Water and Adaptation Interventions in Central and West Asia

Technical Assistance Consultant’s Report

Project Number: 44066-012 (RETA 7532) June 2012

Regional Technical Assistance: Water and Adaptation Interventions in Central and West Asia

Prepared by FCG Finnish Consulting Group Ltd

In association with FutureWater and Finnish Meteorological Institute

For Asian Development Bank

This consultant’s report does not necessarily reflect the views of ADB or the Government concerned, and ADB and the Government cannot be held liable for its contents. (For project preparatory technical assistance: All the views expressed herein may not be incorporated into the proposed project’s design.

FCG Finnish Consulting Group Ltd. in association with FINAL REPORT • FutureWater • Finnish Meteorological Institute Executive Summary

Asian Development Bank

Water and Adaptation Interventions in Central and West Asia

TA 7532

June 2012

ADB / TA 7532

WATER AND ADAPTATION INTERVENTIONS IN CENTRAL AND WEST ASIA - SUMMARY OF RESULTS

ABSTRACT. Central Asia is one of the regions suffering most from the impacts of climate change on its nature and economy. The report describes results from a study which developed models on the impacts of climate change on water resources and natural hazards in the Aral Sea Basin. The past climate was reconstructed using data from global climate databases compiled by using field and satellite observations. Several international climate change scenarios were used to model the changes of temperature, precipitation, glaciers and snow cover in the future up to 2050. The final goal of the project is to support Central Asian countries to prepare national strategies, policies and investment plans for climate resilience and adaptation. Modeling of future hydrology of the rivers helps to plan water management considering agriculture, hydropower, environment and municipal water services. The project made forecasts on climate change induced hazards as floods, droughts, mudflows, landslides and avalanches. The purpose was also to present proposals for adaptation interventions and capacity building in order to increase climate resilience in different regions. In 2050, river flows in the Amu Darya Basin will be 30 % less compared with the average flow of the past 10 years. Because the glaciers will diminish 45 - 60 %, spring flow remains high and late summer flow will radically decrease. This means floods in spring and droughts during hotter summer months and growing season. Thawing of permafrost in the higher mountains will generate landslides and mudflows. Annual total water demand in the Syr Darya Basin increases by 3.0 - 3.9 % in 2050. Annual unmet demand increases from 8.8% currently to 31.6 - 39.7% in 2050. Annual total water demand in the Amu Darya basin increases by 3.8 - 5.0 % in 2050. Annual unmet demand increases from 24.8 % currently to 45.8 - 54.5 % in 2050. Most cost-effective adaptation measures are (i) improving agricultural practice, (ii) deficit irrigation, (iii) increasing the reuse of water in agriculture, and (iv) the reduction of irrigated areas. Costs for closing the entire water gap (43,000 Mm3) are estimated at US$ 1,730 million per year in 2050. Closing the additional water shortage caused by climate change only (25,000 Mm3) will cost US$ 550 million per year in 2050.

Climate Change

Finnish Consulting Group is implementing an ADB financed Climate Change Adaptation project in five Central Asian countries together with Finnish Meteorological Institute and FutureWater from the Netherlands. The objective of the project is to generate models describing impacts of climate change on water resources and natural hazards in the Aral Sea Basin as well as to develop strategies for adaptation interventions.

In the project, past climate was reconstructed using data from global climate databases compiled by using field and satellite observations. There are several international programs generating satellite-based climate data from the world and this data can be downloaded from Internet.

Several international universities and research organizations have compiled complex climate change scenarios forecasting future climate (Fig. 1). Five different scenarios were used to model the changes of temperature, precipitation, glaciers and snow cover in the future up to 2050. Outputs will be applied to the development of climate resilience and risk management strategies at the regional, national, and river-basin levels. The activities of the project include:

1) Collection of existing field records and new satellite-based data on climate, hydrology and land-use in order to reconstruct the past and present situation; 2) Development of methodologies relating to downscaling of climate change scenarios, hydrological modeling, glacial melt, climate impact assessment, climate vulnerability and risk analysis; 3) Modeling of future hydrology of the rivers and identification of water management problems considering agriculture, hydropower, environment and municipal water services; 4) Generating forecasts on climate change induced hazards as floods, droughts, mudflows, landslides and avalanches; 5) Presenting proposals for adaptation interventions and capacity building in order to increase climate resilience in different regions; and 6) Development of capacity in the use and application of hydroclimatic models in the region.

Figure 1. Climate warming of the study area (Aral Sea Basin) downscaled from five different climate change models (A1B emission scenario).

Hydrologic Impacts of Climate Change

The modeling results showed rather remarkable changes in the hydrology of the Aral Sea Basin. In the Tien Shan and Pamir Mountains the volume of glaciers will dramatically decrease as a result of climate warming. Until 2050, 45 - 60 % of the glacier cover has disappeared. During the 20 coming years, the runoff from melting glaciers will increase only in the high mountain areas, but the total discharge of the main rivers will decrease. The small ice caps especially from Tien Shan will disappear and also the melting water will be dramatically reduced (Fig. 2).

In 2050, river flows in the Amu Darya Basin will be 30 % less compared with the average flow of the past 10 years (Figure 3 and 4). The situation will be worst in the late summer and early autumn when the river discharges are 45 % less than today. Without regulation capacity and reservoirs there will be shortage of irrigation water in the downstream areas. This fact should especially be considered in the management of the downstream part of the Amu Darya River and in the Karakum Canal.

In the Syr Darya Basin (Figure 5 and 6), river flows will be 25 % less compared to the average of the past 10 years. For this basin, major flow reductions also occur in late summer and early autumn. The diminishing water resources together with the increasing temperatures in the plains create major risk for agricultural production.

Figure 2. Also geomorphological evidence, moraines indicating bigger size of glaciers, shows that glaciers have already receded substantially during the last 150 years (KAZ/KGZ border, Almaty; Google Earth).

Figure 3. Origin of water in the past ten years. In the Nurek reservoir in Amu Darya basin glacier melt is very important contributor to total runoff. Therefore, diminishing glaciers will cause major shortage of water in the downstream areas.

Figure 4. Origin and volume of water in the past and future in the Nurek reservoir in Amu Darya basin. Reducing glaciers and glacier melt have major effect on the total volume of discharges in the river.

Figure 5. In the Toktogul reservoir in Syr Darya basin, glacier melt is less important. Snow melt and rain-runoff are the most important contributors to total runoff.

Figure 6. Origin and volume of water in the past and future in the Toktogul reservoir in Syr Darya basin.

Impacts on Water Resources Management

People living in the upper stretches of rivers will certainly continue to construct reservoirs to ensure water allocation to irrigated fields there. This may lead to a situation, where farming in the downstream areas becomes difficult. The situation can clearly be seen from the inflow figures of the reservoirs in the both basins (Fig. 7).

