated within a new socio-economic scenario, A2r, developed at the International Institute for Applied Systems Analysis (IIASA) with and without climate change, with and without mitigation of greenhouse
99 gas emissions. Water deficits of crops were developed with the Food and Agriculture Organization (FAO)–IIASA Agro-ecological Zone model, based on daily water balances at 0.5° latitude×0.5° longitude and then aggregated to regions and the globe. Future regional and global irrigation water requirements were computed as a function of both projected irrigated land and climate change and simulations were performed from 1990 to 2080. Future trends for extents of irrigated land, irrigation water use, and withdrawals were computed, with specific attention given to the implications of climate change mitigation. Renewable water-resource availability was estimated under current and future climate conditions. Results suggest that mitigation of climate change may have significant positive effects compared with unmitigated climate change. Specifically, mitigation reduced the impacts of climate change on agricultural water requirements by about 40%, or 125–160 billion m3 (Gm3) compared with unmitigated climate. Simple estimates of future changes in irrigation efficiency and water costs suggest that by 2080 mitigation may translate into annual cost reductions of about 10billion US$ (Fischer et al. 2007, p. 1083).
Zhang, Valentine and Kemp (2007) explored climate change effects on pasture production in the Orth Island of New Zealand. According to climate change scenarios, Zhang, Valentine and Kemp (2007) reported increase in temperature by 1–2°C and change in rainfall by -20 to +20. They stated that change in temperature and rainfall would lead significant changes in pasture production ranging from −46.2 to +51.9% relative to normal climate (1961-1990) values. Zhang, Valentine and Kemp (2007) explained that warming has a positive effect on pasture production in the south and southeast of the North Island. Moreover, they explained that central, south and southeast of the North Island gain from rainfall increases. On the other hand, Zhang, Valentine and Kemp (2007) expressed that north of North Island is affected negatively by rainfall increases. Finally, Zhang, Valentine and Kemp (2007) revealed that overall effect of increased temperatures and decreased rainfalls is decrease for pasture production except some central regions with high precipitation.
100 Yang et al. (2007) explained agricultural adaptation to warming in Northeast China. They stated that Northeast China consists of Heilongjiang, Jilin and Liaoning provinces. They revealed population of this region as 107 million, while its total area is 790 000 km2. They expressed that Northeast Chinas is located in high latitudes which makes the region one of the coolest areas in China. Thus, they expressed that agriculture suffers from cool weather. They announced that this region has long and cold winters. In addition, they explained that it has a short growth season and frequent cold extreme events. Nonetheless, Yang et al. (2007) reported significant warming by 1.0–2.5°C in annual mean temperature for Northeast China. They revealed that increases in temperature, growth period extensions had positive effects on agriculture of Northeast China. In addition to warming weather condition, they expressed that adaptation to warming involving crop composition and structure regulations and using more advanced technologies in agriculture led rapid increase in grain production in Northeast China. Yang et al. (2007) presented benefits of adaptation and changing conditions as follows:
Because of the adjustments of crop composition and structure, the enlargement of sown areas and the adoption of advanced technologies, the total grain output in Northeast China has increased substantially. The total crop sown area increased 8.6% during the period of 1982 to 2002, as the growing belts of rice and maize moved 200 300 km northward. Total grain output in
Northeast China increased by 155.7%∼ while for the whole of China it increased only by 29.3%. In Northeast China, rice yield per hectare grew by 18.1%, maize yield by 54.7%, and wheat yield by 143.0%. Per capita annual grain possession increased by 73.6%. Nowadays, the outputs of maize and soybean in Northeast China account for 30% and 40% of the nation’s totals respectively (Bi & Li 2004) (Yang et al. 2007, p. 53).
Alongside the development of agriculture and other beneficial policies, per capita annual net income of rural households greatly increased. Per capita annual net income of rural households increased by 222.6% between 1987 and
101 2002, from 510.5 Yuan (about *$96.3, in 1987) to 2517.0 Yuan (about *$310.7, in 2002). The poverty ratio in Northeast China decreased as farmers’ living situations greatly improved. The number of rural telephone subscribers rapidly increased, for example, in Heilongjiang Province, only 0.7% rural households had telephones in 1985, increasing to 37.4% in 2002 (The Statistics Bureau of Heilongjiang Province 1992–2003). Paved roads now reach most villages, educational conditions are much better than before, and hospitals and clinics are available in towns or villages. Every township has an extension center for agricultural technologies, providing timely service for farmers (Yang et al. 2007, p. 53).
Fleischer, Lichtman and Mendelsohn (2008) investigated economic effects of climate change on agriculture of Israel by using Ricardian Model. An economic survey at Israel farms was done to accomplish analysis of this study. Then, net annual income relation with climate and other control variables was investigated. Authors performed analysis with and without irrigation water quotas. Fleischer, Lichtman and Mendelsohn (2008) stated that Ricardian model analysis without irrigation water for regions where irrigated agriculture is being done, tend to overestimate benefits and losses of warming. Fleischer, Lichtman and Mendelsohn (2008) expressed that conducting an agricultural warming impact study by not considering climate change impacts on water availability is not realistic for Israel conditions. They reported significant benefit of climate change on agriculture of Israel based on models with irrigation water omitted. On the other hand, when water included, they reported benefits of climate change under mild climate change and agricultural losses was modelled under long term severe climate change. They also explained that irrigation technology is very important for agriculture in Israel and they showed that regions where temperatures are above 20°, net income is higher per hectare due to efficient irrigation systems. Fleischer, Lichtman and Mendelsohn (2008) explained that agriculture of many countries were affected by climate change and intensity of impact level depends on level of development, technological advancement, and the institutional setting in the countries.
102 Kucharik and Serbin (2008) made a research on recent climate change influence on Wisconsin corn and soybean yield trends. They explained that due to Wisconsin’s location (northern), it has a cool climate system so warming owing to climate change may be beneficial for corn and soybean yields. They indicated that precipitation and temperature trends during growing season over 1976-2006 defined 40% and 35% of county corn and soybean yield trends, respectively. They reported yield increases for both corn and soybeans for the counties where cooler and wetter summer trends were observed. According to their study, Kucharik and Serbin (2008) expressed that each °C warming for summer months would potentially cause decrease by 13% and 16% for corn and soybeans respectively. Moreover, they explained that modest increases in total precipitation (50 mm) would increase yields by 5–10%. Kucharik and Serbin (2008) concluded that warmer climate would be beneficial in spring and autumn seasons.
Thomas (2008) investigated agricultural irrigation demand under changing climate in China. Firstly, he implied the importance of crop production in China for global food supply. He stated that more than 30% of agricultural land in China is irrigated and major part of agricultural production has been provided by this area. Thomas (2008) coupled soil data with high resolution monthly grided climate data including temperature, precipitation and potential evapotranspiration in an effort to model climate change impacts on irrigation requirements for crop production. Thomas (2008) combined observed long term trends and interannual variations for the year 2030 climate change scenarios. He generated average conditions, best case and worst case scenarios. He used FAO water balance model with a view to compute irrigation amounts to reach maximum yields for the period 1951–1990 and the climate scenarios. As a result, he reported remarkable spatial and temporal variations in irrigation demand for the period of 1951-1990. In addition, he revealed increase in irrigation water demand according to results of future simulations.
Malla (2008) examined climate change impacts on Nepal agriculture by using a crop simulation model (DSSAT). He stated that 1.8°C increase in temperature with an average increasing rate of 0.06°C/year between 1975 and 2006. He explained that 103 frequent drought, severe floods, landslides and combination of these effects have been causing problems for agricultural production of Nepal. In addition, he expressed that climate change has contribution on above mentioned factors. He reported that
CO2 enrichment technology at Khumaltar resulted increase in rice and wheat yields by
26.6% and 18.4% owing to double CO2 and yield increase by 17.1% and 8.6% because of temperature increases, respectively. He stated that changing climate and CO2 concentrations have positive effects on rice and wheat yields over Nepal, however, negative effects on maize yield was obtained specifically in Terai.
Almaraz et al. (2008) investigated climate variability, corn yield and climate change relation at a higher latitude in Southwestern Quebec. They stated that nations at higher latitudes such as Canada will be more affected by climate change. Almaraz et al. (2008) explored climate variability including temperature & precipitation changes and corn yield association for the period of 33 years at the Monteregie region of south- western Quebec. Almaraz et al. (2008) used historical yield and climate observation data with statistical models. They expressed increase in growing season temperature owing to increased September temperature. They explained that precipitation did not demonstrate clear trend over study period. Almaraz et al. (2008) explained yield increases by 118 kg ha-1 year-1 from 1973 to 2005 under normal climate conditions. They explained yield increase mainly according as technological changes such as genetics and management. They reported strong relation between July temperature & May precipitation and corn yield variability. Almaraz et al. (2008) expressed that these two variables define more than a half of yield variability relation with weather. Almaraz et al. (2008) explained that July temperature less than normal July temperature and May precipitation over normal values have negative influences on corn yield. Moreover, they stated that growing seasons showed warming mainly because of September temperature increases.
Seo and Mendelsohn (2008) examined the adaptation of South American farmers to climate change by changing crops. They developed a “multinomial logit” model to evaluate crop choices of farmers. Seo and Mendelsohn (2008) applied the model
104 across 949 farmers in seven countries. They reported that changing climate has an effect on South American farmers’ crop selection because temperature and precipitation influence their crop selection. For instance, farmers’ choice in warmer locations is fruits and vegetables. On the other hand, Seo and Mendelsohn (2008) revealed that farmers choose wheat and potatoes in cooler locations. Rice, fruits, potatoes, and squash are produced in wetter locations while maize and wheat are the main crops in drier locations. Seo and Mendelsohn (2008) stated based on their study that due to global warming, farmers in South America will convert maize, wheat, and potatoes to squash, fruits and vegetables. Finally, they explained that climate change impact studies on net revenue must consider not only changes in yields per crop but also crop switching.
Ortiz et al. (2008) did a research on climate change impact and wheat production relation at worldwide. They summarized their study in following paragraphs:
Climate change could strongly affect the wheat crop that accounts for 21% of food and 200 million hectares of farmland worldwide. This article reviews some of the approaches for addressing the expected effects that climate change may likely inflict on wheat in some of the most important wheat growing areas, namely germplasm adaptation, system management, and mitigation. Future climate scenarios suggest that global warming may be beneficial for the wheat crop in some regions, but could reduce productivity in zones where optimal temperatures already exist. For example, by 2050, as a result of possible climate shifts in the Indo-Gangetic Plains (IGPs) – currently part of the favorable, high potential, irrigated, low rainfall mega-environment, which accounts for 15% of global wheat production – as much as 51% of its area might be reclassified as a heat-stressed, irrigated, short-season production mega-environment. This shift would also represent a significant reduction in wheat yields, unless appropriate cultivars and crop management practices were offered to and adopted by South Asian farmers. Under the same climate
105 scenarios, the area covered by the cool, temperate wheat mega-environment could expand as far as 65°N in both North America and Eurasia.
To adapt and mitigate the climate change effects on wheat supplies for the poor, germplasm scientists and agronomists are developing heat-tolerant wheat germplasm, as well as cultivars better adapted to conservation agriculture. Encouraging results include identifying sources of alleles for heat tolerance and their introgression into breeding populations through conventional methods and biotechnology. Likewise, agronomists and extension
agents are aiming to cut CO2 emissions by reducing tillage and the burning of crop residues. Mitigation research promises to reduce emissions of nitrous oxide by using infrared sensors and the normalized differential vegetative index (NDVI) that determines the right times and correct amounts of fertilizer to apply. Wheat geneticists and physiologists are also assessing wild relatives of wheat as potential sources of genes with inhibitory effects on soil nitrification. Through the existing global and regional research-for-development networks featuring wheat, technology and knowledge can flow to allow farmers to face the risks associated with climate change (Ortiz et al. 2008, p. 46).
Haim, Shachter and Berliner (2008) investigated impacts of climate change on economic aspects of agricultural production in Israel. They applied “production function” approach to two crops of Israel, wheat which is the major crop in Israel’s dry southern region and cotton which is grown in more humid climate in northern Israel. They explained that these two crops cover approximately 35% of all field crops grown in Israel. They expressed the advantages of production function as direct relation to scientific experiment and capability of linking climate, crop yield, and market equilibrium conditions. Haim, Shachter and Berliner (2008) adjusted HadCM3 global climate model outcomes to the research areas and they produced climate projection over the period of 2070-2100 based on two climate scenarios. As explained in some parts of thesis, due to coarse spatial resolution of global climate models, outputs of global climate models should be downscaled. Haim, Shachter and Berliner (2008) used the LARS-WG (Long Ashton Research Station-Weather Generator) weather generator 106 for the purpose of generating future climate projections with suitable spatial resolutions for regional impact study in Israel. They generated temperature and precipitation projections by using above methodology. According to pessimistic climate scenario, net wheat revenues were affected by climate change intensively by -145 to - 273%. On the other hand, optimistic climate scenario gave moderate effects on wheat net revenue changing from -43 to 35%. Haim, Shachter and Berliner (2008) implied the importance of rain events distribution on wheat yields. On the other hand, very significant yield reductions causing substantial economical losses ranging from -240 to -173% were obtained for cotton crop. They finally stated that using additional irrigation and nitrogen may help to decrease farming losses.
Brassard and Singh (2008) studied effects of climate change and changing ambient carbon dioxide levels on crop yields in Quebec, Canada. Summary of methodology and findings are summarized by them as follows:
The methodology involves coupling the transient diagnostics of two Atmosphere-Ocean General Circulation Models, namely the Canadian CGCM1 and the British HadCM3, to the Decision Support System for Agrotechnology Transfer (DSSAT) 3.5 crop models to simulate current (1961–1990) and future (2040–2069) crop yields and changes. This is done for four different crop species, namely spring wheat, maize, soybean, and potato, and for seven agricultural regions of Southern Quebec. The results of this study focus on the
main causative factors influencing crop yields, namely the direct CO2 fertilization effect, the influence of the increase in growing season temperature, including optimal thermal conditions and acceleration in crop maturation, soil moisture availability, as influenced by precipitation and evapotranspiration, and nitrogen uptake by crops. Our results show that crop yield changes may vary according to climate scenario, crop species, and agricultural region. Consistent with other similar research, it would seem that these multiple causative factors very often seem to cancel each other out and
107 dilute the impacts of climate change on crop yields (Brassard & Singh 2008, p. 241).
Lobell and Burke (2008) focused on uncertainty of climate change-agricultural- impact studies. They stated that climate change effect prediction studies include large amount of uncertainties induced by ignorance of many physical, biological, and socio-economic processes. They explained that it is very important to reduce these uncertainties to get more realistic future estimation. They expressed that reducing uncertainties of climate change impact studies are dependent on better understanding of the relative contributions of individual factors. Lobell and Burke (2008) studied on projections of climate change impacts on crop production for 94 crop–region combinations. In particular, they “focused on the relative contributions of four factors: climate model projections of future temperature and precipitation, and the sensitivities of crops to temperature and precipitation changes” (Lobell & Burke 2008, p. 1). Despite the vital significance of precipitation for crop yields, they unexpectedly found that “uncertainties related to temperature represented a greater contribution to climate change impact uncertainty than those related to precipitation for most crops and regions, and in particular the sensitivity of crop yields to temperature was a critical source of uncertainty” (Lobell & Burke 2008, p. 1). Finally, they recommended developing studies with the aim of better understanding of crop responses to temperature and the magnitude of regional temperature changes for successful adaptation efforts of agriculture to climate change.
Mizyed (2009) studied on potential impacts of climate change on water resources and agricultural water demand in the West Bank where is under Mediterranean climate. It is a fact that Mediterranean climate is very sensitive to climate change and Mizyed (2009) explained that 2 to 6°C temperature increase is projected for this region. Moreover, decrease in precipitation up to 16% is predicted for Mediterranean Basin. Mizyed (2009) selected West Bank as a part of Mediterranean Basin for the purpose of investigating climate change impacts on water resources and agricultural water demand in this region. He used different future climate scenarios because of
108 uncertainty in projecting future climate variables including temperature and precipitation. He used three temperature change scenarios with an increase of 2, 4 and 6°C. In addition, he applied two precipitation change scenarios. In first scenario he assumed no change in precipitation while he assumed 16% precipitation decrease in the second one. According to above scenarios, Mizyed (2009) calculated monthly evapotranspiration and monthly precipitation excess depths at seven weather stations distributed over the different climatic and geographical areas of the West Bank. Mizyed (2009) claimed that temperature increases may increase agricultural water demand up to 17%. In addition, it may cause decrease in groundwater recharge by up to 21% relative to current levels. Mizyed (2009) stated that basin is more sensitive to precipitation change and he reported that according to 16% precipitation decrease scenario, annual groundwater recharge in the West Bank may decrease by about 30% of current level. Finally, he concluded that while 16% precipitation decrease coupled with 6°C temperature increase scenario, it results reduction in groundwater recharge up to 50%.
Mendelsohn (2009) reviewed several current studies on relationship between climate change and agriculture in developing countries. He explained that studies are generally in agreement on a hypothesis that climate sensitivity of tropical and subtropical agriculture is larger than temperate agriculture. He explained that even marginal and mild warmings result yield losses in Africa and Latin America. He expressed that despite rainfall is an advantage for semi-arid locations, precipitation increase in wet places may cause harms for agriculture. He underlined that intensive warming may conclude significant problems for agriculture of low latitude countries. On the other hand, Mendelsohn (2009) stated that mild warming will result modest damages even some benefits. He expressed that the severity of damage is highly dependent on climate change scenario. Moreover, he stated no simple rule between climate change impact level and farm size. However, he added that small farmers may be less impacted by climate change influences than commercial farmers in developing countries. Mendelsohn (2009) implied the importance of irrigation on agricultural sensitivity to warming. He stated that irrigated agriculture will be economically less
109 affected by climate change than rain-fed agriculture. Even irrigated agriculture will gain due to climate change at some locations. He showed irrigated farms in Africa and China as an example to less vulnerability of irrigated agriculture. However, he reminded that climate change has negative impacts on water availability for many parts of the world. So, despite irrigated agriculture is less vulnerable than rain-fed agriculture, reduction in water availability can cause significant problems for irrigated agriculture at locations where available water will decrease due to climate change.
Mo et al. (2009) assessed the regional crop yield, water consumption and water use efficiency in the North China plain and they investigated potential climate change influences on above parameters. They defined the North China as one of the most significant agricultural land for food production in China. They underlined that agricultural system of this regions is very sensitive to climate change due to its vulnerable structure to water stress. Mo et al. (2009) implemented the Vegetation Interface Processes (VIP) model in an effort to assess crop yield, water consumption (ET) and water use efficiency (WUE) for winter wheat and summer maize crops. In addition, Mo et al. (2009) investigated responses to climate change which is projected by HadCM3 global climate model under A2 and B1 emission scenarios. They expressed that the results indicated a rapid enhancement of crop yield during the last 56 years with slight increase of ET and noticeable improvement of WUE. They reported spatial variances in crop yield due to soil quality and irrigation facilities. They explained the model findings under future climate conditions as follow:
By fixing Vcmax to the value for the current cultivars, the crop yield, ET and WUE under SRES A2 and B1 scenarios are predicted. Compared with the yield of 1990s, it is found that the predicted yield of winter wheat is enhanced under both A2 and B1 scenarios. However, mostly the increment is higher than B1. The maximum increment is 19% under A2 occurring in 2070s and 13% under B1 in 2060s, the former is noticeably larger than the later. The results show that as
winter wheat is a C3 crop, it will benefit more from CO2 enrichment. However, the grain enhancement is also affected by both precipitation and temperature patterns, which make the increments fluctuate in different decades. 110 Different from the variation of yield, cumulative ET in the growing stage of winter wheat seems to be affected only slightly by climate change. The cumulative ET amounts gently increase for both A2 and B1 scenarios, which is less than 6%. As it is known, the air warming will intensify evapotranspiration, whereas both lower stomatal conductances resulted from higher CO2 concentration and growing period shortened by warming will mitigate the rising of total ET amount. As a consequence, the change of ET is not remarkable. For summer maize, the yield is reduced gradually with air warming under both A2 and B1 scenarios from 2020s to 2090s. In 2090s, yield will fall by
15% for A2 and 12% for B1 scenario respectively. The effect of CO2 enrichment on C4 maize is weak, which is predicted nearly 10% by the model with double air CO2 concentration under current climate condition. The thermal warming effect on grain yield is significantly larger than that of CO2 fertilization in maize, leading to a net reduction of grain yield under both scenarios.
By sensitivity analysis, it is found that the yield reduction is mainly caused by shortened growing period and higher maintenance respiration. Different from the ET in winter wheat, cumulative ET in maize growing period is significantly increased over 10% since 2050s. At the end of 21st century, the cumulative ET amounts under A2 and B1 scenarios will respectively be 37% and 20% higher than the current values over the maize growing period. Because the percentage of wheat yield enhancement is larger than that of ET decrease, the WUE is slightly improved by 10% and 7% under A2 and B1 scenarios, respectively. The WUE values for maize are reduced more than 25% under both A2 and B1 scenarios in 2090s, resulted from decreased yield and noticeably increased ET. The negative response of maize to climate change implies that new maize cultivars should be bred with higher heat tolerance and larger growing degree– days to mitigate the warming effects. There are at least three reasons to be able to explain why maize yield will decrease under climate change condition even though rainfall will increase 16–48% and CO2 increase from 280 and 450
111 ppm across the two emission scenarios considered in 2090. Firstly, for summer maize as a C4 crop, because of its inner structure, it is mostly sensitive to temperature about 2.8–4.5 °C. Therefore, summer maize gets yield loss with
the increase of temperature. The effect of CO2 enrichment on C4 maize is small and weak. The thermal warming effect on grain yield is significantly larger than
that of CO2 fertilization in maize, leading to net reduction of grain yield under both scenarios.
Secondly, the growth season of summer maize is matched with the wet season in the area. Precipitation in the normal year can almost satisfy the water demand of maize. Therefore even though there is an expected increase of precipitation, the maize yield responds a little. Thirdly, the variation of variables
like temperature, precipitation, CO2 concentration as shown in the study is not a gradual change. It is possible that our conclusion is just true only for the amount of the change we assigned. More study should focus on the continuous response with the driving variables changing gradually. For winter wheat, as it
is a C3 crop, it is very sensitive to the CO2 concentration. Besides, even in normal years, precipitation cannot satisfy winter wheat’s water needs. The increase of precipitation will help the yield production. In this way, under the climate change scenarios of A2 and B1, wheat gets yield gain (Mo et al. 2009, p.67).
Iglesias et al. (2009) investigated impacts of climate change on agriculture in Europe under the project named Projection of Economic impacts of climate change in Sectors of the European Union based on boTtom-up Analysis (PESETA). They used dynamic process-based crop growth models for major agro-climatic regions of Europe. Iglesias et al. (2009) explained the major variables regarding to simulated yield variance as crop water (combination of precipitation and irrigation) and temperature during the growing season. Iglesias et al. (2009) defined nine agro-climatic regions according to K- mean cluster analysis of temperature and precipitation data from 247 meteorological stations, district crop yield data, and irrigation data. They implemented the yield
112 functions which were derived from validated crop model for the purpose of better understanding of crop production vulnerability to climate change. They used three climate change scenarios produced by “prudence HIRHAM RCM nested in the HadCM3 GCM under the A2 and B2 forcing and from the Rossby Centre RC4 nested in the ECHAM4 GCM under the A2 scenario” (Iglesias et al. 2009, p.5). They defined optimal yield in their study as the potential yield given non-limiting water applications, fertilizer inputs, and management constraints.
They calculated adapted yields for each country as a fraction of potential yield. Mentioned fraction equals to ratio of current yields to current yield potential. They performed crop yield projections for the period of 2071-2100 under the HadCM3/HIRHAM A2 and B2 scenarios and for the period of 2011-2040 under ECHAM4/RCA3 A2 scenario. Iglesias et al. (2009) expressed that despite the fact that each scenario generates different results, all scenarios are in agreement on spatial distribution of influences. They projected increases in crop suitability and productivity in Northern Europe as a result of lengthened growing season, decreasing cold effects on growth, and extension of the frost-free period. On the other hand, they predicted decreases in crop productivity of Southern Europe due to shortening of the growing period, with subsequent negative effects on grain filling. They underlined that models performed in this study did not consider climate change impacts on water availability which is dominant factor for irrigation and restrictions in the application of nitrogen fertilizer. Thus, they expressed that results should be assessed optimistic in terms of production point of view and pessimistic in terms of environmental point of view.
Park et al. (2009) examined climate change influences on the temporal variation of paddy rice irrigation reservoir water level based on future watershed inflow. They tried to improve efficient adaptation approaches for the future reservoir water level management to provide stable water supply with the aim of meeting paddy irrigation demands. Park et al. (2009) studied on 366.5 km2 watershed which involve two irrigation reservoirs located in the upper middle part of South Korea. They calibrated SLURP model by using 9 years daily reservoir water level and streamflow observation data at the watershed outlet. They found Nash-Sutcliffe model efficiencies for 113 calibration and validation as 0.69 and 0.65, respectively. They constituted future climate projections by using NIES MIROC3.2 hires data under A1B and B1 greenhouse gases emission scenarios. Due to overcome coarse spatial resolution problem, they implemented change factor statistical method through bias-correction using 30 years past weather data. Park et al. (2009) reported decreases for future reservoir storages of autumn and winter season after completion of irrigation period by 2080s under A1B scenario. They explained that due to reductions in summer and autumn season inflows, it is important to develop adaptation policies. They stated that reservoir operation should be more conservative. They added that with the aim of securing the reservoir water level in autumn and winter season, August and September water releases should be done carefully and it should be decreased where it is a requirement.
Lippert, Krimly and Aurbacher (2009) conducted a study with the aim of investigation influences of climate change on German agriculture. They used “Ricardian analysis accounting for spatial autocorrelation and relying on recent climate change forecasts at a low spatial scale” (Lippert, Krimly & Aurbacher 2009, p. 593). They implemented Ricardian analysis by using data from the 1999 agricultural census and data from the weather observation stations of Germany. Lippert, Krimly and Aurbacher (2009) summarized their methodology and findings as follows:
The cross-sectional analysis yields an increase of land rent along with both a rising mean temperature and a declining spring precipitation, except for in the Eastern part of the country. The subsequent simulation of local land rent changes under three different IPCC scenarios is done by entering into the estimated regression equations spatially processed data averages for the period between 2011 and 2040 from the regional climate model REMO. The resulting expected benefits arising from climate change are represented in maps containing the 439 German districts; the calculated overall rent increase corresponds to approximately 5–6% of net German agricultural income. However, in the long run, when temperature and precipitation changes will be
114 more severe than those simulated for 2011–2040, income losses for German agriculture cannot be excluded (Lippert, Krimly & Aurbacher 2009, p. 593).
Tingem and Rivington (2009) conducted a research on adaptation for crop agriculture to climate change in Cameroon. They expressed that agriculture of Cameroon is vulnerable to climate change and further food security problems are expected due to climate change. They stated that implementing agricultural adaptation policies to climate change may decrease impacts of climate change on agriculture in Cameroon. Tingem and Rivington (2009) modelled current and future (2020, 2080) yields of key crops including bambara nut, groundnut, maize, sorghum and soybean in Cameroon’s eight agricultural regions by using cropping system simulation model (CropSyst). They provided future climate data to model from two general circulation models, the NASA/Goddard Institute GISS and the British HadCM3.
Based on their models, they resulted substantial future yield increases bambara groundnut, soybean and groundnut. On the other hand, they projected very slight decrease and no change for future yields of maize and sorghum yields. Moreover, Tingem and Rivington (2009) focused on adaptation strategies for maize, sorghum and bambara groundnut under GISS A2 and B2 scenarios. They explained that changing sowing dates may be not successful adaptation strategy due to narrow rainfall band which is major factor to decide the timing of farm operations in Cameroon. However, they implied the possibility of developing later maturing new cultivars which may be very effective to offset negative impacts. In addition, Tingem and Rivington (2009) underlined that highest increase in productivity under different scenario projections without management changes can be obtained by this way. They supported their ideas by modelling results as follows: 14.6% reduction in maize yield according to GISS A2 by 2080 was transformed to 32.1% increase. In addition, same climate scenario showed that 39.9% reduction in sorghum yield was converted to a 17.6% increase by applying above adaptation strategy. Moreover, future yields of bambara groundnut were approximately trebled because of increase in growing period and possible positive
115 influences of higher CO2 concentrations under implementing later maturing new cultivars adaptation.
Ludwig, Milroy and Asseng (2009) investigated relation between recent climate change and wheat production in Western Australia. Ludwig, Milroy and Asseng (2009) stated that whet production region in Western Australia has characteristics of a distinct Mediterranean climate with most of the rainfall occurring in the winter months. Rainfall is the dominant factor which is putting limitation on wheat production in this area. Ludwig, Milroy and Asseng (2009) expressed another significant problem in this region as dryland salinity because of native vegetation clearance. They reported substantial reductions in winter rainfall since mid 1970s for study region. They revealed that according to nine sites, a reduction by an average of 11% was observed in growing season rainfall (May to October) in study area. They added that total rainfall in June and July decreased by 20%. Ludwig, Milroy and Asseng (2009) implemented the ASPIM-Nwheat model which is driven by historic climate data in an effort to comprehend recent climate change influence on hydrology and wheat production. They purposed to compare cases before and after 1975. Although significant rainfall reductions were reported for the area, wheat yields did not decline based on simulations performed by actual weather data. Moreover, they stated decrease by up to 95% in drainage which decrease dryland salinity spread. Ludwig, Milroy and Asseng (2009) explained that main reason of stability of wheat yields despite winter rainfall decrease is the seasonal characteristic of winter and crop demand in their study region. In winter season, rainfall generally more than crop demand, so decreases in precipitation in winter season does not affect wheat yield negatively.
Chavas et al. (2009) conducted a study on productivity of 5 main crops including canola, corn, potato, rice, and winter wheat in Eastern China under changing climate. Moreover, they defined vulnerable and emergent regions those have a greater than 10% decrease and increase in productivity, respectively. They used ICTP RegCM3 regional climate model data to accomplish modelling studies. They selected baseline period as 1961-1990 and future period was 2071-2100. IPCC A2 greenhouse emission scenario was used by them. Chavas et al. (2009) used RegCM3 outputs as an input to 116 the EPIC agro-ecosystem simulation model in the domain [30°N,108°E] to [42°N,
123°E]. They achieved their simulations both with and without the enhanced CO2- fertilization effect. Chavas et al. (2009) reported potential productivity increases by 6.5% for rice, 8.3% for canola, 18.6% for corn, 22.9% for potato and 24.9% for winter wheat under with enhanced CO2-fertilization effect. On the other hand, without enhanced CO2-fertilization effect, reductions in potential productivity ranging from 2.5 to 12% were projected. They also stated that climate variables are more significant factors on productivity than soil properties except in the case of potato production. The influences of wind erosion are more significant for potato production in the northwest. Finally, Chavas et al. (2009) concluded that “corn and winter wheat benefit significantly in the North China Plain, rice remains dominant in the southeast and emerges in the northeast, potato and corn yields become viable in the northwest, and potato yields suffer in the southwest with no other crop emerging as a clear beneficiary from among those simulated in this study” (Chavas et al. 2009, p. 1118).
Mozny et al. (2009) investigated climate change influences on the yield and quality of Saaz hops in Czech Republic. They simulated climate change impacts on Saaz hops by using crop model named CORAC. Mozny et al. (2009) validated CORAC model by using 52 year historical data. They stated input requirements of CORAC model as minimum and maximum air temperatures, relative air humidity, precipitation and solar radiation. Based on performed future simulations, Mozny et al. (2009) reported reductions in yield up to 7-10% and decrease in a-acid content which is a main variable to determine quality by 13–32%.
Xiong et al. (2010) explored the water availability for agriculture in China in the 2020s and 2040s by applying regional climate model, PRECIS, outputs to drive crop and water simulation models. They purposed to have better understanding on impacts of different factors on future water availability to support China’s food production. They used Variable Infiltration Capacity (VIC) model for hydrological simulations. Hydrological simulations concluded increases in total water availability in parallel to projected increase in precipitation. Xiong et al. (2010) presented their findings as follows: 117 Hydrological simulations driven by PRECIS climate scenarios suggest increases of around 20% in total water availability for the 2020s, and 18% for the 2040s. The two climate change scenarios produce different effects on cereal irrigation demand, with significant increases under A2 and slight increases under B2. The effects of climate change on water availability for agriculture are small compared to the effects of socio-economic development on demand, and have different directions between A2 and B2. More pessimistic impacts occur with A2 climate and socioeconomic development: as climate change increases the shortfall in planned and feasible expansion in irrigation area and exacerbates water stress for agriculture. In contrast, with B2 conditions climate change increases the water reliability for agriculture and decreases the shortfall in irrigation. However, these impacts may be underestimated because PRECIS projects a wetter climate in the future compared to a multi-model GCM average for China. Future cereal irrigation demand was found to be highly sensitive to the characteristics of daily precipitation. The two climate scenarios produced similar changes (18-20%) in TWA for the 2020s and 2040s, however, they produced significant differences in simulated irrigation demand due to differences in frequency and intensity of daily precipitation events.
Overall, the results suggest that there will be insufficient water for agriculture in China in the coming decades, due to increases in water demand for non- agricultural uses, especially with A2 scenario. This shortfall is greater in southern China, because of a significant decline in the share of water for agriculture and intensive planting of paddy rice. Although a policy option to increase the area of irrigated cereal is assumed under both development pathways in order to maintain food security (consistent with current national policy), the simulated irrigated cereal area decreases, particularly for rice, if the present irrigation efficiency is maintained. The study identified significant spatial differences in impacts at the river basin and provincial level. In broad terms water availability for agriculture declines in southern China and remains 118 stable in northern China. These patterns go against the present situation and are likely to have significant implications for adaptation strategies or policies for agricultural production and water management (Xiong et al. 2010, p. 68).
Stockle et al. (2010) investigated potential impacts of climate change and the atmospheric CO2 increase on eastern Washington State agriculture. They generated future climate projections by using four general circulation model. They focused on apple, potato and wheat crops which have the larger economic value for the region. Stockle et al. (2010) stated that apples and potatoes are irrigated and they focused on only at dryland wheat, which is the dominant dryland crop. For the purpose of crop performance evaluation, a cropping system simulation model (CropSyst) was used with historical and future climate sequences. CropSyst is “a cropping system simulation model that represents the response to weather and management of an array of annual and perennial crops and tree fruit crops” (Stockle et al. 2010, p. 5). It is assumed that crops receive adequate water (irrigated crops), nutrients, and control of weeds, pests and diseases. Stockle et al. (2010) explained that climate change impacts on eastern Washington State agriculture will be mild based on short term (around 20 years) projections. Nonetheless, intensity of impact will increase in time and yield losses may reach 25% for some crops by the end of the century. On the other hand, they revealed that increasing CO2 concentration will provide benefit for some crops. They concluded that if effective adaptation policies can be applied for the region, yield increases can be obtained by the help of positive effect of increasing CO2 concentrations. It should be noted that one of the most important drawback of this study is an assumption of sufficient water for irrigated crops. However, some other studies showed that climate change may result decrease in available water for agriculture.
Al-Bakri et al. (2010) investigated climate change effects on rain-fed agriculture in a semi-arid basin in Jordan. They explained their study and outcomes of their study as follows:
119 Rainfed agriculture in Jordan is one of the most vulnerable sectors to climate change, as the available water and land resources are limited and most of the country’s land is arid. In this study, a crop simulation model (DSSAT) was used to assess the impact of different climate change scenarios on rainfed wheat and barley in the Yarmouk basin in Jordan. Analysis of observed crop data showed differences between cultivated and harvested areas for both crops in the study area with variations among years. Results from DSSAT model for years showed that it was able to capture the trend of yield over the years realistically well. The model predicted an average yield of wheat of 1176 kgha-1, which was close to the average (1173 kg ha-1) obtained from the data of department of statistics (DOS), and an average predicted yield of barley was 927 kg ha-1 while the DOS average was 922 kg ha-1, with higher RMSE for barley (476 kg ha-1) than for wheat (319 kg ha-1 ). Results for predicting future yield of both crops showed that the responses of wheat and barley were different under different climate change scenarios. The reduction of rainfall by 10–20% reduced the expected yield by 4–8% for barley and 10–20% for wheat, respectively. The increase in rainfall by 10–20% increased the expected yield by 3–5% for barley and 9–18% for wheat, respectively. The increase of air temperature by 1, 2, 3 and 4 °C resulted in deviation from expected yield by - 14%, -28%, - 38% and -46% for barley and -17%, +4%, +43% and +113% for wheat, respectively. These results indicated that barley would be more negatively affected by the climate change scenarios and therefore adaptation plans should prioritize the arid areas cultivated with this crop (Al-Bakri et al. 2010, p.1).
Prato et al. (2010) assessed potential climate change impacts on enterprise net returns and annual net farm income (NFI) for Flathead Valley in Northwest Montana. They evaluated crop enterprise net returns and NFI by using historical climate and future climate periods. They selected period of 1960-2005 as historical climate and they used 2006-2050 period as future climate. They produced future climate projections by using A1B, B1, and A2 greenhouse gases emission scenarios. Prato et al. (2010)
120 presented steps of study as ”(1) specifying crop enterprises and agricultural production systems (APSs) (i.e., combinations of crop enterprises) in consultation with locals producers; (2) simulating crop yields for two soils, crop prices, crop enterprises costs, and NFIs for APSs; (3) determining the dominant APS in the historical and future climate periods in terms of NFI; and (4) determining whether NFI for the dominant APS in the historical climate period is superior to NFI for the dominant APS in the future climate period” (Prato et al. 2010, p.577). They modelled crop yields by using the Environmental/Policy Integrated Climate (EPIC) model. They performed dominance comparisons for NFI based on “the stochastic efficiency with respect to a function (SERF) criterion”. Prato et al. (2010) reported decrease in crop enterprises by 24% and in mean NFI for APSs by 57% between historical and future climates.
Wu, Jin and Zhao (2010) studied on relation between climate change and irrigation technology advancement & agricultural water use in China. They performed their analysis for the period of 1949–2005. They used Palmer Drought Severity Index (PDSI) for the purpose of characterizing climate change and they used Gross Irrigation Quota (GIQ) which is defined as the ratio of the agricultural irrigation water quantity to the effective irrigation area, to investigate agricultural water use and climate change relationship. Wu, Jin and Zhao (2010) applied a statistical regression method to the PDSI and GIQ to quantify the relationship between agricultural water use and climate conditions over the China mainland. According to PDSI analysis, they detected downward trends over the irrigated areas of China which means irrigated areas have been drier over study period. They also stated that drying is more intensive between 1990 and 2005 due to more intensive global warming. They focused on agricultural water use under changing climate in China. They explained three major factors for change in irrigation quantity: expansion of agricultural land, climate change and advancement in irrigation technology. They stated that they eliminated expansion of agricultural land by using GIQ and they concentrated on last two factors. Firstly, they applied a statistical regression method to evaluate the relationship between the GIQ and PDSI over the period 1949–1990 when irrigation technology is at minimum level and irrigation water quantity were basically based on climate change. They found a
121 significant relation between GIQ and PDSI. They explained that relation between GIQ and PDSI showed that a drier period refers to larger GIQ while wetter period corresponds to smaller GIQ. Moreover, Wu, Jin and Zhao (2010) expressed important technological improvement in irrigation systems including soil seepage control and drip and sprinkler irrigation methods since 1991. Thus, after 1991, the actual GIQ is driven by both climate change and irrigation technology. They revealed that since 1991, much smaller actual GIQ values were calculated than the regressed GIQ with the climate information. They explained the reason of smaller actual GIQ by reduction effect of irrigation technology advancement on agricultural water use intensity in China. Finally, they expressed that without advances in irrigation technology, the current level of agricultural water use intensity would double owing to current intensive global warming.
Vano et al. (2010) investigated potential effects of climate change on water management and irrigated agriculture in the Yakima River Basin, Washington, US. They explained that Yakima River basin is a significant agricultural basin for central Washington State. They expressed that most crops in this basin are irrigated so irrigation water is very important for his basin. They stated that 32% of irrigated land is covered by tree crops and vineyards. Remaining part include forage, pasture, and annual vegetable and field crops and some special crops such as mint and hops. Vano et al. (2010) implied the importance of snowpack for the basin. They stated that water need of the basin is met by five reservoirs at the basin. Vano et al. (2010) referred previous studies which investigated climate change impacts on agricultural production. Vano et al. (2010) explained that Lobell et al. (2006) reported reduction in yields of almond, walnut and table grapes due to projected temperature increases around 1°C to 3°C by 2050. They stated that Lobell et al. (2006) found yield reductions even without possible impacts of climate change on irrigation water. Vano et al. (2010) gave more detailed information about study of Lobell et al. (2006). Lobell et al. (2006) projected yield decreases ranging from very slight levels to more than 20%. In addition, Vano et al. (2010) cited to Scott et al. (2004 a,b) and presented their study’s findings. Scott et al. (2004 a,b) assessed impacts of periodic droughts in the Yakima Basin and
122 they reported significant decreases in crop yields and economic losses based on analysis for both in dry years with current climate and in a projected future climate with 2°C temperature increase and no change in annual precipitation. Vano et al. (2010) used Variable Infiltration Capacity (VIC) model for hydrological modelling and they force VIC by using gridded historical climate and projected future climate data. Vano et al. (2010) coupled water management model which is a reservoir operation model with VIC model. Vano et al. (2010) reported water shortages in 14% of years based on historical observations. If required adaptation policies are not put into practice, water shortages were projected to increase by 27% in 2020s, 33% in the 2040s and 68% in the 2080s according to IPCC A1B global emission scenario. On the other hand, based on IPCC B1 global emission scenario, Vano et al. (2010) predicted water shortages 24% of years in the 2020s, 31% in the 2040s and 43% in the 2080s. Finally, they reported economical losses including production decreases by 5% to 16%.