Figure 7. Accelerated decrease of discharges to the Nurek reservoir (Tajikistan, Amu Darya Basin) and linear decrease in Toktogul reservoir (Kyrgyzstan, Syr Darya Basin) 2001 – 2050 modeled using five different climate change scenarios [1]. The models also predict possible annual variations between different years based on the reference period 2001 - 2010.

We analyzed the past and present annual hydrology and modeled the impact of climate change on the rivers and generated hydrographs comparing the present discharges with the predicted future situation (Figure 8). The most significant changes can be seen in the rivers which are flowing out from areas dominated by glaciers.

Figure 8. Annual hydrographs for the Nurek reservoir (Tajikistan, Amu Darya Basin) and Toktogul reservoir (Kyrgyzstan, Syr Darya Basin) in the past (field observations) and in the future modeled using five different climate change scenarios [1]. In January-May the discharges (Q) will remain similar as today, but major reduction will take place in June- November discharges because of diminishing glaciers.

Because the glaciers will diminish, their run-off regulation capacity will be lost. In general, the models do not predict higher discharges for future spring times, but sudden changes in temperature and rainfall may cause abrupt run-off peaks. Also in the future, major snowstorms and fast snow melting may easily generate major floods in spring time.

Impacts on Mountain Environment

Because of the rapid melting and retreat of the of the margin of glaciers, new proglacial lakes will be generated. It is common that ice-cored moraines create dams in front of fast receding valley glaciers (Fig. 9). Such lakes may have extensive amount of water and whenever the dam collapses, a catastrophic flood may occur. Flood protection interventions are needed in areas identified to be vulnerable to flood damages.

Figure 9. A catastrophic flood may occur when an ice-cored dam of a proglacial lake will collapse (Petrov Lake, Kyrgyzstan. Google Earth).

The snowline will raise 200 - 300 meters in average until 2050 (Fig. 10). Hill slopes which have always been covered by snow will be exposed to erosion. Thawing of permafrost in the higher mountains will make the slopes instable and this will generate landslides and mudflows (Fig 11). Such calamities may be dangerous to the settlements and infrastructure downstream. The modeling results do not indicate increasing occurrence of spring-time floods in the future, but due to the increasing mudflows, the impacts of floods may probably be more catastrophic than today.

Figure 10. The snowline will raise 200 - 300 meters in average until 2050.

Figure 11. Thawing of permafrost in the higher mountains will make the slopes instable and this will generate landslides and mudflows (Mailuu Suu, Kyrgyzstan. M. Punkari).

Developing Climate Resilience and Adaptation Strategies

The final goal of the project is to support countries to prepare national strategies, policies and investment plans for climate resilience. This new information will give a good basis for risk assessments and planning of interventions to climate change adaptation (Fig. 12). The consequences of climate change will be very different in different Central Asian countries. Turkmenistan will mostly suffer from increasing droughts and decreasing water resources especially during the growing season.

Figure 12. Cost-benefit analysis of the interventions aiming at saving water in the Aral Sea basin. The analysis shows that agriculture is far the most important sector from where results can be expected.

The analyses show that adaptation costs will be as follows: 1) Full recovery: US$ 1,730 million per year (2050); and 2) Climate Change recovery: US$ 550 million per year (2050). The most efficient means to solve the problem include:

Increased reservoir capacity: Reservoir capacity is already large in the region. Increasing the reservoir capacity by 25 % will decrease the unmet demand by 3437 Mm3, corresponding to an 8 % reduction in unmet demand.

Improving agricultural practice reduces the agricultural demand by 15 %. The unmet agricultural demand is reduced by 30 %. Subsequently unmet domestic demand decreases 13 %. In 2041-2050 this reduces the total unmet demand by 12682 Mm3 (29.4 % reduction).

Increased reuse of water in irrigated agriculture: The overall irrigation efficiency, taking into account reuse, is increased from 95 % to 97 % in the downstream areas and from 90 % to 92 % in the upstream areas. Increasing the irrigation efficiency in agriculture reduces the agricultural demand by 2.1 %. The unmet agricultural demand is reduced by 4.4 %. Subsequently unmet domestic demand decreases 1.7 %. In 2041-2050 this reduces the unmet demand by 1871 Mm3 (4.3 % reduction).

Increasing the reuse of water in domestic use reduces the domestic demand by 20 %.The domestic unmet demand is reduced by 20.3 %. Subsequently, the agricultural unmet demand decreases by 64 Mm3. In 2041-2050 the total unmet demand is reduced with 203 Mm3 (0.5 % reduction).

Reducing the irrigated areas reduces the agricultural demand by 10 %. The agricultural unmet demand decreases by 20.3 %. Subsequently unmet domestic demand decreases 8.8 %. In 2041-2050 this reduces the unmet demand by 8697 Mm3 (20.2 % reduction).

Reducing the domestic demand reduces the unmet domestic demand 10.1 %. The agricultural unmet demand is reduced by 32 Mm3. Reducing the domestic demand by 10 % reduces the total unmet demand 102 Mm3 in 2041-2050 (0.2 % reduction).

Applying deficit irrigation to agricultural areas reduces the agricultural demand with 20 %, because evapotranspiration rates decrease strongly. The unmet agricultural demand is reduced by 39 %. Subsequently unmet domestic demand decreases 18 %. In 2041- 2050 this reduces the total unmet demand by 16617 Mm3 (38.5 % reduction).

FCG Finnish Consulting Group Ltd. in association with FINAL REPORT • FutureWater Part 1 • Finnish Meteorological Institute Executive Summary & Climate Change Asian Development Bank

Water and Adaptation Interventions in Central and West Asia

TA 7532

June 2012

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Table of Contents

EXECUTIVE SUMMARY 4

1. CLIMATE CHANGE MODELING 11 1.1. General 11 1.2. Processing of climate data 12 1.3. Baseline Climate 12 1.4. Climate Scenarios 17 1.5. Downscaling of Climate Scenarios 20 1.6. Results of Climate Change Modeling Work 21 1.7. Summary of Climate Change Modeling Work 29 1.8. References 30

2. MOUNTAIN ENVIRONMENT 31 2.1. Geography 31 2.2. Evidence for Melting Glaciers 31 2.3. Natural Hazards Accentuated by Climate Change 33 2.4. References 36

TABLE OF FIGURES

Figure 1-1. IPCC 4th Assessment Report results for Asia (Christensen et al. 2007). .... 11 Figure 1-2. Annual mean precipitation during 2001-2010 (mm/year) based on PERSIANN (http://chrs.web.uci.edu/) and TRMM (2006, 2007) datasets. .. 14 Figure 1-3. Elevation data (meters above sea level). Grey contours indicate country borders...... 14 Figure 1-4. July, January and annual mean (2001-2010) temperatures (°C) based on measurements made at observing stations interpolated to 0.2°*0.2° grid using Kriging-intepolation method...... 15 Figure 1-5. Correlation between observed and predicted daily mean temperature. Each dot present correlation of one day calculated for the all the available observing stations...... 16 Figure 1-6. The 95% percentiles and median of the variance of interpolated daily average temperatures (°C)...... 17 Figure 1-7. a) CO2 emissions and b) cumulative atmospheric CO2 concentrations of the SRES scenarios A2, A1B and B1 (Nakicenovic et al. 2000)...... 19 Figure 1-8. Global GHG emissions (in GtCO2-eq) in the absence of climate policies (left): six illustrative SRES marker scenarios (coloured lines) and the 80th percentile range of scenarios published since SRES (post-SRES) (gray shaded area). (IPCC 2007)...... 19 Figure 1-9. Average change of annual (A) January (B) and July (°C) temperature (°C) and precipitation (%) between the control simulations (simulation period: 1971-2000) and the simulations into the future (five simulation period: 2045-2065). Results are based on the runs made using the five models given in Table 1...... 27 Figure 1-10. Annual (A) January (B) and July (°C) temperature (°C) and precipitation (mm) mean values in 2041-2050. The values are mean of the five climate models (Table 1) used in the Project...... 28

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Figure 1-11. The most important issue in the modeling work is to predict the fate of the mountain glaciers in the upper watersheds. During the recent century, the snouts of almost all the valley glaciers in Central Asia have receded (NE Tajikistan, Kurumdy region, Google Earth)...... 29 Figure 2-1. The retreat of glacier snouts in eastern Tien Shan, Kyrgyzstan, show an accelerating trend especially after 1940 (Kutuzov & Shahgedanova 2009). 32 Figure 2-2. Also geomorphological evidence, marginal moraines indicating bigger size of glacier, shows that glaciers have already receded substantially during the last 150 years (KAZ/KGZ border, Almaty; Google Earth)...... 33 Figure 2-3. The Petrov proglacial lake in Kyrgyzstan has formed as a result of the melting of the glacier. Former 150 years old marginal moraines are damming the lake (Google Earth)...... 34 Figure 2-4. The altitudes of climatic snowline (monthly average T = 0 C) and its rising as a result of climate change in the study area...... 35 Figure 2-5. Rising permafrost altitude will expose new areas for erosion. Therefore, landslides and mudflows become common in the upper parts of the mountain ranges (Northern Pakistan; Google Earth)...... 35

TABLE OF TABLES

Table 1-1. Global climate models used in the Project...... 18

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

ABREVIATIONS AND ACRONYMS

ADB Asian Development Bank ASB Aral Sea Basin ASTER Advanced Spaceborne Thermal Emission and Reflection Radiometer (Terra satellite) CACILM Central Asian Countries Initiative for Land Management CAC Central Asian Countries CAR Central Asian Republics CAREWIB Central Asia Regional Water Information Base Project CRU Climatic Research Unit (University of East Anglia; climate database holder) EIA Environmental Impact Assessment FCG FCG Finnish Consulting Group Ltd FMI Finnish Meteorological Institute GAM General Additive Model GCM General Circulation Model GEF Global Environment Facility GIS Geographical Information System GLIMS Global Land Ice Measurements from Space GPS Geographical Positioning System (satellite based) GWP Global Water Partnership ICARDA International Center for Agricultural Research in the Dry Areas ICSD Interstate Commission on Sustainable Development (IFAS) ICWC Interstate Commission for Water Coordination (IFAS) IFAS International Fund for the Aral Sea IFI International Financing Institution IPCC International Panel for Climate Change IWRM Integrated Water Resource Management MODIS Moderate Resolution Imaging Spectroradiometer (Aqua/Terra satellite) M&E Monitoring and evaluation NGO Non-governmental Organization QA Quality Assessment QMS Quality Management System SIC Scientific Information Center of IFAS SLIMIS Sustainable Land Management Information System (in CACILM) SLM Sustainable Land Management (CACILM program component) TL Team Leader TRMM Tropical Rainfall Measuring Mission (Multisatellite Precipitation Analysis) UNEP UN Environment Program WB World Bank WEAP Water Evaluation And Planning (modeling system developed by Stockholm Environment Institute) WMO World Meteorological Organization

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

EXECUTIVE SUMMARY

Project Background

Central Asia is one of the regions suffering most from the impacts of climate change on its nature and economy. The Final Report of the project ADB / TA 7532: "Water and Adaptation Interventions in Central and West Asia" describes a study which developed models on the impacts of climate change on water resources and natural hazards in the Aral Sea Basin. The past climate was reconstructed using data from global climate databases compiled by using field and satellite observations. Several international climate change projections were used to model the changes of temperature, precipitation, glaciers, snow cover and river discharges in the future up to 2050. Modeling of future hydrology of the rivers helps to plan water management considering agriculture, hydropower, environment and municipal water services.

Water resources management in the Central Asia region faces big challenges. The hydrological regimes of the two major rivers in the region, the Syr Darya and the Amu Darya, are complex and vulnerable to climate change. Water diversions to agricultural, industrial and domestic users have reduced flows in downstream regions, resulting in severe ecological damages. The administrative-institutional system is fragmented, with six independent countries sharing control, often with contradicting objectives.

The discharge in both the Syr Darya and the Amu Darya rivers is driven mainly by snow and glacial melt. The impact of a warming climate on these key hydrological processes has not sufficiently been understood and only fragmentary mitigation and adaptation strategies are in place. Whereas changes in precipitation levels are hard to predict for the future, there is a solid consensus that average global temperatures are rising. As a result, more precipitation will fall as rain in the upstream and the ice volume in the Tien Shan and Pamir mountain ranges will likely shrink. The former will impact the seasonality of the runoff whereas the latter will at least temporarily increase average annual flows. Furthermore, changes in sediment loads may pose additional problems. The impacts have not sufficiently been quantified and adaptation and mitigation strategies are not in place.

The final goal of the project is to support Central Asian countries to prepare national strategies, policies and investment plans for climate resilience and adaptation. The unfavorable developments in this geopolitically important and fragile region call for urgent attention of the international community. Interdisciplinary research can critically inform decision making in the region for better risk management and the design of mitigation and adaptation strategies against negative climate change impacts.

Project Activities

In the project, past climate was reconstructed using data from global climate databases compiled by using field and satellite observations. There are several

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change international programs generating satellite-based climate data from the world and this data can be downloaded from Internet. Several international universities and research organizations have compiled complex climate change scenarios forecasting future climate. Five different scenarios were used to model the changes of temperature, precipitation, glaciers and snow cover in the future up to 2050. Outputs will be applied to the development of climate resilience and risk management strategies at the regional, national, and river- basin levels. The activities of the project included:

This report presents a two-way modeling study assessing the impact of climate change for future water availability in the Amu Darya and Syr Darya river basins. The Part I describes the reconstructed climate change impacts in the basin and Part II focused on the impact of climate change for future runoff generation in the mountainous upstream parts of the basins. The projected inflow from the upstream parts into the downstream areas serves as input for a water allocation model developed in the Water Evaluation and Planning system (WEAP) to simulate water demands and resources in the downstream areas (Part III). The effects of future changes in temperature and precipitation for the future water availability and demand were simulated and the effects of possible adaptation measures explored.