Zhang, Zhu and Wassmann (2010) explored the response of rice yields to recent climate change in China at selected regions for the period of 1981–2005. They determined climate factors which are effective on yield trends at each spatial scale. Based on empirical results over 20 experiment stations, solar radiation is a dominant climate variable on rice yields and rice yield were positively correlated to solar radiation. In addition, Zhang, Zhu and Wassmann (2010) presented strong relation between temperature and rice yields however opposite of often-cited hypothesis (temperature increases have negative influences on yields), they did not detect any negative correlation between them. Moreover, they reported positive correlation between temperature and rice yields for some regions. In places where positive correlation between yields and temperature was detected, also positive correlation was reported between solar radiation and yield. Zhang, Zhu and Wassmann (2010) expressed that irrigation water availability have an importance to determine climatic effects (radiation or rainfall) on yield variability at a regional scale in China.
Abdrabbo et al. (2010) assessed sensitivity of potato to climate change in Egypt. They selected potato due to its key importance for Egypt. They defined potato as one of the
123 major crops in Egypt. They expressed that national potato production is enough for local demand and excess is exported. In addition, they explained that due to rapid population increase and constraint on agricultural land, potato production is affected negatively in Egypt. Thus, they implied the importance of better understanding of how climate change will affect potato production in Egypt. Abdrabbo et al. (2010) explained that total potato cultivated area in Egypt is 89 thousands hectare (two million tons production) and total potato export was expressed as 296 thousand ton/year in 2005. They used SUBSTOR potato to perform simulations of physiological processes and yield of potato production. They used both present climate and future climate produced by two general circulation models (CSIRO and HadCM3) for A1 greenhouse gases scenario for 2050 in modelling studies. Based on future simulations, Abdrabbo et al. (2010) reported reductions in potato yield from 11 to 13%. They expressed that by using HadCM3 global climate model output, they projected higher potato yields than those with CSIRO model. Finally, Abdrabbo et al. (2010) concluded that selection of better cultivar and correct irrigation level are two main factors which is important to produce maximum benefit.
Liu et al. (2010) investigated crop yield responses to climate change in the Huang-Huai- Hai Plain of China. They focused on the grain production of winter wheat–summer maize cropping system. They stated future climate projections as follows: 2 and 5°C temperature increase and precipitation increase and decrease by 15%. In addition, they took into account C02 enriching to 500 and 700 ppmv. Liu et al. (2010) selected two typical counties in the Huang-Huai-Hai (3H) Plain (covering most of the North China Plain), Botou in the north and Huaiyuan in the south. They considered both irrigated and rain-fed conditions. They generated climate change scenarios by using general circulation models and the historical trend from 1996 to 2004. They were preceded in a bio-geo-physically process-based dynamic crop model, Vegetation Interface Processes (VIP). They expressed that “The projected crop yields are significantly different from the baseline yield, with the minimum, mean (±standardized deviation, SD) and maximum changes being −46%,−10.3±20.3%, and 49%, respectively” (Liu et al. 2010, p.1195). They presented overall yield decrease by −18.5±22.8% under
124 5°C increase. They stated that it is remarkably larger than change by −2.3±13.2% under
2 °C warming. They explained that CO2 fertilization have a critical role to reduce negative effects of temperature increase on crop yields. Liu et al. (2010) stated that “response of a C3 crop (wheat) to the temperature rise is significantly more sensitive to CO2 fertilization and less negative than the response of C4 (maize), implying a challenge to the present double wheat–maize systems” (Liu et al. 2010, p.1195). They explained that precipitation increase has positive effects on yields. Thus, they expressed the effect of decreased rainfall as negative which decreasing yields and increasing losses. In addition, they implied the importance of irrigation due to its effect on mitigating decreased crop yield. Liu e al. (2010) explained that the crops in the wetter southern 3H Plain (Huaiyuan) are markedly more vulnerable to climate change than crops in the drier north (Botou). Liu et al. (2010) expressed that CO2 fertilization effects may be more important under drier climate. Finally, they underlined the importance of implementing adaptation strategies and developing efficient water resources management policies.
Boomiraj et al. (2010) explored climate change and mustard association in India. They explained that Rapeseed-mustard is a major group of oilseed crop in the world and is an important crop for India. They expressed that India is the second largest producer of Mustard after China worldwide. Boomiraj et al. (2010) explained that despite the substantial increase of oilseed production in India since 1960s, because of population growth and increase in income levels, demand for oilseed may increase in India. They stated that owing to sensitive characteristic of mustard to climate, climate change may have significant influences on mustard production. Boomiraj et al. (2010) explained that although there are many studies investigating climate change impacts on production of cereals, there are limited studies evaluating climate change and oilseed crop relation. They implemented InfoCrop model, which is a generic dynamic crop model. InfoCrop is a beneficial tool to perform integrated assessment of the effect of weather, variety, pests and soil management practices on crop growth and yield. By using InfoCrop, Boomiraj et al. (2010) expressed the sensitivity of crop to carbon dioxide (CO2) and temperature. They revealed that projected future climate will very
125 likely result decreases in mustard yields for both irrigated and rain-fed conditions. Nonetheless, decreases in mustard yields show regional differences over India. Boomiraj et al. (2010) explained that decrease in mustard yield would be more significant in eastern India with 67 and 57% for irrigated and rain-fed conditions. They projected second most sensitive region as central India with 48 and 14% decreases and finally northern India with 40.3 and 21.4% reductions. They explained that maximum temperature increase was projected for eastern India, thus, largest yield decrease was modelled for eastern India. Finally, they recommended adaptation strategies such as late sowing and growing long-duration varieties to prevent yield loss of irrigated mustard at different regions of India.
2.7.3. Climate Change Impacts on Agriculture of Turkish Republic
There is a consensus by scientists that developing countries are more vulnerable and will be affected more by climate change effects. Most of the developing countries’ economies are basically dependent on agriculture sector (Tubiello & Fischer 2007). Thus, it is mandatory to better understanding of climate change impacts on agriculture and develop adaptation policies to encounter these effects particularly in developing countries.
As a developing country, the agricultural area of Turkish Republic is about 27 million ha with 35% of country’s total surface area. Nevertheless, only 18 million ha area was under cultivation in Turkey in 2004 (Ministry of Environment and Forestry 2007). Economically and technically irrigable land of Turkey is almost 25 million ha and 58% of this area is irrigated by 2005 (Agriculture and Rural Affairs Ministry 2008). The sectoral Gross Domestic Product (GDP) ratio of agriculture sector is around 12% by 2005 in Turkey. Furthermore, the value of crop production in Turkey is 22.5 billion euros in Turkey by 2004 (Ministry of Environment and Forestry 2007). Major agricultural export products of Turkey are hazelnut, tobacco, raisins, apricot, wheat flour, tomatoes and cherries with total production value of 2.381 million euros. On the other hand, cotton, sheep, lamb, wool, wheat, palm oil, corn and tobacco are major agricultural import products with total production of 2.565 million euros (Ministry of Environment and 126 Forestry 2007). However, Dellal and McCarl (2010) stated that currently GDP rate of agriculture sector is 8% and employment rate of Turkish agriculture is 25%. They stated that cereal production occupies 75% of Turkey’s cropland and cereal production is significantly dependent on seasonal precipitation. While wheat production equals to 20 million tons in Turkey, barley production is 9 million tons in Turkey.
Agriculture sector uses approximately 75% of entire water potential in Turkey. 85% of total irrigation water demand in Turkey is met by surface water while 15% is provided by ground water. In Turkey, around 94% of irrigation is done by open channel systems and 6% is achieved by pressurized irrigation systems (Agriculture and Rural Affairs Ministry 2008).
One of the most important challenges regarding to irrigation in Turkey is difficulty of changing farmers’ traditional irrigation methods. Moreover, lack of financial support for farmers to adapt themselves to new irrigation technologies is another significant problem.
Most parts of Turkey experience scarcity and unreliability of precipitation during growing season, thus, irrigation has a huge significance for agricultural production in Turkey. Mengu, Sensoy and Akkuzu (2008) explained that if all irrigable lands in Turkey were opened to irrigation, approximately 200 km3 water deficit would be experienced in Turkey. This effect is more important in the Mediterranean, Central and Southeast Anatolia regions (Mengu, Sensoy & Akkuzu 2008).
One of the most dramatic climate change impact on Turkish agriculture is drought. More frequent and intensive droughts are expected in Turkey particularly in Mediterranean Basin. Mengu, Sensoy and Akkuzu (2008) reported 20% precipitation decrease in the last 25 years in Mediterranean Basin. In similar to other climate projections that are explained in other part of this study, Mengu, Sensoy and Akkuzu (2008) explained that decreasing precipitation trend will continue and will cause significant drought consequences in Turkey’s semi-arid Mediterranean, Aegean and
127 Central Anatolian regions. Drought in 2007 in Turkey is a good example to understand the importance of droughts to Turkish agriculture and agriculture economy. The economical harm of 2007 drought to Turkey’s agricultural sector was calculated as $ 4.2 billion. Production losses have been estimated as 20% for wheat and seedless raisins, 24% for watermelons, 25% for tomatoes, and 17% for sunflowers (TZOB 2008).
Moreover, Mengu, Sensoy and Akkuzu (2008) expressed increase in growing season length (GSL) in Turkey. They reported GSL increase by 35 days in 100 years. They explained that increase in GSL will be beneficial for agricultural products except orchard crops due to requirement of period of cool conditions during winter known as vernalization (Sensoy, Alan & Demircan 2007). GSL increase in Turkey (except coastal regions) is shown in Fig. 2.24.
Fig.2.24: Growing season length increase in Turkey (Mengu, Sensoy & Akkuzu 2008)
As stated before in this study, climate change will affect water resources and water availability significantly because of changing climate alteration effect on hydrological cycle. Climate change effects on climate variables and streamflows in Turkey was discussed at current study’s different parts. Climate change impacts on streamflows will be negative at almost entire Turkey. Decreasing runoffs coupled with increasing 128 water demands of agriculture, industry and urban will cause a serious competition between water sectors. Agriculture is the largest water consumer in Turkey. However, prior water user is residents. Moreover, industrial activities in Turkey have been developing very dramatically, particularly during last decade. In summary, agriculture sector is the one which must improve some policies to reduce its water demand. First of all, climate change impact studies should be performed for different parts of Turkey. Current study aimed this purpose in Eastern part of Turkey. But, there are still many parts in Turkey where climate change impacts on water resources and hydrology is uncertain. According to hydrological impact studies’ findings, some policies (discussed before) such as developing efficient water use technologies or additional water storage reservoirs should be applied in Turkey. Despite the importance of climate change, very limited number of studies was done to investigate climate change influences on different sectors in Turkey. Agricultural impact studies are also very few in Turkey. Following paragraphs are review of some papers which investigated climate change effects on agriculture and/or irrigation in Turkey.
Ozkan and Akcaoz (2002) purposed to investigate relation between climate variables and yields of three crops, wheat, maize and cotton in the Cukurova region of Turkey where intensified agricultural activities are conducted. Ozkan and Akcaoz (2002) used time series data for the period of 1975-1999 while analysing crop yields across various climate factors. They arranged climatic variables based on periods of the examined crops such as planting, flowering and harvesting time. They used linear perturbation model (LPM) for determining significance levels of climate variables for selected crop types. Ozkan and Akcaoz (2002) found R2 values of 46.1%, 57.2% and 74.5% for wheat, maize and cotton respectively. They detected highest variation coefficient (CV) in maize production (43.4%) followed by cotton (23.14%) and wheat (15.29%). Actually, this study did not purpose to study directly climate change impacts on agricultural production. But, they investigated significance of different climate variables on yield of wheat, maize and cotton. It may be used as an initial step to climate change studies at the same region for above mentioned products. They indicated that the most effective climate variable on deviations in crop yields is temperature. Moreover, it is known that
129 although some climate variables can show decrease or increase for different parts of Turkey, all climate studies are consistent in temperature increases over Turkey. Thus, significant yield decrease may be observed in Cukurova region.
Nagano et al. (2008) improved a new physical model named Irrigation Management Performance Assessment Model (IMPAM) and applied this model to the Lower Seyhan Irrigation Project. IMPAM is a quasi three dimensional water balance model and it is used for assessing irrigation districts’ adaptive capacity to changes in climate and social conditions. Nagano et al. (2008) reported increases in irrigation demand between 100 and 170 mm and extend in irrigation duration from early springs to late autumn in 2070s because of significant winter season precipitation decline which is projected by climate models. Moreover, Nagano et al. (2008) projected decreases in water table which is extremely high currently. They also explained the reason of high water table as inefficient irrigation practices in the region and they strongly recommended using water efficiently by using drip irrigation technology in the study area.
Topcu et al. (2008) investigated climate change impacts on agricultural water use in Seyhan Basin. They used RegCM regional climate model outputs to evaluate influences of climate change on irrigation water. Topcu et al. (2008) reported temperature increases of 3.2°C to 4.6°C and they stated larger temperature increases in high latitudes (above 500m) more than 4 °C in upper basin, nonetheless, in the lower basin namely Lower Seyhan Plain, about 3-4 °C increase was detected by them for the period of 2071-2100. Moreover, they revealed precipitation decrease by 16.3%. In addition, based on temperature increases and precipitation decreases, they detected increases in reference and actual evaporations with a percentage of 16.4% and 9.1% respectively. Finally, they found 206.9% increase in net irrigation requirement throughout the basin over the period of 2071-2100.
Yano, Aydin and Haraguchi (2007) studied climate change effects on irrigation demand and crop growth for a wheat-maize cropping sequence in Adana which is in Seyhan Basin, under effect of Mediterranean climate. They used data of the three general
130 circulation models—GCMs (CGCM2, ECHAM4 and MRI)—for the period of 1990 to 2100 and a regional climate model for the period of 2070 to 2079. They selected A2 emission scenario to generate future climate outputs. They used Soil-Water- Atmosphere-Plant (SWAP) model to predict the effects of climate change on water demand and on wheat & maize yields. Yano, Aydin and Haraguchi (2007) reported precipitation decrease of about 163, 163 and 105 mm during the period of 1990 to 2100 under the A2 scenario of the CGCM2, ECHAM4 and MRI models, respectively. Furthermore, the CGCM2, ECHAM4 and MRI models predicted a temperature increase by 4.3, 5.3 and 3.1 °C, respectively by 2100. Opposite of expectation, they reported decreases in actual evapotranspiration by 28 and 8% for the period of 2070 to 2079 according to CGCM2 and RCM respectively. They stated increasing irrigation demand for wheat due to decreases in precipitation. Moreover, Yano, Aydin and Haraguchi (2007) reported acceleration of crop development but shortened growing period by 24 days for wheat and 9 days for maize according to the CGCM2 data owing to temperature increases. They explained that shortened growth duration with a higher temperature decreased the biomass accumulation of both crops regardless of CO2- fertilization effect. Finally, they expressed increase by 16 and 36% in grain yield of wheat and a decrease by about 25% and an increase by 3% in maize yield, respectively.
Kanber, Kapur and Tekin (2007) investigated impacts of climate change on agricultural production under the project named impact of climatic change on agricultural production in arid areas (ICCAP). They discussed the adaptation policies which can be applied to the research area (Seyhan Basin) according to climate change impact simulations. They selected Seyhan Basin as a study area because of its significant role on food production in Turkey. Moreover, it is in Mediterranean region so it is very sensitive to climate change. Furthermore, there are different types of agricultural activities including rain-fed agriculture, irrigated agriculture and stock farming/pasturage so it is very ideal study region to investigate climate change impacts on agriculture. They obtained important findings such as the significant decrease of precipitation for the period of 2070-2100, important changes in snowfall and melting time, changes in planting time of crops such as wheat & corn and areal changes of
131 crops. Moreover, Kanber, Kapur and Tekin (2007) found substantial decreases in water potential and important changes in climate variables. They projected 3 °C monthly average temperature increase in 2070 in Seyhan River Basin. Moreover, they predicted 25% decrease in annual precipitation. In addition, 14% increase in evaporation and 17% decrease in evapotranspiration due to precipitation reductions were projected by Kanber, Kapur and Tekin (2007). They also explained reduction in water potential up to 30% in Seyhan River Basin.
Kanber, Kapur and Tekin (2007) investigated climate change influences on agricultural production especially on wheat and maize in Seyhan Basin. They stated that due to temperature increase and precipitation decrease, wheat yield will decrease in Seyhan River Basin. Moreover, increasing temperature will shorten plant growing season length. Furthermore, due to reduction in winter precipitation, wheat cultivation will be more challenging and middle & north parts of basin will be more suitable for wheat cultivation. Thus, Kanber, Kapur and Tekin (2007) recommended performing irrigation investments to the northern parts of basin. Moreover, they stated that increase in CO2 concentration will accelerate photosynthesis activities of wheat so will contribute to wheat yield in Seyhan River Basin. Kanber, Kapur and Tekin (2007) projected decrease in maize yield approximately by 13% in 2040-2060 relative to 1981-2000.
Caldag and Saylan (2009) applied an explanatory crop growth simulation model (CERES-Rice) under different climate scenarios with the aim of investigating climate change impacts on agriculture in Thrace region of Turkey. Firstly, they explained possible future climate expressed by Onol (2007) in Thrace region. They stated that average temperatures would increase between 3.5°C and 5°C for the summer season in Thrace. Summer season includes almost the whole of the mean paddy rice growing season. They stated that rice is one of the most important agricultural productions in Thrace region. Caldag and Saylan (2009) summarised their study’s findings in Table 2.11.
132 Table 2.11: Grain yield estimations of paddy rice for possible climate change scenarios (Caldag &Saylan 2009)
Table 2.11 shows the grain yield variations under different climate scenarios in Thrace region. In Table 2.11, T corresponds to temperature, P refers to precipitation and RG refers to global solar radiation. Caldag and Saylan (2009) stated that even an increase of only 1°C (T+1) in daily average temperatures would result 7% decrease in grain yield. Moreover, 23.5% decrease of yield was found for the T+5 scenario. They also explained that 30% increase in the global solar radiation (Rg+%30) caused a yield increase by
21.1%. Furthermore, increasing CO2 concentration (CO2 x3) resulted 48% increase of grain yield. Finally, they expressed that their study assumed that there will be no water stress in Thrace region however; future water stresses are expected for many parts of Turkey.
133 Dellal and McCarl (2010) investigated effect of drought in Turkish agriculture. Climate models and climate projections are consistent about more frequent and intense future droughts owing to changing climate in many parts of Turkey. So, Dellal and McCarl (2010) aimed to demonstrate how agriculture is vulnerable to droughts in Turkey. They presented drought influence on agricultural productivity under the example of 2007 drought. They presented precipitation change in 2007 relative to normal years’ rainfall by region. It is shown in Fig. 2.25.
Fig.2.25: Regional rainfall in 2007 in Turkey (Dellal & McCarl 2010)
Moreover, they explained yield response of different crops to drought. They stated that the most sensitive crop to drought in terms of yield is sunflower. Yield responds of different crop types to 2007 drought in percentage change is demonstrated graphically in Fig. 2.26.
134 Fig.2.26: Yield changes in 2007 in Turkey (Dellal & McCarl 2010)
Moreover, they gave regional yield response of specific crops including wheat, corn, sun flower and cotton to 2007 drought. It is demonstrated graphically in Fig. 2.27. In Fig. 2.27, E, W, N and C refer to east, west, north and central respectively.
135 Fig.2.27: Regional yield percentage changes of wheat, corn, sun flower and cotton crops due to 2007 drought (Dellal & McCarl 2010)
136 Furthermore, Dellal and McCarl (2010) presented price change response in percentage to 2007 drought. In Fig. 2.28, Dellal and McCarl (2010) showed change in production and price for wheat, corn, rice, lentil, sun flower, tomato, beef and milk.
Fig.2.28: Production and price percentage change of wheat, corn, rice, lentil, sun flower, tomato, beef and milk (Dellal & McCarl 2010)
Moreover, by using Turkish Agricultural Sector Model (TARSEM), they performed economical analysis of drought on Turkish agriculture and they summarized their findings in Fig. 2.29.
137 Fig.2.29: Production change of wheat, barley, corn, cotton and sun flower owing to drought conditions (Dellal & McCarl 2010)
Consequently, Dellal and McCarl (2010) provided very beneficial information which shows sensitivity level of Turkish agriculture to climate change. Especially, large increases in crop prices can cause very significant socio-economic problems in Turkey. Thus, it is necessary to implement agricultural adaptation policies which were expressed above as soon as possible in Turkey.
138 Chapter 3: Study Area
Headwater of Euphrates Basin which is the largest basin of Turkey was selected as study area for this study. Headwater of Euphrates Basin is named Karasu Basin and it is located in Eastern Turkey. Firstly, information regarding to Turkey is presented and then Karasu Basin is explained specifically in following sections.
3.1. Turkey
Turkey is located in Southern Europe and southwestern Asia spreading from a longitude of 36° to 42° North and a latitude of 26° to 45° East. Bulgaria and Greece are the neighbours of Turkey on the European side and Iran, Iraq, Syria, Azerbaijan, Armenia and Georgia are the neighbours on the Asian side. Location of the Turkey is shown in Fig. 3.1.
Fig.3.1: Location map of Turkey (Wikimedia 2010)
As a land bridge between Europe and Asia, Turkey is surrounded by seas on three sides, Black Sea in the north, Mediterranean in the South and Aegean Sea in the west. Turkey is the thirty-fourth largest country over world with an area of 783 562 km2. 759507 km2 of total area lies in Asia named Anatolia, while remaining part is located in
139 Europe named Thrace. 35% of total land is classified as agricultural land, while 27% forests, 18% pastures and meadows and 20% for other purposes (Ministry of Environment and Forestry 2007).
Population of Turkey is 72.6 million currently (TurkStat 2010). It increased from 56.5 million in 1990 to 71.2 million in 2004 with a growing rate of 1.2%. Population of Turkey is greater than all its neighbours with high population density. There is a high urbanization where the ratio of urbanisation reached 62.7% in 2006, while it was 52.9% in 1990 (Ministry of Environment and Forestry 2007).
1.6% of Turkey’s total surface water consists of inland water. Main rivers of Turkey are Euphrates, Tigris, Kizilirmak, Yesilirmak, Sakarya, Gediz, Menderes, Meric, Ceyhan and Goksu rivers. In Turkey, 200 natural lakes cover an area of 906 000 hectares, while dam lakes are covering 380 000 hectares. The largest lake in Turkey is Lake Van with an area of 3713 km2 (Ministry of Environment and Forestry 2007).
Turkey is divided into seven geographical regions which are, in order of size, East Anatolia, Central Anatolia, Black Sea, Mediterranean, Aegean, Marmara and Southeast Anatolia. Geographical regions of Turkey is demonstrated in Fig. 3.2.
140 Fig.3.2: Geographical regions in Turkey (Dellal et al. 2004)
Turkey has approximately 501 billion m3 precipitation annually. About 35% of this, 186 billion m3 is being runoff; however, 95 billion m3 of total streamflow is economically usable. Annual average precipitation of Turkey varies between 630-643 mm. Renewable water potential of Turkey is totally 234 billion m3, 193 billion m3of this amount is originated from rivers (186 billion m3 inland, 7 billion m3 outland) and 41 billion m3 is coming from groundwater (UCTEA 2009).
Turkey is not a water abundant country or in other words Turkey is not a rich country in the way of water resources. Value of water per person is under world average, while world average is 7600 m3. Although water per person is not at a critical level currently, when it is considered that Turkey’s population is increasing in a large growth rate and Turkey’s population is estimated as 100 million in 2030 by Turkey Statistic Institute, water per person will decrease significantly to the value of 1000 m3. Thus, it is very important for Turkey to manage water efficiently (Ministry of Agriculture and Rural Affairs 2008).
141 Turkey is divided into 26 hydrological basins with different water efficiencies. Among these, Euphrates and Tigris Basins are the largest ones with 28% of total water potential of Turkey (Ministry of Environment and Forestry 2007).
3.1.1. Turkey’s Climate
Turkey is located in Mediterranean macroclimate zone, which results large regional and/or seasonal variations on country’s climate (Ministry of Environment and Forestry 2007). Turkey experiences widely diverse climates extremely strong winter conditions to very hot and dry summers. The south and west parts of Turkey are under Mediterranean climate influenced by hot and dry summers and cool and rainy winters. The climate of Black Sea is colder and more rainy than Mediterranean cost. Northeast Anatolia shows the characteristics of continental climate with long and severe winter and short & cool summers. Furthermore, Central Anatolia is under the influence of a steppe climate where experiences arid & hot summers and cold winters (Kocman 1993; Ministry of Environment and Forestry 2007).
Turkey is divided into five main climate zones as shown in Fig. 3.3.
Fig.3.3: Climate Zones of Turkey (Ministry of Environment and Forestry 2007 [sourced by TSMS])
142 In Fig. 3.3, I corresponds to Mediterranean climate where Ia is humid, Ib is semi-humid Mediterranean climates. Mediterranean climate is mostly seen in Mediterranean and Aegean regions in Turkey. The basic characteristic of this climate is hot & dry summers and rainy & cool winters. Mediterranean coast experiences the highest mean temperatures in Turkey. In humid Mediterranean climate, snow and frost are very rare. Mean January temperature shows variation between 8°C and 10°C, while mean July temperature is around 27-28 °C. Winter is rainy season and annual average rainfall is approximately 1000 mm. Semi-humid Mediterranean climate is colder than humid climate with mean temperature of 5°C -8°C in January. Similar to humid Mediterranean climate, winter is rainy season in semi-humid Mediterranean climate and annual total precipitation is about 600-800 mm.
In Fig. 3.3, II refers to Black Sea climate. Rain, observed during all seasons, is the primary feature of this climate. Maritime influence is so strong in this climate zone. The mean annual temperature in Black Sea climate zone is around 8 °C -12 °C (Ministry of Environment and Forestry 2007).
Amount of precipitation depends on distance from coasts, wind impacts and elevation of area. In Turkey, Black Sea region is the most rainfall dominated zone and the most arid part of Turkey is Southeast Anatolia region. Precipitation mostly falls in winter and autumn seasons in Turkey.
III in Fig. 3.3 refers to semi humid Marmara climate which affects all Marmara region without Black Sea cost of region. Mean air temperature in July, hottest month, is about 23-24 °C while it is about 3-5°C in the coldest month, January. Winter is the most precipitated season of this climate and its total annual amount varies from 500 mm to 700 mm.
Distribution of mean annual temperature and precipitation over Turkey are shown in Fig. 3.4 and 3.5.
143 Fig.3.4: Distribution of mean annual temperature in Turkey (Ministry of Environment and Forestry 2007)
Fig.3.5: Distribution of mean annual rainfall in Turkey (Ministry of Environment and Forestry 2007)
Central Anatolia, mid-western Anatolia, western regions of Eastern Anatolia and Southeast Anatolia is under the effect of steppe climate (IV in Fig. 3.3) or semi-arid climate and it is divided into two types, semi-arid Central Anatolia(a in Fig. 3.3) and semi-arid Southeast Anatolia (b in Fig. 3.3) in Turkey. In semi-arid Central Anatolia
144 climate, winters are cold with mean temperature of 0-3°C in January. Mean July- August temperatures vary around 20-22 °C. Precipitation is seen mostly in spring and winter seasons. Annual precipitation is around 350-500 mm. On the other hand, in semi-arid Southeast Anatolia climate, summers are hot and dry, where mean temperature exceeds 30 °C in the hottest months, July-August. Minimum mean temperatures are between 2 °C and 5 °C. Annual precipitation varies between 350-800 mm.
Finally, in Fig. 3.3, V refers to continental Eastern Anatolia climate. It is the coldest climate zone in Turkey with minimum temperatures of -8 & -10 °C in the coldest month. Moreover, mean winter temperature is less than 0°C in this region. Furthermore, mean temperature of the warmest months is less than 20 °C. The maximum precipitation is observed from late spring to mid-summer and it is more than 500 mm. Snow is very important water source in this region (Ministry of Environment and Forestry 2007).
3.1.2. Greenhouse Gases Emission Sectors in Turkey
3.1.2.1. Energy
Turkey’s energy demand, especially electricity, increased with an annual rate of 3.7% and 7.2% respectively over the period of 1990-2005, which resulted increase in fossil fuel burning (MENR 2006). For instance, the natural gas consumption showed an increase by 2% in 1990 to 13% in 2004. Currently, energy resources of Turkey for the purpose of covering increasing demand is shown in Table 3.1.
145 Table 3.1: Primary energy resources of Turkey (Ministry of Environment and Forestry 2007 [Sourced by MERN 2004])
Total primary energy supply (TPES) and total final consumption (TFC) is both increasing by rate of 3.7% for the period of 1990-2004 in Turkey (Ministry of Environment and Forestry 2007). Increasing trend based on historical data is illustrated in Fig. 3.6.
Fig.3.6: Trend of energy use in Turkey (Ministry of Environment and Forestry 2007 [Sourced by MENR 2006])
146 The most significant factor influencing human-induced climate change is fuel consumption. In Turkey, important increases have been seen in fuel consumption particularly natural gas consumption. The share of natural gas consumption has risen from 2% to 13% from 1990 to 2004. Nonetheless, decrease in oil share which is the largest share in TFC, (from 48% 1990 to 39% in 2004) is a positive impact to climate change (Ministry of Environment and Forestry 2007). The distribution of energy consumption by fuels in 1990 and 2005 is shown in Fig. 3.7.
Fig.3.7: The distribution of energy consumption by fuels in 1990 and 2004 (Ministry of Environment and Forestry 2007 [Sourced by MERN, 2006])
3.1.2.2. Transportation
98% of passenger and almost %100 of freight transportation were supplied by road and railway transportation in Turkey. There is a good network between cities consisting of 64 000 km highways. Fig. 3.8 shows the road network in Turkey over the period of 1990-2004.
147 Fig.3.8: Road network in Turkey (Ministry of Environment and Forestry 2007)
Moreover, Turkey has 8333 km coastline which has an importance for tourism and sea transport. Moreover, there are 34 airports for civilian air traffic in Turkey. By the end of 2004, 14 of total airports are international. Finally, there is a railway connection of 10984 km long in Turkey (Ministry of Environment and Forestry 2007).
3.1.2.3. Industry
Industrial activities are substantial contributor to global warming due to large amount of greenhouse gases emission. Industry showed great development in Turkey since 1980s. Industry share in gross domestic product (GDP) is 25.4% currently, which was 8.3% in 1980. Goods manufacturing share in total exports was 93.7% in 2005 (Ministry of Environment and Forestry 2007).
148 Table 3.2: Manufacturing industry indicators (Ministry of Environment and Forestry 2007)
3.1.2.4. Residential Impacts
As mentioned before in this report, population of Turkey has been growing with a remarkable growing rate and it is expected to be 100 million in 2030. Heating houses in particularly highly populated cities act as an important contributor to global warming due to fossil fuel burning. Turkey is divided into four zones in terms of heating requirement according to TS825 Heat Insulation Standards in Buildings and it is shown in Fig. 3.9 (Ministry of Environment and Forestry 2007).
Fig.3.9: Building heating requirement zones in Turkey (Unit: kwh/m2) (Ministry of Environment and Forestry 2007)
149 Fig. 17 shows that heating requirement of north east and eastern central parts of Turkey is more than other parts.
3.1.2.5. Solid Waste
Solid waste is an important source for greenhouse gases; particularly for methane. Total solid waste collected by government bodies in Turkey was 25 mt, while it was 17.8 mt in 1994. It means %41 increase in total solid waste resulting 1.31 kg per capita solid waste production in Turkey (Ministry of Environment and Forestry 2007).
Based on manufacturing industry data in 2004, 119 mt hazardous waste was produced annually. 6% of total produced waste was recycled, while 21% was sold or granted and 73% was abolished (Ministry of Environment and Forestry 2007).
3.1.2.6. Agriculture
Agricultural activities such as fertilizer usage, rice cultivation influence greenhouse amount in the atmosphere. While agriculture was the most important economical sector before 1980s, industrial development in Turkey decreased the agricultural activities in Turkey. The sectoral GDP ratio of the agricultural economy decreased from 24% in 1980 to less than 12% in 2005. Similarly, employment in agricultural sector decreased from 53.6% in 1990 to 48% in 2000 (Ministry of Environment and Forestry 2007). Detailed information regarding to climate change and agriculture relation can be found climate change impacts on agriculture and irrigation part of this thesis.
3.1.2.7. Livestock Sector
Livestock sector is another important sector which is impacting climate by emitting greenhouse gases. Number of total livestock animals has risen 106% since 1990 and reached to approximately 345 million in Turkey (Ministry of Environment and Forestry 2007).
150 3.1.2.8. Forestry
Forests have an important role in carbon cycle by storing CO2 naturally. So, deforestation activities impact climate negatively. Turkey has rain forests in the north, Mediterranean forests in the south, dry and semi-dry forests in East and Southeast Anatolia. Turkey had around 21.2 million ha forest area in 2004 (Ministry of Environment and Forestry 2007).
3.1.3. Historical Changes in Climate Variables in Turkey
Before performing future climate projections, it is significant to understand if climate change effects started to be observed by detecting changes in historical data set. Moreover, it is important to detect how climate has been changing. Thus, it can be possible to compare historical trends and future climate projections. There are some studies in Turkey aiming to detect climate variables’ trends based on historical data in country scale. One of them was conducted by TSMS by using historical minimum, maximum & average air temperatures and precipitation data. Trends have been determined both annually and seasonally. Moreover, Dalfes, Karaca and Sen (2007) conducted a study to document how climate variables changed in the last century in Turkey. They obtained long term climate data from TSMS and they applied quality control procedures and homogenisation tests to enhance data quality. Dalfes, Karaca and Sen (2007) performed non-parametric Mann-Kendall test to determine trends for temperature (minimum, maximum and average) and precipitation series. Demir et al. (2008) investigated historical temperature and precipitation trends to have a better understanding on how climate changed in Turkey. They used data between 1952 and 2006 from 57 stations where homogeneity and data qualification test were performed. They applied Mann-Kendall and Wald-Wolfowitz trend analysis to data set for the purpose of detecting temperature trends. In addition to non-parametric tests, they also implemented least squares linear regression analysis. Moreover, they used precipitation data for the period of 1940-2006 from 88 stations. Similar to temperature trend analysis study, homogeneity test was implemented to precipitation data. Non- parametric Mann-Kendall and Wald-Wolfowitz trend analysis tests and least square
151 linear regression analysis were applied to precipitation data set in parallel to temperature.
Tayanc et al. (2009) investigated how climate changed in Turkey for the last half century. Similar to Dalfes, Karaca and Sen (2007), Tayanc et al. (2009) performed quality control process to climate data obtained for the period of 1950-2004 from 52 meteorological stations, geographically distributed over Turkey, which are classified as urban (S1) and rural (S2) stations to detect urban heat island effect. In addition to non- parametric trend test, Tayanc et al. (2009) applied signal analysis for better understanding of trends. Tayanc et al. (2009) used low pass filter (LPF) method and moving averages in climate data to remove the shorter term fluctuations from the moving averages signal. Moreover, with the aim of eliminating large-scale effects and dismiss the local changes or possible anomalies for a particular station or station groups in a region of concern, a Relative Difference (RD) signal is performed by Tayanc et al. (2009).
3.1.3.1. Changes in Temperature
According to TSMS analysis, there is an increasing trend of 0.64 °C/100 years in average temperature of Turkey between 1941 and 2007 (Ministry of Environment and Forestry 2008).
152 Fig.3.10: Temperature trend in Turkey between 1941 and 2007 (Ministry of Environment and Forestry 2008)
TSMS did not consider short term trends during analysis period. Tayanc et al. (2009) reported that average temperature trends showed cooling starting in the early 1960s and ending in 1993 then an increase starting from 1993 till to 2004 with the lowest temperatures in 1992-1993 for majority of meteorological stations in Turkey due to Mount Pinatubo eruption, when significant amount of particles were “released into the stratosphere, acting as anti-greenhouse agents”. Temperature trends according to moving averages and LP filtered moving averages are shown in Fig. 3.11. Moreover, a significant warming trend was detected by Tayanc et al. (2009) with maximums in 2000-2002.
153 Fig.3.11: Average air temperature trend in Turkey between 1950 and 2004 (Tayanc et al. 2009)
Furthermore, Tayanc et al. (2009) obtained relative difference (RD) and low pass filtered relative difference (LPFRD) signals in 1950–2004 period for regional average temperature series and assessed them for seven geographical regions of Turkey. Analysis result graph is demonstrated in Fig. 3.12.
154 Fig.3.12: Local relative temperature differences and their low pass filtered signals (Tayanc et al. 2009)
155 According to Fig. 3.12, Inner Anatolia regions, Middle Anatolia, East Anatolia and Southeast Anatolia regions have similar characteristics in respect to average temperature. They showed high variability in temperature which can be explained with their climate type (continental climate). These regions demonstrated warming trend in the 1950s, then a cooling period until 1976 for Central Anatolia region then again cooling period during 1980-1993. Central Anatolia is still showing warming trend while East and Southeast Anatolia regions show cooling in 2000s. Because of close location to sea, Mediterranean, Aegean, Marmara and Black Sea regions show lower variability in mean temperatures. Mediterranean region’s temperature increases slightly except the period of 1974-1981. Marmara and Aegean regions showed cooling trend during the first half of 1950s and then they showed very slight variability. Black sea region showed a significant decrease in temperature between 1960s and 1986 and a slight increasing trend until 1998 and currently average temperature is decreasing in Black Sea region (Tayanc et al. 2009).
Tayanc et al. (2009) stated a significant warming trend in the majority of the Mediterranean region stations, some East Anatolia stations and in Luleburgaz and some Istanbul stations in the northwest according to Mann-Kendal trend analysis. The result graph of Mann-Kendal trend analysis for maximum, minimum and average temperature is illustrated in Fig. 3.13.
156 Fig.3.13: Mann-Kendal trend analysis results a) Maximum temperature b) Minimum temperature c) Average temperature (Tayanc et al. 2009)
In Fig. 3.13, big red and blue circles denote to significant increasing and decreasing trends respectively (red correspond to warming while blue is cooling), while small red and blue circles correspond to insignificant increasing and decreasing trends. It can be seen in Fig. 3.13 that Turkey is mostly under increasing temperature trend with some exceptions such as northern parts of Turkey. Moreover, increasing trend is clearer and more significant in minimum temperatures. Minimum temperature increase is very dominant in urbanized and high populated parts of Turkey. Relation between urbanization & high population and minimum temperature increase can be explained by urban heat island effect. Urban heat island effect can be defined as an artificial 157 heating source originated from absorption of heat absorbed by buildings and air pollutants directly proportional to increase in residential and industrial areas. Its effect can be noticed during night time when the minimum temperature is seen (Tayanc et al. 2009). When average temperature change is considered, it is seen in Fig. 3.13 that there is an insignificant increasing trend for most parts of Turkey except south & southeast parts and very high populated locations in Marmara region. These regions showed significant increasing trends.
Demir et al. (2008) figured out that average temperatures of Turkey showed a significant increase in south which is in agreement with Environment and Forestry Ministry Report (2008) and study of Tayanc et al. (2009). They also reported average temperature increases in south western parts of Turkey. According to least square linear regression, increase rate of average temperature over Turkey is between 0.121°C/ten years and 0.312°C/ten years.
Demir et al. (2008) reported insignificant cooling trend in average temperatures during winter season. They stated significant increasing average temperature trends in Mediterranean, Southeast Anatolia and Marmara regions during spring season. In summer season, majority of meteorological stations showed rising trends in average temperatures. Moreover, insignificant cooling and heating trends were observed in autumn season. Demir et al. (2008) stated that Turkey’s average temperature shows similar trend to global average surface temperatures with a difference, global average surface temperature has shown rapid increasing trend since 1980s, while 1990s is starting time of Turkey average temperature rising trend.
TSMS analysis on historical temperature trends are mostly in agreement with the study of Tayanc et al. (2009). TSMS stated that there are significant increases in temperature in south and southeast parts of Turkey. Significant increases are also observed in high urbanized parts such as Istanbul and Kocaeli which is shown in Fig. 3.14.
158 Fig.3.14: Changing trend in annual average temperatures (1952-2006) (Ministry of Environment and Forestry 2008)
According to TSMS study, annual maximum temperatures have been showing increasing significant trends for many regions in Turkey including south, west, east and western part of southern east Anatolia. Fig. 3.15 illustrates changing trend in annual maximum temperatures.
159 Fig.3.15: Changing trend in annual maximum temperatures (1952-2006) (Ministry of Environment and Forestry 2008)
Demir et al. (2008) reported that annual maximum temperature showed a general increasing trend over Turkey, especially significant in Mediterranean and south parts of East Anatolia. According to least square regression analysis, increasing rate is between 0.102°C and 0.399°C/10 years in maximum annual temperature. Maximum temperatures showed reduction over Turkey except south parts of Turkey in winter season. In spring season, increasing trend indicated substantiveness in Marmara, Aegean, Mediterranean and South Eastern Anatolia regions. General maximum temperature trend was upward over Turkey in summer season, particularly meaningful in Aegean, Mediterranean, South East Anatolia and north parts of Central Anatolia. Increasing or decreasing maximum temperature trends are not meaningful in autumn season.
TSMS reported that most significant changes in temperature were observed in annual minimum temperatures where 47% of all meteorological stations showed significant increasing trends, which is demonstrated in Fig. 3.16.