Climate Change Projections

Climate data to be applied in hydrological modeling included daily mean, maximum and minimum temperatures and daily precipitation data in a 0.2°*0.2° grid covering the Central Asian countries. On one hand the hydrological modeling required data of present day climate - for calibration of the model - and on the other hand projections of the future climate - for simulations into the future. The gridded data representing present-day daily mean, maximum and minimum temperatures were produced by surface observations of temperature (years 2001-2010) and kriging interpolation. The present-day daily precipitation data were based on PERSIANN satellite based data except years 2006-2007 that were based on TRMM satellite based precipitation data. Projections of future climate for the period 2011-2050 were processed using simulations of five different global climate models of the IPCC's CMIP3 project. The global model simulations into the future were

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change downscaled to the dense grid using the gridded present-day climate data and delta change method.

Simulations with global climate models suggest that the annual mean temperatures are going to rise in the Central Asian region by about three degrees during the coming 40 years. The warming is projected to be strongest in the mountains and in the northern parts of Central Asia. On annual basis the precipitation changes are projected to be small during the coming 40 years. There is variation from model to model, however, most climate models suggest that the already dry south-western parts of Central Asia will become even drier especially during summer, and that in some places in the mountains the precipitation rates might increase.

Mountain Environment

Glaciers cover 18,128 km2 from the Aral Sea Basin and they have an important role in hydrology as they release melt water also during the dry summer months. Mountain glaciers include: 1) Small cirque glaciers resting on rather steep mountain slopes; 2) Large ice caps covering mountain tops associated with valley glaciers, narrow and long ice tongues flowing down in U-shaped valleys.

The percentage of glaciated area of the two catchments differs significantly. In the Amu Darya glaciers cover 15,500 km2 (2% from the area) and 1.800 km2 in the Syr Darya (0.15% from the area). The biggest glaciers are located in the Pamir Mountains in Tajikistan.

The glaciers in the Tien Shan and Pamir are retreating and the rates of retreat vary between regions and time periods. The largest retreat rates have been observed in the northern Tien Shan where glaciated area has declined by 30- 40% during the second half of the 20th century. The rates of deglaciation were slower further east and south, however, acceleration of glacier retreat has been noted in the eastern Pamir from 7.8% over 1978–1990 to 11.6% over 1990–2001 periods. Glaciers have lost 12.6% (0.33% /yr) of their 1965 area in the 1965-2003 period. One purpose of the project was to predict melting of glaciers in the future using modeling of climate change impacts on mountain hydrology.

New proglacial lakes will be generated when maybe ice-cored moraines create dams in front of fast receding valley glaciers. Such lakes may have extensive amount of water and when the dam collapses, a catastrophic flood may occur. Flood protection interventions are needed in areas identified to be vulnerable to flood damages.

The snowline will raise 200 - 300 meters in average until 2050. Hill slopes which always have been covered by snow will be exposed to erosion. Thawing of permafrost in the higher mountains will make the slopes instable and this will generate landslides and mudflows. Such calamities may be dangerous to the settlements and infrastructure downstream. The modeling results do not indicate increasing occurrence of spring-time floods in the future, but due to the increasing mudflows, the impacts of floods may probably be more catastrophic than earlier.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Hydrology of Upstream Areas

A raster based highly detailed cryospheric-hydrological model for the upstream parts of the Amu and Syr Darya basins was developed. The upstream hydrology and the role of snow and ice are crucial for understanding the downstream impact. The impacts of climate change for the upstream hydrology are modeled in high detail including dynamic changes in glacier and snow cover. The contributions of glacier melt, snow melt, rainfall and base flow to the total runoff are determined by the model. The model was calibrated for a historical reference period (2001-2010), using historical climate data and reservoir inflow observations. The calibrated model was subsequently forced with downscaled future temperature and precipitation scenarios for the A1B scenario based on five different Global Circulation Models for 2011-2050.

The results of the hydrological modeling in the upstream areas included the following:

• The developed model is able to mimic observed streamflows and can be used to explore the impact of climate change on the hydrological cycle. • There are large differences in the role that melt water plays in runoff generation in the Amu Darya and Syr Darya river basins. Melt water has a higher contribution to runoff in the Amu Darya basin compared to the Syr Darya river basin. • It is very likely that glacier extent in the Pamir and Tien Shan mountain ranges will decrease by 45 to 60% by the year 2050. • The composition of the four components of stream flow (rainfall-runoff, snow melt, glacier melt, base flow) is very likely to change in the future. This will have major impacts on total runoff, but especially on seasonal shifts in runoff. The runoff peak will shift from summer to spring and decrease in magnitude. Model output when forced with climate projections generated with five Global Circulation Models shows decreasing runoff generation in the upstream parts of the two basins until 2050. The changes differ strongly spatially. The runoff generation decreases most significantly in upstream areas of glacier retreat. • Total annual runoff into the downstream areas is expected to decrease by 22-28% for the Syr Darya and 26-35% for the Amu Darya by 2050. • Strongest decreases in stream flow are expected for the late summer months (August, September, October), where inflow into downstream areas decreases around 45% for both river basins.

Water Management in Downstream Areas

The WEAP tool was used as an integrated framework for water supply-demand balancing and modeling of adaptation scenarios in the Amu and Syr Darya river basins. Water demand and water stress were evaluated at the basin level for the current situation and the future. The model is set up for a historical reference period (2001-2010) using modeled inflow from upstream, historical data regarding climate, agricultural land use, population, and water

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change management structures like reservoirs and canals. The model is calibrated using observed reservoir inflows, outflows and volumes, and Aral Sea inflow. Subsequently, the calibrated model is forced with downscaled future temperature and precipitation scenarios for the A1B scenario based on five different Global Circulation Models for 2011-2050 and the corresponding inflow generated by the upstream model.

The key messages resulting from the downstream modeling study are:

• The developed models are able to mimic observed streamflows and water use rates and can be used to explore the impact of climate change on water resources in the region. • There are large differences in the role that melt water plays in runoff generation in the Amu Darya and Syr Darya river basins. Melt water has a higher contribution to runoff in the Amu Darya basin compared to the Syr Darya river basin. • Annual total water demand in the Syr Darya basin increases by 3.0 - 3.9% in 2050. Annual unmet demand increases from 8.8% currently to 31.6 - 39.7% in 2050. • Annual total water demand in the Amu Darya basin increases by 3.8 - 5.0% in 2050. Annual unmet demand increases from 24.8% currently to 45.8 - 54.5% in 2050. • The total extent of the Aral Sea will reduce from about 17,000 km2 currently to 13,800 km2 in 2050. Differences between the North and South Lake are striking. The North Lake has the potential to expand by about 72% in terms of projected inflow (but will be less given spilling to South Lake), while the South Lake will shrink by about 38%.

Adaptation Strategies

To overcome current and future water shortage countries have a range of options at their disposal to respond and adapt. These options can be summarized into three broad categories: (i) expanding supply, (ii) increasing the productivity, and (iii) reducing demand. For each of these three categories typical options were explored in the study resulting in the following framework:

Objective Technology Expanding supply: A: Increased reservoir capacity Increasing the productivity: B: Improved agricultural practice C: Increased reuse of water in irrigated agriculture D: Increased reuse of water for domestic use Reducing demand: E: Reduction of irrigated areas F: Reduction of domestic demand G: Deficit irrigation

Each of these options is associated with certain marginal unit costs, ranging from US$ 0.02 per m3 for improving agricultural practices to US$ 2.00 per m3

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change in case of reducing supply to domestic demand. It is clear that in general the cheapest options will be introduced first, but at the same time might not be sufficient to overcome water shortage completely and more expensive options are required to bridge the water gap. By ranking the adaptation options by their unit costs country specific water marginal costs curves are constructed. The water availability cost curve’s use is limited to comparing measures’ financial costs to close the gap. It is important to note that these might be different from the economic costs for society as a whole. The cost curve should be therefore considered as a guide for comparing the financial costs of measures for decision-making.

Most cost-effective adaptation measures are improving agricultural practice, deficit irrigation, increasing the reuse of water in agriculture and the reduction of irrigated areas. Besides, the measures applied to agriculture are much more effective in terms of unmet demand reduction because the domestic water use in the basins is negligible compared to the water use for agriculture. Applying the most cost-effective adaptation measures will close the water gap and cost US$ 1,730 million per year in 2050. Closing the water gap caused by climate change only will cost US$ 550 million per year in 2050.

• Most cost-effective adaptation measures are (i) improving agricultural practice, (ii) deficit irrigation, (iii) increasing the reuse of water in agriculture, and (iv) the reduction of irrigated areas. • Costs for closing the entire water gap (43,000 Mm3) are estimated at US$ 1,730 million per year in 2050. • Closing the additional water shortage caused by climate change only (25,000 Mm3) will cost US$ 550 million per year in 2050.

The ongoing construction of new dams in Kyrgyzstan and Tajikistan is adding tension to the existing situation. The Soviet-era designed hydropower projects Kambarata I and II in Kyrgyzstan and the Rogun dam in Tajikistan are on the table again as a result of an increased access to international donor money with Russia and China investing in these projects. For the downstream countries, these developments have raised concern because this can mean that the upstream states can decouple themselves the necessity to receive energy deliveries in the winter from Kazakhstan, Uzbekistan and Turkmenistan. The upstream countries could lose their will to abide to summer operation rules with severe impacts to irrigated agriculture and the overall economy. From this perspective, it is not surprising that certain tensions between the countries exist. Although the new infrastructure will be effective at damming river flow and in adding management options that are direly needed, measures need to be taken so that further flow impediment does not equal impediment to regional integration.

Capacity Building

The project identified national organizations and 5 - 8 experts from each country to participate to the capacity building program aiming at developing skills in hydroclimatic modeling. The experts represented hydrometeorological institutes, other authority or research organizations responsible for water

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change management or technical institutes specialized to modeling and information systems. The project team organized several seminars on climate change in the countries and experts from different ministries, academic institutes and other projects were invited. The presentations made by the project experts and national authorities provided information on climate change, its impacts on the water management and on possible adaptation strategies.

The national experts received from the project powerful laptop computers, software and other equipment needed for modeling and information management. The international experts trained the national experts to use climate data and modeling (Downscaling, R programming), hydrologic modeling (PCRaster) and water allocation modeling (WEAP). The national experts will continue the work of the project e.g. on national scales by using more detailed reference statistics.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

1. CLIMATE CHANGE MODELING

1.1. General

In the 4th Assessment Report of the IPCC large changes in climate are projected across Central Asian region (Christensen et al. 2007). Figure 1-1 shows the projected changes in temperature (top row) and precipitation (bottom row) between the 1980-1999 and 2080-2099 periods for the A1B SRES scenario as a mean of 21 global climate models (GCM). Figures are shown for annual, December to February (DJF) and June to August periods (JJA). In the Pamir mountains, temperature increases up to 4 degree in 2100 combined with a modest increase in precipitation are projects. The increased temperature results in a higher evapotranspiration demand and will outweigh the modest increase in precipitation. In the Central Asian plains where agricultural production takes place the climate will be drier and hotter.

Figure 1-1. IPCC 4th Assessment Report results for Asia (Christensen et al. 2007).

Like indicated in Figure 1-1 considerable changes in climate conditions may happen in the future. In order to be able to assess the impacts of climate change on hydrological conditions and water resource management meteorological data is needed. Observed data is needed for the estimation of present/past climate and projections based on climate model simulations give the best estimate for the future climatological conditions. In this project the so called baseline (observed data) period is 2001-2010 and the future period to be examined is 2011-2050. The hydrological models applied in this Project require as input continuous timeseries of daily temperature and daily precipitation data. All the data has to be in a common grid covering the Project area.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

1.2. Processing of climate data

Processing and analyzing the large meteorological and hydrological dataset require advanced analyzing tools. In this Project processing the climate observations and climate projections was mainly done using the R environment (http://www.r-project.org/). R is a free software environment for statistical computing and graphics. The use of R has expanded recently and is has become almost a standard in statistical computing and data processing. It can be run on a variety of operating systems like UNIX platforms, Windows and MacOS. It is a GNU project which is similar to the S language and environment which was developed at Bell Laboratories (formerly AT&T, now Lucent Technologies) by John Chambers and colleagues. R provides a many kinds of statistical and graphical techniques, and is highly extensible. For example using R it is easy to produce graphs and plots including mathematical symbols and formulae. R is available as Free Software under the terms of the Free Software Foundation's GNU General Public License in source code form. This means that there are no installation or maintenance costs related to R.

R is designed around a true computer language and this enables users to add additional functionality by defining new functions. For computationally- intensive tasks, C, C++ and Fortran code can be linked and called at run time. R includes useful feature e.g. such as: an effective data handling and storage facility, a suite of operators for calculations on arrays, in particular matrices, and a well-developed, simple and effective programming language which includes conditionals, loops, user-defined recursive functions and input and output facilities.

R can be regarded as an environment within which statistical techniques are implemented. R can be extended easily via packages. There are about eight packages supplied with the R distribution and many more are available through the CRAN family of Internet sites covering a very wide range of modern statistics (source: http://www.r-project.org/).

Training to R programming was one of the main capacity building component of this Project. To the trainees were introduced basics of R, as well as the tools to make spatial interpolations and downscale climate scenario data.