160 Fig.3.16: Changing trend in annual minimum temperatures (1952-2006) (Ministry of Environment and Forestry 2008)
Demir et al. (2008) stated that minimum temperatures showed rising trend over Turkey in which 27 stations showed significant increases. Increase between 0.103°C and 0.679 °C /ten years was reported by Demir et al. (2008) based on least square linear regression analysis. Demir et al. (2008) explained the significant upward trend in minimum temperatures by urban heat island effect, which is the result of urbanization and population growth, similar to Tayanc et al. (2009). The most significant increase in minimum temperatures was observed in summer season, 41 of all stations showed meaningful increasing trend. In spring season, minimum temperatures are substantive in Marmara, Aegean, Mediterranean and South Eastern Anatolia. Finally, in autumn season, significant growing trend was detected by Demir et al. (2008) in Central Anatolia, Mediterranean and Southeast Anatolia.
In addition to TSMS, Tayanc. et al. (2009) and Demir et al. (2008), Dalfes, Karaca and Sen (2007) investigated historical changes in temperature seasonally. Similar to Tayanc 161 et al. (2009), Dalfes, Karaca and Sen (2007) used Mann-Kendal trend analysis to determine seasonal trends. They reported that the most important change in average temperature based on data between 1951 and 2004 is increase in summer temperatures especially in western and southwestern parts. It is shown in Fig. 3.17.
Fig.3.17: Seasonal annual average temperature trends (Dalfes, Karaca & Sen 2007)
Dalfes, Karaca and Sen (2007) also stated based on seasonal analysis that winter temperatures have a tendency to decrease especially in coastal stations.
162 3.1.3.2. Changes in Precipitation
Precipitation shows large difference spatially and temporally in Turkey. Moreover, seasonality on precipitation is very important in Turkey. Approximately, 40% of total precipitation falls in winter, while %27 in spring, %10 in summer and %24 of all precipitation are falling in autumn season. There is a decreasing trend of 29 mm/100 years in precipitation according to analysis of 1941-2007 precipitation data. 658. 5 mm annual average precipitation was observed between 1941 and 1970, while it was 635 mm between 1971 and 2000. Finally, annual average precipitation was 627.2 mm in 1980-2006 (Ministry of Environment and Forestry 2008).
Fig.3.18: Turkey annual average precipitation (1941-2007) (Ministry of Environment and Forestry 2008)
Fig. 3.19 demonstrates the long term geographical precipitation distribution of Turkey. The high amount of precipitation has occurred in small regions at seaside. On the other hand, central and eastern parts’ precipitation is less than 400 mm.
163 Fig.3.19: Long term geographical precipitation distribution of Turkey (Ministry of Environment and Forestry 2008)
The decreasing trends in precipitation have been observed particularly in winter season in Turkey. On the other hand, autumn precipitation had increasing trend. Decrease in winter precipitation can be related to increase in high pressure conditions owing to decrease in frequency of middle latitude and Mediterranean low pressures in winter season. Winter precipitation decreases correspond to Northern Atlantic Oscillation (NAO) strong positive anomaly terms. Moreover, in strong El-Nino years or one year later, decline in winter precipitation has been observed (Ministry of Environment and Forestry 2008).
Regionally, significant precipitation decreases in southern and western parts in winter season; significant increases northern parts of central Anatolia in autumn season; and insignificant precipitation increases in spring season have been observed.
164 Fig.3.20: 1951-2004 seasonal precipitation trends a) Winter b) Spring c)Summer d)Autumn (Ministry of Environment and Forestry 2007)
Moreover, by using precipitation data between 1940 and 2006, a trend analysis performed for 30 year time slices by TSMS to detect any regional trend in precipitation. It is shown in Fig. 3.21.
165 Fig.3.21: Regional precipitation trend analysis for 30 year slices (Ministry of Environment and Forestry 2008)
No trend in precipitation was found in Marmara, Central Anatolia and East Anatolia. On the other hand, increasing trend in Black sea and decreasing trend in Mediterranean, Aegean and Southeast Anatolia was detected. Mediterranean region’s average precipitation decreased from 831.1mm in 1940-1969 to 795.9 mm in 1980- 2006. Aegean region’s precipitation declined from 712.3 mm to 655.9 mm for the same period and it went down from 661.2mm to 619.6 mm in Southeast Anatolia region (Ministry of Environment and Forestry 2008).
Furthermore, decrease in rainy days in winter season has been observed especially in Mediterranean region based on historical data. In addition, in summer season, significant reduction in rainy days has been reported particularly for western parts of Turkey (Ministry of Environment and Forestry 2008).
166 Demir et al. (2008) also conducted a study for the purpose of precipitation trend detection. They presented increasing precipitation trend in Black Sea & central parts of East Anatolia and decreasing trend in Mediterranean regions. They state that they could not detect any trend in Marmara and Central Anatolia. 78 stations of 88 showed decreasing trends in precipitation in winter season over Turkey while 24 of them were meaningful. They also reported that the most important changes in precipitation were assigned in winter season. Regionally, Mediterranean, Central Anatolia and East Anatolia showed downward precipitation trend. In spring season, there are very few significant increasing trends over Turkey. Increasing trends were observed in Eastern Anatolia, Central Anatolia and Mediterranean regions. Black Sea and Marmara regions indicated declining trends in spring season. There is not a clear precipitation trend in summer season in Turkey and an increasing trend in 75 stations of 88 were observed in autumn season where as only 10 of them statistically significant.
In conclusion, according to seasonal and annual analysis of precipitation data, although some insignificant increases have been observed in autumn season, for the majority of Turkey, decreases in precipitation have been detected particularly in winter season. Studies and analysis showed that especially Mediterranean, Aegean and Southeast Anatolia regions are under risk of climate change with regard to precipitation.
3.1.3.3. Changes in Streamflow
Streamflow data has less uncertainty when it is compared with precipitation. So, it is more trustworthy data than precipitation. Nevertheless, water taken from rivers for irrigation or domestic purposes can alter over time and it influences homogeneity and quality of streamflow data (Dalfes, Karaca & Sen 2007).
Dalfes et al. (2007) conducted a streamflow trend investigation for Turkey based on flow data between 1969 and 1998. Summary graph of their findings is Fig. 3.22.
167 Fig.3.22: Streamflow trend analysis in Turkey (Dalfes, Karaca & Sen 2007)
Dalfes, Karaca and Sen (2007) detected significant decreases in streamflow in western and southwestern parts of Turkey and significant increases for a few stations in northern parts. They stated that this trends more and less similar in all seasons.
3.1.4. Future Climate Projections for Turkey
There are limited climate studies to forecast future climate and investigate effects of changing climate on different sectors in Turkey. One of the most important studies conducted for the purpose of predict future climate in Turkey is a project named Climate scenarios for Turkey. This project was performed in cooperation of TSMS and Istanbul Technical University (ITU). In this project, Global Climate Models and RegCM regional climate model were coupled for future climate prediction. Project findings are as follows:
According to pessimistic greenhouse gases emission scenario (A2), temperatures in Turkey will increase 2-6 °C in 2070-2100 relative to 1961-1990 averages. In winter season, increases will be clearer in the eastern parts of Turkey, while temperature
168 increases will be 3-4°C more in the western parts than eastern parts of Turkey in summer season. Average annual temperature increase over entire country is expected to be around 2-3 °C. It is shown in Fig. 3.23.
Fig.3.23: Expected seasonal temperature change in 2070-2100 relative to 1961-1990 (a) Winter season (b) Spring Season (c) Summer season (d) Autumn season (Ministry of Environment and Forestry 2008)
According to climate projections, it is expected to be a decrease of 20% in precipitation over Turkey. However, it is expected to observe clearer decreases in winter season
169 especially for Aegean and Mediterranean costs and increase in northern part of Turkey. According to project results, significant changes in precipitation will not occur in Central Anatolia region. The most severe reduction in precipitation will be observed in south western coast (Ministry of Environment and Forestry 2007). Projected seasonal changes in precipitation are shown in Fig. 3.24.
Fig.3.24: Expected seasonal precipitation change in 2070-2100 relative to 1961-1990 (a) Winter season (b) Spring Season (c) Summer season (d) Autumn season (Ministry of Environment and Forestry 2008)
170 Snow is very important water resource particularly in East Anatolia region. Snowfall will reduce in direct proportion to temperature increase and more rainfall than snowfall will be observed. It is expected to observe up to 200 mm snow depth reductions in higher elevations of East Anatolia and eastern parts of Black Sea region (Ministry of Environment and Forestry 2008). It means decreases in Euphrates and Tigris’s river flows which are very significant for Middle East countries.
Coastal sides of Black Sea region will have more precipitation. However, there will be no significant increase in summer precipitation but significant increase in autumn precipitation. When precipitation and evaporation difference is considered, an increase has found in northern Turkey, and a decrease has detected in southern parts of Turkey. Moreover, it is expected to see more floods in northern parts and droughts in southern parts of Turkey (Ministry of Environment and Forestry 2008).
Sea level rise was around 12 cm for Mediterranean and Black Sea region in the last century. According to IPCC scenarios, increase in Mediterranean Basin sea level was projected by 12-18 cm by 2030, 14-38 cm by 2050 and 35-65 cm by 2100 (Kadioglu 2008).
In addition to above future climate projection studies, Demir et al. (2008) aimed to define detailed future climate projections for Turkey by using British Meteorology Office, Hadley Centre for Climate Prediction and Research’s Regional Climate Model PRECIS (Providing Regional Climates for Impacts Studies). The model was run by using Hadley Centre’s Global Climate Model HadAMP3 outputs. Reference period was determined as 1961-1990 and corresponding future period was selected as 2071-2100. Moreover, A2 scenario was selected for model application. According to model results, 5-6 °C increases in mean annual temperature is expected over Turkey except coastal regions while 4-5°C temperature increases are expected over coastal region. Mean temperature difference map between 1960-1990 and 2071-2100 periods is shown in Fig. 3.25. Moreover, mean temperature change graph is shown in Fig. 3.26.
171 Fig.3.25: Mean temperature difference map of Turkey between 1960-1990 and 2070- 2100 periods (Demir et al. 2008)
Fig.3. 26: Mean temperature difference graph of Turkey between 1960-1990 and 2070-2100 periods (Demir et al. 2008)
172 According to Demir et al. (2008), in winter season, mean temperature increase in eastern Turkey is expected as 4-6°C while it is 3-4 °C in western Turkey. In spring season, 4-5°C increases are expected. Increase values will be 3-4 °C in Black Sea coasts and it will be 5-6 °C inner parts of East Anatolia. In summer season, temperature increase will be observed around 4-5°C, while it is around 6-7 °C in inner Aegean, south Mediterranean, inner west Black Sea. Other parts showed 5-6°C temperature increase. In autumn season, 4-5°C increases in temperature are expected over Turkey. Increase in maximum temperature is expected around 5-6 °C in Turkey, while increase in minimum temperatures is around 4-6 °C. Minimum night temperatures showed an increase of 5-6 °C in eastern parts in winter while increase was larger with 7-8 °C in the inner parts of Aegean region during summer season.
Demir et al. (2008) projected precipitation reductions in Turkey. However, regional differences are expected in precipitation trends. Eastern Black Sea, Aegean and Mediterranean regions will experience precipitation decrease at the rate of 100-400 mm/year. Average annual precipitation difference in mm/year between 1960-1990 and 2071-2100 periods and annual precipitation change rate in percentage are shown in Fig. 3.27 and 3.28.
Fig.3. 27: Average annual precipitation difference in mm/year between 1960-1990 and 2071-2100 periods (Demir et al. 2008) 173 Fig.3. 28: Annual precipitation change rate in percentage between reference and future climate (Demir et al. 2008)
Decreases in precipitation will be more intensive in Aegean, Mediterranean, Central Anatolia, Trachea, and Southeast Anatolia with 30-40%. On the other hand, from west to east, in East Anatolia and east Black Sea regions, decrease in precipitation will be less with 5% (Demir et al. 2008).
Moreover, there will be decreases in snow depth up to 300 mm, which will be effective on water budget, in eastern Anatolia and north east Black Sea regions. Snow depth changes in Turkey are shown in Fig. 3.29.
174 Fig.3. 29: Winter season snow depth change (mm) (Demir et al. 2008)
Finally, in parallel to increase in temperature and decrease in precipitation, surface evaporation will increase and will result more water loss in Turkey (Demir et al. 2008).
3.1.5. Exposure Potential of Turkey to Climate Change Impacts and Adaptation Need
Turkey is among the high risk group countries in terms of climate change impacts due to the fact that it is surrounded by seas by three sides and it is in East Mediterranean Basin & significant part of country shows Mediterranean climate characteristics. Owing to characteristics of Mediterranean climate, which is severe summer droughts, sudden and intensive precipitation, floods and strong winds, Turkey is sensitive to changes in climate.
When increases in summer temperatures particularly in western parts of Turkey, increasing trends in minimum temperatures for entire Turkey, significant decreases in winter precipitations, increase in extreme weather events such as drought & floods and high population growth rate are considered, it is possible to see obvious threat of climate change on Turkey. Environmental and socio-economic effects of climate change on Turkey can be stated as follows: 175 . Increase in hot weathers resulting longer and more frequent droughts will be observed. It will influence capacity and duration of forest fires. . Shifts in climatic zones will occur and ecosystems which could not adapt to new climatic zone will disappear. . Additional water problems in semi drought and drought areas will be seen and agricultural & domestic water demand will increase. . Large impacts will be seen on agricultural production. Agricultural production potential will change. Even a small change in drought zone will impact area very intensively in the way of agricultural production. . Increase in duration and intensity of summer droughts will contribute to desertification, salinity and erosion events. . Increases in extreme weather events such as heat waves will affect human health. Children, old, poor, disabled people and people with health problems are among those who are in high risk group in terms of climate change effects. . Under the impact of urban heat island effect, night temperatures in summer period will increase particularly in large populated cities, it will result more energy consumption with the aim of cooling. . Water shortages and temperature increases will lead health problems especially in big cities. . Wind characteristics, intensity and duration of incoming solar radiation will possibly change resulting alterations in renewable energy sources. . Changes in sea ecosystems will be experienced, which will result significant socio-economic problems for the regions where fishery is main livelihood source. . Due to sea level increases, high populated regions, tourism and agriculture lands can remain under water (Ministry of Environment and Forestry 2008).
Due to droughts, groundwater will go deeper so for using groundwater in irrigation activities more investment will be necessary. It will also result more energy
176 consumption. Agriculture related sectors will be affected from climate change so unemployment problem (already very important problem in Turkey) will increase.
Climate change will influence not only water quantity but also water quality. Increasing temperatures, decrease in precipitation and flow will enhance pollution concentration. Increasing pollution concentration will make already existing water quality problems more severe.
In Turkey, conservation of water resources, generalization of modern irrigation systems which decreases water consumption, constructing flood warning systems, generalization of renewable energy resources in many sectors can be shown examples to precautions against water related climate change problems (Ministry of Environment and Forestry 2008).
Main reason of desertification in Turkey is erosion (86%). Climate change induced sudden and severe rainfall events and drought will enhance the intensity of problem in Turkey. This means along with climate change effects, lands’ vulnerability to erosion will increase and more financial investment will be required for erosion control studies. Decreasing land yield will increase the immigration from agricultural rural regions to high populated industrialised urban areas, which will later increase socio- economic problems obviously.
Around 30 million people in Turkey are living in coastal parts, where are the most sensitive parts to climate change. Any impact, small or large, will influence Turkey’s economy significantly. Because, 70% of all industrial activities is done at coastal regions of Turkey (Ministry of Environment and Forestry 2008).
Tourism will be affected intensively from climate change. For instance, decreases in snow depth or early melting due to increasing temperatures will have significant influences on snow tourism. Tourism sector has to be aware of this approaching danger and long term plans have to be decided in terms of changing conditions to
177 supply sustainability. Moreover, master plans have to prepared for the regions where are vulnerable to floods and inundations which can be seen owing to increasing sea level.
Rapid urbanization with irregular industrialization and economic development has been contributing to climate change, enhancing pressure on natural resources & environment and influencing sustainable development of Turkey. It is not easy to allocate finance to adapt or mitigate climate change impacts in developing countries such as Turkey. There is no any formal international sanction or binding for Turkey (Kyoto Protocol), however due to sensitivity of Turkey to climate change, government should allocate some investment regarding to climate change.
In conclusion, climate change has many significant influences over world. As a country which has experiencing Mediterranean climate at many parts, Turkey is very delicate to changes in climate. Several majors including water resources, agriculture, industry, ecosystems, biodiversity, health, tourism, energy have been affected and will be affected in the future in Turkey. Thus, it is very important to conduct mitigation and adaptation studies in large and small scales. First of all, impacts of climate change on different sectors have to be understood clearly then effective solution steps can be achieved. Hence, it is significant to perform impact studies to have a better understanding of change in climate on Turkey.
3.2. Karasu Basin
Euphrates River located in the mountainous Eastern Anatolia in Turkey, is one of the major rivers within Middle Eastern countries and has a significant importance for the countries including Turkey, Syria and Iraq. Euphrates Basin is the largest basin of Turkey and it has 17% of entire country’s water potential. Ozdemir et al. (2008) stated that approximate water potential of Euphrates Basin is 30 billion m3. However, due to uncertainty of observations in Syria and Iraq, it is very difficult to give accurate value for Euphrates Basin’s total water potential. Ozdemir et al. (2008) purposed to calculate accurate water potential of Euphrates Basin and they reported 37 km3/year water 178 potential for Euphrates Basin. They explained that 31.9 km3/year of total water potential is coming from Turkey from a drainage area of 121 560 km2, 4.3 km3/ year is in Syria’s board with a drainage area of 87 300 km2, 0.8 km3/ year is in Iraq with a drainage area of 182 300 km2. In addition, Ozdemir et al. (2008) expressed that it is possible to irrigate 1 600 000 ha land area in Turkey and 800 000 ha land area in Syria by using the water of Euphrates Basin. They also stated that if Syria uses much of its potential, Iraq can have significant water problems.
Under Southeastern Anatolia Project (GAP) of Turkey, more dams and water structures are planned to be constructed in Euphrates Basin which can reduce quantity and quality of water released to downstream countries. GAP is very substantial project for socio-economic development of Southeast Anatolia region. However, more dams and more water structures for the purposes of more irrigation opportunity, more water supplies for residents and more hydropower generation can cause important political problems with downstream countries. Moreover, due to increasing number of dams, evaporation from reservoir surface will increase.
Snow is the main water source of Euphrates Basin, particularly for headwater of Euphrates Basin, which is also called Karasu Basin. Major purpose of this study is to investigate climate change impacts on snow hydrology in Euphrates Basin. However, it is very challenging and not realistic to succeed accurate modelling for entire Euphrates Basin. Hence, Karasu Basin which is one of the most snow dominated part of Euphrates Basin was selected with the aim of investigating climate change influences on snow hydrology in Turkey. Very high amount of Karasu Basin annual flow consists of snow melt runoff. Geographical location of Karasu Basin is longitudes from 38º 58’013’’E to 41o 38’28’’ E and latitudes from 39 o 23’18’’ N to 40 o 24’26’’ N. Basin location in Turkey is shown in Fig. 3.30.
179 Fig.3.30: Location map of Karasu Basin (Tekeli et al. 2005)
Karasu Basin has an area of 10215 km2. It is the most mountainous part of Euphrates Basin with elevation range from 1125 m to 3487 m. Fig. 3.31 shows the elevation map of the basin.
Fig.3.31: Karasu Basin elevation map (Sensoy et al. 2008)
180 Meteorological data in Upper Euphrates Basin was obtained from Turkish State Meteorology Service (TSMS), while flow data was provided by General Directorate of Electrical Power Resources Survey and Development Administration (EIE). In Euphrates Basin, daily meteorological data including precipitation, minimum-maximum & average air temperatures, humidity and wind speed is available between 1 January 1975 and 31 December 2008. Meteorological stations in Karasu Basin are Erzurum, Erzincan and Tercan. Moreover, station 2119, which is at the outlet of Karasu Basin, provides daily flow data for the study. Flow data is available from 1 October 1975 to 30 September 1987 and from 1 October 1994 to 31 December 2004. Flow data is not available between 1 October 1987 and 30 September 1994 as the station 2119 was closed during this period.
Meteorological stations in Karasu Basin are geographically distributed over Karasu Basin. Latitude of Erzurum Meteorological station is 39° 54’ while longitude is 41° 17’. Moreover, elevation of Erzurum station is 1869 m. Latitude of Erzincan station is 39° 45’ while longitude is 39° 30’ and elevation of Erzincan station is 1218 m. Latitude of Tercan Meteorological station is 39° 47’ while longitude is 40° 23’ and its elevation is 1425 m. Physical locations of meteorological and flow stations in Karasu Basin are shown in Fig. 3.32.
Fig. 3.32: Meteorology and flow stations in Karasu Basin
181 In Karasu Basin, minimum air temperature between 1975 and 2008 was -30 °C, while maximum temperature was 28.3 °C. Observed average air temperature was 5 °C. Maximum rainfall measured between 1975 and 2008 is 59.6 mm observed on 23 February, 2004. Basin streamflows showed variation between 12.3 and 734 m3/s. Maximum streamflow values were observed in spring seasons, when snow starts melting.
182 Chapter 4: Literature Review
Importance of climate change on snow has been realized for around last two or three decades. For instance, as initial steps, Gleick (1987), Lettenmaier and Gon (1990) indicated that a few °C increases in temperature would remarkably influence runoff timing. It may cause increased streamflow in the cool season and reduced streamflow in the warm season and it shifts peak flows towards cool season. Despite the improvements in hydrological and climate models, the basic results of past and current climate change studies on snow remains quite similar. Both observations and model based studies are in agreement that global warming caused significant influences on mountainous basins where snow is dominated water resource. For example, Hamlet et al. (2005, 2007), Mote et al. (2005) and Stewart, Cayan and Dettinger (2005) illustrated that global warming resulted important changes in spring snow accumulation and runoff timing in the western US (Adam, Hamlet &Lettenmaier 2009).
It is possible to find large number of studies which are done for the purpose of investigation climate change influences on snow hydrology in literature. Following paragraphs are including sample studies which are explaining climate change and snow hydrology relation for different regions of the world.
Andreasson et al. (2004) analysed global warming influences on hydrology of six test basins which are distributed over Sweden. They used future climate variables generated under Swedish Regional Climate Modelling Programme (SWECLIM). They used HBV hydrological model for future steamflow prediction. They stated spring flood peak decreases, very significant annual runoff volume decline in southeastern Sweden, increase in autumn and winter runoff, increases in annual runoff volumes in northern Sweden and increasing frequency of high flow events during autumn.
Majority of climate change impact studies at snow dominated basins have been using limited number of scenarios while investigating climate change influences on hydrology. Jasper et al. (2004) underlined this issue and utilized 23 regional climate 183 scenarios for the purpose of explaining possible climate change effects on hydrology of 2 Alpine river basins, the Thur basin and the Ticino basin. They reported temperature increases between 1.3 and 4.8°C & precipitation change between –11 and +11%, with substantial variability between months and catchments. They also presented strongly decreased snowpack and shortened duration of snow cover, resulting in time-shifted and reduced runoff peaks. Moreover, significant decreases in summer flows and soil- water availability, specifically at lower elevations were stated by Jasper et al. (2004).
Dibike and Coulibally (2005) implied the importance of hourly or daily climate data set for hydrological studies and they explained the deficiencies of GCMs to provide hourly or daily data. To overcome this problem, they implemented two statistical downscaling methods, a stochastic and a regression based, to generate the possible future local meteorological variables such as precipitation and temperature in the Chute-du-Diable sub-basin of the Saguenay watershed in northern Que´bec, Canada. Dibike and Coulibally (2005) stated that downscaled future climate variables were used as input to two conceptual hydrological models, the Swedish HBV-96 and the Canadian CEQUEAU for the purpose of understanding future streamflows of the basin. Both downscaling methods are in agreement on temperature increases. However, the regression based downscaling technique and stochastic weather generator are not consistent about precipitation predictions. The regression based downscaling technique concluded an increasing trend in the mean and variability of daily precipitation values, while stochastic weather generator did not result an obvious precipitation trend. In addition, Dibike and Coulibally (2005) expressed an overall increasing trend in mean annual river flow and reservoir inflow as well as earlier spring peak flows in the basin based on hydrological impact simulation results.
Barnett et al. (2005) reviewed three case studies to comprehend the magnitude of the potential regional water problems in snowmelt-dominated regions. First of them was conducted in western US by the Accelerated Climate Prediction Initiative (ACPI) in 2000. The clearest impact of climate change in this simulation was an increase in temperature projected to be between 0.8 and 1.7 °C by the middle of the 21st century.
184 However, a little or no change in precipitation was projected for the same region. The most important influence of warming was reported as a large reduction in mountain snowpack and up to one month earlier shift in peak flow by 2050. It is also implied that there is not sufficient reservoir capacity to manage ‘early water’ so most of this water will not to be used and waste to oceans. Climate change has important results on future water allocation. Reductions in future water availability will increase competition among water sectors. For instance, ACPI research showed that residents of the Columbia River system have to make a choice between substantial (10-20%) reduction in hydroelectric power generation for summer & autumn seasons and spring & summer releases for salmon runs resulting serious harm to the federally protected salmon population (Barnett et al. 2005).
The second case study explained by Barnett et al. (2005) was conducted in the Rhine River in Europe by Middelkoop et al. (2001). It is expressed that climate models projected a temperature increase of 1.0–2.4 °C by the middle of the century for this region. Hydrological models suggest that this region will change to rainfall dominated basin from combination of rainfall and snowmelt dominated basin leading increase in winter flows and decrease in summer flows. Moreover, increases in the frequency and height of peak flows, and longer & more frequent periods of low flow during the summer are expected due to warming. Climate change effects in the Rhine River will likely to result socio-economic conclusions as follows: a decrease in water availability for industry, agriculture and domestic use during the peak demand season, an increase in transportation costs because of not to load ships fully on major transport routes due to rise in the number of low-flow days, a decline in the level of flood protection, a reduction in annual hydropower generation in some parts of the region and a reduction in snow tourism incomes.
The third regional case study reported by Barnett et al. (2005) is on Canadian prairies. They explained that climate studies are in agreement on temperature increase in Canadian prairies up to 8 °C in winter time, a decrease in snowpack & summer soil moisture, and earlier snowmelt. It is high possibility to experience increase in
185 frequency and intensity of droughts owing to long low flows during summer and autumn. They reported that almost half of water is used for agricultural purposes in Canadian prairies and most agricultural water demand is met by surface water. Moreover, Barnett et al. (2005) reported that streamflows of the basin is limited and extremely variable from year to year. Thus, agriculture in Canadian prairies is very sensitive to droughts. Furthermore, early snow melting and changing runoff patterns due to warming climate can be an additional disadvantage for the agriculture of Canadian prairies.
Kleinn et al. (2005) investigated streamflow responses to climate variations and anthropogenic climate change. They defined a model chain which was developed for the Rhine basin upstream of Cologne, a 145,000 km2 river basin in Central Europe north of the Alps. They used a regional climate model at grid spacing of 56 and 14 km. Moreover, Kleinn et al. (2005) utilized a distributed hydrological model with a grid spacing of 1 km. It is one-way nested into the RCM through a downscaling interface. They explained the process and findings of the study as follows:
Biases in precipitation and temperature are accounted for by catchment- dependent but seasonally constant correction factors. Apart from these bias corrections, the hydrological model is forced by hourly RCM data. In the evaluation we compare a 5-year integration driven by observed lateral boundary conditions (ECMWF reanalysis) against daily analysis of high-density rain gauge data and streamflow data. The regional climate model is found to qualitatively reproduce the main mesoscale precipitation patterns and their seasonal evolution. Systematic biases are, however, found in the distribution of precipitation with topographic height in the Alpine region and at adjacent hill ranges. The RCM also reproduces intercatchment variations in the frequency distribution of daily precipitation. Simulated runoff resembles closely the mean annual cycle, and daily runoff agrees well with observations in timing and amplitude of runoff events for lowland gauges. Larger model errors are found for high-altitude Alpine catchments. The 14-km RCM provides much finer and
186 more realistic precipitation fields compared to the 56-km RCM, but these improvements did not have a significant impact on the skill of the hydrological model to simulate streamflow. The model chain was found to reproduce observed month-to-month variations of basin-mean winter precipitation and streamflow with correlations between 0.85 and 0.95. This result provides confidence that the model chain is able to represent key processes related to streamflow variations in response to climate variations and climate change (Klein et al. 2005, p.1).
Horton et al. (2006) also implemented multiple models and scenarios approach as similar to Jasper et al. (2004). Horton et al. (2006) searched climate change impacts on the hydrology of 11 mountainous catchments in the Swiss Alps. By using 19 RCMs and 3 GCMs, future climate projections were generated and they are used in a conceptual reservoir-based precipitation-runoff transformation model called GSM-SOCONT to predict basins’ responds to changing climate. Major findings of their study are earlier snowmelt resulting changes in runoff regimes and substantial annual runoff decrease.
Menzel et al. (2006) purposed to achieve impact analysis of global climate change on regional hydrology with special interest on discharge conditions and floods. They performed their study in the major part of the German Rhine catchment with a drainage area of approximately 110 000 km2. They divided study area into 23 sub- catchments. Firstly, Menzel et al. (2006) used HBV-D hydrological model to simulate present runoff conditions by using present climate variables. Then, by using two different GCM driven by business as usual emission scenario, they achieved hydrological impact simulations. Menzel et al. (2006) utilized ECHAM4/OPYC3 model of the Max Planck Institute for Meteorology and the HadCM3 model of the Hadley Centre for Climate Prediction and Research GCMs with the aim of determining future climate. Due to the fact that GCMs outputs’ spatial resolution is too coarse to use in hydrological model, they downscaled GCMs outputs by the technique of expanded downscaling (EDS). They expressed the most important advantage of EDS in comparison to other regression methods as the capability of modelling extreme
187 precipitation. Menzel et al. (2006) found potential increase in precipitation, mean runoff and flood discharge for small return intervals. Nonetheless, they emphasized the uncertainty chain ranging from GCM to hydrological model and they end up with importance of reducing uncertainties of climate change impact studies.
Mall et al. (2006) stated that although there are large number of studies on climate change and hydrology relation, only a few studies were conducted for this purpose in India. Thus, they reviewed studies which are done on hydrological impacts of possible climate change for Indian regions/basins. Some important points of their review as follows: The effect of climate change using different climatic scenarios on snow-water equivalent, snowmelt run-off, glacier melt run-off and total stream flow and their distribution is examined for Spiti River, which is a high altitude Himalayan river located in the western Himalayan region. It is found that annual snowmelt run-off, glacier melt run-off and total stream flow increase linearly with changes in temperature (1–3°C), but the most prominent effect of increase in temperature has been noticed on glacier melt run-off. Gosain and Rao (2003) projected that the quantity of surface run-off due to climate change would vary across the river basins as well as sub-basins in India. However, there is general reduction in the quantity of the available run-off. An increase in precipitation in the Mahanadi, Brahimani, Ganga, Godavari and Cauvery is projected under climate change scenario; however, the corresponding total run-off for all these basins does not increase. This may be due to increase in ET on account of increased temperature or variation in the distribution of rainfall. In the remaining basins, a decrease in precipitation was noticed. Sabarmati and Luni basins show drastic decrease in precipitation and consequent decrease of total run-off to the tune of two-thirds of the prevailing run-off. This may lead to severe drought conditions in future. The analysis has revealed that climate change scenario may deteriorate the condition in terms of severity of droughts and intensity of floods in various parts of the country (Mall et al. 2006, p. 1620- 1623).
188 Chen et al. (2006) studied hydrological response to climate change in the Tarim River Basin by analysing the hydro-meteorological data of the past 50 years. By using both parametric and non-parametric tests, they performed long-term trend analysis of the hydrological time-series, including air temperature, precipitation, and streamflow. Moreover, they investigated association between streamflow and climate change by using grey correlation analysis. Chen et al. (2006) reported 5% monotonic increase increase in air temperature. They also stated a significant precipitation decrease in the 1970s and a major precipitation increase in the 1980s and 1990s. They reported a step change around 1986 for both temperature and precipitation “with mean temperature and precipitation increasing from 6.7 °C and 146 mm before 1986 to 7.3 °C and 180 mm respectively after 1986”. Chen et al. (2006) expressed around 1 °C temperature increase over the past 50 years very likely due to global climate change. Streamflows of Aksu River and the Yarkant River have demonstrated statistically significant increasing trend. Chen et al. (2006) explained that Aksu River streamflow increase is more important than Yarkant River. They stated the coefficients of streamflow increase in the Aksu and Yarkant Rivers as 0.41 and 0.13 respectively. Finally, Chen et al. (2006) explained based on grey correlation analysis that for Aksu River, which is located in the northwest of the basin, precipitation is more dominant than temperature on streamflows. Nonetheless, in the Hotan River, which is located in the southwest of the basin, temperature is more dominant on streamflow.
Merritt et al. (2006) developed climate change scenarios for the Okanagan Basin, a snow-driven semi-arid basin located in the southern interior region of British Columbia. They used three global climate models (GCMs) with high (A2) and low (B2) emission scenarios. GCMs which are used by Merritt et al. (2006) were the CGCM2, the Australian developed CSIROMk2 and the HadCM3 model developed at the Hadley Centre in the United Kingdom. They selected three time periods for simulations. Time periods are 2010–2039 (2020s), 2040–2069 (2050s) and 2070–2099 (2080s). Merritt et al. (2006) presented results of study as follows:
189 An increase in winter temperature of 1.5–4.0 °C and a precipitation increase of the order of 5-20% is predicted by the 2050s. Modelled summer precipitation is more variable with predicted change ranging from zero to a 35% decrease depending on the GCM and emission scenario. Summer temperatures were simulated to increase by approximately 2–4 °C. The UBC watershed model was used to model the hydrologic response of gauged sub watersheds in the basin under the altered climates. All scenarios consistently predicted an early onset of the spring snowmelt, a tendency towards a more rainfall dominated hydrograph and considerable reductions in the annual and spring flow volumes in the 2050s and 2080s. Of the three climate models, the CGCM2 model provided the most conservative predictions of the impacts of climate change in Okanagan Basin. Simulations based on the CSIROMk2 climate model suggested greatly reduced snowpack and flow volumes despite a sizeable increase in the winter precipitation. The scenarios raise questions over the availability of future water resources in the Okanagan Basin, particularly as extended periods of low flows into upland reservoirs are likely to coincide with increased demand from agricultural and domestic water users (Merritt et al. 2006, p. 79).
Nohara et al. (2006) investigated 24 major rivers’ discharge projections over world during the twenty-first century by using 19 coupled atmosphere–ocean general circulation models based on the A1B scenario. They used a weighted ensemble mean (WEM) to decrease model bias and uncertainty. They stated that WEM is very beneficial to generate accurate reproduction for most rivers, except those impacted by human-induced water usage. Nohara et al. (2006) expressed increase in annual mean precipitation, evaporation, and runoff in high latitudes of the Northern Hemisphere, southern to eastern Asia, and central Africa at the end of the twenty-first century. On the other hand, same variables are expected to decrease in the Mediterranean region, southern Africa, southern North America, and Central America. They stated discharge increases for high-latitude rivers such as Amur, Lena, MacKenzie, Ob, Yenisei, and Yukon and peak flows shifts to earlier dates due to temperature increases induced early snow melting. In addition, according to their findings, discharges of rivers in
190 Europe to the Mediterranean region (Danube, Euphrates, and Rhine) and southern United Sates (Rio Grande) are expected to decrease.
Christensen et al. (2007) investigated the potential effects of climate change on the hydrology and water resources of the Colorado River basin by comparing simulated hydrologic and water resources scenarios obtained from downscaled climate simulations of the U.S. Department of Energy/National Center for Atmospheric Research Parallel Climate Model (PCM) to scenarios driven by observed historical (1950–1999) climate. Downscaled temperature and precipitation outputs were extracted from PCM simulations, and were preceded in the Variable Infiltration Capacity (VIC) macroscale hydrology model to generate hydrological responses of basin to climate change. Results for the business as usual scenario are summarized into Periods 1, 2, and 3 (2010–2039, 2040–2069, 2070–2098). Average annual temperature increases of 1.0, 1.7 and 2.4 °C were reported for periods 1-3 by Christensen et al.
(2007) for the Colorado River Basin relative to the historical climate. Furthermore, they reported decreases in precipitation by 1% for the control climate relative to observed historical climate and they projected decrease in precipitation by 3, 6 and 3% for future periods 1–3, respectively. They projected 10% decline in annual runoff in the control run and they modelled annual runoff decrease of 14, 18 17% for periods 1–3, respectively. Christensen et al. (2007) also performed analysis of water management operations using a water management model driven by simulated streamflows and they detected 7% total basin storage decrease for the control climate and 36, 32 and 40% decrease for periods 1–3, respectively.
Hagg et al. (2007) aimed to estimate how water availability will change under climate change conditions as using GISS GCM future climate outputs in two simple conceptual hydrological models, HBV-ETH and OEZ for three test sites ‘‘Ala Archa’’, ‘‘Abramov’’ and ‘‘Oigaing’’. They constituted baseline scenario by using data from 10 stations between 1951 and 1980 then they tested performance of the following Global Circulation Models (GCMs): GFDL and GFDL-T (Geophysical Fluid Dynamics Laboratory, University of Princeton), UKMO (Meteorological Agency of the UK), CCC (Canadian
191 Climate Center) and GISS (Goddard Institute for Space Studies) by using baseline climate. Hagg et al. (2007) performed regional climate change scenarios by using linear regression approach. Among the climate models, GISS model showed the best agreement to observed temperatures so its output was selected with the aim of hydrological modelling. GISS model resulted temperature increase of 4.2 °C and the precipitation increase by 17%. Increase in spring and summer runoff found by Hagg et al. (2007) is not in agreement of most climate change impact studies’ consequences at snow dominated basins. Most studies showed that global warming will result decrease in spring and especially summer flows. However, most likely due to precipitation increase in mentioned seasons, Hag et al. (2007) projected increasing future spring and summer flows.
Novotny and Stefan (2007) studied on streamflow records of five major river basins of Minnesota from 36 USGS gauging stations up to the year 2002. They analysed seven annual stream flow statistics: mean annual flow, 7-day low flow in winter, 7-day low flow in summer, peak flow due to snow melt runoff, peak flow due to rainfall as well as high and extreme flow days. Novotny and Stefan (2007) corrected time series data for serial correlation by using Trend Free Pre-Whitening (TFPW) method and they performed trend analysis by using Mann–Kendal non-parametric test over time windows from 90 to 10 years. They reported streamflow variations over the period of observation record and they stated that river basins showed significantly different trends. Novotny and Stefan (2007) detected periodicity in the trends of Red River of the North, the Mississippi River, and the Minnesota River basins. However, they did not find any statistically significant trend in peak flows due to snow melt. On the other hand, they found increases in peak flows due to rainfall events in the summer season. In addition, they reported increase in number of days with higher flows. They also found increase in wetter summers and more frequent snow melt events due to warmer winters. Novotny and Stefan (2007) implied increases in intense precipitation events, more days with precipitation and earlier and more frequent snowmelt events. They suggested flooding risks owing to increasing rainfall events, however, they did not report any flooding risk because of snowmelt. According to results of their study,
192 they stated possible water quality increase due to higher summer and winter base flows.
Huntington, Sheffield and Hayhoe (2007) investigated climate change impacts on precipitation, snow melt regime, surface runoff, and infiltration in the north eastern US by using statistically downscaled future temperature and precipitation projections which are generated in climate models. They used future climate predictions as an input into Variable Infiltration Capacity (VIC) model. They stated that increase in air temperature for the period of 2070-2099 will be 2.9 ° C in low emission scenario while surface air temperature will show 5.9°C increase in high emission scenario. Huntington, Sheffield and Hayhoe (2007) also found increasing winter precipitation, earlier snowmelt, and a shift in snow melting time lead to increases in surface runoff, infiltration, and subsurface flow during the winter.
Chang et al. (2007) assessed the freshwater vulnerability for five major Korean river basins for 2015 and 2030 under changing climate and population growth. They used a regional climate model, MM5, based on the IPCC SRES A2 scenario. They utilized US Geological Survey’s Precipitation Rainfall Simulation Model (PRMS) to achieve hydrological modelling. Moreover, they used population and industrial growth scenarios for impact assessment. They found increase in mean annual temperatures by 1.6°C by 2015 and 2.2°C by 2030, respectively. They also reported that precipitation change shows variation over different parts of South Korea. They projected wetter climate for northern central regions and drier climate for southern regions. According to study results, they revealed that freshwater is more vulnerable to population growth than the climate change. They stated that climate change alone could result decrease in mean annual runoff by 10% in four major river basins by 2030.
Kim et al. (2007) investigated climate change impact on the runoff and water resources of Yongdam Dam, Korea. They used YONU GCM to predict future climate and they statistically downscaled GCM outputs to get better spatial resolution for the purpose of using them for hydrological impact study. They selected semi-distributed
193 hydrological model named SLURP to simulate the streamflows. By applying SLURP model to Yongdam dam which is located in the southern Korea, Kim et al. (2007) projected 7.6% annual mean runoff decrease. According to seasonal analysis, they reported streamflow increases in winter & autumn seasons and they projected decreasing streamflows in summer season.
Maurer (2007) performed a hydrologic model by using climate data obtained from 11 GCMs under two emissions scenarios (the higher emission SRES A2 and the lower emission SRES B1). He stated his findings as follows:
There are highly significant average temperature increases by 2071–2100 of 3.7°C under A2 and 2.4°C under B1; July increases are 5°C for A2 and 3°C for B1. Two high confidence hydrologic impacts are increasing winter streamflow and decreasing late spring and summer flow. Less snow at the end of winter is a confident projection, as is earlier arrival of the annual flow volume, which has important implications on California water management. The two emissions pathways show some differing impacts with high confidence: the degree of warming expected, the amount of decline in summer low flows, the shift to earlier streamflow timing, and the decline in end-of-winter snow pack, with more extreme impacts under higher emissions in all cases. This indicates that future emissions scenarios play a significant role in the degree of impacts to water resources in California (Maurer 2007, p. 309).
Baltas (2007) evaluated climate change impact on hydrological regimes and water resources in the basin of Siatista, a sub-basin of the Aliakmon river basin in Northern Greece. He used monthly conceptual water balance model for hydrological modelling purpose after calibrating it by using historical hydro-meteorological data. He determined changes in streamflow runoff under two different equilibrium scenarios (UKHI, CCC) for the years 2020, 2050 and 2100. He reported a future decrease in mean winter runoffs and very significant summer runoff reductions. In addition, Baltas (2007) stated an increase in maximum annual runoff and a decrease in minimum
194 annual runoff values. Moreover, he reported an increase in potential and actual evapotranspiration which lead soil moisture reduction. Furhermore, Baltas (2007) explained decreases in snowpack and early melting owing to temperature increases. Baltas (2007) expressed spring flow decreases based on snowpack decreases and early snow melting. He also explained a shift of the wet period towards December and increase duration and intensity of dry periods.
Jiang et al. (2007) implied a significant issue that different GCMs usage and basins’ response to changing climate were extensively discussed in literature. However, different hydrological responses of distinct hydrological models to the climatic scenarios have had less attention. Thus, Jiang et al. (2007) investigated influences of changing climate on the water availability in the Dongjiang basin, South China, using six different monthly water balance models, namely the Thornthwaite–Mather I, Vrije Universitet Brussel (VUB), Xinanjiang (XAJ), Guo (GM), WatBal (WM) and Schaake (SM) models. The ability of the six models in simulating the present climate water balance components is first tested and then hydrological impact simulations were achieved by each models and compared. The results of analysis was stated by Jiang et al. (2007) as follows: “(1) all six conceptual models have similar capabilities in reproducing historical water balance components; (2) greater differences in the model results occur when the models are used to simulate the hydrological impact of the postulated climate changes; and (3) a model without a threshold in soil moisture simulation results in greater changes in model-predicted soil moisture with respect to alternative climates than the models with a threshold soil moisture” (Jiang et al. 2007, p. 316-317). The study provides insights into the plausible changes in basin hydrology due to climate change. It shows that there can be significant implications for the investigation of response strategies for water supply and flood control due to climate change.
Elgaali and Garcia (2007) used Artificial Neural Network (ANN) methodology to investigate climate change impacts on monthly and seasonal surface water supplies in Colorado’s Arkansas River Basin under two transient climate change scenarios, the HAD from the Hadley Centre for Climate Prediction and Research and the CCC from the Canadian Climate Centre. They reported that HAD and CCC are not in agreement on 195 available water for irrigation due to large degree of uncertainty coming from future climate estimation. While HAD scenario suggests increase in available water for irrigation, CCC projects constant water shortages in the region and decreased water available for irrigation in almost every month.
Shi et al. (2007) studied on impacts of climate change on northwest China’s hydrology. Similar to current study, they expressed influence of changing climate for both current and future climate. They reviewed the papers regarding to future climate change impacts on northwest China. Shi et al. (2007) explained the conclusion of their study as follows:
Based on an analysis of the hydrological and meteorological data base, basic circulation patterns and 196 modelling studies we think that climate change from warm-dry to warm-wet started in 1987 in northwest China. Eight facts support this thesis: (1) A continuously rising air temperature with an annual temperature 0.7 °C higher from 1987–2000 compared to 1961– 1986. (2) A notable increase of precipitation with annual precipitation 10–30% higher from 1987–2000 compared to 1961–1986. (3) Glacier retreat and glacial melt water increased. Glacier area is reduced by 1400 km2 from 1960 to 1995 and the annual runoff of glacial melt water was 84.2% higher in 1985–2001 than during 1958–1985 at glacier of the Urumqi River, Tianshan. (4) Increased of river runoff. The annual total runoff of the Xinjiang area was 7% higher in 1987–2000 than in 1956–1986. (5) Water level rise and area expansion of the inland lakes. The Bosten Lake in the central Tianshan Mountains showed a descending trend from 1955 on which reversed to rise since 1977. The lake area grew by more than 1000 km2. (6) Frequency of flood disasters largely increased. Extraordinary flood occurred 7 times during 1956–1986 and 21 times from 1982–2000 in Xinjiang area. (7) Vegetation cover increases in the west and north Xinjiang, linked with a notable increase in precipitation and with better management on water usage in several oasis (8) Less sand dust storm days. In northern part of China, the 68 and 89 days with strong and extraordinary sand-dust storms in
196 the 1960s and 1970s were reduced to 47 in the 1980s and 36 in the 1990s (Shi et al. 2007, p. 391).
Shi et al. (2007) classified the climate change discussed above into three types: “(1) The notably changed region covers largely northern part of Xinjiang and west part of Gansu, and small part of Qinghai. (2) The slightly changed region covers the Taklamakan Desert, the borderland of Xinjiang, Gansu and Qinghai and (3) the unchanged region covers mainly the east part of northwest China” (Shi et al. 2007, p. 391). They used a regional climate model RegCM2 nested with a global coupled ocean- atmospheric model from CSIRO to predict the climate under doubling of CO2 in northwest China. They reported annual air temperature increase by 2.7 °C and annual precipitation increase by 25%. Shi et al. (2007) stated that due to the complexity of climate change and large area of extensive deserts and high mountains without meteorological observations, future climate projections have large uncertainties. Finally, they indicated next aim as reduction of uncertainties in climate studies. They recommended using water resources and energy more efficient, to construct more reservoirs in the alpine area for adapting to climate change in Northeast China.
Thodsen (2007) studied impact of climate change on river discharges of five major Danish rivers. He divided his focus area into 29 sub-catchments and he investigated future river discharges for the period of 2071-2100. He used HIRHAM regional climate model with the aim of climate change modelling and he used IPCC A2 greenhouse emission scenario. Thodsen (2007) implemented NAM rainfall runoff model to convert precipitation to river discharges. Similar to most climate change studies, he used an interface to adjust regional climate model outputs before applying them into NAM model. Thodsen (2007) projected an increase of 7% in mean annual precipitation, potential evapotranspiration increase by 3% and river discharge increase by 12% on average between a control period (1961–1990) and the future period. Owing to precipitation increases from October to March and precipitation decreases from July to September, the monthly river discharges are projected to increase from December to August and decrease in September and October. Moreover, he expressed that the
197 amount of precipitation exceeded 0.1% of all days increases by 7% and the river discharge exceeded 0.1% of all days increases approximately 15%.
Steele-Dunne et al. (2008) aimed to estimate climate change impacts on Ireland’s hydrology by applying hydrological impact study on 9 Irish catchments. They modelled future climate by using the Rossby Centre Atmosphere Model (RCA3) regional climate model which is forced by the European Centre Hamburg Model Version 5 (ECHAM 5) general circulation model. Then, Steele-Dunne et al. (2008) used dynamically downscaled climate variables as input to HBV-Light conceptual rainfall-runoff model for the purpose of modelling streamflows for reference period (1961–2000) and for the future (2021–2060) under the Special Report on Emissions Scenarios (SRES) A1B scenario. Steele-Dunne et al. (2008) presented winter flow increases and summer flow decreases between 20% and 60% for Irish catchments.
Hlavcova et al. (2008) conducted a study to predict the potential impact of climate change on the mean monthly runoff in the upper Hron River basin which is a mountainous region in Central Slovakia. They calibrated a conceptual hydrological balance model, developed at the Slovak University of Technology, with data from the period 1971–2000 and used it for modelling changes in runoff with monthly time steps. Future predictions of climate variables were explained by using two different climate change scenarios developed within the framework of the CECILIA (Central and Eastern Europe Climate Change Impact and Vulnerability Assessment) project. ECHAM4/OPYC3 and HadCM2 GCMs were used to determine future climate. The runoff changes at the selected basin in the future time horizons of 2025, 2050 and 2100 show alterations in the runoff distribution within a year. Furthermore, Hlavcova et al. (2008) compared changes in the seasonal runoff distribution with previous results, which were obtained with climate change scenarios developed from the outputs of the CCCM97 and GISS98 global circulation models. They projected monthly mean runoff changes for spring , summer and autumn seasons changing between 4% and 50%. Furthermore, they projected runoff increases in winter season varying between 32% and 54%.
198 Stahl et al. (2008) studied on sensitivity of streamlow to climate and glacier cover changes in Bridge River basin, British Columbia. They used a semi-distributed conceptual hydrological model coupled with a glacier response model. Stahl et al. (2008) utilized mass balance data to constrain model parameters. They reported significant decrease in glacier area and summer streamflow even under the steady climate scenario.
Li et al. (2008) studied the climate change influence on runoffs in the head region of the Yellow River. Future climate data from seven GCMs involving CCCma, CCSR, CSIRO, ECHAM4, GFDL, Hadley, and NCAR were downscaled and used in distributed model to comprehend impacts of climate change on runoffs in Yellow River basin. They selected A2 and B2 emission scenarios for model application. Finally, Li et al. (2008) indicated that amount of runoff will change slightly until 2020 and then it will reduce approximately 5% per year.
Marshall and Randhir (2008) used a continuous simulation model to investigate the hydrology of the Connecticut River Watershed which has an area of approximately 28 500 km2. They used SWAT model as a hydrological model and they assessed the outputs of two IPCC GCM, MRI2 and CCSR Nies2, as an input to hydrological model. Marshall and Randhir (2008) explained that:
The patterns and fluxes of water in the study watershed were found to be significantly influenced by snowmelt and evaporation. Watershed processes change under climate scenarios by impacting the quality and quantity of water throughout the study period through decreases in annual stream flow, increase in winter and spring runoff, and increases in annual sediment loading. The simulations of climate change for high and low temperature regimes showed impacts on timing and magnitude of water and sediment yield. The variability in surface runoff and sediment loading decreased under both the warming scenarios, while the variability in water yield and evapotranspiration increased
199 during low warming scenario and decreased for the high warming scenario. The simulation results show that climate change can have significant impacts on the quantity of water available throughout the year. Climate change decreased water storage during the winter months because of a decline in snowfall and snow pack volume. This impacted surface runoff rates. The change in water availability can reduce river flows during periods of high water demand. This can place severe strain on spring anadramous fish runs and to peak demand for hydroelectric power during summer months. Climate change had significant impacts on water quality in the study watershed. Sediment loading increased from June through October by up to 50%, while the volume of receiving waters decreased by up to 19%. Climate change impacted nutrient cycles and the N: P ratio of annual loading in the watershed that may lead to increased algal and plant biomasses. The climate-induced changes in the timing and magnitude of nutrient loads have implications on the watershed system; under both climate change scenarios, the study watershed is more nitrogen limited, and is faces the higher risk of eutrophication (Marshal & Randhir 2008, p. 277-278).
Rauscher et al. (2008) used a high-resolution nested climate model for the purpose of investigating future changes in snowmelt runoff over the western US. They reported approximately 3 to 5°C temperature increase which can lead occurrence of runoffs as much as two months earlier than present. They discussed consequences of earlier runoffs as influences on water storage in reservoirs and hydroelectric generation, with substantial results for land use, agriculture, and water management in the American West.
Fujihara et al. (2008) investigated influences of climate change on hydrology and water resources of the Seyhan River Basin in Turkey. They expressed the study and their findings as follows:
A dynamical downscaling method, referred to as the pseudo global warming method (PGWM), was used to connect the outputs of general circulation
200 models (GCMs) and river basin hydrologic models. The GCMs used in this study were MRI-CGCM2 and CCSR/NIES/FRCGC-MIROC under the SRES A2 scenario, and the downscaled data covered two 10-year time slices corresponding to the present (1990s) and future (2070s). The hydrologic models along with a reservoir model were driven using the downscaled data for the present period. As a result, the temperature and precipitation, which were dynamically downscaled through bias-correction, were in good agreement with the observed data. The hydrologic simulation results also matched the observed flow, reservoir volume, and dam discharge. Therefore, we concluded that the PGWM combined with bias-correction is extremely useful to produce input data for hydrologic simulations. The simulation results for the future were compared with those for the present. The average annual temperature changes in the future relative to the present were projected to be +2.0 °C and +2.7 °C by MRI and CCSR, respectively. The annual precipitation decreased by 157 mm (25%) in MRI-future and by 182 mm (29%) in CCSR-future, and the annual evapotranspiration decreased by 36 mm (9%) in MRI-future and by 39 mm (10%) in CCSR-future; the annual runoff decreased by 118 mm (52%) in MRI- future and by 139 mm (61%) in CCSR-future. The analysis of water resource systems was conducted by using a simple scenario approach to take into account changes in water use. This analysis indicated that despite the impacts of climate change, water scarcity will not occur in the future if water demand does not increase. However, if the irrigated area is expanded in the future under the expectation of current flow, water scarcity will occur due to the combination of decreased inflow and increased water demand. Thus, in the Seyhan River Basin, water use and management will play more important roles than climate change in controlling future water resource conditions (Fujihara et al. 2008, p. 33-34).
Adam, Hamlet and Lettenmaier (2009) expressed the impacts of climate change on snow hydrology at global scale. Firstly, they investigated global warming effects for snow dominated regions of US, western and northern east US. Then they generalized the findings for northern high latitudes and Alaska. Adam, Hamlet and Lettenmaier 201 (2009) presented some literature review on snow studies performed in US. For example, Mote (2003) stated significant downward trend for SWE based on the time- series analysis of 1 April snow water equivalent (SWE) measurements from 230 manual snow courses in the Pacific northwest (defined as the states of Washington, Oregon, Idaho, and Montana, west of the Continental Divide, and southern British Columbia) for the period 1950–2000. Mote et al. (2005) improved the analysis of Mote (2003) by using combination of modelling and data analysis for the period 1916–2003. They used VIC model for modelling purposes to simulate SWE over the entire western US. Analogously to Mote (2003), they found a general downward trend in SWE over most of the region. Hamlet et al. (2005) extended Mote et al. (2005) study by demonstrating that strong elevational gradients in observed SWE trends can be explained by temperature, rather than precipitation changes. Barnett et al. (2008) presented similar results to Mote (2003). Moreover, Barnett et al. (2008) explained by using a climate fingerprinting technique that up to 60% of the observed winter trends in streamflow, winter air temperature, and SWE are due to human activities. Stewart, Cayan and Dettinger (2005) investigated changes in the spring snowmelt runoff timing in the western US. They stated that the timing of the centre (CT) of mass of annual runoff has moved towards cool season (earlier dates) over the last half century at snow dominated basins. Burns, Klaus and McHale (2007) expressed that peak runoff from 1952 to 2005 in the Catskills, which is the source of New York City’s water supply, shifted from early April to late March. Hodgkins, Dudley and Huntington (2003) and Hodgkins and Dudley (2006) reported that spring snowmelt runoff has shifted 1–2 weeks earlier, with most of the change occurring in the last 30 years in New England (Adam, Hamlet &Lettenmaier 2009).
Van Pelt et al. (2009) underlined the significance of precipitation input in hydrological modelling and aimed to investigate influence of selecting appropriate precipitation bias correction method to achieve accurate hydrological impact studies. They implemented bias correction methods to outputs of RCM called Regional Atmospheric Climate Model (RACMO2) which was forced by ECHAM5/MPIOM under the condition of the SRES-A1B emission scenario, with a 25 km horizontal resolution. The first bias
202 correction method implemented by Van Pelt et al. (2009) corrects for the wet day fraction and wet day average (WD bias correction) and the second method corrects for the mean and coefficient of variance (MV bias correction). They stated that despite initial performance of the WD bias correction is good and corrects well for the average, it removed too many successive precipitation days. On the other hand, MV bias correction showed lower performance on average bias correction but the temporal precipitation of MV bias correction is better than WD bias correction. Then, Van Pelt et al. (2009) performed hydrological predictions by using HBV-96 hydrological model. They used the uncorrected RACMO2 output, the WD bias corrected output and the MV bias corrected output as an input to hydrological model to understand how different bias correction approaches impact hydrological respond of basin and they concluded a large difference between the simulated discharge which shows the importance of an appropriate bias correction.
Akhtar, Ahmad and Booij (2009) used meteorological station observations and results of the PRECIS (Providing Regional Climate for Impact Studies) RCM, which was driven by the outputs of reanalysis ERA 40 data and HadAM3P general circulation model (GCM), as input into the HBV hydrological model. They defined the objective of their study to conduct hydrological impact study at three river basins in the Hindukush- Karakorum-Himalaya (HKH) region by using PRECIS RCM precipitation and temperature output in six HBV model types: HBV-Met, HBV-ERA, HBV-Had, HBV-MetCRU−corrected, HBV-ERABenchmark and HBV-HadBenchmark. They state that according to the calibration and validation results of the HBV model experiments, the performance of HBV-Met is better than other HBV models driven by other data sources. More discussion regarding to HBV models are expressed by Akhtar, Ahmad and Booij (2009) as follows:
Using input data series from sources different from the data used in the model calibration shows that HBV-Had is more efficient than other models and HBV- Met has the least absolute relative error with respect to all other models. The uncertainties are higher in least efficient models (i.e. HBV-MetCRU−corrected
203 and HBV-ERABenchmark) where the model parameters are also unrealistic. In terms of both robustness and uncertainty ranges the HBV models calibrated with PRECIS output performed better than other calibrated models except for HBV-Met which has shown a higher robustness. This suggests that in data sparse regions such as the HKH region data from regional climate models may be used as input in hydrological models for climate scenarios studies (Akhtar, Ahmad & Booij 2009, p. 1075).
Ma et al. (2010) assessed the impact of climate change on snowfall in Japan. They used outputs of RCM named WRF as an input to SVAT&HYCY one dimensional hydrological model and performed hydrological simulation in the Agano River basin. The future hydrological response to climate change in the 2070s was evaluated by using a pseudo- global-warming method. Ma et al. (2010) projected discharge increases by approximately 43% in January and 55% in February, but to decrease by approximately 38% in April and 32% in May. Moreover, they reported that the flood peak in the hydrograph was shifted earlier by approximately one month, changing from April in the 1990s to March in the 2070s. Furthermore, they stated that 10-year average snowfall amount will decrease approximately 49.5% in the 2070s relative to the 1990s.
Elsner et al. (2010) investigated influences of changing climate on hydrology and water resources of the Pacific Northwest (PNW). They reported that PNW is very sensitive to climate change because seasonal runoff is highly depend on snowmelt from cool season mountain snowpack and temperature changes influence the balance of precipitation falling as rain and snow. They stated the future climate variable predictions based on results from 39 global simulations performed for the fourth assessment report of the Intergovernmental Panel on Climate Change (IPCC, 2007). An increase of approximately 0.3°C per decade over the 21st century are projected for PNW temperatures while annual mean precipitation changes are predicted to be modest with an increase of 1% by the 2020s and 2% by the 2040s. Elsner et al. (2010) used results from 20 global climate models (GCMs) and two emissions scenarios from the Special Report on Emissions Scenarios (SRES), A1B and B1 to simulate 21st century
204 hydrology of PNW regions including Washington, Columbia River basin, the Yakima River basin, and those Puget Sound river basins. They performed hydrological simulations by using two hydrological models, DHSVM and VIC, and they assessed projected changes in snow water equivalent, seasonal soil moisture and runoff for the entire state and case study watersheds for A1B and B1 SRES emissions scenarios for the 2020s, 2040s, and 2080s. Moreover, they investigated future projected changes in seasonal streamflow in Washington. They found reductions in April 1 snow water equivalent (SWE) by an average of around 27-29% across the State by the 2020s, 37- 44% by the 2040s and 53-65% by the 2080s, based on the B1 and A1B scenarios, respectively. April 1 SWE is projected to almost disappear by the 2080s in three watersheds west of the Cascade crest. Moreover, they detected substantial seasonal streamflow timing shifts in both snowmelt dominant and rain-snow mixed basins. Finally, they reported increase of annual runoff across the State by 0-2% by the 2020s, 2-3% by the 2040s, and 4-6% by the 2080s and they expressed the increase with projected winter precipitation.
Yuanqing et al. (2010) applied trend analysis to climate variables and streamflow of Lijang basin. They used hydro-meteorological data from 1979 to 2006. They purposed to follow fingerprints of climate change by implementing trend analysis in Lijang basin. They used linear approach to detect trends which is not recommended by scientists due to non-distributed structure of hydro-meteorological data. Yuanqing et al. (2010) stated that temperature has risen significantly from 1979 to 2006 in Lijang basin. Moreover, they revealed that winter temperature increases are larger than the other seasons. Moreover, they detected precipitation increases in spring and summer seasons. On the other hand, they reported decreasing precipitation trend in autumn and winter seasons. Furthermore, Yuanqing et al. (2010) detected significant upward streamflow trends at Yanggong River. Increase is specifically significant in spring season. Among four seasons, the least increase in streamflow is determined for autumn season.
205 Methodology and findings of climate change impact studies at snow dominated basins are very similar. It is possible to increase number of studies which investigated climate change influences at snow dominated basins. However, it will only result repeating same consequences. In summary, climate change will lead early snow melting, changes in hydrological regimes, changes in total runoff volume, increasing frequency in floods and droughts due to early melting.
Moreover, increasing water demand will constitute a competition among domestic water, hydroelectric power generation, agriculture and industry sectors. According to many studies summarized above and other climate & hydrology studies, it will be challenging to meet future water demands under future climate conditions for many parts of the world. The larger populations and economies will face to larger impacts. Expected results of climate change on snowmelt dominated regions are generally temperature driven not precipitation driven and all climate models are in agreement on a warmer future climate. Another important factor impacting water availability is insufficient reservoir storage to manage the shifts in seasonal streamflow cycle. Precautions against warming have to be taken very urgently because warming is already started according to observations.
206 Chapter 5: Methodology
5.1. Trend Analysis Methods
Time series of hydrological data are widely used for water resources design and management. In literature, there are several statistical tools to evaluate trends in hydrological or other time series data (Birsan et al. 2005; Novotny & Stefan 2007; St. George 2007; Jiang, Su & Hartmann 2007; Caloiero et al. 2009; Gautam, Acharya & Tuladhar 2010). However, statistical tools to detect time series trends are grouped in two: parametric and non-parametric trend analysis (Bouza-deano, Ternero-Rodriquez & Fernandez-Espinosa 2008). The main reason of preference of non-parametric tests over parametric tests is it’s suitability for non-normally distributed data and censored data, which are very frequent in hydro-meteorological time series (Yue, Pilon & Cavadias 2002). In hydrological and meteorological studies two most commonly used non-parametric trend tests are Mann-Kendal and Spearman’s Rho trend analysis. In this study, these tests were selected and applied to detect trends of hydro- meteorological time series data in Karasu Basin.
5.1.1. Mann-Kendal Non-Parametric Trend Analysis
The rank-based Mann-Kendall (M-K) trend analysis was applied to meteorological and streamflow time series to determine if there is a significant changing trend in climate variables. M-K is a popular non-parametric trend analysis method and commonly used in environmental studies (Onoz & Bayazit 2003; Demir et al. 2008; Tayanc et al. 2009; Mohsin & Gough 2009).
The formula of Mann-Kendall test statistic S is as follows:
(1)
S = ∑ ∑ sgn (x − x )
207 (2)
In formulas, x1, x2, …, xn are defining time series while n is number of observations (Basistha, Arya & Goel 2009).
Variance of S and normalized test statistic of z are calculated as follows:
(3) ( )( ) Var(S) =
(4)
Critical test statistic values for various significance levels can be found from normal probability tables. After calculating z-statistic, it is possible to determine any trend and its significance level by comparing z-statistic and critical test statistic values for different significance levels.
5.1.2. Spearman’s Rho Trend Analysis
Spearman’s Rho test is another popular non-parametric test which is widely used in hydrological trend analysis studies (Yue, Pilon & Cavadias 2002; Bouza-deano, Ternero- Rodriquez & Fernandez-Espinosa 2008). Spearman’s Rho test is a rank-based test determines if the correlation between two variables is significant or not. When it is used in trend analysis, one variable is time itself, and the second one is corresponding to time series data. Similar to Mann-Kendal trend test, the n time series values are ranked and replaced by their ranks.
208 Spearman’s Rho test statistic ρs is the correlation coefficient, which is calculated by using following equations.
0.5 ρs= SXY / (Sx Sy) (5)
(6)
S = ∑ ((x X)) (7)
S = ∑ ((y Y)) (8)
S = ∑ (x − X)(y − Y)
In above equations, xi refers to time, yi is variable of interest, and correspond to the ranks.
5.2. Hydrological Models
Snow melt is an important water resource to many aspects of hydrology including water supply, flood control and erosion. Because of its significant contribution to hydrological cycle, it is important to simulate and forecast snow runoffs by using hydrological models.
Hydrological models have been used by hydrologists for the comprehension of hydrologic processes including simulation and estimation of hydrologic unknowns such as catchment flows for many decades. Snow melt runoff models have been improved to model complex snow runoff relation. Complex nature of snow runoff process motivated hydrologists to develop simplified hydrological models including snow melt processes. Both physically meaningful and simplified snow runoff models are available in literatures.
Snow melt runoff models can be classified as either deterministic or stochastic. These models are often classified as lumped, distributed and semi distributed model.
209 Moreover, hydrological models can be evaluated under three titles: statistical model also known as parametric or empirical model, conceptual model and physically based model. Physically based models are based on mass-energy balance and have physical meaning. On the other hand, conceptual models are mainly based on mass conservation in association with simplified representation of momentum and energy equations (Malikov 2004).
Temperature index or degree day factor models are based on relations between snow or ice melt and air temperature. Although simple in concept, if model parameters are estimated correctly, they are often a suitable tool for simplified snow runoff studies. Temperature index models have more common usage than physically based model due to four main reasons: availability of air temperature data; easy interpolation and forecasting possibilities of air temperature; good performance and finally computational simplicity (Hock 2003).
There are several studies on both physically based models (Liston & Elder 2006; Mernild et al. 2008; Dressler et al. 2006) and conceptual index models (Nagler et al. 2008; Parajka & Bloschl 2008; Prasad & Roy 2008) for the purpose of snow runoff modelling. However, considerable data requirements, compared to index model approaches, limits usage of physically based models for snow runoff studies in developing countries. Thus, number of conceptual index snow runoff models is higher than physically based snow runoff models in the literature.
Correct snow melt runoff model type selection is necessary to achieve reliable and satisfactory model results for a particular research area. For the purpose of correct model selection, there are important points that must be considered. Tarboton and Luce (1996) stated that complexity of model does not necessarily increase model accuracy. Charbonneau and Lardeau (1981) used different snow melt runoff models in France’s Alpine regions. They stated that correct distribution of meteorological data such as temperature and precipitation makes snow runoff model powerful and correct distribution is more important than model complexity. Generally, lumped models use
210 lapse rate parameter to distribute temperature and precipitation over the basin. Successful estimation of lapse rate parameters affects model performance strongly. Catchment size is another important criterion that must be taken into consideration while determining snow runoff model. Chanseng and Croley (2007) expressed that application of physically based models into basins with an area more than 1000 km2 is not feasible, especially if it is distributed physically based models. Distributed physically based models need massive amounts of data to compute spatial and temporal distributions of energy and water balances in the soil–plant–atmosphere system. So, they are not appropriate to simulate or predict hydrological processes in large basins.
It is obvious from the literature that calibration ability of models must be taken into consideration while selecting a snow melt runoff model for a catchment. All conceptual or physically based snow runoff models require parameters to explain snowmelt runoff process. Correct parameter choice has a large impact on simulating or forecasting runoff reliably.
In current study, snow melt runoffs of Karasu Basin were simulated by using two lumped conceptual models which are using temperature index approach to calculate snow melt process, The Hydrologic Engineering Center- Hydrologic Modeling System (HEC-HMS) and Large Basin Runoff Model (LBRM). The main reason for selecting these models was relevant data availability for the study area. They require only precipitation and temperature data to perform hydrological simulations. Moreover, these models were used to model future streamflows of Karasu Basin.
Both physically based and conceptual models require parameters associated with the basin characteristics which are not always available particularly for study regions in developing countries. In addition, complicated mathematical interactions, large amount of calibration data, overparameterisation effects and parameter redundancy impacts are the drawbacks of physically based and conceptual models. Hence, it is reasonable to apply alternative tools which can model relation between input and
211 output data set without absolute physical meaning. Artificial Neural Networks (ANNs) can be used as an alternative modelling tool to physically based and conceptual models.
ANNs are flexible mathematical structure with the ability of defining sophisticated nonlinear relationship between input and output parameters without a need to solve complex partial differential equations. As a black box model, ANNs can be evaluated under empirical hydrological models title. ANNs were introduced to literature early 1940s by McCulloch and Pitts while trying to understand human brain and emulate its working processes mathematically. However, ANNs have gained attraction after characteristic details of computational processes were explained by Hopfield (1982). Learning process of ANNs were presented by Rumelhart and McClelland (1986) and Dolling and Varas (2002). Idea behind improving ANNs is simply developing a tool that can adapt itself changes in its environment for the purpose of solving nonlinear problems (Birikundavyi et al. 2002). ANNs provide advantages over conventional hydrological models on successful identification of nonlinear hydrologic relationship between input output parameters, adaptation capability to changing circumstances, improved model performance, shorter calculation times with faster model development.
Due to several advantages of ANNs mentioned above, ANNs have been widely used in hydrological applications during last two decades. ANNs have been used for rainfall runoff modelling (Smith & Eli 1995; Shamseldin 1997; Tokar & Johnson 1999; Sudheer, Gosain & Ramasastri 2002; De Vos & Rientjes 2005; Akhtar et al. 2009) for streamflow forecasting (Imrie, Durucan & Korre 2000; Riad et al. 2004; Firat & Gungor 2007; Turan & Yurdusev 2009) for ground water modelling (Ranjithan, Ehearta & Rarret 1993; Rogers & Dowla 1994; Coulibally et al. 2001; Daliakopoluos, Coulibalya & Tsanis 2005) for water quality modelling (Basheer & Najar 1995; Maier & Dandy 1996) for precipitation forecasting (Tohma & Igata 1994; Hsu, Gupta & Sorooshian 1995; Luk, Ball & Sharma 2001; Rami’rez, Velho & Ferreira 2005) for sediment prediction (Kisi 2005).
212 Finally, three models, HEC-HMS, LBRM and ANN model, were used to simulate historical streamflows in Karasu Basin. Model performances are very similar to each other. However, only HEC-HMS and LBRM were used for future runoff prediction due to future climate data availability. ANN model with highest performance requires precipitation, temperature, humidity, wind speed and temperature range data for the purpose of modelling complex snow processes without physical meaning. However, only future temperature and precipitation data are available so ANN model could not be utilized for future runoff modelling.
5.2.1. HEC-HMS Model
The Hydrologic Engineering Center- Hydrologic Modeling System (HEC-HMS) was originally developed to simulate the precipitation-runoff processes of dendritic watershed systems. Later, it was improved to solve significant hydrological problems including large river basin water supply, flood hydrology and small urban or natural watershed runoff (USACE 2009).
There are three main components of HEC-HMS: basin component, meteorology component and control specification component. Watershed physical description is accomplished by basin component. Hydrologic elements involving sub-basin, reach, junction, reservoir, diversion, source and sink are connected in a dendritic network from upstream to downstream to simulate runoff processes. Furthermore, basin component includes loss, transform and baseflow calculations through different approaches to determine catchment runoffs. In this study, initial and constant loss method, Snyder unit hydrograph, constant monthly baseflow and lag approaches were selected as loss, transform, baseflow and routing methods.
The second main component of HEC-HMS is meteorology. Meteorologic data analysis is performed by the meteorology component. Meteorological model consists of precipitation, evapotranspiration and snowmelt processes. For Karasu Basin, a significant portion of runoff comprises of snow melt water. In HEC-HMS, two snow melt methods, i) temperature index, ii) gridded temperature index, can be used to
213 model snow melt process. A temperature index method is performed to compute the melt rate based on current atmospheric conditions and past conditions in the snowpack. A cold content is integrated snow melt part to take into account of a cold snowpack to freeze liquid water entering the pack from rainfall. Basin can be defined with elevation bands or grid cells in snow melt part of meteorology component (USACE 2009). As mentioned earlier, elevation range in the study basin is from 1125 m to 3487 m. Basin is divided into five elevation zones for the purpose of snow melt calculation and lapse rate parameter was used to distribute temperature over the basin.
In current study, meteorological model was built with selection of monthly average, specified hyetograph and temperature index methods to model evapotranspiration, precipitation and snow melt.
Final HEC-HMS component to accomplish hydrologic simulation is control specification. The time period of a simulation is managed by control specification component. Control specification consists of a starting date and time, ending date and time, and time interval information.
5.2.2. LBRM
The Large Basin Runoff Model (LBRM) was developed by the National Oceanic and Atmospheric Administration (NOAA)’s Great Lakes Environmental Research Laboratory (GLERL) in the 1980s to perform hydrologic simulations and water resources applications in the Great Lakes Basin. LBRM has been implemented to basins draining into the Laurentian Great Lakes for the purpose of simulation and forecasting runoffs (Croley et al. 1998; Croley & He 2005).
The LBRM is based on serial and parallel cascade of linear reservoirs (outflows proportional to storage) to represent moisture storages within a watershed: surface, upper soil zone (USZ), lower soil zone (LSZ), and groundwater zone (GZ) (Croley & He 2005). Total available heat is calculated by model each day, “indexed by daily air temperature, to become potential evapotranspiration (ETP) or actual 214 evapotranspiration (ET), a complementary approach” (He & Croley 2007, p. 1005). Model divides available heat between potential evapotranspiration and actual evapotranspiration according to total available heat. Model takes ET as proportional both ETP and storage. The model utilizes “variable-area infiltration (infiltration proportional to the unsaturated fraction of USZ, daily precipitation and degree-day snowmelt (Croley, 2002)” (He & Croley 2007, p. 1005).
Croley (2002) stated that LBRM uses basic climatological data with few parameters. The LBRM uses mass continuity equations to compute the water storages in each storage zone. He also indicated that LBRM with its few data requirements and structure is suitable for large basin applications. This is a significant characteristic for current study because of relatively large basin area.
The LBRM structure schematic is illustrated in Fig. 5.1. Precipitation reaches to snowpack, if exists, is then available as snow melt based on primarily air temperature and solar radiation. Some part of snow melt and rainfall infiltrate to soil while the other part transforms into direct runoff based on soil moisture content. In terms of vegetation type, the season, solar radiation, air temperature, humidity and wind speed, soil moisture evaporates or is transpired by vegetation. The remaining part percolates into deeper basin storages that feed surface storage through interflows and groundwater flows (Croley 2002; Croley & He 2005).
215 Fig. 5.1: Tank cascade schematic (Croley & He 2005)
In Fig. 5.1, snow pack accumulations and net supply can be determined by using daily precipitation, temperature and insolation. “The net supply is divided into surface runoff, sU/C, and infiltration to the upper soil zone, s−(sU/C), in relation to the upper soil-zone moisture content, U, and the fraction it represents of the upper soil-zone capacity, C. Percolation to the lower soil zone, αp× U, and evapotranspiration, βu×ep×U, are taken as outflows from a linear reservoir (flow is proportional to storage)” (Croley & He 2005, p.174) .
Other storage zones use the same approach with upper soil zone, the main idea is linear proportionality to zone moisture content.
Interflow from the lower soil zone to the surface, αi × L; evapotranspiration, βl
×ep× L and deep percolation to the groundwater zone, αd×L are linearly
216 proportional to the lower soil-zone moisture content, L. Groundwater flow, αg×
G, and evapotranspiration from the groundwater zone, βg×ep×G, are linearly proportional to groundwater-zone moisture content, G. Finally, basin outflow,
αs× S, and evaporation from the surface storage βs ×ep×S, depend on its content, S. Additionally, evaporation and evapotranspiration are dependent on
potential evapotranspiration, ep, as determined by joint consideration of the available moisture and the heat balance over the watershed. The alpha coefficients (α) are used to represent linear reservoir proportionality factors, and the beta coefficients (β) are used to represent partial linear reservoir coefficients associated with evapotranspiration. (Croley & He 2005, p. 174- 175).
The LBRM outputs are surface runoff and storages in upper soil zone, lower soil zone, groundwater zone and surface.
5.2.3. Artificial Neural Network (ANN) Model
5.2.3.1. Introduction to Artificial Neural Networks
ANNs are relatively new nonlinear statistical approach which has the capability to model complex nonlinear hydrological processes without physical expression. It is an alternative modelling approach that is inspired by brain and nervous systems. Fundamental theories of ANNs are the massive interconnections and parallel processing architecture of biological neuron systems (Riad et al. 2004).
ANNs can be defined as a network of interconnected neurons, also referred to as nodes, units or cells. Information processing is performed by nodes. Signals are transmitted between nodes by connection links. Connection strength of each link is explained with associated weight. For the purpose of determining each node’s output signal, an activation function which is nonlinear transformation is implemented (ASCE 2000). A neural network is designed according to its architecture that consists of nodes, their connection weights and an activation function. There are different ways to
217 classify ANNs. One of them is classification with respect to number of layers: single layer, bilayer and multilayers. Another way of classifying neural networks is based on direction of information flow and processing: feed-forward network and a recurrent ANN. A feed-forward network processing structure is parallel and distributed. It is comprised of three layers: input layer that introduces the input data to the network, hidden layer which is a collector of feature detectors and output layer that generates network outputs dependently on given input (Aqil 2007). In a feed-forward network, processing in nodes is achieved through first input layer to final output layer. A number of hidden layers with one or more nodes can be existed. “The number of hidden layers and number of nodes in each hidden layer are usually determined by a trial-error procedure” (ASCE 2000, p.116). Information is transferred from input to output side. In a feed-forward neural network, the nodes in a layer have to be connected to the one in the next layer not to the one in the same. Therefore, output of a layer is determined with respect to previous layer’s input and corresponding weight. Unlike feed forward neural network, information is preceded through the nodes in both directions (from input to output and from output to input) in recurrent ANNs. With a view to success both sided information flow, recurrent ANNs use previous network output as current input (ASCE 2000).
The nodes in neighbouring layers are connected each other with links referred to synaptic weight that explains connection strength between nodes. Fig. 5.2 schematically illustrates feed-forward neural networks.
Fig. 5.2: Feed-forward network architecture (Turan & Yurdusev 2009)
218 Mathematical aspect of ANNs can be defined on a sample single neuron named j as follows. Fig. 5.3 is a schematic illustration of node j. The inputs coming to node j comprise input vector X= (x1,...,xi,.....,xn) in ANN. Similarly, group of weights coming to node j form weight vector W=(W1j,....,Wij,...Wnj). Weight vector, Wij corresponds to a connection weight between I and j nodes. The effective incoming signal to node j is sum of all incoming signals and bias. Node j output, yj is acquired by calculating the value of activation function, f, which determines the response of node to the effective incoming signal receiving to node j in terms of input vector, weight vector and bias of node j (ASCE 2000). Following equation explains the mentioned process.
Yj= f(XWj-bj) (9)
Fig. 5.3: Schematic diagram of single node (ASCE 2000)
Non-linearity is provided by activation function in ANNs. The mostly used activation functions in literature are linear, sigmoid and hyperbolic tangent functions (Imrie, Durucan & Korre 2000). Among these three, sigmoid function is the most commonly used (Riad et al. 2004). Sigmoid function is a bounded, monotonically increasing continuous function as shown below.
(10)
5.2.3.2. Training
Training process is performed by adjusting weights that connects nodes in a neural network. Some part of data set is split for training purpose. This is a similar process, which is called calibration in hydrological models. Training is a repeating process that
219 consists of number of epochs until the underlying function is learned (Dawson & Wilby 2001). Primary aim of training is to minimize error function by adjusting ANN network connection weights and threshold values or bias with a continuous stimulation process. Error function is minimized by generating equal or closer network outputs to targets. There are two basic training types, supervised and unsupervised. An external teacher is necessary for supervised training, which is recommended when large input and output training data sets are used, to guide learning process. On the other hand, unsupervised learning is achieved without an external teacher by using only input data set to adapt ANN connection strengths to cluster those input pattern into classes according to similar properties (ASCE 2000).
5.2.3.3. Popular ANNs Architectures and Training Algorithms in Hydrological Studies
The multilayer perceptron (MLP) is the most popular neural network architecture in hydrology studies (Riad et al. 2004, Wang et al. 2009). Fig. 5.2 is a simple schematic representation of MLP. In the most MLP based ANNs studies, MLP is trained by utilizing error back-propagation algorithm which is also most common used algorithm for training ANNs. Back-propagation algorithm is a gradient descent method that performs training by minimizing error function. Error back-propagation algorithm is achieved by two phases: a feed forward and a back-forward. In the feed-forward phase, each input variable in training data set is transmitted in the direction from input layers to output layers in the network. Then, the network output and observed output in training output data set are compared. Difference between these two, which is also named error, is computed. Finally, in the back-ward phase, error is propagated back to each neuron in the network to adjust weights and bias (ASCE 2000). Training process is performed until obtaining optimum weight vector (Turan & Yurdusev 2009). The radial basis function (RBF), generalized regression neural network (GRNN), conjugate gradient algorithm and cascade correlation algorithm are other commonly used ANN algorithms and detailed information on them can be found in Jayawardena and Fernando (1998), Dawson and Wilby (2001), ASCE (2000) and Cigizoglu (2005).
220 5.3. Climate Models
Climate models are mathematical representation of the major features of earth’s climate, explained as computer codes based on physical laws such as conservation of mass, energy and momentum coupled with observations. The major features of the climate system relevant to climate modelling and water resources are energy cycle, hydrologic cycle and carbon cycle (IPCC 2007).