1.3. Baseline Climate

The baseline dataset produced in this Project includes daily average temperature, daily maximum temperature, daily minimum temperature and daily precipitation data for 2001-2010. The unit for temperature is degrees Celsius and for precipitation mm/day. The area coverage is: from 57.12083°E to 78.92083°E longitude and from 34.10833°N to 49.90833°N latitude. The spatial resolution is 0.2 x 0.2 degrees. This means that the dataset is 109 columns x 79 rows. The coordinates of the first value are such that the value is centered at 57.22083 °E, 34.20833°N and the second value at 57.42083°E, 34.40833°N. The last two values are centered at 78.62083°E, 49.60833°N and 78.82083°E, 49.80833°N (e.g. Figure 1-2).

Precipitation data (Figure 1-2) were bilinearly interpolated to the 0.2° x 0.2° resolution grid from the PERSIANN 0.25° neural network based

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change precipitation data (http://chrs.web.uci.edu/) (Adler et al. 2003; Huffman et al. 2009; Hsu et al. 1997; Sorooshian et al. 2000). PERSIANN precipitation estimation is based on cloud texture information from longwave infrared images (~10.2-11.2 µm). This data is obtained from geostationary satellites and updated using the higher quality rainfall estimates from low-orbit passive microwave sensors (Hsu et al. 2007; Sorooshian et al. 2000). The reduction of bias and still preserving spatial and temporal patterns in high resolution, PERSIANN precipitation is adjusted based on Global Precipitation Climatology Project (GPCP) rainfall at 2.5° monthly resolution (Adler et al. 2003; Huffman et al. 2009). Missing data in PERSIANN estimation is filled with passive microwave rainfall estimation at each 30 minutes time step. In the subsequent step, the data is aggregated to 2.5° monthly scale and a correction factor is computed based on the ratio of GPCP rainfall and PERSIANN rainfall at a grid of 2.5° monthly scale. This ratio is then used to calculate the PERSIANN rainfall fine spatial and temporal scale (hourly) within the 2.5° coverage. The final product is provided to users at 3-hour and 0.25° resolution.

The GPCP dataset (Adler et al. 2003) used for adjustment of PERSIANN data is a global monthly analysis of surface precipitation at 2.5° latitude 2.5° longitude resolution. The data starts from January 1979 onwards. The data incorporates precipitation estimates from low-orbit satellite microwave data, geosynchronous-orbit satellite infrared data, and surface rain gauge observations. The higher accuracy of the low-orbit microwave observations is used to calibrate, or adjust, the more frequent geosynchronous infrared observations. Microwave measurements are available only after mid-1987, the period before that is processed by using only infrared observations. These observations have been calibrated using the microwave based analysis available of for the later years. The combined satellite-based product is finally adjusted by the rain gauge analysis. To the dataset archive is included also the individual input fields, a combined satellite estimate, and error estimates for each field.

According preliminary analyses for years 2006 and 2007 PERSIANN data was giving too high values for some regions of interest when compared with the surface observations. For that reason for those two years a dataset obtained from the TRMM (http://trmm.gsfc.nasa.gov/) mission was used. TRMM is a joint mission between NASA and JAXA. It was launched on November 27, 1997 from Tanegashima, Japan and it monitors the rainfall in the tropics. TRMM is part of the NASA Mission to Planet Earth. TRMM instruments include Microwave Imager (TMI), Precipitation Radar (PR), Visible and Infrared Scanner (VIRS), Cloud and Earth’s Radiant Energy System (CERES) and Lightning Imaging Sensor (LIS) (see e.g. Kaushik et al. 2010).

Elevation data was interpolated from the DEM_15s data into 0.2° x 0.2° resolution (Figure 1-3). Coordinates are as in temperature and precipitation data. This data is needed for the spatial interpolation of temperature.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Figure 1-2. Annual mean precipitation during 2001-2010 (mm/year) based on PERSIANN (http://chrs.web.uci.edu/) and TRMM (2006, 2007) datasets.

Figure 1-3. Elevation data (meters above sea level). Grey contours indicate country borders.

Temperature data includes daily minimum, maximum and average temperatures (degree Celsius). The gridded dataset needed for the Project was obtained by interpolating at meteorological stations measured data into the 0.2° x 0.2° grid. The interpolation was done using a spatial interpolation

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change method known as “kriging” (e.g. Ripley 1981). The theory behind interpolation by kriging is based on the work of D. Krige, This method is widely used also in meteorological and climatological applications (e.g. Venäläinen and Heikinheimo 1997; Haylock et al. 2008). Kriging is a spatial interpolation method, which gives the best linear unbiased predictors of unobserved values. The method also provides an estimate of the variance of the prediction error. This interpolation method can take into account external forcing, like terrain properties or altitude.

The programming tool applied in this project is the R environment that is created for statistical computing and graphics. More precisely R package called “gstat” is applied here. Package “gstat” was programmed by Pebesma (2004).

Temperature is very much dependent on elevation, the higher the location is the cooler the temperatures typically are. This is why terrain elevation was used to predict the spatial variation of temperature also in this project. As an example of results obtained July, January and annual mean tempeatures are given in Figure 1-4.

Figure 1-4. July, January and annual mean (2001-2010) temperatures (°C) based on measurements made at observing stations interpolated to 0.2°*0.2° grid using Kriging- intepolation method.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Quality checks of the kriging

The quality of interpolated fields is affected by the availability of input data (observations), the terrain properties like variation of elevation and naturally the explanatory power of e.g. elevation on temperature. The number of observing station available for the interpolation in the current project was small when the areal extend and complexity of the region is taken into account. To get an estimate of the performance of the method applied the kriging interpolated data was checked for each day by calculating the correlation between observed and to the nearest grid box predicted temperature for each station location (Figure 1-5). For most days the correlation is above 0.95 that can be regarded as a good accuracy when the above mentioned challenges are taken into account.

Correlation

Day from 1.1.2001

Figure 1-5. Correlation between observed and predicted daily mean temperature. Each dot present correlation of one day calculated for the all the available observing stations.

Variance of the kriging as an indicator of the uncertainty of the predicted values

Variance can be used also as an indicator of the quality of interpolation. The smaller the variance is the better the interpolated fields are. The mean, maximum and minimum variances for interpolated daily mean temperatures can be seen in (Figure 1-6). The variance of the predicted values were best near observation stations. The variance on the south-east corner of the kriging area were occasionally large indicating the inaccuracy of the data on the mountain area with high variation in elevation However, the quality of interpolated fields can be regarded relatively good when the limitations related to the availability of data are taken into account.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Figure 1-6. The 95% percentiles and median of the variance of interpolated daily average temperatures (°C).

1.4. Climate Scenarios

Projections of future climate are based simulations made using atmospheric Global Climate Models (GCM). These state-of-the-art comprehensive GCMs with modules simulating the marine and land biospheres and sea ice are the tools for the study of past, present and future climates. The resolution of such GCMs is usually 100–300 km (IPCC 2007). For the regional studies the spatial resolution of CGMs’ is still quite coarse and also not all the processes in land- sea-atmosphere system can be fully taken into account. Spatially a more detailed estimate can be gained by using regional climate models (RCM) or statistical methods for downscaling the output of the GCMs. However, even for

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

RCMs there are processes that can only be approximated and the statistical methods have their limitation, as well. The uncertainty related to the climate scenarios can to some extent be taken into account by using not only one but several GCMs as input for the impacts models. In this Project results of five separate GCMs we applied.