Fig. 5.4: Schematic of the climate system (IPCC 2007)
Climate models simulate significant aspects of current climate and their simulations are assessed by comparing their outcomes with observations of the atmosphere, ocean, cryosphere and land surface. Climate models can be used to simulate ancient and current climate. Furthermore, they have commonly been used to predict future climate (IPCC 2007). Schematic illustration of a climate model is illustrated in Fig. 5.5.
221 Fig. 5.5: Graphical illustration of climate model (Ruddiman 2001)
Climate models have showed important and upward ability to represent significant mean climate variables such as temperature, precipitation, radiation, wind and ice cover. For instance, observed instrumental global temperature and fourteen reliable climate models global temperature output comparison is shown in Fig. 5.6.
222 Fig. 5.6: Global mean near-surface temperatures over the 20th century (IPCC 2007)
In Fig. 5.6, black line shows temperature observations, yellow lines refer to 58 simulations produced by 14 different climate models driven by both natural and human-caused factors that influence climate and red line is mean of climate model simulations. In Fig. 5.6, temperature anomalies are shown relative to the 1901- 1950 mean. It is seen in Fig. 5.6 that global temperature observations and climate model global temperature outputs are in agreement.
Global Climate Models (GCMs) can be defined as representations of the coupled atmosphere-land-ocean-ice systems and their interactions. They are full three dimensional representation of the climate system. A typical demonstration of GCM is given in Fig. 5.7.
223 Fig. 5.7: Schematic illustration of global climate model (NOAA 2010)
They are the primary way used by IPCC to evaluate climate change impacts. GCMs were introduced to literature by Philips (1956) and they have evolved over the past 50 years. Basic equations that are solved in GCMs are conservation of mass, energy and momentum. The earth must be divided into grid cells to solve each of these questions. Grid box division of world is demonstrated schematically in Fig. 5.7. Each equation must be solved at each grid cell at each time step which is causing computational limitations for climate models (Johnson 2009).
There are important uncertainties of GCMs. Type of uncertainties can be stated as structural, parameter and boundary/initial values uncertainties. Structural uncertainty corresponds to the structural selections of GCMs while developing a model such as “how to represent the land surface in a climate model” or which convective rainfall scheme is more suitable in climate model (Johnson 2009, p. 7). Parameter uncertainty refers to ambiguity to determine what values parameters should take in a model and 224 finally boundary & initial conditions uncertainties refer to fuzziness to select the most accurate values to force the model and initial values at the start of the model run (Johnson 2009).
Despite the fact that GCMs are very helpful tool to evaluate climate, they still have significant errors. Although, errors are more important at smaller scales, large scale problems also remain. Main reasons of most errors are as follows:
-Limitations in computing power - Limitations in scientific understanding -Limitations in availability of detailed observations of some physical processes -Uncertainties in representation of clouds and clouds responses to climate change (IPCC 2007).
GCMs can be defined as the best tool to assess climate change effects on water resources and hydrological cycle by coupling them with hydrological models. However, there are substantial questions that should be answered to trust the confidence of hydrological impact studies. These questions include how much agreement global climate models demonstrate in climate variables, and for which regions for which variables climate models demonstrate good consistency (Johnson 2009). For the purpose of finding reasonable answers to questions above, Johnson and Sharma (2009) improved the variable convergence score (VCS) technique to test performance of different GCMs outputs and they stated that GCMs showed best agreement on surface pressure, temperature and humidity and they showed worst performance on precipitation which is one of the most important variables for hydrological studies.
Although, GCMs have showed important improvement with respect to resolution (see Fig. 5.8), spatial and temporal resolutions of GCMs are not sufficient to use them directly in hydrological models. Fig. 5.8 demonstrates the improvement of spatial resolution of GCMs in IPCC assessment reports. In Fig. 5.8, FAR corresponds to first
225 assessment report, SAR is second assessment report, TAR is third assessment report and finally FAR is fourth assessment report.
Fig. 5.8: Improvement of GCMs resolutions at IPCC assessment reports (Le Treut et al. 2007)
226 When one focuses on the hydrological cycle, spatial resolution of GCMs are too coarse to provide regional and local details of the climate to achieve hydrological impact simulations. Therefore, different downscaling methods including dynamical, statistical or combination of these two were developed for the purpose of obtaining climate data with high spatial resolution (Beldring et al. 2008). Using regional climate models is currently very popular and commonly preferred by climate change researchers for downscaling GCMs into finer resolution. For example, Beldring et al. (2008) used Max Planck Institute atmosphere-ocean general circulation model (AOGCM) ECHAM4/OPYC3 with 125×125 km2 spatial resolution and Hadley Center HadAM3H as boundary conditions for dynamical downscaling with the HIRHAM regional climate model which has a spatial resolution of 55×55 km2. Thus, spatial resolution was downscaled into better resolution to provide more detailed regional hydrological information.
Downscaling approaches, which are developed to convert coarse GCMs resolutions to convenient resolution for impact studies, can be stated as follows.
1. Delta/ratio methods 2. Stochastic/statistical downscaling 3. Dynamic downscaling (Vicuna & Dracup 2007).
Delta/ratio method is based on determining delta change between current and future GCMs and applying this change to observed climate data set. Although, delta/ratio approach is the most common one among three downscaling methods due to simplicity, more complicated statistic and dynamic downscaling approaches have been used in more current studies (Vicuna & Dracup 2007).
Statistical downscaling models perform statistical relations between large scale forcing and station observed variables by using statistical approaches such as multiple regression, canonical correlation analysis or circulation pattern analysis. Although they are computationally very efficient, a consistency of statistical relations under future
227 climate conditions has to be considered. On the other hand, the RCMs are three dimensional atmospheric models with higher resolution than GCMs, usually coupled atmosphere-land surface systems based on physical laws as well as parameterizations for subgridscale processes. Disadvantage of the RCMs is the high computation cost of them. Although, RCMs are stated as a better technique to downscale GCMs outputs by some scientists (Schmidli et al. 2007), there is not any absolute evidence showing overall priority of any downscaling method to other ones (Smiatek 2009). Abdo et al. (2009) explained advantage of statistical model over dynamic model as being less data intensive and computational less demanding. Dawson & Wilby (2007) expressed that in cases where low-cost, rapid evaluation of regional climate is required, statistical downscaling is a better option despite the limitation of them about assuming that the statistical relationship developed for the current climate also used for future climate which is under different forcing conditions (Abdo et al. 2009).
In summary, physical basis of GCMs and their skill in representing observed and past climate change are the confidence proofs of GCMs. They are vital method both understanding & simulation of current climate and estimating credible quantitative estimates of future climate change particularly at larger scales. Nonetheless, they are insufficient to represent regional details due to spatial resolution deficiencies. Even so, as global climate models continue to improve, they are also becoming more powerful to investigate climate change even at smaller scales (IPCC 2007).
In this study, future climate data, which was generated by downscaling ECHAM5 GCM outputs dynamically by using RegCM3 regional climate model, was used for hydrological projections.
5.3.1. Global Climate Model – ECHAM5
Max Planck Institute of Meteorology developed ECHAM GCM. It is the modified version of global forecast models developed by The European Centre for Medium- Range Weather Forecasts (ECMWF). This model has been modified specifically for climate research. ECHAM5 is the most current version (Roeckner et al. 2003). 228 5.3.2. Regional Climate Model – RegCM3
Regional Climate Model version 3 called ICTP-RegCM3 has been developed and distributed by The Earth Systems Physics group of the Abdus Salam International Centre for Theoretical Physics (ICTP). RegCM3 has been utilized for wide range climate applications in many distinct countries. Climate applications with RegCM3 was used to improve solutions about paleoclimate, climate change projections, aerosol effects, agricultural impacts, water resources, extreme events, biosphere-atmosphere interactions, lake and ocean interactions with the atmosphere over various regions. The current version of RegCM has been expressed by Pal et al. (2006) (Onol 2007). Details of RegCM3 were explained by Onol (2007) as follows:
The first version of the RegCM was applied at the National Center of Atmospheric Research (NCAR) (Dickinson et al., 1989; Giorgi and Bates, 1989; and Giorgi 1990). In this version, the radiative transfer package (Kiehl et al., 1987) and the Biosphere- Atmosphere Transfer Scheme (BATS) (Dickinson et al., 1986) was embedded into the dynamical core of NCAR Pennsylvania State University (PSU) Mesoscale Model version 4 (MM4) (Anthes et al., 1987) to run long-term climate simulations. Major improvements in the model physics have been made in the second version. The radiative transfer package (Community Climate Model version 2: CCM2; Briegleb 1992), convective parametrization (Grell, 1993) and the explicit cloud and precipitation scheme (Hsie et al., 1984) were upgraded. In addition, non local PBL parametrization (Holtslag et al., 1990) and latest version of BATS (Dickinson et al. 1993) were included in the hydrostatic version of the MM5.
Before the last version of RegCM was released, the radiative transfer package
of CCM3 with improvements including the effects of NO2, CH4 and CFCs was implemented instead of the package of CCM2. Other significant features of the model were application of a stretched grid model configuration and application of coupling with a lake model, which were tested by Qian et al. (1999) and Small et al. (1999), respectively. 229 In the newest version of the model (RegCM3), precipitation physics, surface physics,atmospheric chemistry and aerosols and the required model input fields have been improved and released in the 2004 workshop organized by ICTP. New large-scale cloud and precipitation (Pal, 2000) and cumulus convection scheme (Betts, 1986) were included in the RegCM3. In addition, parameterization for ocean surface fluxes (Zeng et al., 1998) has been added for a new option. Also, dynamical code has been adopted to parallel computing in the last version. Details of the model physics are presented in the next sections.
The radiative transfer package based on NCAR/CCM3 (Kiehl et al., 1996) uses delta- Eddington approximation over 18 discrete spectral intervals from 0.2 to 5 m. Cloud properties related to liquid water content and droplet radius, are determined in terms of optical depth, single-scattering albedo, and asymmetry parameter (Slingo, 1989). Fractional cloud cover which is total cover in the column from the model-computed cloud-base level to the cloud-top level, is a function of horizontal grid point spacing (Elguindi et al., 2004).
Precipitation calculation has two main forms in most of the climate models. In RegCM3, large scale component of precipitation which generally occurs in winter hemisphere at mid and high latitudes is represented by SUB-grid Explicit moisture scheme (SUBEX) (Pal et al., 2000). Convective component of precipitation which generally occurs in summer hemisphere over tropics, is represented by more than one parametrization schemes (Onol 2007, p. 10-11).
5.4. Interface between Climate Model and Hydrological Model
Meteorological data observations based on site measurements such as temperature and precipitations have been used traditionally in hydrological models. Therefore, meteorological estimates of regional climate models must be transferred from climate models to selected locations. This process is designated as interface between regional
230 climate model and hydrological model. In other words, interface can be defined as transferring signal of climate change from the regional climate model to the hydrological model. Steps of hydrological impact studies and interface usage in it is shown schematically in Fig. 5.9.
Among these methods, delta change approach is the widely used one (Lettenmair et al. 1999; Hay, Wilby & Leavesly 2000; Reynard, Prudhomme & Crooks 2001; Beldring et al. 2008). Delta change approach was also used in this study as an interface.
Fig. 5.9: Schematic illustration of hydrological impact study steps (Andreasson et al. 2004)
Delta change approach requires three time series for each variable. The first one is observed climate data time series, the second and the third ones are time series of climate model outputs for the current climate and future climate. For delta change approach application, firstly mean of each climate variable for each of the four seasons (winter, spring, summer, autumn) must be calculated for current climate model data and future climate model data. Then one should compute the changes between future
231 and current climate data and finally apply the changes to observed (historical) climate. It is also possible to perform same analysis in monthly scale instead of seasonal scale.
232 Chapter 6: Results and Discussion
6.1. Historical Hydro-meteorological Data Trend Analysis Results
6.1.1. Precipitation Trends
Spring is the rainiest season in Karasu Basin according to all meteorological stations. On the other hand, summer is the least rainy season for Tercan and Erzincan areas, whereas winter is the least rainy season in Erzurum. Summer is the second least rainy season in Erzurum. Average total annual precipitation between 1975 and 2008 in Erzurum was 407.4 mm while it was 379 mm in Erzincan and 449 mm in Tercan.
When precipitation data between 1975 and 2008 was investigated, there is an overall decreasing precipitation trend in Erzurum and Tercan. On the other hand, a very slight increasing precipitation trend was observed in Erzincan. Time series graph of total annual precipitation at meteorological stations are shown in Fig. 6.1.
233 Erzurum Annual Total Erzincan Annual Total Precipitation (mm) Precipitation (mm) 650 650
600 600
550 550 500 500 450
450 400
400 350 300 350 250 300 200
250 150 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Tercan Annual Total Precipitation (mm)
700
600
500
400
300
200 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Fig. 6.1: Total annual precipitation graphs of meteorology stations in Karasu Basin
234 All annual precipitation trends in Karasu Basin are statistically insignificant based on Mann-Kendal and Spearman’s Rho non-parametric trend tests. Nonetheless, the most important trend among three stations is Erzurum precipitation decline trend. Moreover, between 1979 and 1993, statistically significant downward trend was detected in Erzurum at 0.05 significance level. Mann-Kendal & Spearman’s Rho trend analysis results for annual precipitation is shown in Table 6.1.
Table 6.1: Mann-Kendal & Spearman’s Rho trend analysis results for annual precipitation Station Mann- Spearman’s Critical Values of Kendal Rho Significance Levels Mean Median Standard z z statistic 0.1 0.05 0.01 Deviation statistic
Erzurum -1.097 -1.024 407 396 63.32
Erzincan 0.296 0.347 1.64 1.96 2.57 379 368 78.99
Tercan -0.148 -0.032 448 438 87.51
There are downward trends in winter precipitations at all meteorological stations. It is shown in Fig. 6.2.
235 Erzurum Winter Precipitation (mm) Erzincan Winter Precipitation (mm) 135 180
160 115 140
95 120
100 75 80
55 60
40 35 20
15 0 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1979 1983 1987 1991 1995 1999 2003 2007
Tercan Winter Precipitation (mm) 180
160
140
120
100
80
60
40
20
0 1970 1975 1980 1985 1990 1995 2000 2005 2010
Fig. 6.2: Winter season precipitation trends at meteorological stations
236 Although all stations’ winter precipitation showed downward trend, only Erzurum station winter precipitation trend is statistically significant at 0.05 significance level. Trend analysis results for winter season precipitation is demonstrated in Table 6.2.
Table 6.2: Winter season precipitation trend analysis results at meteorological stations Mann-Kendal Spearman’s Rho Critical Values of Significance Levels z statistic z statistic
0.1 0.05 0.01 Erzurum Winter -1.986↓↓ -1.976↓↓ 1.645 1.96 2.576 Precipitation
Erzincan Winter -0.089 -0.101 1.645 1.96 2.576 Precipitation
Tercan Winter -0.904 -0.933 1.645 1.96 2.576 Precipitation
Moreover, all meteorological stations’ summer season precipitation demonstrated insignificant decreasing trends. Nevertheless, summer precipitation of Erzurum station showed insignificant upward trend since 2000. Summer season precipitation trend graphs and trend analysis results are shown in Fig. 6.3 and Table 6.3.
237 Erzurum Summer Season Precipitation Erzincan Summer Season Precipitation 180 (mm) (mm) 100 160
140 80
120
100 60
80 40 60
40 20 20
0 0 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Tercan Summer Season Precipitation (mm) 160
140
120
100
80
60
40
20
0 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007
Fig. 6.3: Summer season precipitation trends at meteorological stations
238 Table 6.3: Summer season precipitation trend analysis results at meteorological stations Mann- Spearma Critical Values of Kendal n’s Rho Significance Levels z statistic z statistic 0.1 0.05 0.01 Erzurum Summer -0.815 -0.668 1.645 1.96 2.576 Precipitation Erzincan Summer -0.845 -0.782 1.645 1.96 2.576 Precipitation Tercan Summer -0.964 -1.061 1.645 1.96 2.576 Precipitation
Furthermore, decreasing trends in spring season precipitations were observed at Erzurum and Erzincan. However, they are very slight decreasing trends. Moreover, in Tercan, very slight increasing trend was observed in spring season precipitation. It is shown in Fig. 6.4. Trend analysis results are illustrated in Table 6.4.
239 Erzurum Spring Season Precipitation Erzincan Spring Season Precipitation (mm) (mm)
300 250
250 200
200 150
150
100 100
50 50
0 0 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Tercan Spring Season Precipitation (mm) 350
300
250
200
150
100
50
0 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Fig. 6.4: Spring season precipitation trends at meteorological stations
240 Table 6.4: Spring season precipitation trend analysis results at meteorological stations Mann-Kendal Spearman’s Rho Critical Values of Significance Levels z statistic z statistic
0.1 0.05 0.01 Erzurum Spring -0.208 -0.175 1.645 1.96 2.576 Precipitation Erzincan Spring -0.03 -0.03 1.645 1.96 2.576 Precipitation Tercan Spring 0.415 0.291 1.645 1.96 2.576 Precipitation
Unlike from other seasons, an increasing precipitation trend was observed in autumn precipitation at all meteorological stations in Karasu Basin. Autumn season increasing trend at Erzincan station is more significant than the others, however, all trends are statistically insignificant. Autumn precipitation trend graphs and trend analysis results are shown in Fig. 6.5 and Table 6.5.
241 Erzurum Autumn Season Total Precipitation (mm) Erzincan Autumn Season Total Precipitation (mm) 250 250
200 200
150 150
100 100
50 50
0 0 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010
Tercan Autumn Season Total Precipitation (mm) 250
200
150
100
50
0 1970 1980 1990 2000 2010
Fig. 6.5: Autumn season precipitation trends at meteorological stations
242 Table 6.5: Autumn season precipitation trend analysis results at meteorological stations Mann-Kendal Spearman’s Rho Critical Values of Significance Levels z statistic z statistic
0.1 0.05 0.01 Erzurum Spring 0.489 0.626 1.645 1.96 2.576 Precipitation Erzincan Spring 1.319 1.451 1.645 1.96 2.576 Precipitation Tercan Spring 1.141 1.159 1.645 1.96 2.576 Precipitation
6.1.2. Temperature Trends
In this study, average, minimum and maximum air temperature trends in Karasu Basin were evaluated. Investigations were performed in both annual and seasonal perspectives. From the temperature data between 1975 and 2008 in Erzurum, it is clear that 1992 was the critical year when significant warming trend started to be observed. Annual average and minimum temperatures showed significant cooling trend until 1992, then average and minimum temperatures commenced to rise and showed significant warming trend. On the other hand, maximum temperatures of Erzurum station showed constant significant upward trend. Annual average, minimum and maximum temperatures time series graphs of Erzurum station are shown in Fig. 6.6.
243 8 Erzurum- Annual 3 Average Temperature (°C) Erzurum- Annual 2 Minimum 7 Temperature (°C) 1
6 0
-1 5 -2
4 -3
-4
3 -5
-6 2 1975 1980 1985 1990 1995 2000 2005 1975 1980 1985 1990 1995 2000 2005
15 Erzurum- Annual Maximum 14 Temperature (°C)
13
12
11
10
9 1975 1980 1985 1990 1995 2000 2005
Fig. 6.6: Annual average, minimum and maximum temperatures of Erzurum station
244
Table 6.6: Average, Maximum and Minimum Temperature Trends of Erzurum Station
Median Dev Sta
Erzurum Mann-Kendal Spearman’s Rho Critical Values of Mean . z statistic z statistic Significance Levels .
0.1 0.05 0.01 Annual 1975- 1992- 1975- 1992- Average 1991 2008 1991 2008 1.316 1.568 2.061 5.36 5.28 1.037 Temperature -1.68↓↓ 0.865 -2↓↓ 0.882 Annual 1975- 1992- 1975-1991 1992- Minimum 1991 2008 2008 1.316 1.568 2.061 -1.42 -1.875 1.749 Temperature -1.93↓↓ 1.31↑ -2.10↓↓↓ 1.43↑
Annual 1975-2008 1975-2008 1.645 1.96 2.576 12.11 12.19 0.944 Maximum 1.809↑ 1.756↑ Temperature
In Table 6.6, significance levels were shown in bold characters. Moreover, arrows were added to represent increasing or decreasing trends. In Table 6.6, ↑ corresponds to significant trend with 0.1 significance level, while ↑↑means stas cal l y si gnificant trend with 0.05 significance level and ↑↑↑ represents significant trend with 0.01 significance level.
Furthermore average, minimum and maximum temperature observations at Erzincan station showed statistically upward trends. It is demonstrated in Fig. 6.7.
245 Erzincan Annual Average Erzincan Annual Minimum Temperature (°C) Temperature (°C)
13 7.00
12 6.00
11 5.00
10
4.00 9
3.00 8
2.00 7 1975 1982 1989 1996 2003 1975 1982 1989 1996 2003
Erzincan Annual Maximum Temperature (°C) 20.00
19.00
18.00
17.00
16.00
15.00
14.00 1975 1982 1989 1996 2003 Fig. 6.7: Annual average, minimum and maximum temperatures of Erzincan station
246
Table 6.7: Average, maximum and minimum temperature trends of Erzincan station
Deviation Standard
Erzincan Mann- Spearman’s Critical Values of Median Mean Kendal Rho Significance Levels z statistic z statistic 0.1 0.05 0.01
Annual 1975-2008 1975-2008 Average Temperature 1.645 1.96 2.576 10.91 10.88 0.92 2.95↑↑↑ 2.989↑↑↑
Annual 1975-2008 1975-2008 Minimum 1.645 1.96 2.576 4.94 4.91 0.96 Temperature 3.365↑↑↑ 3.438↑↑↑
Annual 1975-2008 1975-2008 Maximum 1.645 1.96 2.576 17.33 17.37 1.03 Temperature 2.491↑↑ 2.529↑↑
It is seen in Table 6.6 and 6.7 that temperatures in Erzurum and Erzincan showed upward trends, however, according to Mann-Kendal and Spearman’s Rho tests, warming trends in Erzincan station is more significant than Erzurum station. Moreover, Erzincan station showed continuous upward trend, while Erzurum had mostly warming trend after 1992.
Finally, Tercan meteorological station temperatures were investigated for the purpose of understanding significance levels of temperature increases. Temperature trends of Tercan station are shown in Fig. 6.8 and Table 6.8.
247 Tercan Annual Average Tercan Annual Maximum Temperature (°C) Temperature (°C) 10 18
17 9 16
15 8
14
7 13
12 6 11
5 10 1975 1982 1989 1996 2003 1975 1982 1989 1996 2003
Tercan Annual Minimum Temperature (°C) 5
4
3
2
1
0
-1 1975 1982 1989 1996 2003
Fig. 6.8: Annual average, minimum and maximum temperatures of Tercan station
248 Table 6.8: Average, maximum and minimum temperature trends of Tercan station Tercan Mann- Spearman’s Critical Values of Significance Kendal Rho Levels Mean Median Standard z statistic z statistic Deviation
0.1 0.05 0.01 Annual 1975-2008 1975-2008 Average Temperature 1.645 1.96 2.576 8.47 8.54 0.954 1.69↑ 1.798↑
Annual 1975-2008 1975-2008 Minimum 1.645 1.96 2.576 2.256 2.335 0.952 Temperature 0.637 0.552
Annual 1975-2008 1975-2008 Maximum 1.645 1.96 2.576 14.924 15.025 1.116 Temperature 3.024↑↑↑ 3.01↑↑↑
According to Table 6.8, annual average and maximum temperatures showed significant warming, nevertheless, increase in maximum temperature is much more significant in Tercan.
Moreover, seasonal trend analyses of minimum, average and maximum temperatures in Karasu Basin were performed based on non-parametric trend tests. Seasonal temperatures showed similar trends for minimum, maximum and average temperatures.
Erzurum seasonal average temperature trend graphs and analysis results are shown in Fig. 6.9 and Table 6.9.
249 Erzurum Winter Season Average 8 Erzurum Spring Season -1 Temperature (°C) Average Temperature(°C) 7 -3 6 -5 5 -7 4 -9 3 -11 2
-13 1
-15 0 1975 1980 1985 1990 1995 2000 2005 1970 1975 1980 1985 1990 1995 2000 2005 2010
10 21 Erzurum Autumn Erzurum Summer Season Season Average Average Temperature 9 Temperature(°C) 20 (°C)
8 19 7 18 6
17 5
4 16
3 15 1970 1975 1980 1985 1990 1995 2000 2005 2010 1970 1975 1980 1985 1990 1995 2000 2005 2010
Fig. 6.9: Seasonal average temperature trends in Erzurum
250 Table 6.9: Trends analysis results of seasonal average temperature in Erzurum
Erzurum Mann-Kendal Spearman’s Rho Critical Values of Median
Mean
Stand Dev
z statistic z statistic Significance Levels . 0.1 0.05 0.01 Erzurum Winter 1975-2008 1975-2008 -8.4 -8.7 2.7 Season Average -2.283 -2.376 1.645 1.96 2.57 Temperature (°C) 1975-1992 1993- 1975- 1993-2008 1975-1992 Erzurum Spring 2008 1992 1.316 1.56 2.06 4.5 4.73 1.0 Season Average Temperature (°C) -1.03 1.31↑ -0.784 1.461↑ 1993-2008 1.316 1.56 2.06 4.2 4.0 1.6
Erzurum Summer 1975-1992 1993- 1975- 1993- 1.398 1.66 2.19 17.8 17.8 0.9 Season Average 2008 1992 2008 Temperature (°C) -1.3↓ 1.5↑ -1.3↓ 1.3↑ 1.234 1.47 1.93 17.9 17.8 0.9 ↑ Erzurum Autumn 1975-1992 1993- 1975- 1993- 1975-1992 Season Average 2008 1992 2008 7.7 8.2 1.2 Temperature 1.316 1.56 2.06 (°C) -1.6↓↓ 2↑↑ -1.8↓↓ 1.9↑↑ 1993-2008 7.0 7.1 0.9
Seasonal minimum temperature trend graphs in Erzurum are demonstrated in Fig. 6.10. Trend analysis results of seasonal minimum temperatures in Erzurum are presented in Table 6.10.
251 1 Erzurum Spring Season -7 Erzurum Winter Season Minimum Temperature(°C) Minimum Temperature(°C) 0 -9 -1 -11 -2 -13 -3 -15 -4
-17 -5
-19 -6
-21 -7 1975 1980 1985 1990 1995 2000 2005 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
12 Erzurum Summer Season 5 Erzurum Autumn Minimum Temperature(°C) Season Minimum 11 4 Temperature(°C)
3
10 2
1 9 0
-1 8 -2
7 -3 -4
6 -5 1975 1980 1985 1990 1995 2000 2005 1975 1980 1985 1990 1995 2000 2005
Fig. 6.10: Seasonal minimum temperature trends in Erzurum
252 Table 6.10: Trends analysis results of seasonal minimum temperature in Erzurum
Mann-Kendal Spearman’s Rho Critical Values of Median
Deviat
Stand. Mean Erzurum z statistic z statistic Significance Levels
0.1 0.05 0.01 Erzurum Winter 1975-2008 1975-2008 Season Minimum -2.995↓↓↓ -3.22↓↓↓ 1.645 1.96 2.576 -13.7 -14.0 3.5 Temperature (°C) 1975- 1993- 1975- 1993- 1975-1992 1992 2008 1992 2008 1.398 1.666 2.19 Erzurum Spring -0.9 -0.32 1.7 Season Minimum Temperature (°C) - 1993-2008 2.7↓↓ 1.84↑ - 2.02↑ 1.234 1.47 1.932 ↓ ↑ 2.72↓ ↑↑ -2.0 -1.5 1.7 ↓↓ 1975- 1993- 1975- 1993- 1975-1992 1992 2008 1992 2008 1.398 1.666 2.19 Erzurum Summer 9.7 9.9 1.4 Season Minimum Temperature (°C) - 1993-2008 2.235↓ 2.16↑ - 2.34↑ 1.234 1.47 1.932 ↓↓ ↑↑ 2.53↓ ↑↑ 8.3 8.1 0.8 ↓↓ 1975- 1993- 1975- 1993- 1975-1992 1992 2008 1992 2008 1.398 1.66 2.19 1.2 1.6 2.0 Erzurum Autumn Season Minimum 1993-2008 Temperature - (°C) 2.083↓ 1.846 - 1.743 1.234 1.47 1.932 -0.8 -0.5 1.4 ↓↓ ↑↑ 2.31↓ ↑↑ ↓↓
Moreover, seasonal maximum temperature trend graphs and trend analysis results in Erzurum are illustrated in Fig. 6.11 and Table 6.11.
253 Erzurum Winter Season Maximum 16 Erzurum Spring 2 Temperature(°C) 15 Season Maximum Temperature(°C)
14 0 13
-2 12
11 -4 10
9 -6 8
-8 7 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
30 Erzurum Summer Season 18 Maximum Erzurum Autumn Season Maximum 29 Temperature(°C) 17 Temperature(°C)
28 16
27 15
26 14
25 13
24 12
23 11 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Fig. 6.11: Seasonal maximum temperature trends in Erzurum
254
Table 6.11: Trends analysis results of seasonal maximum temperature in Erzurum
Deviation Standard
Erzurum Mann-Kendal Spearman’ Critical Values of Median Mean z statistic s Rho Significance Levels z statistic 0.1 0.05 0.01
Winter Season 1975-2008 1975-2008 Maximum -0.46 -0.51 1.645 1.96 2.576 -2.8 -3.2 2.2 Temperature Spring Season 1975-2008 1975-2008 Maximum 1.645 1.96 2.576 13.3 13.1 1.6 Temperature 1.972↑↑ 2.153↑↑
Summer Season 1975-2008 1975-2008 Maximum 2.654↑↑↑ 2.601↑↑ 1.645 1.96 2.576 25.5 25.4 1.2 Temperature ↑ Autumn Season 1975-2008 1975-2008 Maximum 1.645 1.96 2.576 15.0 15.0 1.1 Temperature 2.105↑↑ 2.093↑↑
In Erzurum, for average and minimum temperatures, decreasing temperature trends were observed from 1975 to 1992, followed by warming trends from 1992 to 2008 in all seasons except winter season which showed constant decreasing trend. However, maximum temperatures in spring, summer and autumn seasons showed warming trends which are more significant since 1992.
Seasonal temperature analyses of average, minimum and maximum temperatures were also applied to Erzincan and Tercan. Erzincan station’s seasonal average temperature trend graphs are shown in Fig. 6.12 and trend analysis results are demonstrated in Fig. 6.12.
255 Erzincan Winter Season Average Erzincan Spring Season Average Temperature (°C) Temperature (°C) 3 13 2 12 1
0 11 -1 10 -2
-3 9
-4 8 -5
-6 7 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Erzincan Summer Season Average Erzincan Autumn Season Average Temperature (°C) Temperature (°C) 14 25 14 25
24 13
24 13 23 12 23
22 12
22 11 21 11 21
20 10 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Fig. 6.12: Seasonal average temperature trends in Erzincan
256
Table 6.12: Trends analysis results of seasonal average temperature in Erzincan
Deviation Standard
Erzincan Mann-Kendal Spearman’s Critical Values of Median Mean z statistic Rho Significance Levels z statistic 0.1 0.05 0.01
Winter Season 1975-2008 1975-2008 Average 0.385 0.452 1.645 1.96 2.576 -1.4 -1.1 2.23 Temperature Spring Season 1975-2008 1975-2008 Average 1.645 1.96 2.576 10.2 10.1 1.15 Temperature 1.957↑↑ 2.288↑↑
Summer Season 1975-2008 1975-2008 Average 4.106↑↑↑ 3.917↑↑↑ 1.645 1.96 2.576 22.6 22.7 1.06 Temperature Autumn Season 1975-2008 1975-2008 Average 2.372↑↑ 2.25↑↑ Temperature 1.645 1.96 2.576 11.8 11.7 1.02
Moreover, Erzincan’s seasonal minimum temperature trend graphs are demonstrated in Fig. 6.13.
257 0.00 Erzincan Winter 7 -1.00 Season Minimum Erzincan Spring Season Temperature(°C) Minimum Temperature(°C) -2.00 6 -3.00
-4.00 5 -5.00
-6.00 4
-7.00
-8.00 3
-9.00
2 -10.00 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
18 8 Erzincan Summer Erzincan Autumn Season Minimum Season Minimum Temperature(°C) Temperature(°C) 17 7
16 6
15
5 14
4 13
12 3 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Fig. 6.13: Seasonal minimum temperature trends in Erzincan
258 Maximum temperature trend graphs in four seasons are demonstrated for Erzincan in Fig. 6.14. Moreover, trend analysis results for Erzincan’s maximum temperatures are shown in Table 6.13.
9 20 Erzincan Spring Erzincan Winter Season Maximum Season Maximum Temperature(°C) Temperature(°C) 19 7 18 5 17
3 16
15 1 14 -1 13
-3 12 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
33 Erzincan Summer 22 Season Maximum Erzincan Autumn Temperature(°C) Season Maximum 32 21 Temperature(°C)
31 20
19 30 18 29 17
28 16
27 15 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Fig. 6.14: Seasonal maximum temperature trends in Erzincan
259 Table 6.13: Trends analysis results of seasonal maximum temperature in Erzincan
Erzincan Mann- Spearman’s Critical Values of Median
Stan Mean
Kendal Rho Significance Levels Dev. d
z statistic z statistic 0.1 0.05 0.01 .
Winter Season 1975-2008 1975-2008 Maximum 0.682 0.675 1.645 1.96 2.576 3.23 3.59 2.46 Temperature Spring Season 1975-2008 1975-2008 Maximum 1.645 1.96 2.576 16.22 16.11 1.31 Temperature 1.853↑ 1.988↑↑
Summer Season 1975-2008 1975-2008 Maximum 3.424↑↑↑ 3.427↑↑↑ 1.645 1.96 2.576 30.11 30.23 1.18 Temperature Autumn Season 1975-2008 1975-2008 Maximum Temperature 1.645 1.96 2.576 19.38 19.45 1.28 0.815 0.766
In summary, there are significant warming trends in spring, summer and autumn average temperatures in Erzincan. Especially, in summer season, average temperature increase is very clear. Also, there is an increasing trend in average temperature in winter season, however not statistically significant. Moreover, there are significant warming trends in spring, summer and autumn seasons’ minimum temperatures in Erzincan. Summer upward trend is very significant. Furthermore, increasing trend was observed in minimum temperature of winter season. In addition, there are insignificant upward trends in maximum temperatures in Erzincan for autumn and winter seasons. Nonetheless, warming trends in maximum temperature are significant in spring and particularly in summer season.
Tercan station seasonal average temperature trend graphs are shown in Fig. 6.15 Moreover, trend analysis results of Tercan’s average temperatures are illustrated in Table 6.14.
260 Tercan Winter Season Average Tercan Spring Season Average Temperature (°C) Temperature (°C) 1 11
10 -1
9 -3
8 -5 7
-7 6
-9 5
-11 4 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Tercan Autumn Season Average Tercan Summer Season Average Temperature (°C) Temperature (°C) 12 23
22 11
21 10
20
9 19
18 8 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Fig. 6.15: Seasonal average temperature trends in Tercan
261 Table 6.14: Trends analysis results of seasonal average temperatures in Tercan
Tercan Mann- Spearman’s Critical Values of Median
Stand. Mean Kendal Rho Significance Levels Dev. z statistic z statistic 0.1 0.05 0.01
Winter Season 1975-2008 1975-2008 Average 0.07 0.045 1.645 1.96 2.576 -5.11 -4.84 2.84 Temperature Spring Season 1975-2008 1975-2008 Average 1.645 1.96 2.576 7.56 7.57 1.33 Temperature 1.393 1.609
Summer Season 1975-2008 1975-2008 Average 2.624↑↑ 2.429↑↑ 1.645 1.96 2.576 20.69 20.73 0.87 Temperature ↑ Autumn Season 1975-2008 1975-2008 Average Temperature 1.645 1.96 2.576 10.33 10.36 0.84 1.26 1.207
Moreover, Fig. 6.16 demonstrates seasonal minimum temperature trend graphs in Tercan. Furthermore, Table 6.15 presents trend analysis results of seasonal minimum temperatures in Tercan.
262 -2 Tercan Winter Season 5 Minimum Tercan Spring Temperature(°C) Season Minimum Temperature(°C) -4 4
-6 3
-8 2
-10 1
-12 0
-14 -1
-16 -2 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
14 Tercan Summer 6 Season Minimum Tercan Autumn Temperature(°C) Season Minimum Temperature(°C)
5 13
4
12
3
11 2 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Fig. 6.16: Seasonal minimum temperature trends in Tercan
263 Table 6.15: Trends analysis results of seasonal minimum temperature in Tercan
Tercan Mann- Spearman’ Critical Values of Median
Stand. Mean Kendal s Rho Significance Levels Dev. z statistic z statistic 0.1 0.05 0.01
Winter Season 1975-2008 1975-2008 Minimum -0.193 -0.168 1.645 1.96 2.576 -9.46 -9.17 3.089 Temperature Spring Season 1975-2008 1975-2008 Minimum 1.645 1.96 2.576 1.995 2.325 1.294 Temperature 0.682 0.775
Summer Season 1975-2008 1975-2008 Minimum 0.682 0.717 1.645 1.96 2.576 12.19 12.23 0.663 Temperature Autumn Season 1975-2008 1975-2008 Minimum Temperature 1.645 1.96 2.576 3.96 4.03 0.854 0.341 0.499
Finally, Tercan station seasonal maximum temperature trend graphs are shown in Fig. 6.17. In addition, Tercan station seasonal maximum temperature trend analysis results are illustrated in Table 6.16.
264 6 18 Tercan Winter Tercan Spring Season Season Maximum Maximum Temperature(°C) Temperature(°C) 4 16 2
0 14
-2 12 -4
-6 10 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
32 Tercan Summer 21 Season Maximum Tercan Autumn Temperature(°C) Season Maximum 31 Temperature(°C) 20
30 19
29
18 28
17 27
16 26
25 15 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Fig. 6.17: Seasonal maximum temperature trends in Tercan
265 Table 6.16: Trends analysis results of seasonal maximum temperature in Tercan
Tercan Mann- Spearman’s Critical Values of Median
Stand. Mean Kendal Rho Significance Levels Dev. z statistic z statistic 0.1 0.05 0.01
Winter Season 1975-2008 1975-2008 Maximum 1.067 1.014 1.645 1.96 2.576 -0.20 0.09 2.709 Temperature Spring Season 1975-2008 1975-2008 Maximum 1.645 1.96 2.576 13.30 13.13 1.604 Temperature 1.972↑↑ 2.153↑↑
Summer Season 1975-2008 1975-2008 Maximum 3.573↑↑ 3.547↑↑↑ 1.645 1.96 2.576 28.46 28.56 1.215 Temperature ↑ Autumn Season 1975-2008 1975-2008 Maximum 1.645 1.96 2.576 17.69 17.72 1.085 Temperature 2.164↑↑ 2.2↑↑
There is no significant trend in winter season average temperature in Tercan. There is a slight warming trend in spring and autumn seasons’ average temperatures in Tercan. On the other hand, there is a significant upward trend in summer season’s average temperature. Moreover, there is a slight decreasing trend in minimum temperature in Tercan during winter season. There is an insignificant upward trend in spring, summer and autumn seasons’ minimum temperatures. Furthermore, there is an insignificant warming trend in winter season maximum temperature in Tercan. Nonetheless, maximum temperature between 1992 and 2001 showed substantial warming trend. There are significant increasing trends in spring, summer and autumn seasons’ maximum temperatures in Tercan.
Maximum temperature warming trends are dramatic and occurred in all seasons for all stations. Summer season warming trend is more significant than all other seasons at all stations for minimum, maximum and average temperatures; however, maximum temperature warming is the most significant one. Summer season maximum temperature trends of meteorological stations are shown in Fig. 6.18.
266 Erzurum-Summer Season Maximum Erzincan-Summer Season Temperature Trend Maximum Temperature Trend 29 33
28 32 27 31 26
30 25
29 24
23 28
22 27 1975 1979 1983 1987 1991 1995 1999 2003 2007 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Tercan-Summer Season Maximum Temperature Trend 32
31
30
29
28
27
26
25 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Fig. 6.18: Summer season maximum temperature trends
Table 6.17 shows the significance levels of maximum temperature increasing trends in summer season at all meteorological stations.
267 Table 6.17: Significance levels of maximum temperature increasing trends in summer season
Mann- Spearman’s Critical Values of
Deviation
Standard Median Station Kendal Rho Significance Levels Mean z statistic z statistic 0.1 0.05 0.01
Erzurum 2.654↑↑↑ 2.601↑↑↑ 1.645 1.96 2.576 25.60 25.49 1.29
Erzincan 3.424↑↑↑ 3.427↑↑↑ 30.11 30.23 1.18
Tercan 3.573↑↑↑ 3.547↑↑↑ 28.46 28.56 1.22
Table 6.17 shows that summer season maximum temperature increases are so substantial at all stations at all significance levels based on Mann-Kendal and Spearman’s Rho tests.
6.1.3. Flow Trends
Flow data for the period of 1975-1987 and 1995-2004 is available at the outlet point of Karasu Basin, station 2119. There are two main impacts of climate change on streamflows in snow dominated basins: first one is decreasing trend in streamflow amount and the second one is shifts to earlier snow melting time because of increase in temperatures.
There is a very slight decreasing trend in annual average runoff in Karasu Basin. Nevertheless, decrease in streamflow amount is not statistically significant. Time series graph of annual average runoff in Karasu Basin is shown in Fig. 6.19.
268 Average Annual Runoff (m3/s) 120 110 100 90 80 70 60 50 40 1976 1981 1986 1999 2004 Fig. 6.19: Time series graph of annual average runoff in Karasu Basin
Table 6.18: Flow trend analysis in Karasu Basin
Deviation Standard
Mann- Spearman Critical Values of Median Mean 2119 Kendal ’s Rho Significance Levels z statistic z statistic 0.1 0.05 0.01
Annual Average -0.695 -0.645 1.645 1.96 2.576 81.81 84 16.741 Flow (m3/s)
It is shown in Table 6.18 that although there is a downward flow trend in Karasu Basin, it is not statistically significant according to both Mann-Kendal and Spearman’s Rho non-parametric tests.
Moreover, peak flow dates were investigated at flow station to detect any shift to earlier time due to climate change. It is established by temperature trend analysis that 1992 was the year when climate change impacts started to be experienced in Karasu Basin. So, available flow data after 1992 were considered for the purpose of determining peak flow dates. Dates of peak flows after 1995 are demonstrated in Table 6.19
269 Table 6.19: Peak flow dates Flow Year (m3/s) Peak Date 1995 461 3-May 1996 375 16-May 1997 434 4-May 1998 718 19-Apr 1999 244 8-May 2000 293 7-May 2001 211 7-Apr 2002 291 18-Apr 2003 524 27-Apr 2004 456 11-May
As shown in Table 6.19, there is a very slight shifting trend to earlier melting date. Finally, based on runoff amount trend and peak flow date analysis, it can be stated that climate change impact on Karasu Basin catchment flows is not significant currently.
6.2. Historical Runoff Simulations
6.2.1. HEC-HMS Results
HEC-HMS model was applied to Karasu Basin in lumped structure at daily time step. Meteorological data was collected from Karasu Basin meteorological stations. Calibration process was conducted by using observed outflow data at the outlet of the basin.
For study area, model was run between 1997 and 2004. Calibration was performed from 1 January, 1997 to 31 December, 2002. By using calibration trial, it is possible to calibrate basin component parameters including loss, transform, baseflow and routing methods. However, there is no option to calibrate meteorological component parameters in HEC-HMS.
Comparison hydrograph of modelled and observed outflow at basin outlet after calibration run is shown in Fig. 6.20.
270 800
700 Modeled Flow (m3/s)
600 Observed Flow (m3/s)
500
400
300
200
100
0 1-Jan-97 1-Jan-98 1-Jan-99 1-Jan-00 1-Jan-01 1-Jan-02
Fig. 6.20: HEC-HMS comparison hydrograph for calibration run at outlet point of Karasu Basin
Calibration period model performance was evaluated in terms of three criteria: Nash- Sutcliffe coefficient of determination (R2), mean squared relative error (RMSE) and linear regression coefficient. Nash- Sutcliffe coefficient of determination and mean squared relative error formulas are as follows:
R2= (11) ⅀( ) ⅀ ̇
RMSE= ⅀ (12) ⅀
In formula 11 and 12, Qobs corresponds to observed flow where Qmod corresponds to modelled flow. Moreover, refers to mean observed flow and n equals to number of observations.
Performance evaluation results are shown in Table 6.20. 271 Table 6.20: Calibration run performance in HEC-HMS
1997-2002 Calibration Run Nash-Sutcliffe Coefficient of 0.70 Determination Mean Squared 0.28 Relative Error Linear Regression 0.70
Validation process was accomplished between 1 January, 2003 and 31 December, 2004. Basin outlet point flow comparison hydrograph for validation run is demonstrated in Fig. 6.21. Validation run results according to Nash-Sutcliffe coefficient of determination, mean squared relative error (RMSE) and linear regression coefficient are shown in Table 6.21.
600 Modeled Flow (m3/s) 500 Observed Flow (m3/s) 400
300
200
100
0 1-Jan-03 1-Jul-03 1-Jan-04 1-Jul-04
Fig. 6.21: HEC-HMS comparison hydrograph for validation run at outlet point of Karasu Basin
Table 6.21: Validation run performance in HEC-HMS 2003-2004 Validation Run Nash-Sutcliffe Coefficient of 0.76 Determination Mean Squared 0.18 Relative Error Linear Regression 0.77
272 According to performance assessment results, it is possible to state that HEC-HMS model has an adequate success to simulate runoffs in Karasu Basin.
6.2.2. LBRM Results
The LBRM was applied to Karasu Basin at daily time step and calibration run was performed from 1 January, 1997 to 31 December, 2002 similar to HEC-HMS application. Comparison hydrograph of outlet outflow is shown in Fig. 6.22.
7.00E-01
6.00E-01 Modeled Flow (cm) 5.00E-01 Observed Flow (cm)
4.00E-01
3.00E-01
2.00E-01
1.00E-01
0.00E+00 1-Jan-97 1-Jan-98 1-Jan-99 1-Jan-00 1-Jan-01 1-Jan-02
Fig. 6.22: LBRM comparison hydrograph for calibration run at outlet point of Karasu Basin Calibration run model performance was evaluated according to Nash-Sutcliffe coefficient of determination, mean squared relative error and linear regression. Results are shown in Table 6.22.
Table 6.22: Calibration run performance in LBRM 1997-2002 Calibration Run Nash-Sutcliffe Coefficient of 0.73 Determination Mean Squared 0.23 Relative Error Linear Regression 0.75
273 Moreover, validation was performed between 1 January, 2003 and 31 December, 2004. Validation results in terms of Nash Sutcliffe coefficient of determination, relative means square error and linear regression coefficient are shown in Table 6.23. Furthermore, flow hydrograph at basin outlet is shown in Fig. 6.23.
Table 6.23: Validation run performance in LBRM 2003-2004 Validation Run Nash-Sutcliffe Coefficient of 0.72 Determination Mean Squared 0.29 Relative Error Linear Regression 0.75
5.00E-01 Modeled Flow (cm) 4.50E-01 Observed Flow (cm) 4.00E-01 3.50E-01 3.00E-01 2.50E-01 2.00E-01 1.50E-01 1.00E-01 5.00E-02 0.00E+00 1-Jan-03 1-Jul-03 1-Jan-04 1-Jul-04
Fig. 6.23: LBRM comparison hydrograph for validation run at outlet point of Karasu Basin
In terms of Nash Sutcliffe coefficient of determination, linear regression and mean squared relative error, it can be concluded that LBRM has an adequate ability to simulate runoffs in Karasu Basin.
Although overall performances of HEC-HMS and LBRM are acceptable for flow simulation in Karasu Basin, both HEC-HMS and LBRM showed insufficient performance 274 to simulate flows in 1998. It is due to unexpected extreme peak flow jump in 1998 with 718 m3/s, which is the second largest peak flow value between 1954 and 2004. Rapid flow increase is expected to be owing to rapid increase in precipitation but this is not the case for current problem. Long term precipitation time series including precipitation data for the period of 1995-2004 is shown in Fig. 6.24. In addition, Fig. 6.25 demonstrates the observed flow time series for modelling period (1995-2004) to comprehend increase level of flow in 1998.
70
60 Rainfall(mm)
50
40
30
20
10
0
Fig. 6.24: Precipitation time series from January 1995 to December 2004
275 800.0
700.0 Observed flow time series (m3/s)
600.0
500.0
400.0
300.0
200.0
100.0
0.0
Fig. 6.25: Observed flow time series for modelling period
Although precipitation in 1998 is slightly higher than the average precipitation, it is not that extreme to produce such an extreme runoff. Average peak flow during modelling period (1995-2004) is 400.7 m3/s while it is 718 m3/s in 1998. If Fig. 6.24 and 25 are considered together, flow versus precipitation comparison in 1998 and 2004 can give an idea of reason to underestimate flow in 1998. Peak precipitation events were observed in 1998 and 2004. However, although precipitation peak in 1998 is lower than precipitation peak in 2004, flow peak in 1998 is higher than flow peak in 2004. Moreover, it is not possible to explain extreme flow peak in 1998 with snow melt. It suggests a mistake while recording data. Thus, underestimation of flow in 1998 may be explained with high likely chance of a mistake made by observation station than HEC- HMS and LBRM incapability to simulate flows.
Moreover, LBRM underestimated flows in late February and early March of 2004 due to the largest precipitation event of the basin based on historical records (1975-2004) with 59.6 mm on 23 February 2004. Normally in Karasu Basin, peak flows have been observed in April & May because of snow melt and high flows have not been 276 experienced until spring except 2004. In 2004, due to extreme precipitation event on late February, high flows in March were observed. LBRM failed to model unexpected extreme early flow in 2004 corresponding to unexpected extreme precipitation peak in 2004. Nevertheless, overall performances of both HEC-HMS and LBRM are good enough to simulate snow runoffs in Karasu Basin.
6.2.3. ANN Results
6.2.3.1. Data Set Selection and Standardization
Data set was divided into training and test parts for the purpose of ANN model application. The test part is similar to validation process in conventional hydrological modelling. Data from 1 October 1975 to 30 September 1987 was selected as training data and from 1 October 1994 to 31 December 2004 was selected as test data. The basic statistical characteristics of used data in Erzurum are summarized in Table 6.24 for the training and test period separately. Table 6.24: Basic statistical characteristics of used data Minimum Maximum Average Standard Deviation Rainfall (mm) Training Period(1975-1987) 0 58.2 1.20 3.32 Test Period (1994-2004) 0 59.6 1.08 3.14
Temperature (°C) Training Period(1975-1987) -23 26.4 6.18 10.39 Test Period (1994-2004) -30 26 5.02 11.33
Wind speed (m/s) Training Period(1975-1987) 0 13.2 2.45 1.53 Test Period (1994-2004) 0 12.1 2.74 1.84 Humidity (%) Training Period(1975-1987) 7 97.3 64.56 15.69 Test Period (1994-2004) 18 98.3 63.87 15.27 Flow (m3/s) Training Period(1975-1987) 12.3 734 82.96 94.24 Test Period (1994-2004) 17.9 718 82.48 81.85
277 On the other hand, Karasu Basin streamflows varied between 12.3 and 734 m3/s. Maximum streamflow values were observed in spring seasons, when snow starts melting. Flow time series graphs for training and test phases are shown in Fig. 6.26.
Fig. 6.26: Flow time series graphs for training and test periods
Suitable data selection is one of the most important issues for successful ANN model application. Suitable data set means sufficient data to achieve meaningful ANN model results. As it is stated before, Upper Euphrates Basin-Karasu Basin is a snow dominated area. Peak flows during a year are seen at the end of spring and beginning of summer seasons due to snow melt. Seasonal variation of flows is shown in Fig. 6.27 for a sample year,1995.
A reasonable set of snow ground data is not available in Turkey. Although some snow depth data are available for study basin, it is not sufficient for use in hydrological model. Thus, it is aimed to model snow runoffs using precipitation, air temperature, humidity, wind speed and temperature range data.
278 Fig. 6.27: Runoff graph in Karasu Basin in 1995
There are three most significant energy fluxes in the physics of snow melt; these are shortwave radiation, longwave radiation and turbulent fluxes including sensible and latent heat fluxes. Tarboton and Luce (1996) claimed that it is possible to explain physics of snow accumulation and melt with input data set involving precipitation, air temperature, wind speed, humidity and temperature range. Utah Energy Balance Model (Tarboton & Luce 1996) has been developed based on this data set.
In addition, four different ANN models in terms of different input data set were generated and simulated. It is found that based on linear regression criteria precipitation, air temperature, humidity, wind speed and temperature range is the best input data set to simulate Karasu Basin runoffs.
It is important to standardize data to provide equal attention during the training process in successful ANN application. If data is not standardized, “input variables in different scales will dominate training to a greater or lesser extent due to randomization of initial weights within a network to the same finite range” (Dawson & Wilby 2001). Moreover, it is significant to standardize data for the efficiency of training
279 algorithms (Dawson & Wilby 2001). Thus, all data set was standardized into [0 -1] range by using following equation.
(13)
In Equation 13, X refers to standardized data where Xmin and Xmax are minimum and maximum values of any particular data.
6.2.3.2. Network Type and Training Algorithm Selection
Feed Forward Back Propagation (FFBP) algorithm is obviously most popular training algorithm for ANN based hydrological studies. Thus, FFBP is selected as a training algorithm for the current study. Due to widespread usage of Multilayer Perceptron (MLP) ANN network type in hydrology studies, it is decided to use MLP as ANN network type in this study.
Furthermore, it is very significant to consider seasonality in Karasu Basin to achieve a better application of ANN based runoff model. Major portion of runoffs consist of snow melt and peak flows are seen late spring and early summer periods. Thus, it is also necessary to perform seasonal analysis using data for only spring and summer periods. So, ANN model in Karasu Basin was developed and run for both annual (whole year) and seasonal (spring –summer) periods.
6.2.3.3. Annual ANN Model
Four different models (Model 1, Model 2, Model 3, Model 4) were generated to simulate daily Karasu Basin runoffs for whole year using different input data sets. First model uses precipitation and air temperature to model runoffs while the last one uses precipitation, air temperature, humidity, wind speed and temperature range to simulate runoffs. Model 4 showed the best performance with a linear regression coefficient of 0.78 for training phase and 0.52 for test phase. The first model showed
280 the lowest performance with a linear regression coefficient of 0.59 for training phase and 0.51 for test phase. Summary of different ANN model outcomes are presented in Table 6.25.
Table 6.25: Summary of different annual ANN runoff models
Input Set Output Linear Regression I Flow (m3/s) Training:0.59 Model 1 Precipitation(mm)+Air Temperature (°C) Test:0.51
Precipitation (mm)+Air Temperature (°C) Flow (m3/s) Training:0.65 Model 2 +Humidity (%) Test:0.51
Precipitation (mm)+Air Temperature Flow (m3/s) Training:0.67 Model 3 (°C)+ Test:0.52 Humidity (%)+Wind Speed (m/s)
Precipitation (mm)+Air Temperature Flow (m3/s) Training:0.78 Model 4 (°C)+Humidity (%)+Wind Speed Test:0.52 (m/s)+Temperature Range (°C)
Number of hidden layers was decided as 5 and number of neurons in each hidden layer was selected as 10 based on trial-error procedure in ANN models. Flow scatter diagrams of best ANN model (Model 4) are shown in Fig. 6.28.
281 Fig. 6.28: Flow scatter diagrams of training and test phases in annual analysis
Moreover, comparison of observed and modelled flow time series graphs at the outlet of Karasu Basin for training and test phases are shown in Fig. 6.29.
282 Fig. 6.29: a) Observed-Modelled flow graph at Karasu Basin outlet point for training phase b) Observed-Modelled flow graph at Karasu Basin outlet point for test phase
Model performance was evaluated in terms of Nash Sutcliffe coefficient of determination (R2). R2 for training and test phases was found 0.71 and 0.50 respectively. Model performance is not very good owing to impacts of large variations in seasonal flows. Other than snowmelt season, flow consists of baseflow and limited amount of rainfall. With the start of snowmelt, flow increases significantly. Large variation between low flows and peak flows is the cause of ANNs incapability to simulate such a large variations.
283 6.2.3.4. Seasonal ANN Model
Daily meteorological data between 1976 and 2004 was used to develop seasonal ANN models. Unlike annual model, meteorological and flow data between March and August months were used for the purpose of developing seasonal models. Similar to annual model, four models according to different input set were generated and comparisons between observed and modelled flows were assessed through linear regression. Model details and performances based on linear regression are shown in Table 6.26.
Table 6.26: Summary of different seasonal ANN runoff models
Input Set Output Linear Regression I Training:0.74 Model 1 Precipitation (mm)+Air Temperature (°C) Flow (m3/s) Test:0.65
Precipitation (mm)+Air Temperature (°C) Training:0.78 Model 2 +Humidity (%) Flow (m3/s) Test:0.68
Precipitation (mm)+Air Temperature Training:0.78 Model 3 (°C)+ Flow (m3/s) Test:0.68 Humidity (%)+Wind Speed (m/s)
Precipitation (mm)+Air Temperature Training:0.88 Model 4 (°C)+Humidity (%)+Wind Speed Flow (m3/s) Test:0.71 (m/s)+Temperature Range (°C)
As it can be seen in Table 6.26, best model performance was obtained for Model 4 with a linear regression coefficient of 0.88 for training phase and 0.71 for test phase. It means precipitation, air temperature, humidity, wind speed and temperature range is the best input data set to simulate flows in Karasu Basin. The lowest performance was observed for Model 1, however still it is reasonable and may be acceptable to simulate runoffs in Karasu Basin.
284 Analogous to annual model, feed forward back propagation learning algorithm was used for seasonal model application. Different numbers of neurons and hidden layers were experienced in this case. As a result, 5 hidden layers with 10 neurons each provided the best results and used in seasonal ANN models. Flow scatter diagrams of training and test phases for Model 4 are shown in Fig. 6.30.
Fig. 6.30: Seasonal flow scatter diagrams of training and test phases in seasonal analysis
Model performance is also evaluated in regards to Nash-Sutcliffe coefficient of determination (R2). R2 is calculated as 0.81 for training phase and 0.70 for test phase for Model 4. Comparison of modelled and observed flow time series graphs at the outlet of Karasu Basin are shown in Fig. 6.31 for training and test phases.
285 (a)
(b)
Fig. 6.31: a) Seasonal observed-modelled flow graphs for training phase b) Seasonal observed-modelled flow graphs for test phase
It is to be noted that Fig. 6.31 (a) and (b) are the comparisons of observed flow and modelled flow for March-August periods only.
6.3. Future Climate and Flow Projections
6.3.1. Temperature and Precipitation Projections in Karasu Basin
Historical precipitation data trend analysis in Karasu Basin between 1975 and 2008 years showed statistically insignificant precipitation decreases in winter, spring & summer seasons and insignificant increases in autumn season. RegCM3 outputs are in agreement with historical data trends. The largest precipitation decrease for time slice of 2070-2100 is expected in summer season with 15%. The second largest precipitation decrease is expected in spring season with 11%. Moreover, decline by 5% in winter
286 precipitation and an increase by 2% in autumn precipitation were projected by RegCM3.
One should note that one of the most important drawbacks of global and regional climate models is precipitation estimations (Johnson & Sharma 2009). Most climate models are not consistent in future precipitation prediction due to the fact that precipitation depend on many factors and not easy to predict simply. Although climate models are not very trustworthy for future precipitation prediction, they still provide a coarse idea about future precipitation.
Historical temperature data trend analysis including 1975-2008 period showed that especially after 1992 climate change finger prints have been commenced to be observed in temperature of Karasu Basin. There are significant warming trends at all meteorological stations in Karasu Basin after 1992. Future temperature projections, which are generated based on RegCM3 temperature output, present that air temperature will continue to increase in Karasu Basin. In parallel to historical data analysis, the largest warming is expected in summer temperature with an increase of 5°C for 2070-2100 time period. Autumn season temperature increase with 4.2°C is the second largest warming. Spring and winter seasons’ increase are lower than autumn and summer seasons but still quite significant with 3.8°C and 2.3°C respectively.
Modelled temperature and precipitation changes by RegCM3 in Karasu Basin for 2070- 2100 time slice are summarized in Table 6.27.
287 Table 6.27: Modelled changes in precipitation and air temperature in Karasu Basin over 2070-2100
Average Air Precipitation Change Temperature Change Winter +2.3 °C -5%
Spring +3.8 °C -11%
Summer +5 °C -15%
Autumn +4.2 °C +2%
6.3.2. Streamflow Projections in Karasu Basin
As proportional to precipitation decrease and the largest temperature increase in summer season, the largest flow decrease was projected in summer season. HEC-HMS projected 34% decrease in summer season streamflow, while LBRM estimated 45% decrease in summer flow in Karasu Basin. The second largest impact of climate change on Karasu Basin streamflow was projected for spring season according to both hydrological models. While LBRM predicted 32% decrease, HEC-HMS estimated 13% decrease in Karasu Basin’s spring streamflow. Moreover, both models reported similar amount of decreases in winter season flow with 10%. Finally, HEC-HMS projected 28% increase in autumn flows while LBRM predicted 20% increase in autumn season flows. Streamflow increase of autumn season can be explained by projected precipitation increase by RegCM3 and increasing amount of rainfall instead of snowfall resulting faster runoff relative to snow melt runoff in autumn season. Flow estimations of HEC- HMS and LBRM are shown graphically in Fig. 6.32.
288 250
200
150 HEC-HMS 2070-2100 (m3/s) LBRM 2070-2100 (m3/s) 100 Observed Flow (m3/s)
50
0 Jan Jun Oct Apr Feb Sep July Dec Aug Nov Mar May
Fig. 6.32: HEC-HMS and LBRM monthly average flow estimations for 2070-2100
Fig. 6.32 demonstrates that HEC-HMS runoff estimations are larger than LBRM runoff estimations. Groundwater component of models may be the main modelled flow difference between HEC-HMS and LBRM. Monthly constant groundwater values were used in HEC-HMS modelling. It is very challenging issue to predict (even make an assumption) future groundwater discharge. Thus, monthly groundwater values used in current simulations were utilized for future simulations in HEC-HMS. On the other hand, LBRM considers changes in climate while representing groundwater discharge. Groundwater representation difference of HEC-HMS and LBRM very likely resulted difference between HEC-HMS and LBRM flow estimations.
The most important impact of climate change on catchment flow at snow dominated basins is decrease in quantity and quality of the water. It is shown in Fig. 6.32 that both HEC-HMS and LBRM estimated considerable decreases in runoff amount in Karasu Basin. The second remarkable climate change impact that should be addressed at snow dominated basins is shifts in snow melting time due to changes in climate. It
289 means that peak flows in spring season are observed at earlier dates due to early snow melting. Historical flow data evaluation resulted that there is a slight shifting trend to earlier peak flow dates. According to available data (historical), earliest annual peak flow date is 7 April in Karasu Basin. HEC-HMS and LBRM showed strong agreement in peak flow date prediction in Karasu Basin and they projected that annual peak flow in Karasu Basin over 2070-2100 time period will shift around a month earlier and will be observed in early March. Effects of early snow melting on basins are increasing spring flooding and summer drought risks. In Karasu Basin, early snow melting influence will appear as summer drought. In addition to precipitation decrease, which is already projected by climate model in summer season, early snow melting will result crucial consequences in supplying sufficient amount of water to many sectors particularly in summer season.
6.3.3. Influences of Streamflow Decrease in Karasu Basin
Approximate annual water potential of Euphrates Basin is 30 billion m3. Akin and Akin (2007) stated Euphrates Basin’s water potential as 31.6 billion m3. However, annual water potential is computed as around 2.41 billion m3 in Karasu Basin. Technically and economically usable water potential is less than total water potential. As it is stated before, urban, agriculture, industry, hydro-electric power generation are the main sectors which use available water in water catchments. In Karasu Basin, urban & rural residents and agriculture are underlying water consuming factors. Industrial activities are quite poor that can be ignored in Karasu Basin. Furthermore, hydro-electrical power generation is not very significant in Upper Euphrates Basin.
Total population in Karasu Basin is currently 497 228 (TurkStat 2010). There are totally 10 residential regions in Karasu Basin. Increasing population coupled with decrease in available water due to climate change is crucial problem all over the world. Current population of Turkey is 72.6 million and it is foreseen 100 million in 2030 by Turkey Statistic Institute (Ministry of Environment and Forestry 2007). However, eastern parts of Turkey is not showing rapid increasing population trend. Socio-economic reasons in Turkey restrict eastern Turkey’s population. The most significant constrain on eastern 290 Turkey population is unemployment. As stated before, eastern Turkey is not heavily industrialized region so people especially young population immigrates to western Turkey. However, population of Karasu Basin region has not showed significant decrease so it is expected that population of Karasu Basin will be stable forthcoming two-three decades. It is not reasonable to make any further population prediction for Karasu Basin. It means domestic water demand will remain similar to current levels. Although domestic water demand is not expected to rise in Karasu Basin, decrease in streamflows owing to climate change can limit domestic water use.
There are many dams in Euphrates Basin which are used for agriculture, hydro-power generation and domestic water supply purposes. However, only four of them which are Kuzgun, Palandoken, Tercan and Erzincan are in Karasu Basin. Annual hydropower generation capacity of Kuzgun dam is 36 GWh, while hydropower generation capacity of Tercan is 51 GWh. Palandoken and Erzincan dams are not being used for the purpose of power production. Moreover, Askale Dam which is in Karasu Basin is projected and when it starts to operate, it is expected to generate 43.24 GWh hydropower annually (SHW 2010). Climate change has substantial influences on hydro-electrical power generation. Hence, changes in Karasu Basin streamflows quantity owing to climate change will directly influence hydropower generation in Karasu Basin. Vicuna et al. (2008) stated that increasing ratio of rainfall relative to snowfall due to increasing temperatures; earlier spring snow melt and changing flow seasonality owing to climate change may influence hydropower reservoirs. They expressed that a timing mismatch between energy generation and energy demand may be observed, moreover, higher inflows in winter season may decrease overall energy generation because of increasing spillage. Some more papers related to climate change impacts on hydropower generation can be found in literature (Bergstrom et al. 2001; Harrison & Whittington 2002; Schaefli, Hingray & Musy 2007). However, dams in Karasu Basin which are generating hydropower are quite small and total annual generation is currently 87 GWh in Karasu Basin. Moreover, only one more low capacity dam have been planned to construct in Karasu Basin. However, Yenigun, Gumus and Bulut (2008) and Kahya and Kalayci (2004) reported significant decreasing trends based
291 on historical streamflow data analysis at some flow stations in middle and lower Euphrates Basin where the largest dams of Turkey are located. Current study showed that despite there is no evidence of current streamflow decrease in Karasu Basin according to historical data trend analysis, climate change will have important impacts on Karasu Basin’s future streamflows particularly in spring and summer seasons. Although any modelling study is not available in middle and lower Euphrates Basin, significant future decreases in streamflow at middle and lower Euphrates Basin due to climate change can be certainly predicted.
Another important sector which will be affected by climate change is agriculture in Karasu Basin. Hydrological impact predictions stated significant decreases in summer and spring seasons. Irrigation water is vital for agriculture sector especially in summer season. RegCM3 projected the largest precipitation decreases in summer season. Moreover, hydrological models reported the largest streamflow decreases in summer season. Kuzgun, Palandoken, Tercan and Erzincan dams have been used for irrigation purposes in Karasu basin and total irrigation area of these dams are 81 522 ha. Moreover, there are irrigation structures in Daphan Valley, Tercan, Erzincan and Uzumlu and total irrigated area is 37 488 ha. Hence, total irrigated area in Karasu Basin is 119 010 ha. In addition to current irrigation plants, it has been planned to operate two more irrigation structures which are Daphan Valley second section irrigation and additional irrigation structure to Erzincan. Projected total additional irrigation area in Karasu Basin is 17739 ha (SHW 2010).
Minimum amount of water which is required for human activities including drinking, cooking, washing and laundry purposes is 50 litres per person per day (UCTEA 2009). In Karasu Basin, according to total population of 497 228 people, total annual domestic water demand can be stated as 2.5 107 m3 in Karasu Basin. However, it is not possible to make assumptions on agricultural and industrial water demands. Moreover, it is not plausible to make predictions on future agriculture, domestic, industry and hydropower demands. Current water potential of Karasu Basin is adequate to meet
292 water demand. Hydrological models projected significant declines in annual surface water potential which is demonstrated in Table 6.28.
Table 6.28: Annual surface water potential estimation of hydrological models HEC-HMS LBRM 109 109 billion m3 billion m3 2070-2100 2.17 1.79
According to Table 6.28, total annual surface water potential will decrease 26% based on LBRM flow predictions while it declines 10% according to HEC-HMS flow projections.
Finally, even there will be no increase in total water demand in Karasu Basin, due to decrease in future streamflow, some water problems can be expected in Karasu Basin. One should note that at least irrigation and hydropower generation water demand will increase in Karasu Basin. Industrial water need is a big question mark but it can be absolutely maintained that it will not decline, will be stable in an optimistic assumption. Hence, correct discussion of future water limitation in Karasu Basin can be performed by only estimating future water demands (domestic, agricultural, industrial, hydropower) correctly.
6.4. Uncertainties of Study
Although current study gives a beneficial idea about future climate and hydrological impacts of climate change on catchment streamflow, there are still uncertainties inherit in hydrological models, climate models and interface between these two. The most important uncertainty regarding to hydrological models is linked to hydrological calibration in hydrological impact studies. It is assumed that calibration will hold in the future. Although, calibration and validation were performed in HEC- HMS and LBRM by using warmer years (climate change finger print started to be observed after 1992), it is not reasonable to state no future change in calibrated parameters owing to changing climate. Using more physically based model can reduce 293 the calibration uncertainty; however, Jones et al. (2006) showed that using 10- parameter lumped model can perform as well as more physical complex distributed hydrological model. Georgakakos et al. (2004) recommended multi-model approach for better understanding of uncertainty coming from hydrological model calibration. However, Minville, Brissette and Leconte (2008) explained that uncertainty linked to global & regional climate models and interface is much more significant then uncertainty linked to hydrological model calibration.
Furthermore, drawbacks of HEC-HMS and LBRM are the inconsideration of land use and vegetation while performing simulations. They only use precipitation and temperature data as input. Moreover, both models are applied in lumped structure. It means 10 215 km2 catchment area was considered as a single unit. It is a common implementation when required data for distributed model is not available. Due to lack of required data for distributed model application in Karasu Basin, models are performed in lumped structure.
The most challenging issue in using climate model outputs for hydrological models is spatial resolution of climate models. Despite both global climate models and regional climate models contain the representation of hydrology, hydrological cycle is not resolved by them in suitable detail for hydrological applications (Bergstrom et al. 2001; Fowler, Blenkinsop & Tebaldi 2007). Due to low resolution of global climate models resulting insufficient representation of catchments, different methods have been developed to downscale resolution for the purpose of better representation of regions’ details. In current study, data generated by dynamic downscaling approach was used. Hence, spatial resolution of ECHAM5 was downscaled to 27 km by using RegCM3. Nonetheless, spatial resolution of RegCM3 is still coarse. For instance, if resolution of the climate model increases, representation of elevation, which is very dominant parameter for correct distribution of climate variables such as temperature and precipitation, increases and it results more accurate climate input for hydrological impact studies. In summarize, the finer the climate model resolution, the more accurate hydrological simulations and estimations.
294 Moreover, inconsistency and failure to predict precipitation, which is the most important input to perform accomplished hydrological simulations, is very substantial problem regarding to climate models. The largest impacts of future climate change on human will most likely occur owing to change in precipitation patterns and variability (Dai 2006). Because of complexity of precipitation processes in the atmosphere involving cloud microphysics, cumulus convection, planetary boundary layer processes and many others, it is very challenging for climate models to realistically predict the regional patterns, temporal variations, frequency and intensity of precipitation (Trenberth et al. 2003; Meehl et al. 2005).
Using multiple climate projections constituted by combination of more than one GCMs and greenhouse gases emission scenarios has been currently popular in hydrological impact studies and it is a reasonable approach to generate an impact interval of climate change on water resources (Minville, Brissette & Leconte 2008; Vicuna et al. 2008; Maurer 2007). In current study, data obtained by only one GCM downscaled by RCM and one greenhouse gases scenario (A2) is utilized due to data availability. Hence, it was not possible to present a change interval either for climate variables or flow.
Moreover, using delta change approach as an interface between RCM output and hydrological model limits the representation of RCM details in large catchments. In other words, the main deficiency of this approach is not to take into consideration of spatial and temporal structure of temperature and precipitation (Fowler et al. 2005). However, because of its stability, simplicity and its ability to give outcomes which can be connected to present conditions, it is commonly used in hydrological impact studies (Graham, Andreasson & Carlsson 2007).
295 Chapter 7: Conclusion and Recommendations
Climate change has substantial influences on hydrological cycles and water resources. Snow dominated basins are more sensitive to climate change. Owing to global warming, less snowfall has been observed in snow dominated basins resulting less snow melt runoff, which is very important especially in late spring and summer seasons.
Climate change has two basic effects on snow melt runoffs. Firstly, it causes alterations in quantity & quality of runoff and secondly, it results shifts in snow melting time which leads floods and/or droughts in snow dominated basins. Decrease in total water potential will result very crucial problems to allocate enough water to sectors including agriculture, industry, hydro-electric power generation and urban water supply.
It is aimed to predict future water availability of mountainous Upper Euphrates Basin (Karasu Basin) by predicting future streamflows in this study. Firstly, climate change fingerprint were investigated by using two popular non-parametric trend analysis methods, Mann-Kendal and Spearman’s Rho for climate variables including average, maximum, minimum temperatures, precipitation and streamflow data. Meteorological data from three meteorological stations within Karasu Basin and flow data from a station at the outlet of basin were analysed. Findings of historical data trend analysis are summarized as follows:
- There are statistically insignificant decreasing trends in precipitation in Karasu Basin for all seasons except autumn season. All meteorological stations showed upward precipitation trend in autumn season.
- Significant warming trends were detected for Karasu Basin’s temperatures. In Erzurum station, it is possible to discuss two distinct periods; one from 1975 to 1992 and the other from 1992 to 2008. In first period, decreasing trends were observed except maximum temperature, while there are significant warming trends since 1992.
296 - Maximum temperature warming trends are dramatic and occurred in all seasons for all stations.
- Summer season warming trend is more significant than all other seasons.
- Erzincan station temperature trends are more important and clear than Erzurum and Tercan stations.
- Tercan station temperature trends are the least significant among three meteorological stations.
- Temperature increases are mostly more important since 1992.
- There is an insignificant downward trend in Karasu Basin’s flow; moreover, very slight shift in snow melting time was detected.
Detected trends in Karasu Basin are in agreement with the trends stated by previous trend analysis studies performed in Turkey based on historical meteorological data (Demir et al. 2008; Tayanc et al. 2009). 1992 can be determined as the starting point when human induced climate change impacts have been commenced to be observed in Turkey. Temperature showed the lowest values in 1992 due to Mount Pinatubo eruption, when significant amount of particles were released into the stratosphere, acting as anti-greenhouse agents. It is explained by Tayanc et al. (2009) that most meteorological stations in Turkey demonstrated substantial decrease in temperature due to Mount Pinatubo eruption. After 1992 to date, there is an obvious increasing trend in minimum, maximum and average temperatures in all seasons. Temperature trends in Karasu Basin are quite similar to Turkey average temperature trends. Demir et al. (2008) stated that Turkey average temperature has showed similar trend to global average surface temperatures. Only difference between global average surface temperature trend and Turkey’s average surface temperature is commencing time
297 period of them. Global average surface temperature has shown rapid increasing trend since 1980s, while Turkey average temperature rising trend was observed since 1990s.
It should be noted that expected influences of climate change on middle and lower Euphrates Basin will be more significant than Upper Euphrates Basin. Because of current study focus (snow hydrology), mountainous Upper Euphrates Basin was chosen as a study area. But, largest dams of Turkey and broad irrigation areas are located in middle and lower Euphrates Basin. Dams in Euphrates Basin are used for the purposes of domestic water supply, irrigation, hydroelectric power generation. Ataturk, Keban and Karakaya dams are the most significant ones amongst these. Most of the water in these dams comprises of snow melt (60%-70% of all streamflow).
It is found that there is no statistically significant decreasing trend in streamflows in Karasu Basin. However, previous studies showed that there are already significant decreasing trends in lower Euphrates streamflows based on historical flow data analysis (Yenigun, Gumus & Bulut 2007).
Secondly, snow melt runoffs in Karasu Basin were simulated by using ANN based hydrological model and conventional hydrological models including HEC-HMS and LBRM. Firstly, streamflows of Karasu Basin were modelled by developing an ANN based model which is currently popular method in hydrological studies. ANN was selected owing to its capability to model non-linear hydrological processes successfully. Moreover, use of ANN does not require high expertise on in-depth hydrological processes. Finally, ANN has relatively low computational demands and it is possible to integrate it with other mathematical tools easily (De Vos & Rientjes 2005).
Most of the previous ANN based hydrological simulation studies dealt with rainfall contributed runoff simulations only. It is more difficult to simulate runoff process in snow dominated basin in compared to runoff process in rainfall dominated basin. Because, for snow dominated basin; both runoffs from rainfall and snow melt need to be considered. Moreover, physical processes of snow accumulation and melting are
298 highly complex, including “mass and energy balances as well as heat and mass transport by conduction, vapour diffusion and meltwater drainage” (Tarbaton & Luce 1996).
To find most suitable results in ANN modelling, it is important to select correct input data set to simulate snow melt runoffs. Different combinations of input data sets among precipitation, air temperature, temperature range, humidity and wind speed data were used to explore best input data set. Four different ANN models were developed based on different input data set for both annual and seasonal analyses. It is found that for both annual and seasonal analyses, Model 4 showed the best performance to simulate runoffs in Karasu Basin. For Model 4, input data set consists of precipitation, air temperature, humidity, wind speed and temperature range. However, due to the drawback of ANN to find global optima in complex parameter spaces, performances of models are not too high but good enough. Because of ANN’s incapability to simulate large variations, both low flows and peak flows were not matched to expected accuracy. To overcome this issue, a seasonal analysis and modelling was performed with only peak-flow season (spring-summer) data.
It is found that the ANN model is sensitive to seasonal simulations. Seasonal runoff simulations involving data between March and August were run in addition to annual (whole year) simulations. It is found that model performances increased significantly for seasonal simulations compared to annual simulations. For Model 4, linear regression coefficient increased from 0.78 to 0.88 for training phase and increased from 0.52 to 0.71 for test phase. Also, Nash-Sutcliffe coefficient of determination was increased significantly. For training phase it increased from 0.71 to 0.81 and for test phase increased from 0.50 to 0.70.
ANN model performance to simulate snow runoffs is good enough especially when considering spring – summer periods, which are in fact most important seasons in terms of runoff volumes. As a further step, it is recommended to use snow data such as snow depth or snow water equivalent as input data with other compatible inputs. It
299 will most likely increase model performance in estimating snow runoffs from snow dominated basins.
Then, conventional hydrological models, HEC-HMS and LBRM, were utilized for snow melt runoff simulation in Karasu Basin. Both models were applied to study basin at daily time step from 1997 to 2004. However, for calibration run data from 1 January, 1997 to 31 December, 2002 was selected and for validation run data from 1 January, 2003 to 31 December, 2004 was selected. Model performance assessments were performed by using three popular criterion in hydrology; Nash-Sutcliffe coefficient of determination, mean squared relative error and linear regression coefficient. For calibration run, LBRM has shown slightly better performance than HEC-HMS. However, HEC-HMS performance was better for validation run. In fact, model performances are quite similar for LBRM and HEC-HMS. Maximum Nash- Sutcliffe coefficient was achieved as 0.76 for HEC-HMS model and similarly maximum linear regression coefficient was achieved as 0.77 for HEC-HMS model. For calibration run, where performance is worse with HEC-HMS, both the Nash-Sutcliffe coefficient of determination and linear regression coefficient was 0.70 and mean squared relative error was 0.28. When it is considered that models are applied in lumped structure and basin is relatively large with an area of 10 215 km2, model performances are acceptable.
An application of distributed hydrological model could not be performed in Karasu Basin owing to data limitations. For a distributed model, in addition to more distributed meteorological and flow data, digital elevation map (DEM) is a requirement for the determination of flow path in terms of elevation of each cell. Moreover, detailed soil and land use map is necessary for distributed model application. As a further step, using distributed hydrological model to simulate and forecast runoffs in Karasu Basin can be recommended. It may be either conceptual or physically based distributed model. However, due to data requirement and computational difficulties, it is challenging to apply distributed physically based model in Karasu Basin. Thus, for
300 Karasu Basin, streamflow simulation and forecasting using distributed conceptual hydrological model is recommended.
As a next step, dynamically downscaled climate data, which are generated by GCM designated as ECHAM5 and RCM named RegCM3 was preceded for Karasu Basin. Then, climate variables were assessed in an interface before using them as an input to hydrological models. HEC-HMS and LBRM were selected for future hydrological predictions in Karasu Basin. Data availability and basin conditions were two main factors affected hydrological model selection. ANN model could not be used due to lack of future climate data to perform ANN model.
Hydrological predictions were achieved over the period of 2070-2100. A2 greenhouse emission scenario was preferred to generate climate input of future streamflow prediction models. RegCM3 resulted temperature increases for all seasons between 2.3°C and 5 °C. Largest temperature increase was projected in summer season and the lowest increase was estimated in winter season. Second largest increase was projected in autumn then spring seasons. RegCM3 outcomes precipitation decreases in summer, spring and winter seasons. However, it resulted precipitation increase in autumn season. Precipitation decreases were estimated 5% in winter, 11% in spring and 15% in summer. Precipitation increase was 2% in autumn season. Future temperature and precipitation projections of RegCM3 are in agreement with results of historical climate data trends.
LBRM projected streamflow decreases with percentage of 45, 32 and 10 in summer, spring and winter seasons respectively. It predicted 20% streamflow increase in autumn season. HEC-HMS predicted smaller changes for Karasu Basin’s future streamflow due to very likely different groundwater component methods. HEC-HMS projected 34%, 13% and 10% runoff declines in summer, spring and winter seasons and 28% increase in autumn streamflow.
301 Total available surface water potential of Karasu Basin is enough for basin water needs currently. Agriculture and residents are the primary water consumers in Karasu Basin. It is realized that due to decreases in Karasu Basin streamflows, water limitations can be experienced in future. More irrigation structures will be took into operation by General Directorate of State Hydraulic Works in further years. It means at least agricultural water demand will show increase in Karasu Basin. Industrial water demand of basin is uncertain currently and basin is not an industrial area. In summary, even water demands of different sectors would be stable; Karasu Basin will probably face with water problems in future. But, for accurate water management strategies, it is very significant to develop projects to predict current and future water demands of different sectors in the basin. Furthermore, clarification of groundwater potential of basin and climate change impact evaluation on basin’s groundwater potential is substantial.