The needed GCM simulations were downloaded from the World Climate Research Programe (WCRP) Climate Model Intercomparison Project (CMIP3) Multi-Model Dataset Archive. Altogether 19 climate models’ output have been downloaded. These models belong to the set of GCMs employed by IPCC in composing the global-scale projections published in the 4th assessment report. More details of the models is given in IPCC (2007) table 8.1. Out of the 19 models finally five models were selected to be used in this Project (Table 1-1).

The criteria for the selection of these models was that their spatial resolution is 1.9° or higher and the models originate from different countries. This ensures that the models are genuinely separate models and this way the results obtained by employing these models depict the scale of variation different climate projections possess.

Model Institute, Country Resolution

CGCM3(T63) Canadian Centre for Climate 1.9˚ x 1.9˚ Modeling and Analysis, Canada CNRM-CM3 Météo-France, France 1.9˚ x 1.9˚

ECHAM5/MPI-OM Max Planck Institute for 1.5˚ x 1.5˚ Meteorology, Germany MIROC3.2(HIRES) Centre for Climate System 1.1˚ x 1.1˚ Research (University of Tokyo), Japan NCAR-CCSM3 National Center for 1.4˚ x 1.4˚ Atmospheric Research, USA

Table 1-1. Global climate models used in the Project.

T he duration of the warming phase and the rate of warming will depend on the amount and nature of anthropogenic emissions. In climate change assessments the commonly-applied scenarios are set out in an IPCC SRES report (Nakicenovic et al. 2000). The CO2 emissions and cumulative atmospheric CO2 concentrations of the A2, A1B and B1 scenarios are illustrated in Figure 1-7. In the B1 scenario, the world’s economy will aim at sustainable development, and the growth of the world’s population will level off. A2 is a consumer society scenario in which energy is mainly produced by fossil fuels (leading to high emissions) and the world’s population will grow rapidly. The A1B is a scenario lying in between A2 and B1. The estimates of the atmospheric CO2 concentration reached in the year 2100 ranges in the B1 and A2 scenarios from 550 ppm to 850 ppm, respectively.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Figure 1-7. a) CO2 emissions and b) cumulative atmospheric CO2 concentrations of the SRES scenarios A2, A1B and B1 (Nakicenovic et al. 2000).

The simulations used in this Project are based on the emission scenario A1B. According to A1B emission scenario the global temperature is project to rise 2-3 °C before the end of this century ( Figure 1-8).

Figure 1-8. Global GHG emissions (in GtCO2-eq) in the absence of climate policies (left): six illustrative SRES marker scenarios (coloured lines) and the 80th percentile range of scenarios published since SRES (post-SRES) (gray shaded area). Dashed lines show the full range of post-SRES scenarios. Right Panel: Solid lines are multi-model global averages of surface warming for scenarios A2, A1B and B1. The pink line depicts alternative in which atmospheric concentrations are held constant at year 2000 values. The bars at the right of the figure indicate the best estimate (solid line within each bar) and the likely range assessed for the six SRES marker scenarios at 2090-2099. All temperatures are relative to the period 1980-1999. (IPCC 2007).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

1.5. Downscaling of Climate Scenarios

Downscaling of climate scenarios from the coarse global climate models’ grid to the 0.2° grid was done by applying high resolution (0.2*0.2°) dataset based on observations. The assessment of magnitude of change was based on method known as “delta change” (e.g. Arnell 1996) where the magnitude of the change is assumed to be the difference between the climate model’s control and scenario runs. The steps taken in the downscaling were:

1. Calculation of the changes in mean temperature and daily temperature variance and precipitation of each month between the control run (1971-2000) and the simulation into the future (2046-2065); 2. Calculated changes were added to observed values of temperature and precipitation in 2001-2010 to formulate scenarios for 2011-2050.

For temperature the absolute change (°Celsius) and for precipitation relative change (%) was applied.

In order to take into account possible change in the variation of daily temperatures in the future climate, we considered the changes in the monthly mean temperature and daily variance of temperature. The daily temperatures

(Tske) were calculated from the observed (2001-2010) temperature values (Tpre) by

, , (1) = + ∆ + − 1 − ,

in which the ∆ is the projected monthly mean temperature rise calculated for each calendar month. and are the standard deviations of daily , , temperature calculated for each calendar month based on the temperatures obtained from the model simulations for the baseline (2001-2010) period (pre) and for the future period (ske). is the mean monthly temperature in the observed dataset.

The daily precipitation values (Pske) were calculated from the observed (2001- 2010) daily precipitation values (Ppre) by

∆ , (2) = + % ∗

in which the ∆ is the projected monthly mean precipitation change in percentages (%) calculated for each calendar month, and t is the time step (day).

If the simulated monthly precipitation was in a climate model simulation less than 1 mm/month, the ∆ was set to zero, as the use of the percentual change would have led to unrealistic precipitation changes (e.g., in case the control simulation simulated a monthly precipitation value of 0.005 mm and in the simulation of the future simulated a monthly precipitation of 0.5 mm, hence the ∆ would be 9900%).

Statistical downscaling methods are often applied because they require less computational effort than dynamical downscaling, they offer the opportunity

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

for testing scenarios for many decades or even centuries and include the opportunity to use “ensemble” GCM results. For statistical downscaling long time series of observations needed and one can also ask weather present day climate can depict future conditions. Dynamical downscaling is a physically justified approach and e.g. includes many of the feedbacks of the climate system. However, the method is very dependent on the boundary conditions defined by the GCM and it is sensitive to parameterizations used in the RCM used for the downscaling. It also requires a lot of computing resources and was already in that sense not possible to be applied in this Project.