It should be noted that expected influences of climate change on middle and lower Euphrates Basin will be more significant than Upper Euphrates Basin. Previous studies showed that there is already decreasing trend in lower Euphrates streamflows based on historical flow data analysis. It is also known that lower parts of Euphrates Basin will be more affected by climate change and more warming is expected to be observed in lower Euphrates Basin. Moreover, at lower parts of Euphrates Basin industrial activities are denser than upper parts. In addition, one of the most important projects of Turkish Republic, Southeastern Anatolia Project, is under construction in some parts of Euphrates Basin. Under this project, construction of more dams is planned in different parts of Euphrates Basin. Thus, water demands for hydropower generation and agriculture will definitely increase. Moreover, an increase in industrial activities of lower parts of basin may be expected. So, high probably, industrial water demand will increase as well. In summary, lower parts of Euphrates Basin is more sensitive to climate change than upper parts of Euphrates Basin owing to more intensive water demand. Water limitations and more severe water problems will be very likely monitored in lower Euphrates Basin and water allocation among different sectors will be an important problem at this part of basin.
302 There are uncertainties coming along with climate models, hydrological models and interface between climate and hydrological models. The most significant problem arising from climate models is spatial resolution inadequacy. Another important problem related to climate models is low accuracy to predict precipitation. Structure of hydrological models and calibration problems inherit in future streamflow predictions are uncertainties coming with hydrological models. Finally, inconsideration of spatial and temporal structure of temperature and precipitation is the most substantial deficiency coming from the delta change method to transfer climate change signal from climate model to hydrological model.
This study can be stated as an initial step to investigate climate change impacts on hydrology of Euphrates Basin, particularly Upper Euphrates Basin. Following recommendations should be taken into account for further studies in Euphrates Basin:
-Spatial resolution of regional climate model should be increased. Finer resolution will give better representation of catchment resulting more accurate hydrological simulations and predictions.
-More than one global climate model and regional climate model coupled with multiple greenhouse gases scenarios should be used to determine future climate. Thus, instead of giving solid results of future climate variables, an interval for climate variables can be presented. It will also allow getting multiple streamflow responds from basin. Hence, it will be possible to generate an interval for future catchment streamflow.
- Different hydrological models with varied structures should be used. For instance, if required data can be collected, distributed models should be applied in Karasu Basin. Furthermore, more physically meaningful models which require more data may be experienced in Basin.
303 - Different interface approaches such as delta change approach, daily scaling and bias correction should be applied in Karasu Basin.
- Modelling studies should be extended to other parts of Euphrates Basin. Especially, modelling studies in lower Euphrates Basin are substantial due to its sensitivity to climate change and its intense hydro-electric generation and irrigation activities.
-Studies on investigation of current and future water demand of agriculture, hydropower generation and urban should be developed.
304 References
Abdo, KS, Fiseha, BM, Rientjes, THM, Gieske, ASM & Haile, AT 2009, ’Assessment of climate change impacts on the hydrology of Gilgel Abay catchment in Lake Tana basin, Ethiopia’, Hydrol. Process., vol. 23, no. 26, pp. 3661–3669.
Abdrabbo, MAA, Khalil, AA, Hassanien, MKK & Abou-Hadid, AF 2010, ‘Sensitivity of Potato Yield to Climate Change’, Journal of Applied Sciences Research, vol. 6, no. 6, pp. 751-755.
Abou-Hadid, AF, Mougou, R, Mokssit, A & Iglesias, A 2003, Assessment of Impacts, Adaptation, and Vulnerability to Climate Change in North Africa: Food Production and Water Resources, AIACC AF90 Semi-Annual Progress Report.
Abraha, MG & Savage, MJ 2006, ‘Potential impacts of climate change on the grain yield of maize for the midlands of KwaZulu-Natal, South Africa’, Agriculture, Ecosystems and Environment, vol. 115 pp. 150–160.
Adam, JC, Hamlet, A F & Lettenmaier, DP 2009, ’Implications of global climate change for snowmelt hydrology in the twenty-first century’, Hydrol. Process., vol. 23, no. 7, pp. 962–972.
Adler, RF, Susskind, J, Huffman, GJ, Bolvin, Ferraro, R,D, Nelkin, E, Chang, A, Gruber, A, Xie, P, Janowiak, J, Rudolf, B, Schenider, U, Curtis, S & Arkin, P 2003, ‘The version 2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979– present)’, J. Hydrometeorol., vol. 4, pp. 1147–1167.
Agriculture and Rural Affairs Ministry 2008, Turkiye Tarimsal Kuraklikla Mucadele Stratejisi ve Eylem Plani (Strategy and Action Plan for Agricultural Drought in Turkey), Turkish Government, Ankara.
305 Aqil, M, Kita, I, Yano, A & Nishiyama, S 2007, ’A comparative study of artificial neural networks and neuro-fuzzy in continuous modelling of the daily and hourly behaviour of runoff’, Journal of Hydrology, vol. 337, no. 1-2, pp. 22-34.
Akhtar, MK, Corzo, GA, van Andel, SJ & Jonoski, A 2009, ‘River flow forecasting with artificial neural networks using satellite observed precipitation pre-processed with flow length and travel time information: case study of the Ganges river basin’, Hydrology and Earth System Sciences, vol. 13, pp. 1607-1618.
Akhtar, M, Ahmad, N & Booij, MJ 2009, ‘Use of regional climate model simulations as input for hydrological models for the Hindukush-Karakorum-Himalaya region’, Hydrol. Earth Syst. Sci.,vol.5 , no. 2, pp. 865-902.
Akin, M & Akin, G 2007, ‘Suyun Önemi , Türkiye’de Su Potansiyeli , Su Havzalari ve Su Kirliliği (Importance of Water, Water Potential in Turkey, Water Catchments and Water Pollution)’, Ankara Üniversitesi Dil ve Tarih-Coğrafya Fakültesi Dergisi, vol. 47, no. 2, pp. 105-118.
Al-Bakri, J, Suleiman, A, Abdulla, F & Ayad, J 2010, ‘Potential impact of climate change on rainfed agriculture of a semi-arid basin in Jordan’, Physics and Chemistry of the Earth, doi:10.1016/j.pce.2010.06.001.
Alcamo, J, Flörke, M & Marker, M 2007, ‘Future long-term changes in global water resources driven by socio-economic and climatic change’, Hydrol. Sci. J., vol. 52, pp. 247–275.
Alcamo, J, Dronin, N, Endejan, M, Golubev, G & Kirilenko, A 2007, ‘A new assessment of climate change impacts on food production shortfalls and water availability in Russia’, Global Environmental Change, vol. 17, pp. 429–444.
306 Almaraz, JJ, Mabood, F, Zhou, X, Gregorich, EG, & Smith, DL 2008, ‘Climate change, weather variability and corn yield at a higher latitude locale: Southwestern Quebec’, Climatic Change, vol. 88, pp. 187–197.
Andréasson, J, Bergström, S, Carlsson, B, Graham, LP & Lindström, G 2004, ‘Hydrological change – climate change impact simulations for Sweden’, Ambio, vol. 33, pp. 228–234.
Anisimov, OA, Vaughan, DG, Callaghan, TV, Furgal, C, Marchant, H, Prowse, TD, Vilhjálmsson, H & Walsh, JE 2007, ‘Polar regions (Arctic and Antarctic)’, in ML Parry, OF Canziani, JP Palutikof, PJ van der Linden & CE Hanson (eds) 2007, Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, pp. 653-685.
Anthes, R, Hsie, E & Kuo, Y-H 1987, Description of the Penn State / NCARMesoscale Model Version 4 (MM4), NCAR Tech. Note, NCAR/TN-282, National Center for Atmospheric Research.
Anwar, MR, O’Leary, G, McNeil, D, Hossain, H & Nelson, R 2007, ‘Climate change impact on rainfed wheat in south-eastern Australia’, Field Crops Research, vol. 104, pp. 139–147.
Arnell, NW 2003, ‘Relative effects of multi-decadal climatic variability and changes in the mean and variability of climate due to global warming: future streamflows in Britain’, J. Hydrol., vol. 270, pp. 195-213.
Arnell, NW 2004, ‘Climate change and global water resources: SRES emissions and socio economic scenarios’, Global Environmen. Chang., vol. 14, pp. 31–52.
307 ASCE Task Committee 2000, ‘Artificial neural networks in hydrology— I: preliminary concepts’, J. Hydrol. Eng., vol. 5, 115–123.
Aydinalp, C & Cresser, MS 2008, ‘The Effects of Global Climate Change on Agriculture’, American-Eurasian J. Agric. & Environ. Sci., vol.3, no. 5, pp. 672-676. Baltas, EA 2007, ‘Impact of Climate Change on the Hydrological Regime and Water Resources in the Basin of Siatista’, Water Resources Development, vol. 23, no. 3, pp. 501–518.
Barnett, TP, Malone, R, Pennell, W, Stammer, D, Semtner, B & Washington, W 2004, ‘The effects of climate change on water resources in the West: introduction and overview’, Climatic Change, vol. 62, no. 1-2, pp. 1–11.
Barnett, TP, Adam, JC & Lettenmaier, DP 2005, ‘Potential impacts of a warming climate on water availability in snow-dominated regions’, Nature Publishing Group, vol. 438, pp. 303-309.
Barnett, TP, Pierce, DW, Hidalgo, HG, Bonfils, C, Santer, BD, Das T, BG, Wood, AW, Mirin, AA, Cayan, DR & Dettinger, MD 2008, ‘Human-induced changes in the hydrology of the western United States’, ScienceExpress, 10.1126/science.1152538.
Basheer, IA & Najjar, YM 1995, ‘Designing and 308odelling fixedbed adsorption systems with artificial neural networks’, J. Envir. Syst., vol. 23, no. 3, pp. 291–312.
Basistha, A, Arya, DS & Goel, NK 2009, ‘Analysis of Historical Changes in rainfall in the Indian Himalayas’, International Journal of Climatology, vol. 29, pp. 555-572.
Bates, BC, Kundzewicz, ZW, Wu, S & Palutikof, JP Eds 2008, Climate Change and Water, Technical Paper of the Intergovernmental Panel on Climate Change, IPCC Secretariat, Geneva.
308 Beck, C, Grieser, J & Rudolf, B 2005, ‘A new monthly precipitation climatology for the global land areas for the period 1951 to 2000’, Geophysical Research Abstracts, vol. 7, 07154.
Beldring, S, Engen-Skaugen, T, Forland, EJ & Roald, LA 2008, ‘Climate change impacts on hydrological processes in Norway based on two methods for transferring regional climate model results to meteorological station sites’, Tellus, vol. 60, no. 3, pp. 439– 450.
Benhin, JKA 2006, ‘Climate Change and South African Agriculture: Impacts and Adaptation Options’, CEEPA Discussion Paper No. 21 Centre for Environmental Economics and Policy in Africa, University of Pretoria.
Bergstrom, S, Carlsson, B, Gardelin, M, Lindstrom, G, Pettersson, A & Rummukainen, M 2001, ‘Climate change impacts on runoff in Sweden—assessments by global climate models, dynamical downscaling and hydrological modelling’, Clim. Res., vol. 16, pp. 101–112.
Betts, A 1986, ‘A new convective adjustment scheme Part I: Observational and theoretical basis’, Quart. J. Roy. Meteor. Soc., vol. 112, pp. 677-691.
Bi, W & Li, K 2004, ‘Adjusting industrial structure and constructing green, clean agricultural zone’, http://www.sciencetimes.com.cn/col116/col149/article.htm.
Bindoff, NL, Willebrand, J, Artale, V, Cazenave, A, Gregory, J, Gulev, S, Hanawa, K, Le Quéré, C, Levitus, S, Nojiri, Y, Shum, CK, Talley, LD & Unnikrishnan, A 2007, ‘Observations: Oceanic Climate Change and Sea Level’, in S Solomon, D Qin, M Manning, Z Chen, M Marquis, KB. Averyt, M Tignor & HL Miller (eds) 2007, Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 385-432.
309 Birsan, MV, Molnar, P, Burlando, P & Pfaundler, M 2005, ‘Streamflow Trends in Switzerland’, Journal of Hydrology, vol. 314, no. 1-4, pp. 312-329.
Bobba, A, Singh, V, Berndtsson, R & Bengtsson, L 2000, ‘Numerical simulation of saltwater intrusion into Laccadive Island aquifers due to climate change’, J. Geol. Soc. India, vol. 55, pp. 589–612.
Boomiraj, K, Chakrabarti, B, Aggarwal, PK, Choudhary, R & Chander S 2010, ‘Assessing the vulnerability of Indian mustard to climate change’, Agriculture, Ecosystems and Environment, vol. 138, pp. 265–273.
Bouza-deano, R, Ternero-Rodriguez, M & Fernandez-Espinosa, AJ 2008, ‘Trend study and assessment of surface water quality in the Ebro River (Spain)’, Journal of Hydrology, vol. 361, no. 3-4, pp. 227-239.
Bradley, RS, Keimig, FT & Diaz, HF 2004, ‘Projected temperature changes along the American cordillera and the planned GCOS network’, Geophys. Res. Lett., vol. 31, L16210, doi:10.1029/2004GL020229.
Brassard, JP & Singh, B 2008, ‘Impacts of climate change and CO2 increase on agricultural production and adaptation options for Southern Que´bec, Canada’, Mitig Adapt Strateg Glob Change, vol. 13, pp. 241–265.
Briegleb, B 1992, ‘Delta-Eddington Approximation for Solar Radiation in the NCAR Community Climate Model’, J. Geophys. Res, vol. 97, pp. 7603-7612.
Birikundavyi, S, Labib, R, Trung, HT & Rousselle, J 2002, ‘Performance of Neural Networks in Daily Flow Forecasting’, Journal of Hydrological Engineering, vol. 7,no. 5, pp. 392-398.
310 Brown, RD 2000, ‘Northern hemisphere snow cover variability and Change, 1915-97’, J. Clim., vol. 13, pp. 2339–2355.
Bruinsma, J 2003, World Agriculture: Towards 2015/2030. An FAO Perspective, Earthscan, London.
Burns, DA, Klaus, J & McHale, MR 2007, ‘Recent climate trends and implications for water resources in the Catskill Mountain region, New York, USA’, Journal of Hydrology, vol.336, no. 1-2, pp. 155–170.
Buttle, J, Muir, JT & Frain, J 2004, ‘Economic impacts of climate change on the Canadian Great Lakes hydro-electric power producers: a supply analysis’, Can. Water Resour. J., vol. 29, pp. 89–109.
Caldag, B & Saylan, L 2009, ‘Validating the Adaptation of Paddy Rice to Different Scenarios Using a Climate Change Impact Model in Northwestern Turkey’, Impact of Climate Change and Adaptation in Agriculture, International Symposium, Vienna.
Caloiero, T, Coscarelli, R, Ferraric, E & Mancinia, M 2009, ‘Trend detection of annual and seasonal rainfall in Calabria (Southern Italy)’, International Journal of Climatology, DOI: 10.1002/joc.2055.
Cayan, DR, Luers, AL, Franco, G, Hanemann, M, Croes, B & Vine, E 2007, ‘Overview of the California climate change scenarios project’, Clim. Change, vol. 72, issue 1, pp.1– 6.
Chang, H, Franczyk, J, Im, E-S, Kwon, W-T, Bae, D-H & Jung, I-W 2007, ‘Vulnerability of Korean water resources to climate change and population growth’, Water Science & Technology, vol. 56, no. 4, pp. 57–62.
311 Changchun, X, Yaning, C, Weihong, L, Yapeng, C & Hongtao, G 2007, ‘Potential impact of climate change on snow cover area in the Tarim River Basin’, Environmental Geology, vol. 53, pp. 1465-1474.
Chanseng, HT & Croley, TE 2007, ‘Application of a distributed large basin runoff model in the Great Lakes basin’, Control Engineering Practice, vol. 15, no. 8, pp. 1001-1011.
Charbonneau, R & Lardeau JP 1981, ‘Problems of modelling a high mountainous drainage basin with predominant snow yields’, Hydrological Sciences Bulletin, vol. 26, no. 4, pp. 345-361.
Chavas, DR, Izaurralde, RC, Thomson, AM & Gao, X 2009, ‘Long-term climate change impacts on agricultural productivity in eastern China’, Agricultural and Forest Meteorology, vol. 149, pp. 1118–1128.
Chen, C, Gillig, D & McCarl, BA 2001, ‘Effects of climatic change on a water dependent regional economy: a study of the Texas Edwards Aquifer’, Climatic Change, vol. 49, no. 4, pp. 397–409.
Chen, M, Xie, P & Janowiak, JE 2002, ‘Global land precipitation: a 50-yr monthly analysis based on gauge observations’, J. Hydrometeorol., vol. 3, pp. 249–266.
Chen, Z, Grasby, S & Osadetz, K 2004, ‘Relation between climate variability and groundwater levels in the upper carbonate aquifer, southern Manitoba, Canada’, J. Hydrol., vol. 290, no. 1–2, pp. 43–62.
Chen, Y, Takeuchi, K, Xu, C, Chen, Y & Xu, Z 2006, ‘Regional climate change and its effects on river runoff in the Tarim Basin, China’, Hydrol. Process., vol. 20, pp. 2207– 2216.
312 Choi, O & Fisher, A 2003, ‘The impacts of socioeconomic development and climate change on severe weather catastrophe losses: Mid-Atlantic Region MAR and the U.S.’, Climatic Change, vol. 58, no. 1–2, pp. 149–170.
Ciais, P, Reichstein, M, Viovy, N, Granier, A, Ogee, J, Allard, V, Aubinet, M & Buchmann, N 2005, ‘Europe-wide reduction in primary productivity caused by the heat and drought in 2003’, Nature, vol. 437, pp. 529–533.
Cigizoglu, HK 2005,’ Application of generalized regression neural networks to intermittent flow forecasting and estimation’, Journal of Hydrologic Engineering, vol. 10, no. 4, 336–341.
Clarke, R & King, J 2004, The Atlas of Water, Earthscan, London.
Cline W 2007, Global Warming and Agriculture: Impact Estimates by Country, Center for Global Development and Peterson Institute for International Economics, Washington DC.
Coulibaly, P, Anctil, F, Aravena, R & Bobe’e, B 2001, ‘Artificial neural network modeling of water table depth fluctuations’, Water Resour.Res., vol. 37, pp. 885–896.
Crimp, SJ 2000, Possible Benefits, Possible Drawbacks, Department of Natural Resources, www.longpaddock.qld.gov.au/ClimateChanges/slides/dnrPI2.html.
Croley, TE, Quinn, FH, Kunkel, KE & Changnon, SJ 1998, ‘Great Lakes hydrology under transposed climates’, Climatic Change, vol. 38, pp. 405–433.
Croley, TE 2002, ‘Large basin runoff model’, in V Singh, D Frevert & S Meyer (eds)2002, Mathematical models in watershed hydrology, Water Resources Publications, Littleton, Colo. pp. 717–770.
313 Croley, TE & He, C 2005, ‘Distributed-parameter large basin runoff model I: Model development’, Journal of Hydrologic Engineering, vol. 10, no. 3, pp. 173–181.
Dai, A, Trenberth, KE & Qian, T 2004, ‘A global data set of Palmer Drought Severity Index for 1870–2002: relationship with soil moisture and effects of surface warming’, J. Hydrometeorol., vol. 5, pp. 1117–1130.
Dai, A 2006, ‘Precipitation Characteristics in Eighteen Coupled Climate Models’, Journal of Climate, vol. 19, pp. 4605-4630.
Dalfes, HN, Karaca, M & Sen, OL 2007, ‘Climate Change Scenarios for Turkey’, United Nations Development Programme- Climate Change & Turkey, pp. 11-17.
Daliakopoulos, IN, Coulibalya, P & Tsanis, IK 2005, ‘Groundwater level forecasting using artificial neural networks’, J. Hydrol., vol. 309, pp. 229–240.
Dawson, CW & Wilby, RL 2001, ‘Hydrological modelling using artificial neural networks’, Progress in Physical Geography, vol. 25, pp. 80-108.
Dawson CW & Wilby RL 2007, Statistical Downscaling Model SDSM, version 4.1, Department of Geography, Lancaster University, https://copublic.lboro.ac.uk/cocwd/SDSM/main.html.
De Silva, CS, Weatherhead, EK, Knox, JW &Rodriguez-Diaz, JA 2007, ‘Predicting the impacts of climate change—A case study of paddy irrigation water requirements in Sri Lanka’, Agricultural water management, vol. 93, pp. 19 – 29.
De Vos, NJ & Rientjes, THM 2005, ‘Constraints of artificial neural networks for rainfall- runoff modelling: trade-offs in hydrological state representation and model evaluation’, Hydrology and Earth Sciences, vol. 9, pp. 111-126.
314 Dellal, I, Butt, T, McCarl, B & Dyke, P 2004, ‘Economic Impact of Climate Change on Turkish Agriculture’, Ankara Climate Change Conference, Turkey.
Dellal, I & McCarl, B 2010, ‘The Economic Impact of Drought on Agriculture: The case of Turkey’, Second International Conference on Drought Management, Istanbul.
Demir, I, Kilic, G, Coskun, M & Sumer, UM 2008, ‘Türkiye’de Maksimum, Minimum ve Ortalama Hava Sıcaklıkları ile Yagıs Dizilerinde Gözlenen Degisiklikler ve Egilimler (Observed Changes and Trends in Maximum, Minimum and Average Air Temperatures and Precipitation in Turkey)’, TMMOB Iklim Degisimi Sempozyumu, Ankara, 13-14 Mart 2008.
Dibike, YB & Coulibaly, P 2005, ‘Hydrologic impact of climate change in the Saguenay watershed: comparison of downscaling methods and hydrologic models’, Journal of Hydrology, vol. 307, pp. 145–163.
Dickinson, R, Errico, R, Giorgi, F & Bates, G 1989, ‘A regional climate model for the western united states’, Clim. Change, vol. 15, pp. 383-422.
Dickinson, R, Henderson-Sellers, A & Kennedy, PJ 1993, Biosphere-Atmosphere Transfer Scheme (BATS) version 1E as coupled to the NCAR Community Climate Model, Tech. Rep. TN-387+STR, NCAR, Boulder, Colorado.
Dolling, OR & Varas, EA 2002, ‘Artificial neural networks for streamflow prediction’, Journal of Hydraulic Research, vol. 40, no. 5, pp. 547-554.
Downing, TE, Butterfield, RE, Edmonds, B., Knox, JW, Moss, S, Piper, BS, Weatherhead, EK & the CCDeW Project Team, 2003: Climate change and the demand for water, Research Report, Stockholm Environment Institute, Oxford Office, Oxford.
315 Döll, P 2002, ‘Impact of climate change and variability on irrigation requirements: a global perspective’, Climatic Change, vol. 54, 269–293.
Döll, PM, Flörke, MM & Vassolo, S 2003, ‘Einfluss des Klimawandels auf Wasserressourcen und Bewässerungswasserbedarf: eine globale Analyse unter Berücksichtigung neuer Klimaszenarien (Impact of climate change on water resources and irrigation water requirements: a global analysis using new climate change scenarios)’, Klima–Wasser– Flussgebietsmanagement – im Lichte der Flut, H.-B. Kleeberg, Ed., Proc. Tag der Hydrologie 2003 in Freiburg, Germany, Forum für Hydrologie und Wasserbewirtschaftung, 11–14.
Döll, P & Flörke M 2005, ‘Global-scale estimation of diffuse groundwater recharge’, Frankfurt Hydrology Paper 03, Institute of Physical Geography, Frankfurt University, Frankfurt.
Dressler, KA, Leavesley, GH, Bales, RC & Fassnacht, SR 2006, ‘Evaluation of gridded snow water equivalent and satellite snow cover products for mountain basins in a hydrologic model’, Hydrological Process, vol. 20, pp. 673–688.
Dye, DG 2002, ‘Variability and trends in the annual snow-cover cycle in Northern Hemisphere land areas, 1972–2000’, Hydrolog. Process., vol. 16, pp. 3065–3077.
Easterling, WE, Aggarwal, PK, Batima, P, Brander, KM, Erda, L, Howden, SM, Kirilenko, A, Morton, J, Soussana J-F, Schmidhuber, J & Tubiello, FN 2007, ‘Food, fibre and forest products’, in ML Parry, OF Canziani, JP Palutikof, PJ van der Linden & CE Hanson (eds), Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, UK, pp. 273-313.
316 Elgaali, E & Garcia, LA 2007, ‘Using Neural Networks to Model the Impacts of Climate Change on Water Supplies’ Journal of Water Resources Planning and Management, vol. 133, no. 3, pp. 230-243.
Elguindi N, Bi, X, Giorgi, F, Nagarajan, B, Pal, J & Solmon, F 2004, RegCM Version 3.0 User’s Guide, ICTP, Trieste, Italy.
Elsner, MM, Cuo, L, Voisin, N, Deems, JS, Hamlet, AF, Vano, JA, Mickelson, K EB, Lee, S & Lettenmaier, DP 2010, ‘Implications of 21st century climate change for the hydrology of Washington State’, Climatic Change, DOI: 10.1007/s10584-010-9855-0.
Erda, L, Wei, X, Hui, J, Yinlong, X, Yue, L, Liping, B & Liyong, X 2005, ‘Climate change impacts on crop yield and quality with CO2 fertilization in China’, Phil. Trans. R. Soc., vol. 360, pp. 2149-2154.
Evans, E, Ashley, R, Hall, J, Penning-Rowsell, E, Saul, A, Sayers, P, Thorne, C, & Watkinson, A 2004, Foresight. Future Flooding. Scientific Summary: Volume 1. Future Risks and their Drivers, Office of Science and Technology, London.
Ewert, F, Rounsevell, MDA, Reginster, I, Metzger, MJ & Leemans, R 2005, ‘Future scenarios of European agricultural land use I. Estimating changes in crop productivity’, Agriculture, Ecosystems and Environment , vol. 107, pp. 101–116.
Falarz, M 2002, ‘Long-term variability in reconstructed and observed snow cover over the last 100 winter seasons in Cracow and Zakopane (southern Poland)’, Clim. Res., vol. 19, no.3, pp. 247–256.
Firat, M & Gungor, M 2007, ‘River flow estimation using adaptive neuro fuzzy inference system’, Mathematics and Computers in Simulation, vol. 75, pp. 87-96.
317 Fischer, G, Shah, M, Tubiello, FN & van Velhuizen, H 2005, ‘Socio-economic and climate change impacts on agriculture: an integrated assessment, 1990 -2080’, Phil. Trans. R. Soc.,vol. 360, pp. 2067-2083
Fischer, G, Tubiello, FN, van Velthuizen, H & Wiberg, D 2007, ‘Climate change impacts on irrigation water requirements: global and regional effects of mitigation, 1990– 2080’, Tech. Forecasting Soc. Ch., vol. 74, no. 7, pp. 1083-1107.
Fischlin, A, Midgley, GF, Price, JT, Leemans, R, Gopal, B, Turley, C, Rounsevell, MDA, Dube, OP, Tarazona, J, & Velichko, AA 2007, ‘Ecosystems, their properties, goods, and services’, in ML Parry, OF Canziani, JP Palutikof, PJ van der Linden & CE Hanson (eds) 2007, Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, pp. 211-272.
Fleischera, A, Lichtmana, I & Mendelsohn, R 2008, ‘Climate change, irrigation, and Israeli agriculture: Will warming be harmful?’, Ecological Economics, vol. 65, pp. 508 – 515.
Food and Agriculture Organization (FAO) 2003, World Agriculture Towards 2015/2030, http://www.fao.org/documents/show_cdrasp?url_file=/docrep/004/y3557e/y3557e0 0.htm.
Fowler, HJ, Ekstrom, M, Kilsby, CG & Jones, PD 2005, ‘New estimates of future changes in extreme rainfall across the UK using regional climate model integrations. 1. Assessment of control climate’, Journal of Hydrology, vol. 300, pp. 212–233.
Fowler, HJ, Blenkinsop, S & Tebaldi, C 2007, ‘Linking climate change modelling to impacts studies: recent advances in downscaling techniques for hydrological modelling’, Int. J. Climatol., vol. 27, pp. 1547–1578.
318 Fujihara, Y,Tanaka, K, Watanabe, T, Nagano, T & Kojiri, T 2008, ‘Assessing the impacts of climate change on the water resources of the Seyhan River Basin in Turkey: Use of dynamically downscaled data for hydrologic simulations’, Journal of Hydrology, vol. 353, pp. 33– 48.
Fuhrer, J, Beniston, M, Fischlin, A, Frei, CH, Goyette, S, Jasper, K & Pfister, CH 2006,’ Climate risks and their impact on agriculture and forests in Switzerland’, Climatic Change, vol. 79, pp. 79–102. Garnaut, R 2007, Garnaut climate change review, Issues Paper 1 Climate Change: Land use—Agriculture and Forestry, Australian Greenhouse Office, Canberra.
Gautam, MR, Acharya, K &Tuladhar, MK 2010, ‘Upward trend of streamflow and precipitation in a small, non-snow-fed, mountainous watershed in Nepal’, Journal of Hydrology, vol.387, no. 3-4, pp. 304-311.
Gay, C, Estrada, F, Conde, C, Eakin, H & Villers, L 2006, ‘Potential Impacts of Climate Change on Agriculture: A Case of Study of Coffee Production in Veracruz, Mexico’, Climatic Change, vol. 79, pp. 259–288.
Gbetibouo, GA & Hassan, RM 2005, ‘Measuring the economic impact of climate change on major South African field crops: a Ricardian approach’, Global and Planetary Change, vol. 47, pp. 143–152.
Georgakakos, KP, Seo, D-J, Gupta, H, Schaake, J & Butts, MB 2004, ‘Towards the characterizations of streamflow simulations uncertainty through multimodel ensembles’, Journal of Hydrology, vol. 298, no. 1–4, pp. 222–241.
Giorgi, F & Bates, G 1989, ‘The climatological skill of a regional model over complex terrain’, Mon. Wea. Rev., vol. 117, pp. 2325-2347.
319 Giorgi, F 1990, ‘Simulation of regional climate using a limited area model nested in a general circulation model’, J Clim, vol. 3, pp. 941 963.
Gleick, PH 1987, ‘Regional hydrologic consequences of increases in atmospheric CO2 and other trace gases’, Climatic Change, vol. 10, pp. 137–160.
Gosain, AK & Rao S 2003, ‘Climate Change and India: Vulnerability Assessment and Adaptation’ in PR Shukla, et al (eds), Universities Press (India) Pvt Ltd, Hyderabad, p. 462.
Government of Western Australia 2003, Securing our Water Future: A State Water Strategy for Western Australia, http://dows.lincdigital.com.au/files/State_Water_Strategy_ complete_001.pdf.
Graham , LP, Andréasson, J & Carlsson, B 2007, ‘Assessing climate change impacts on hydrology from an ensemble of regional climate models, model scales and linking methods – a case study on the Lule River basin’, Climatic Change, vol. 81, pp. 293–307.
Grell, GA 1993, ‘Prognostic evaluation of assumptions used by cumulus parameterizations’, Mon. Wea. Rev., vol. 121, pp. 764-787.
Griffiths, GM, Salinger, MJ & Leleu, I 2003, ‘Trends in extreme daily rainfall in the South Pacific and relations to the South Pacific Convergence Zone’, Int. J. Climatol., vol. 23, pp. 847–869.
Groisman, PY, Knight, RW, Karl, TR, Easterling, DR, Sun, B & Lawrimore, JH 2004, ‘Contemporary changes of the hydrological cycle over the contiguous United States: Trends derived from in situ observations’, J. Hydrometeorol., vol. 5, pp. 64–85.
320 Gunasekera D, Tulloh C, Ford M & Edwina H 2008, ‘Climate change: Opportunities and challenges in Australian agriculture’, Proceedings of Faculty of Agriculture, Food & Natural Resources Annual Symposium 2008 (FAFNR ’08), University of Sydney.
Hagg, W, Braun, LN, Kuhn, M & Nesgaard, TI 2007, ‘Modelling of hydrological response to climate change in glacierized Central Asian catchments’, Journal of Hydrology, vol. 332, pp. 40– 53.
Haim D, Shechter, M & Berliner, P 2008, “Assessing the impact of climate change on representative field crops in Israeli agriculture: a case study of wheat and cotton, Climatic Change, vol. 86, pp. 425–440.
Hall, JW, Sayers, PB & Dawson, RJ 2005, ‘National-scale assessment of current and future flood risk in England and Wales’, Nat. Hazards, vol. 36, pp. 147–164.
Hamlet AF, Mote PW, Clark MP & Lettenmaier DP 2005, ‘Effects of temperature and precipitation variability on snowpack trends in the western United States’, Journal of Climate, vol. 18, pp. 4545–4561.
Hamlet, AF, Mote, PW, Clark, MP & Lettenmaier, DP 2007, ‘Twentiethcentury trends in runoff, evapotranspiration, and soil moisture in the western United States’, Journal of Climate, vol. 20, pp. 1468–1486.
Harrison, GP & Whittington, HW 2002, ‘Vulnerability of hydropower projects to climate change, IEE Proceedings Generation, Transmission and Distribution, vol. 149, pp. 249– 255.
Hay, LE, Wilby, RL & Leavesly, HH 2000, ‘Comparison of delta change and downscaled GCM scenarios for three mountainous basins in the United States’, Journal of the American Water Resources Association, vol. 36, no. 2, pp. 387–397.
321 Hayhoe, K, Cayan, D, Field, CB, Frumhoff, PC, Maurer, EP, Miller, NL, Moser, SC, Schenider, SC, Cahill, KN, Cleland, EE, Dale, L, Drapek, R, Hanemann, RM, Kalkstein, LS, Lenihan, J, Lunch, CK, Neilson, RP, Sheridan, SC & Verville, JH 2004, ‘Emissions pathways, climate change, and impacts on California, Proc. Natl. Acad. Sci. USA, vol. 101, no. 34, pp. 12422–12427.
Haylock, MR & Goodess, CM 2004, ‘Interannual variability of extreme European winter rainfall and links with mean large-scale circulation’, Int. J. Climatol., vol. 24, pp. 759– 776.
He, C & Croley, TE 2007, ‘Application of a distributed large basin runoff model in the Great Lakes basin’, Control Engineering Practice, vol.15, pp. 1001–1011. Hennessy, KJ, Whetton, PH, Walsh, K, Smith, IN, Bathols, JM, Hutchinson M & Sharples J 2003, The Impact of Climate Change on Snow Conditions in Mainland Australia, CSIRO Atmospheric Research, Aspendale, Australia.
Herath, S & Ratnayake, U 2004, ‘Monitoring rainfall trends to predict adverse impacts – a case study from Sri Lanka (1964–1993)’, Global Environ. Change, vol. 14, pp. 71–79.
Hlavčova, K, Szolgay, J, Kohnová, S & Hlásny, T 2008, ‘Simulation of Hydrological Response to the Future Climate in the Hron River Basin’, J. Hydrol. Hydromech., vol. 56, no. 3, pp. 163–175.
Hock, R 2003, ‘Temperature index melt modelling in mountain areas’, Journal of Hydrology, vol. 282, pp. 104–115.
Hodgkins GA & Dudley RW 2006, ‘Changes in the timing of winter–spring streamflows in eastern North America, 1913–2002’, Geophysical Research Letters, vol. 33, DOI:10.1029/2005GL025593.
322 Hodgkins GA, Dudley RW & Huntington, TG 2003, ‘Changes in the timing of high flows in New England over the 20th century’, Journal of Hydrology, vol. 278, pp. 244–252.
Hopfield, JJ 1982, ‘Neural networks and physical systems with emergent collective computational abilities’, Proc., Nat. Academy of Scientists, vol. 79, pp. 2554–2558.
Holtslag, A, de Bruijn, E & Pan, H-L 1990, ‘A high resolution air mass transformation model for short-range weather forecasting’, Mon. Wea. Rev., vol. 118, pp. 1561-1575.
Horton, P, Schaefli, B, Mezghani, A, Hingray, B & Musy, A 2006, ‘Assessment of climate- change impacts on alpine discharge regimes with climate model uncertainty’, Hydrol. Process., vol. 20, pp. 2091–2109.
Howden, M & Jones, R 2004, ‘Risk assessment of climate change impacts on Australia’s wheat industry’, In Proceedings for the 4th International Crop Science Congress, Brisbane, Australia, 26 September – 1 October 2004.
Howden, SM, Soussana, JF, Tubiello, FN, Chhetri, N, Dunlop, M & Meinke, H 2007, ’Adapting agriculture to climate change’, PNAS, vol. 104, no. 50, pp. 19691–19696.
Hsie, E, Anthes, R & Keyser, D 1984, ‘Numerical simulation of frontogenisis in a moist Atmosphere’, J. of Atmos. Sci., vol. 41, pp. 2581-2594.
Hsu, K, Gupta, HV & Sorooshian, S 1995, ‘Artificial neural network modelling of the rainfall–runoff process’, Water Resour. Res., vol. 31, pp. 2517–2530.
Huntington, TG, Sheffield, J & Hayhoe, K 2007, ‘Impacts of Climate Change on Wintertime Precipitation, Snowmelt Regime, Surface Runoff and Infiltration in the Northeastern USA during the 21st Century’, 64th Eastern Snow Conference, St. John’s, Newfoundland, Canada.
323 Hyvärinen, V 2003, ‘Trends and characteristics of hydrological time series in Finland’, Nord. Hydrol., vol. 34, no. 1–2, pp. 71–90.
Iglesias, A, Garrote, L, Quiroga, S & Moneo, M 2009, ‘Impacts of climate change in agriculture in Europe, PESETA-Agriculture study ’JRC Scientific and Technical Reports, Spain.
Imrie, CE, Durucan, S & Korre, A 2000, ‘River flow prediction using artificial neural networks: generalisation beyond calibration range’, Journal of Hydrology, vol. 233, pp. 138-153.
Intergovernmental Panel on Climate Change (IPCC) 2000, IPCC Special Report Emission Scenarios (SRES).
Intergovernmental Panel on Climate Change (IPCC) 2007: Climate Change 2007, ‘The Physical Science Basis’, in S Solomon, D Qin, M Manning, Z Chen, M Marquis, KB Averyt, M Tignor & HL Miller (eds) 2007, Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 996 pp.
Isik, Murat & Devadoss, Stephen 2006 'An analysis of the impact of climate change on crop yields and yield variability', Applied Economics, vol. 38, no. 7, pp. 835-844.
Jasper, K, Calanca, P, Gyalistras, D & Fuhrer, J 2004, ‘Differential impacts of climate change on the hydrology of two alpine river basins’, Climate Research, vol. 26, pp. 113–129.
Jayawardena, AW & Fernando, DAK 1998, ‘Use of Radial Basis Function Type Artificial Neural Networks for Runoff Simulation’, Computer-Aided Civil and Infrastructure Engineering, vol. 13, pp. 91–99.
324 Jiang, T, Chen, YD, Xu, C, Chen, X, Chen, X & Singh, VP 2007, ‘Comparison of hydrological impacts of climate change simulated by six hydrological models in the Dongjiang Basin, South China’, Journal of Hydrology , vol. 336, pp. 316– 333.
Jiang, T, Su, B & Hartmann, H 2007, ‘Temporal and spatial trends of precipitation and river flow in the Yangtze River Basin, 1961–2000’, Geomorphology, vol. 85, pp. 143- 154.
Johnson, F & Sharma, A 2009, ‘Measurement of GCM Skill in Predicting Variables Relevant for Hydroclimatological Assesments’, American Meteorological Society, vol. 22, no. 16, pp. 4373-4382.
Jones, RN, Francis, HS, Chiew, WC, Boughton & Zhang, L 2006, ‘Estimating the sensitivity of mean annual runoff to climate change using selected hydrological models’, Advances in Water Resources , vol. 29, no. 10, pp. 1419–1429.
Kadioglu, M 2008, ‘Gunumuzden 2100 Yilina Kuresel Iklim Degisimi (Global Climate Change from current date to 2100)’, TMMOB Kuresel Iklim Degisikligi Sempozyumu (UCTEA Global Climate Change Symposium), Ankara, 13-14 March 2008.
Kahya, E & Kalayci, S 2004, ‘Trend analysis of streamflow in Turkey’, Journal of Hydrology, vol. 289, pp. 128–144.
Kanber, R, Kapur, B & Tekin, S 2007, ‘Kurak Alanlarda Iklim Degisiminin Tarimsal Uretim Yonunden Degerlendirilmesi: ICCAP Projesi (A New Approach to Assess Climate Change impacts on Drought Regions: ICCAP Project)’, Uluslararası “Küresel İklim Değişikliği ve Çevresel Etkileri” Konferansı (International Global Climate Change and Environmental Effects Conference), Turkiye.
Kay, A, Bell, V & Davies, H 2006, Model Quality and Uncertainty for Climate Change Impact, Centre for Ecology and Hydrology, Wallingford.
325 Kharkina, MA 2004, ‘Natural resources in towns’, Energia, vol. 2, pp. 44–50.
Kiehl J, Hack, J, Bonan, G, Boville, B, Breigleb, B, Williamson, D & Rasch, P 1996, Description of the NCAR Community Climate Model (CCM3), Technical report, National Center for Atmospheric Research.
Kim, BS, Kim, HS, Seoh, BH & Kim, NW 2007, ‘Impact of climate change on water resources in Yongdam Dam Basin, Korea’, Stoch Environ Res Ris Assess, vol. 21, pp. 355–373.
Kingwell, R 2006, ‘Climate change in Australia: agricultural impacts and adaptation’, Australasian Agribusiness Review, 14, ISSN 1442-6951.
Kirshen, P, McCluskey, M, Vogel, R & Strzepek, K 2005, ‘Global analysis of changes in water supply yields and costs under climate change: a case study in China’, Climatic Change, vol. 68, no.3, pp. 303–330.
Kisi O 2005, ‘Suspended sediment estimation using neuro-fuzzy and neural network approaches’, Journal of Hydrological Science, vol. 50, pp. 683-696.
Kistemann, T, Classen, T, Koch, C, Dagendorf, F, Fischeder, R, Gebel, J, Vacata, V & Exner, M 2002, ‘Microbial load of drinking water reservoir tributaries during extreme rainfall and runoff’ Appl.Environ. Microbiol., vol. 68, no.5, pp. 2188–2197.
Klein Tank, AMG & Können, GP 2003, ‘Trends in indices of daily temperature and precipitation extremes in Europe, 1946–1999’, J. Clim., vol. 16, pp. 3665–3680.
Kleinn, J, Frei, C, Gurtz, J, Luthi, D, Vidale, PL & Schar, C 2005, ‘Hydrologic simulations in the Rhine basin driven by a regional climate model’, Journal of Geophysical Research, vol. 110, D04102, doi:10.1029/2004JD005143.
326 Kocman, A 1993, Turkiye Iklimi (Climate of Turkey), Ege Universitesi, Izmir.
Kron, W & Berz, G 2007, ‘Flood disasters and climate change: trends and options – a (re-)insurer’s view. Global Change: Enough Water for All? J.L. Lozán, H. Graßl, P. Hupfer, L. Menzel and C.- D. Schönwiese, Eds., University of Hamburg, Hamburg, 268- 273.
Krysanova, V & Wechsung, F 2002, ‘Impact of climate change and higher CO2 on hydrological processes and crop productivity in the state of Brandenburg, Germany’, Climatic Change: Implications for the Hydrological Cycle and for Water Management, M. Beniston, Ed., Kluwer, Dordrecht, pp. 271–300.
Krysanova, V, Hattermann, F & Habeck, A 2005, ‘Expected changes in water resources availability and water quality with respect to climate change in the Elbe River basin (Germany)’, Nordic Hydrol., vol. 36, no. 4–5, pp. 321–333.
Kucharik, CJ & Serbin, SP 2008, ‘Impacts of recent climate change on Wisconsin corn and soybean yield trends’, Environ. Res. Lett, vol. 3, pp. 1-10.
Kundzewicz, ZW, Mata, LJ, Arnell, NW, Döll, P, Kabat, P, Jiménez, B, Miller, KA, Oki, T, Sen, Z & Shiklomanov, IA 2007, ‘Freshwater resources and their management’, in ML Parry, OF Canziani, JP Palutikof, PJ van der Linden & CE Hanson (eds) 2007, Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, UK, pp. 173-210.
Kunkel, KE, Easterling, DR, Redmond, K & Hubbard, K 2003, ‘Temporal variations of extreme precipitation events in the United States: 1895–2000’, Geophys. Res. Lett., vol. 30, 1900, doi:10.1029/2003GL018052.
327 Labat, D, Godderis, Y, Probst, JL &Guyot, JL 2004, ‘Evidence for global runoff increases related to climate warming’, Adv. Water Resour., vol. 27, pp. 631-642.
Lecture on Fundamentals of Climate Change by B McNeil, The University of New South Wales, 2009.
Lecture on Climate Change and Water Resources by F Johnson, The University of New South Wales, 2009.
Lehner, B, Czisch, G & Vassolo, S 2005, ‘The impact of global change on the hydropower potential of Europe: a model-based analysis’, Energ. Policy, vol. 33, pp. 839–855.
Lemke, P, Ren, J, Alley, RB, Allison, I, Carrasco, J, Flato, G, Fujii, Y, Kaser, G, Mote, P, Thomas, RH & Zhang, T 2007, ‘Observations: Changes in Snow, Ice and Frozen Ground’, in S Solomon, D Qin, M Manning, Z Chen, M Marquis, KB Averyt, M Tignor & HL Miller (eds) 2007, Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 337-383.
Lettenmaier, DP & Gan, TY 1990, ‘Hydrologic sensitivities of the Sacramento-San Joaquin River Basin, California, to global warming’, Water Resources Research, vol. 26, pp. 69–86.
Lettenmaier, DP, Wood, AW, Palmer, RN, Wood, EF & Stakhiv, EZ 1999, ‘Water resources implications of global warming: A US regional perspective’, Climatic Change, vol. 43, pp. 537–579.
328 Le Treut, H, Somerville, R, Cubasch, U, Ding, Y, Mauritzen, C, Mokssit, A, Peterson, T & Prather, M 2007, ‘Historical Overview of Climate Change’, in S Solomon, D Qin, M Manning, Z Chen, M Marquis, KB Averyt, M Tignor & HL Miller (eds) 2007, Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 93-127.
Li, L, Hao, Z, Wang, J, Wang, Z & Yu, Z 2008, ‘Impact of Future Climate Change on Runoff in the Head Region of the Yellow River’, Journal of Hydrologic Engineering, vol. 13, no. 5, pp. 347-354.
Liston, GE & Elder, K 2006, ‘A Distributed Snow-Evolution Modeling System (SnowModel)’, Journal of Hydrometeorology, vol. 7, no. 6, pp. 1259-1276.
Liu, S, Mo, X, Lin, Z, Xu, Y, Ji, J, Wen, G & Richey, J 2010, ‘Crop yield responses to climate change in the Huang-Huai-Hai Plain of China’, Agricultural Water Management, vol. 97, pp. 1195–1209.
Lobell, DB, Field, CB, Cahill, KN & Bonfils, C 2006, ‘Impacts of future climate change on California perennial crop yields: model projections with climate and crop uncertainties’, Agric For.Meteorol.,vol. 141, no. 2–4, pp. 208–218.
Lobell, DB & Burke, MB 2008, ‘Why are agricultural impacts of climate change so uncertain? The importance of temperature relative to precipitation’, Environ. Res. Lett. ,vol. 3, 8 pp.
Lofgren, B, Clites, A, Assel, R, Eberhardt, A & Luukkonen, C 2002, ‘Evaluation of potential impacts on Great Lakes water resources based on climate scenarios of two GCMs.’, J. Great Lakes Res., vol. 28, no. 4, pp. 537–554.
329 LOSLR (International Lake Ontario–St. Lawrence River Study Board) 2006, Options for Managing Lake Ontario and St. Lawrence River Water Levels and Flows, Final Report to the International Joint Commission, http://www.losl.org/reports/finalreport-e.html.
Ludwig, F & Asseng, S 2006, ‘Climate change impacts on wheat production in a Mediterranean environment in Western Australia’, Agricultural Systems, vol. 90, pp. 159-179.
Ludwig, F, Milroy, SP & Asseng, S 2009, ‘Impacts of recent climate change on wheat production systems in Western Australia’, Climatic Change, vol. 92, pp. 495–517.
Luk, KC, Ball, JE & Sharma, A 2001, ‘An application of artificial neural networks for rainfall forecasting’, Math. Computer Model, vol. 33, pp. 683–693.
Luo, Q, Bellotti, W, Williams, M, Cooper, I & Bryan, B 2007, ‘Risk analysis of possible impacts of climate change on South Australian wheat production’, Climatic Change, vol. 85, pp. 89–101.
Ma, X, Yoshikane, T, Hara, M, Wakazuki, Y, Takahashi, HG & Kimura, F 2010, ‘Hydrological response to future climate change in the Agano River basin, Japan’, Hydrological Research Letters, vol. 4, pp. 25–29.
Magrin, GO, Travasso, MI & Rodríguez, GR 2005, ‘Changes in climate and crops production during the 20th century in Argentina’, Climatic Change, vol. 72, pp. 229– 249.
Maier, H & Dandy, G, 2000, ‘Neural networks for the predictions and forecasting of water resources variables: review of modelling issues and applications’, Environ. Modell. Software, vol. 15, pp. 101–124.
330 Malikov, M 2004, The Importance of Snowmelt Runoff Modelling for Sustainable Development and Disaster Prevention, Regional Workshop on the Use of Space Technology for Environmental Security, Disaster Rehabilitation and Sustainable Development, Islamic Republic of Iran.
Mall, RK, Gupta, A, Singh, R, Singh, RS & Rathore, LS 2006, ‘Water resources and climate change: An Indian perspective’, Current Science, vol. 90, no. 12, pp. 1610-1626.
Malla, G 2008, ‘Climate Change and its Impact on Nepalese Agriculture’, The Journal of Agriculture and Environment, Vol.9, pp. 62-71.
Markoff, MS & Cullen, AC 2008, ‘Impact of climate change on Pacific Northwest hydropower’, Climatic Change, vol. 87, pp. 451-469.
Marshall, E & Randhir, T 2008, ‘Effect of climate change on watershed system: a regional analysis’, Climatic Change, vol. 89, pp. 263–280.
Maurer, EP 2007, ‘Uncertainty in hydrologic impacts of climate change in the Sierra Nevada, California, under two emissions scenarios’, Climatic Change, vol. 82, pp. 309– 325.
Meehl, GA, Stocker, TF, Collins, WD, Friedlingstein, P, Gaye, AT, Gregory, JM, Kitoh, A, Knutti, R, Murphy, JM, Noda, A, Raper, SCB, Watterson, IG, Weaver, AJ & Zhao, Z-C 2007, ‘Global Climate Projections’ in S Solomon, D Qin, M Manning, Z Chen, M Marquis, KB Averyt, M Tignor & HL Miller (eds) 2007, Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
Mendelsohn, R 2009, ‘The Impact of Climate Change on Agriculture in Developing Countries’, Journal of Natural Resources Policy Research, vol.1, no. 1, pp. 5–19.
331 Mengu, GP, Sensoy, S and Akkuzu, E 2008, ‘Effects of Global Climate Change on Agriculture and Water Resources’, The Third International Scientific Conference, Balwois, Ohrid, Republic of Macedonia.
Menzel, L, Thieken, AH, Schwandt, D & Burger, G 2006, ‘Impact of Climate Change on the Regional Hydrology – Scenario-Based Modelling Studies in the German Rhine Catchment’, Natural Hazards, vol. 38, pp. 45–61.
Mernild, SH, Liston, GE, Kane, DL, Knudsen, NT & Hasholt, B 2008, ‘Snow, runoff, and mass balance modeling for the entire Mittivakkat Glacier (1998–2006), Ammassalik Island, SE Greenland’, Danish Journal of Geography, vol. 108, pp. 121-136.
Merritt, WS, Alila, Y, Barton, M, Taylor, B, Cohen, S & Neilsen, D 2006, ‘Hydrologic response to scenarios of climate change in sub watersheds of the Okanagan basin, British Columbia’, Journal of Hydrology, vol. 326, pp. 79–108.
Middelkoop, H, Daamen, K, Gellens, D, Grabs, W, Kwadijk, JCJ, Lang, H, Parmet, BWAH, Schädler, B., Schulla, J & Wilke, K 2001, ‘Impact of climate change on hydrological regimes and water resources management in the Rhine basin’, Climatic Change, vol. 49, pp. 105–128.
Millennium Ecosystem Assessment 2005a, Ecosystems and Human Well-being: Volume 2 – Scenarios, Island Press, Washington, DC, 515 pp.
Millennium Ecosystem Assessment 2005b, Ecosystems and Human Well-being: Synthesis. Island Press, Washington, DC, 155 pp.
Milly, PCD, Wetherald, RT, Dunne, KA & Delworth, TL 2002, ‘Increasing risk of great floods in a changing climate’, Nature, vol. 415, pp. 514–517.
332 Milly, PCD, Dunne, KA & Vecchia, AV 2005,‘Global pattern of trends in streamflow and water availability in a changing climate’, Nature, vol. 438, no. 7066, pp. 347–350.
Ministry of Energy and Natural Resources (MENR) 2006, Energy Scenario Report under the UNDP GEF FNC Project, Turkish Republic, Ankara.
Ministry of Environment and Forestry 2007, First National Communication of Turkey on Climate Change, Turkish Republic, Ankara.
Ministry of Environment and Forestry 2008, Iklim Degisikligi ve Yapilan Calismalar (Climate Change and Performed Actions), Turkish Republic Ankara, Turkey.
Minville, M, Brissette, F & Leconte, R 2008, ‘Uncertainty of the impact of climate change on the hydrology of a nordic watershed’, Journal of Hydrology, vol. 358, pp. 70– 83.
Mirza, MMQ 2003, ‘Three recent extreme floods in Bangladesh: a hydro- meteorological analysis’, Nat. Hazards, vol. 28, pp. 35–64.
Mitchell, TD & Jones, PD 2005, ‘An improved method of constructing a database of monthly climate observations and associated highresolution grids’, Int. J. Climatol., vol. 25, pp. 693–712.
Mizyed, N 2009, ‘Impacts of Climate Change on Water Resources Availability and Agricultural Water Demand in theWest Bank’, Water Resour. Management, vol. 23, pp. 2015-2029.
Mo, X, Liu, S, Lin, Z & Ruiping Guo, 2009, ‘Regional crop yield, water consumption and water use efficiency and their responses to climate change in the North China Plain’, Agriculture, Ecosystems and Environment, vol. 134, pp. 67–78.
333 Mohamed, AB, Van Duivenbooden, N & Abdoussallam, S 2002, ‘Impact of Climate Change on Agricultural Production in the Sahel – Part 1. Methodological Approach and Case Study for Millet in Niger’, Climatic Change, vol. 54, pp. 327–348.
Mohsin, T & Gough, WA 2009, ‘Trend analysis of long-term temperature time series in the Greater Toronto Area (GTA)’, Theoretical Applied Climatology, DOI: 10.1007/s00704-009-0214-x.
Mote, PW, Canning, DJ, Fluharty, DL, Francis, RC, Franklin, JF, Hamlet, AF, Hershman, M, Holmberg, M, Gray-Ideker, KN, Keeton, WS, Lettenmaier, DP, Leung, LR, Mantua, NJ, Miles, EL, Noble, B, Parandvash, H, Peterson, DW, Snover AK & Willard, SR 1999, Impacts of Climate Variability and Change, Pacific Northwest, National Atmospheric and Oceanic Administration, Office of Global Programs, and JISAO/SMA Climate Impacts Group.
Mote PW 2003, ‘Trends in snow water equivalent in the Pacific Northwest and their climatic causes’, Geophysical Research Letters 30, DOI: 10.1029/2003GL017258.
Mote, PW, Hamlet, AF, Clark, MP & Lettenmaier, DP 2005, ‘Declining mountain snowpack in western North America’, Bulletin of the American Meteorological Society, vol. 86, pp. 39–49.
Mozny, M, Tolasz, R, Nekovar, J, Sparks, T, Trnka, M & Zalud, Z 2009, ‘The impact of climate change on the yield and quality of Saaz hops in the Czech Republic’, Agricultural and Forest Meteorology, vol. 149, pp. 913–919.
Mutke, S, Gordo, J & Gil, L 2005, ‘Variability of Mediterranean Stone pine cone production: Yield loss as response to climate change’, Agricultural and Forest Meteorology, vol. 132, pp. 263–272.
334 Nagano, T, Hoshikawa, K, Donma, S, Kume, T, Onder, S, Ozekici, B, Kanber, R & Watanabe, T 2008, ‘Assessing Impact of Climate Change on the Large Irrigation District in Turkey with Irrigation Management Performance Assessment Model’, International Congress on River Basin Management, Turkey.
Nagler, T, Rott, H, Malcher, P & Muller, F 2008, ‘Assimilation of meteorological and remote sensing data for snowmelt runoff forecasting’, Remote Sensing of Environment, vol. 112, pp. 1408–1420.
National Oceanic and Atmospheric Administration (NOAA), ‘Climate Model’, viewed 2010,
Naylor, RL, Battisti, DS, Vimont, DJ, Falcon, WP & Burke, MB 2007, ‘Assessing risks of climate variability and climate change for Indonesian rice agriculture’, PNAS, vol. 104, no. 19, pp. 7752-7757.
Nicholls, N 2005, ‘Climate variability, climate change and the Australian snow season’, Aust. Meteorol. Mag., vol. 54, pp. 177–185.
Nohara, D, Kitoh, A, Hosaka, M & Oki, T 2006, ‘Impact of Climate Change on River Discharge Projected by Multimodel Ensemble’, Journal of Hydrometeorology, vol. 7, pp. 1076-1089.
Novotny, VE & Stefan, HG 2007, ‘Stream flow in Minnesota: Indicator of climate change’, Journal of Hydrology, vol. 334, no. 3-4, pp. 319-333.
Olesen, JE 2005, 'Climate Change and CO2 Effects on Productivity of Danish Agricultural Systems', Journal of Crop Improvement, vol. 13, no. 1, pp. 257 - 274.
335 Onol, B 2007, ‘Downscaling Climate Change Scenario Using Regional Climate Model Over Eastern Mediterranean’, PhD Thesis, Istanbul Technical University, Istanbul.
Onoz, B & Bayazit, M 2003, ‘The power of statistical tests for trend detection’, Turkish Journal of Engineering & Environmental Sciences, vol. 27, pp. 247–251.
Ortiz, R, Sayre, KD, Govaerts, B, Gupta, R, Subbarao, GV, Ban, T, Hodson, D, Dixon, JM, Ortiz-Monasterio, JI & Reynolds, M 2008, ‘Climate change: Can wheat beat the heat?’, Agriculture, Ecosystems and Environment, vol. 126, pp. 46–58.
Ozdemir, Y, Ozis, U, Baran, T, Fistikoglu, O & Demirci, N 2008, ‘Sinir-Aşan Firat-Dicle Havzasinin Su Potansiyeli ve Yararlanilmasi (Water Potential of Euphrates-Tigris Basin and Benefiting from it)’, TMMOB Su Politikalari Kongresi (UCTEA Congress on Water Policies), Ankara.
Ozkan, B & Akcaoz, H 2002, ‘Impacts of Climate Factors on Yields for Selected Crops in Southern Turkey’, Mitigation and Adaptation Strategies for Global Change, vol. 7, pp. 367–380.
Pal, JS, Small, EE & Eltahir, EAB 2000, ‘Simulation of regional scale water and energy budgets: Influence of a new moist physics scheme within RegCM’, J. Geophys.Res., vol. 105, pp. 579-594.
Pal, JS, Giorgi, F, Bi, X, Elguindi, N, Solmon, F, Gao, Francisco, R, Zakey, A, Winter, J, Ashfaq, M, Syed, F, Bell, JL, Diffenbaugh, NS, Karmacharya, J, Konar, A, Martinez, D, da Rocha1, RP, Sloan, LC & Steiner, A 2005, The ICTP RegCM3 and RegCNET: Regional Climate Modeling for the DevelopingWorld, BAMS.
Palmer, TN & Räisänen, J 2002, ‘Quantifying the risk of extreme seasonal precipitation events in a changing climate’, Nature, vol. 415, pp. 512–514.
336 Parajka, J & Bloschl, G 2008, ‘The value of MODIS snow cover data in validating and calibrating conceptual hydrologic models’, Journal of Hydrology, vol. 358, pp. 240-258.
Park, G-A, Shin, H-J, Lee, M-S, Hong, W-Y & Kim, S-J 2009, ‘Future potential impacts of climate change on agricultural watershed hydrology and the adaptation strategy of paddy rice irrigation reservoir by release control, Paddy Water Environ, vol. 7, pp. 271– 282.
Peterson, TC, Taylor, MA, Demeritte, R, Duncombe, DL, Burton, S, Thompson, F, Avalon, P, Mercedes, M, Villegas, E, Fils, RS, Tank, AK, Martis, A, Warner, R, Joyette, A, Mills, W, Alexander, L & Gleason, B 2002, ‘Recent changes in climate extremes in the Caribbean region’ J. Geophys. Res., vol. 107, no. D21, 4601, doi:10.1029/2002JD002251.
Petkova, N, Koleva, E & Alexandrov, V 2004, ‘Snow cover variability and change in mountainous regions of Bulgaria, 1931-2000’, Meteorol. Z., vol. 13, no. 1, pp. 19–23.
Phillips, N A 1956, ‘The general circulation of atmosphere: a numerical experiment’, Q. J. R.Meteorol. Soc., vol.82, pp. 123–164.
Pittock, B 2003, Climate Change: An Australian Guide to the Science and Potential Impacts, Australian Greenhouse Office, Canberra.
Porter, JR & Semenov, MA 2005, ‘Crop responses to climatic Variation’, Philos. Trans. R. Soc. B: Biological Sciences, vol. 360, pp. 2021–2035.
Pounds, JA, Bustamante, MR, Coloma, LA, Consuegra, JA, Fogden, MPL, Foster, PN, La Marca, E, Masters, KL, Merino-Viteri, A, Puschendorf, R, Ron, SR, Sanchez-Azofeifa, GA, Still, CJ &Young, BE 2006, ‘Widespread amphibian extinctions from epidemic disease driven by global warming’, Nature, vol. 439, no. 7073, pp. 161–167.
337 Prasad, VH & Roy, PS 2005, ‘Estimation of Snowmelt Runoff in Beas Basin, India’, Geocarto International, vol. 20, pp. 41 - 47.
Prato, T, Zeyuan, Q, Pederson, G, Fagre,D, Bengtson, LE & Williams, JR 2010, ‘Potential Economic Benefits of Adapting Agricultural Production Systems to Future Climate Change’, Environmental Management, vol. 45, pp. 577–589.
Prieto, R, Herrera, R, Doussel, P, Gimeno, L, Ribera, P, Garcia, R & Hernandez, E 2001, ‘Interannual oscillations and trend of snow occurrence in the Andes region since 1885’, Aust. Meteorol. Mag., vol. 50, no. 2, pp. 164-168.
Protopapas, L, Katchamart, S & Platonova, A 2000, ‘Weather effects on daily water use in New York City’, J. Hydrol. Eng., vol. 5, pp. 332–338.
Purkey, DR, Joyce, B, Vicuna, S, Hanemann, MW, Dale, LL, Yates, D & Dracup, JA, 2008, ‘Robust analysis of future climate change impacts on water for agriculture and other sectors: a case study in the Sacramento Valley’, Climatic Change, vol. 87, pp. 109-122.
Qian, J-H, Giorgi, F & Fox-Rabinovitz, MS 1999, ‘Regional stretched grid generation and its application to the ncar regcm’, J. of Geophys. Res., vol. 104, pp. 6501-6513.
Ramírez, E, Francou, B, Ribstein, P, Descloitres, M, Guérin, R, Mendoza, J, Gallaire, R, Pouyaud, B & Jordan, E 2001, ‘Small glaciers disappearing in the tropicalAndes: a case study in Bolivia: the Chacaltaya glacier, 16°S’, J. Glaciol., vol. 47, pp. 187–194.
Rami’rez, MCP, Velho, HFC & Ferreira, NJ 2005, ‘Artificial neural network technique for rainfall forecasting applied to the Sao Paulo region’, J. Hydrol., vol. 301, pp. 146–162.
Ranjithan, S, Eheart, JW & Rarret Jr, JH 1993, ‘Neural networkscreening for groundwater reclamation under uncertainty’, Water Resour. Res., vol. 29, no. 3, pp. 563–574.
338 Rauscher, SA, Jeremy, SP, Diffenbaugh, NS & Benedetti, MM 2008, ‘Future changes in snowmelt-driven runoff timing over the western US’, Geophysical Research Letters, vol. 35, L16703, DOI:10.1029/2008GL034424.
Reilly, J, Tubiello, F, McCarl, B, Abler, D, Darwin, R, Fuglie, K, Hollinger, S, Izzaurralde, C, Jagtap, S, Jones, J, Meams, L, Ojima, D, Paul, E, Paustian, K, Riha, S, Rosenberg, N & Rosenzweig, C 2003, ‘U.S. agriculture and climate change: new results’,Climatic Change, vol. 57, pp. 43–69.
Reynard, NS, Prudhomme, C & Crooks, SM 2001, ‘The flood characteristics of large UK rivers: potential effects of changing climate and land use’, Climatic Change , vol. 48, pp. 343–359.
Riad, S, Mania, J, Bouchaou, L & Najjar, Y 2004, ‘Predicting catchment flow in a semi- arid region via an artificial neural network technique’, Hydrological Processes, vol. 18, pp. 2387-2393.
Roeckner, E, Bäuml, G, Bonaventura, L, Brokopf, R, Esch, M, Giorgetta, M, Hagemann, S, Kirchner, I, Kornblueh, L, Manzini, E, Rhodin, A, Schlese, U, Schulzweida, U & Tompkins, A 2003, The atmospheric general circulation model echam5, part I: Model description, Technical Report 349, Max-Planck-Institute for Meteorology, Hamburg,Germany.
Rogers, LL & Dowla, FU 1994, ‘Optimization of groundwater remediation using artificial neural networks with parallel solute transport modelling’, Water Resour. Res., vol. 30, no. 2, pp. 457–481.
Rosenzweig, C, Strzepek, KM, Major, DC, Iglesias, A, Yates, DN, McCluskey, A & Hillel, D 2004, ‘Water resources for agriculture in a changing climate: international case studies’, Global Environmental Change, vol. 14, pp. 345–360.
339 Rosenzweig, C, Casassa, G, Karoly, DJ, Imeson, A, Liu, C, Menzel, A, Rawlins, S, Root, TL, Seguin, B & Tryjanowski, P 2007, ’Assessment of observed changes and responses in natural and managed systems’, in ML Parry, OF Canziani, JP Palutikof, PJ van der Linden & CE Hanson (eds) 2007, Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, UK, pp. 79-131.
Root, TL, Price, JT, Hall, KR, Schneider, SH, Rosenzweig, C & Pounds, JA 2003, ‘Fingerprints of global warming on wild animals and plants’, Nature, vol. 421, no. 6918, pp. 57–60.
Rosenzweig, C, Tubiello, FN, Goldberg, R, Mills, E & Bloomfield, J 2002, ‘Increased crop damage in the US from excess precipitation under climate change’, Global Environ. Chang., vol. 12, pp. 197–202.
Ruddiman, WF 2001, Earth's Climate: past and future, W.H. Freeman & Sons, New York.
Rumelhart, DE, McClelland, JL & the pdp research group 1986, Parallel recognition in modern computers, In Proceeding, Explorations in the microstructure of Cognition, vol. 1, Foundations, MIT Press/Bradford Books, Cambridge Mass.
Schaefli, B, Hingray, B & Musy, A 2007, ‘Climate change and hydropower production in the Swiss Alps: quantification of potential impacts and related modelling uncertainties’, Hydrology & Earth System Sciences, vol. 11, no. 3, pp. 1191-1205.
Schar, C, Vidale, PL, Luthi, D, Frei, C, Haberli, C, Liniger, MA & Appenzeller, C 2004, ‘The role of increasing temperature variability in European summer heatwaves’, Nature, vol. 427, no. 6972, pp. 332–336.
340 Scherrer, SC, Appenzeller, C & Laternser, M 2004, ‘Trends in Swiss alpine snow days – the role of local and large scale climate variability’, Geophys. Res. Lett., vol. 31, L13215, doi:10.1029/2004GL020255.
Schlenker, W, Hanemann, WM & Fisher, AC 2007, ‘Water Availability, Degree Days, and the Potential Impact of Climate Change on Irrigated Agriculture in California’, Climatic Change, vol. 81, pp. 19-38.
Schmidli, J, Goodess, CM, Frei, C, Haylock, MR, Hundacha, Y, Ribalaygua, J & Schmith, T 2007, ‘Statistical and dynamical downscaling of precipitation: an evaluation and comparison of scenarios for the European Alps’, J Geophys Res, vol. 112, D04105, DOI:10.1029/2005JD007026.
Schneeberger, C, Blatter, H, Abe-Ouchi, A & Wild, M 2003, ‘Modelling changes in the mass balance of glaciers of the northern hemisphere for a transient 2× CO2 scenario’, J. Hydrol., vol. 282, no. 1–4, pp. 145–163.
Schreider, SY, Smith, DI & Jakeman AJ 2000, ‘Climate change impacts on urban flooding’, Climatic Change, vol. 47, no. 1–2, pp. 91–115.
Scott, MJ, Vail, LW, Jaksch, JA, Stöckle CO & Kemanian AR 2004a, ‘Water exchanges: tools to beat el niño climate variability in irrigated agriculture’, J Water Resour Assoc, vol.40, no. 1, pp. 15–31.
Scott, MJ, Vail, LW, Jaksch, JA, Stöckle, CO & Kemanian ,AR 2004b, ‘Climate change and adaptation in irrigated agriculture—a case study of the Yakima River’, In: Proceedings of the UCOWR/NIWR annual conference, 20–22 July 2004, Portland, Oregon. PNWD- SA-6448. Pacific Northwest National Laboratory, Richland, WA.
341 Sensoy, S, Alan, I & Demircan, M 2007, ‘Trends in Turkey Climate Extreme Indices from 1971-2004’, IUGG Conference, 02-13 July, 2007 Perugia, Italy.
Sensoy, A, Sorman, AA, Pekkan, E, Sorman, AU, Sezen, N, Akgoz, A, Yazici, A, Keskin, R & Hasimoglu F, 2008, ‘Snow studies and evaluations in the Upper Euphrates Basin’, Snow Hydrology Conference, Erzurum, Turkey.
Seo, SN & Mendelsohn, R 2008, ‘An analysis of crop choice: Adapting to climate change in South American farms’, Ecological Economics, vol. 6 7, pp. 109 – 116.
Shamseldin, A Y 1997, ‘Application of a neural network technique to rainfall-runoff modelling’, J. Hydrol., Amsterdam, vol. 199, pp. 272–294.
Shi, Y, Shen, Y, Kang, E, Li, D, Ding, Y, Zhang, G & Hu, R 2007, ‘Recent and future climate change in northwest China’, Climatic Change, vol. 80, pp. 379–393.
Slingo, J 1989, ‘A GCM Parameterization for the Shortwave Radiative Properties of Water Clouds’, J. of Atmos. Sc., vol. 46, pp. 1419-1427.
Small, E, Giorgi, F & Sloan, L 1999, ‘Regional climate model simulation of precipitation in central asia: Mean and interannual variability’, J. of Geophys. Res., vol. 104, pp. 6563-6582.
Smiatek, G, Kunstmann, H, Knoche, R & Marx, A 2009, ‘Precipitation and temperature statistics in high-resolution regional climate models: Evaluation for the European Alps’, Journal of Geophysical Research , vol. 114, D19107,16.
Smith, J & Eli, RN 1995, ‘Neural-network models of rainfall runoff process’, J. Water Resour. Plng. and Mgmt., ASCE, vol. 121, no. 6, pp. 499–508.
342 Solomon, S, Qin, D, Manning, M, Chen, Z, Marquis, M, Averyt, KB, Tignor, M & Miller HL 2007, Climate Change 2007: The Physical Science Basis, Cambridge Univ Press, Cambridge, UK.
St. George, S 2007, ‘Streamflow in the Winnipeg River basin, Canada: Trends, extremes and climate linkages’, Journal of Hydrology, vol. 332, no. 3-4, pp. 396-411.
Stahl, K, Moore, RD, Shea, JM, Hutchinson, D & Cannon, AJ 2008, ‘Coupled modelling of glacier and streamflow response to future climate scenarios’, Water Resources Research, vol. 44, W02422, doi:10.1029/2007WR005956.
State Hydraulic Works (SHW) of Turkish Republic, viewed 2010,
Statistics Bureau of Heilongjiang Province (1992–2003) Heilongjiang statistics yearbook, China Statistics Press, Beijing.
Steele-Dunne, S, Lynch, P, McGrath , R, Semmler, T,Wang, S, Hanafin, J & Nolan, P 2008, ‘The impacts of climate change on hydrology in Ireland’, Journal of Hydrology, vol. 356, pp. 28– 45.
Stewart, IT, Cayan, DR & Dettinger, MD 2005, ‘Changes towards earlier streamflow timing across western North America’, J. Clim., vol. 18, pp. 1136–1155.
Stockle, CO, Nelson, RL, Higgins, S , Brunner, J, Grove, G , Boydston, R, Whiting, M & Kruger, C 2010, ‘Assessment of climate change impact on Eastern Washington agriculture’, Climatic Change, DOI 10.1007/s10584-010-9851-4.
Stone, RS, Dutton, EG, Harris, JM & Longnecker, D 2002, ‘Earlier spring snowmelt in northern Alaska as an indicator of climate change’, J. Geophys. Res., vol. 107, D10, DOI:10.1029/2000JD000286.
343 Sudheer, KP, Gosain, AK & Ramasastri, KS 2002, ‘A data-driven algorithm forconstructing artificial neural network rainfall–runoff models’, Hydrological Processes, vol. 16, no. 6, pp. 1325–1330.
Svendsen, M & Kunkel, N 2009, ‘Adapting to Hydrological Impacts of Climate Change: An international development perspective’, Irrigation and Drainage, vol. 58, pp. 121- 132.
Tao, F, Yokozawa, M, Hayashi, Y & Lin, E 2003, Changes in agricultural water demands and soil moisture in China over the last half-century and their effects on agricultural production, Agri.Forest Meteorol., vol. 118, pp. 251–261.
Tao, F,Yokozawa, M, Xu, Y, Hayashi, Y & Zhang, Z 2006, ‘Climate changes and trends in phenology and yields of field crops in China, 1981–2000’, Agricultural and Forest Meteorology, vol. 138, pp. 82–92.
Tarboton, DG & Luce, CH 1996, Utah Energy Balance Snow Accumulation and Melt Model (UEB), Computer model technical description and users guide, viewed 2008,
Tayanc, M, Im, U, Dogruel, M & Karaca, M 2009, ‘Climate change in Turkey for the last half century’, Climatic Change, vol. 94, pp. 483–502.
Tekeli, A E, Akyurek, Z, Sorman, AA, Sensoy, A & Sorman, AU 2005, ‘Using MODIS snow cover maps in modeling snowmelt runoff process in the eastern part of Turkey, Remote Sensing of Environment, vol. 97, pp. 216-230.
Thodsen, H 2007, ‘The influence of climate change on stream flow in Danish rivers’, Journal of Hydrology, vol. 333, pp. 226– 238.
344 Thomas, A 2008, ‘Agricultural irrigation demand under present and future climate scenarios in China’, Global and Planetary Change, vol. 60, pp. 306–326.
Thomson, AM, Rosenberg, NJ, Izaurralde, RC & Brown, RA 2005, ‘Climate change impacts for the conterminous USA: an integrated assessment Part 5. Irrigated agriculture and national grain crop production’, Climatic Change, vol. 69, pp. 89-105.
Tingem, M & Rivington, M 2009, ‘Adaptation for crop agriculture to climate change in Cameroon: Turning on the heat’, Mitig Adapt Strateg Glob Change, vol. 14, pp. 153– 168.
Tohma, S & Igata, S 1994, ‘Rainfall estimation from GMS imagery data using neural networks’, in WR Blain & L Katsifarakis (eds), Hydraulic engineering software V, Vol. 1, Computational Mechanics, Southampton, U.K., pp. 121–130.
Tokar, AS & Johnson, PA 1999, ‘Rainfall-runoff modeling using artificial neural networks’, J. Hydrologic Engrg., ASCE, vol. 4, no. 3, pp. 232– 239.
Topcu, S, Sen, B, Giorgi, F, Bi, X, Kanit, EG & Dalkilic, T 2008, ‘Impact of Climate Change on Agricultural Water Use in the Mediterranean Region’, World Water Congress, Montpellier, France.
Trenberth, KE, Jones, PD, Ambenje, P, Bojariu, R, Easterling, D, Klein Tank, A, Parker, D, Rahimzadeh, F, Renwick, JA, Rusticucci, M, Soden, B & Zhai, P 2007, ‘Observations: Surface and Atmospheric Climate Change’, in [S Solomon, D Qin, M Manning, Z Chen, M Marquis, KB Averyt, M Tignor & HL Miller (eds) 2007, Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
345 Tubiello, FN & Fischer, G 2007, ‘Reducing climate change impacts on agriculture: Global and regional effects of mitigation, 2000–2080’, Technological Forecasting & Social Change, vol. 74, pp. 1030–1056.
Turan, ME & Yurdusev, MA 2009, ‘River flow estimation from upstream flow records by artificial intelligence methods’, Journal of Hydrology, vol. 369, no. 1-2, pp. 71-77.
Turkish State Meteorology Service (TSMS) 2006, First National Communication Group Report on Climate Change Impacts, Vulnerability and Adaptation and Research and Systematic Observation, Ankara.
Turkish State Meteorology Service (TSMS) 2007, Turkiye Geneli Icin 2007 Kuraklik Degerlendirmesi (Drought Assessment of Turkey for the case in 2007), Ankara.
Turkish Statistical Institute (TurkStat), viewed 2010,
Union of Turkish Agricultural Chambers (TZOB) 2008 , 2007 Yılında Tarım (Agriculture in 2007), viewed 2009,
US Army Corps of Engineers Hydrologic Engineering Center (USACE) 2009, Hydrologic Modeling System HEC-HMS User’s Manual, viewed 2008,
Van Duivenbooden, Abdoussalam, S & Mohamed, AB 2002, ‘Impact of Climate Change on Agricultural Production in the Sahel – Part 2. Case Study for Groundnut and Cowpea in Niger’, Climatic Change, vol. 54, pp. 349–368.
346 Van Pelt, SC, Kabat, P, Ter Maat, HW, Van den Hurk, BJJM & Weerts, AH 2009, ‘Discharge simulations performed with a hydrological model using bias corrected regional climate model input’, Hydrol. Earth Syst. Sci., vol. 13, pp. 2387–2397.
Vano, JA, Scott, MJ, Voisin, N, Stöckle, CO, Hamlet, AF, Mickelson, KEB, Elsner, MM & Lettenmaier, DP 2010, ‘Climate change impacts on water management and irrigated agriculture in the Yakima River Basin, Washington, USA’, Climatic Change, DOI 10.1007/s10584-010-9856-z.
Vicuna, S & Dracup, JA 2007, ‘The evolution of climate change impact studies on hydrology and water resources in California’, Climatic Change, vol. 82, pp. 327-350.
Vicuna, S, Leonardson, R, Hanemann, MW, Dale, LL & Dracup, JA, 2008, ‘Climate change impacts on high elevation hydropower generation in California’s Sierra Nevada: a case study in the Upper American River’, Climatic Change, vol. 87, pp. 123–137.
Vojtek, M, Fasko, P & St’astny, P 2003, ‘Some selected snow climate trends in Slovakia with respect to altitude’, Acta Meteorologica Universitatis Comenianae, vol. 32, pp. 17–27.
Wang, W, Chau, K, Cheng, C & Qiu, L 2009, ‘A comparison of performance of several artificial intelligence methods for forecasting monthly discharge time series’, Journal of Hydrology, vol. 374, pp. 294-306.
Westerling, AL, Hidalgo, HG, Cayan, DR & Swetnam, TW 2006, ‘Warming and Earlier Spring Increase Western U.S. Forest Wildfire Activity’, Science Express , vol. 313, no. 5789, pp. 940 – 943.
Wikimedia, 2010, viewed 2010,
347 World Disaster Report (WDR) 2003, World Disaster Report: Focus on Ethics in Aid, International Federation of Red Cross and Red Crescent Societies, Geneva.
World Disaster Report (WDR) 2004, WorldDisaster Report: Focus on Community Resilience, International Federation of Red Cross and Red Crescent Societies, Geneva.
Wu, P, Jin, J & Zhao, X 2010, ‘Impact of climate change and irrigation technology advancement on agricultural water use in China’, Climatic Change, vol. 100, pp. 797– 805.
Xiong, W, Holman, I, Lin, E, Conway, D, Jiang, J, Xu, Y & Li, Y 2010, ‘Climate change, water availability and future cereal production in China’, Agriculture, Ecosystems & Environment, vol. 135, no. 1-2, pp. 58-69.
Yao, F, Xu, Y, Lin, E, Yokozawa, M & Zhang, J 2007, ‘Assessing the impacts of climate change on rice yields in the main rice areas of China’, Climatic Change, vol. 80, pp. 395–409.
Yang, X, Lin, E, Ma,S, Ju, H, Guo, L, Xiong, W, Li, Y & Xu, Y 2007, ‘Adaptation of agriculture to warming in Northeast China’, Climatic Change, vol. 84, pp. 45–58.
Yano, T, Aydin, M & Haraguchi, T 2007, ‘Impact of Climate Change on Irrigation Demand and Crop Growth in a Mediterranean Environment of Turkey’, Sensor , vol. 7, pp. 2297-2315.
Ye, HC & Ellison, M 2003, ‘Changes in transitional snowfall season length in northern Eurasia’, Geophys. Res. Lett., vol. 30, no. 5, 1252, DOI:10.1029/2003GL016873
Yenigun, K, Gumus, V & Bulut, H 2008, ‘Trends in streamflow of the Euphrates basin, Turkey’, Proceedings of the Institution of Civil Engineers, vol 161, pp. 189-198.
348 Yuanqing, H, Tao, P, Theakstone, WH & Jiankuo, D 2010, ‘Climate Change and Its Effect on Annual Runoff in Lijiang Basin-Mt. Yulong Region, China’, Journal of Earth Science, vol. 21, no. 2, pp. 137–147.
Yue, S, Pilon, P & Cavadias, G 2002, ‘Power of the Mann–Kendall and Spearman's rho tests for detecting monotonic trends in hydrological series’, Journal of Hydrology, vol. 259, no. 1-4, pp. 254-271.
Zeng, X, Zhao, M & Dickinson, RE 1998, ‘Intercomparison of Bulk Aerodynamic Algoriths for the Computation of Sea Surface Fluxes Using TOGA COARE and TAO data’, J. Climate, vol. 11, pp. 2628-2644.
Zhang, T, Zhu, J & Wassmann, R 2010, ‘Responses of rice yields to recent climate change in China: An empirical assessment based on long-term observations at different spatial scales (1981–2005)’, Agricultural and Forest Meteorology, vol. 150, pp. 1128– 1137.
Zhou, Y & Tol, RSJ 2005, ‘Evaluating the costs of desalination and water transport’, Water Resourc. Res., vol. 41, pp. 1–10.
349