1.6. Results of Climate Change Modeling Work

According to the results the annual mean temperature is projected to rise by about three degrees and the change annual precipitation is very small during the coming 40 years (

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Figure 1-9 A). When we look at the seasonal precipitation changes, the already dry south-western parts become even drier and especially during summer. In some places in the mountains the precipitation might increase. However there is variation from model to model (

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Figure 1-9 B and C). The mean temperatures rise everywhere around the year. The warming is strongest in the mountains and in the northern parts of the area.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

When the rate of change given Figure 1-1 is qualitatively compared with the rate of change given in

Figure 1-9 it is possible to see that the general features are the same, i.e. reduction of precipitation during summer season on mountains and same time rise of temperature. In January the precipitation will increase in most of the

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change region except at the most southern areas. The results given in

Figure 1-9 have more details. It is also good to remember that though the models indicate change in precipitation for some of the dry regions even as large as 20% it means in reality almost non-measurable change as the precipitation amounts in those regions are so small.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

The predicted change means e.g. that the temperature isotherms will move northwards and upwards when compared with past or current climate (Figure 1-10). For example the January -10 C° isotherm is predicted to move about 150-200 km northwards during the coming 40 years. Rising temperature means that the 0°C isotherm moves upwards in mountainous areas and this leads to faster melting of glaciers.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Figure 1-9. Average change of annual (A) January (B) and July (°C) temperature (°C) and precipitation (%) between the control simulations (simulation period: 1971-2000) and the simulations into the future (five simulation period: 2045-2065). Results are based on the runs made using the five models given in Table 1.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Figure 1-10. Annual (A) January (B) and July (°C) temperature (°C) and precipitation (mm) mean values in 2041-2050. The values are mean of the five climate models (Table 1) used in the Project.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

1.7. Summary of Climate Change Modeling Work

Climate data to be applied in hydrological modeling included daily mean, maximum and minimum temperatures and daily precipitation data in a 0.2°*0.2° grid covering the Central Asian countries. The baseline is 2001- 2010 and interpolated temperature values are based on measured data. Baseline precipitation is based on so called PERSIANN data except years 2006- 2007 that are based on TRMM data. Projections of future climate for the period 2011-2050 were processed using five different global climate models. Data was downscaled to the dense grid using statistical downscaling method. The capacity building activities include production of training material and conduction of workshops and presentations on subjects such as: Global climate datasets, climate modeling, spatial interpolation, statistical downscaling and on R programming. The training material have been delivered for the use of the Project trainees. The same subjects were included also to the program of the Project’s final seminar.

The climate change modeling work done produced datasets that were used in the hydrological modeling. The Project, conclusions are drawn to describe the past and present climate and the changes taking place in the future. Such changes are very diverse in different parts of the Aral Sea basin and the future of mountain glaciers is a crucial issue (Figure 1-11). Future changes in temperature and precipitation are important factors, together with river fluxes, whenever the impacts of climate change on e.g. socio-economic and ecological issues are assessed.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

1.8. References

Haylock, M., Hofstra, N, Klein Tank, A., Klok, E., Jones, P and New, M., 2008. A European daily high-resolution gridded data set of surface temperature and precipitation for 1950–2006. Journal of Geophysical Research, Vol. 113, D20119, doi:10.1029/2008JD010201 IPCC, 2007. Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team, Pachauri, R.K. and Reisinger, A. (eds.)]. IPCC, Geneva, Switzerland, 104 pp. Kaushik, G., Wang, N-Y., Ferraro, R., Liu, C., 2010. Status of the TRMM 2A12 Land Precipitation Algorithm Journal of Atmospheric and Oceanic Technology Volume 27, Issue 8 (August 2010) pp. 1343-1354 doi: 10.1175/2010JTECHA1454.1 Abstract Nakicenovic, N. & Swart, R. (eds.). 2000. Special Report on Emission Scenarios. A special report of Working group III of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 599 pp Pebesma, E.J., 2004. Multivariable geostatistics in S: the gstat package. Computers & Geosciences, 30: 683-691

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

2. MOUNTAIN ENVIRONMENT

2.1. Geography

Mountains of Central Asia (Tien Shan and Pamir) are geologically rather young (540 - 250 million yr) are very high, sharp crested and steep sloped. The highest peaks in the Aral Sea Basin of the Tien Shan are c. 7,000 m and frequently 4000 - 5000 m. The Pamir Mountains in the Aral Sea Basin include several peaks that are over 7,000 m. The highest mountains is Ismoil Somoni Peak (Communism Peak) 7,495 m.

Glaciers transport debris which will be deposited as moraines to the sides and edge of the glacier. Lateral moraines are formed on the sides of the glacier. Terminal or end moraines are formed at the foot or terminal end of a glacier. Medial moraines are formed when two different ice streams, flowing in the same direction, coalesce and the lateral moraines of each combine to form a moraine in the middle of the merged streams. Supraglacial debris may be very thick close to the margin of a glacier and it will slow down melting of ice. Because ice tongues are covered by debris, it is difficult to observe where the actual margin is located.

2.2. Evidence for Melting Glaciers

It is a well known fact that glaciers reached their Holocene maximum in 1850s (e.g. Kutuzov & Shahgedanova 2009). In most glaciers the terminal moraine representing the maximum extent can clearly be seen in the valleys. Glaciers respond relatively quickly to changes in climate (temperature, precipitation, humidity and cloudiness). If the snowfall (accumulation) sustaining a glacier declines, or if there is an increase in ice loss (ablation), these will result in recession of the glacier or an overall thinning of the ice mass (or both), generally within a few years.

All the glaciers have their own dynamics and it is common that sometimes their margins are retreating and some other times advancing. A few glaciers have periods of very rapid advancement called surges. These glaciers exhibit normal movement until suddenly they accelerate, then return to their previous state. During these surges, the glacier may reach velocities far greater than normal speed and the margin may advance. For example, the Medvejiy glacier in Tajikistan moves periodically downhill every 12-15 years.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

As temperatures have risen almost everywhere, the retreat of glaciers from mountain valleys is one of the most visible symbols of global warming. The overall picture of widespread recession is unequivocal and reflects the well- known record of global warming. Immerzeel et al. (2010) suggest that the situation in Asia is more complex. Although there seems to be a general loss of ice in this region, some regions with higher altitudes such as the Karakorum show an increase, and uncertainties about rates of change are considerable.

The Fedchenko glacier is a large glacier in the Pamirs in central Tajikistan at altitudes of 2900 - 6300 m. The glacier is long and narrow, currently 77 km long and covering over 649 km2. It is the longest glacier in the world outside of the polar regions. The maximum thickness of the glacier is 1000 meters, and the volume of the Fedchenko and its dozens of tributaries is estimated at 144 km3 (Aizen & Aizen 2010). The edge of this glacier has retreated 1 km in 70 years.

During the last 50 years, in Western Pamir, the glaciated area in the basin of Vandj river has reduced by more than 25-30%. In Eastern Pamir, the climate is colder and the glacial melting takes place less intensively than in all other mountainous areas of the region. Average retreat rate is 1-2 m/yr, whereas in some cases it can be 100 m/yr.

Figure 2-1. The retreat of glacier snouts in eastern Tien Shan, Kyrgyzstan, show an accelerating trend especially after 1940 (Kutuzov & Shahgedanova 2009).

The glaciers in the Tien Shan and Pamir are retreating and the rates of retreat vary between regions and time periods. The largest retreat rates have been observed in the northern Tien Shan where glaciated area has declined by 30-40% during the second half of the 20th century. The rates of deglaciation were slower further east and south, however, acceleration of glacier retreat has been noted in the eastern Pamir from 7.8% over 1978–1990 to 11.6% over 1990–2001 periods. Glaciers have lost 12.6% (0.33% /yr) of their 1965 area in the 1965- 2003 period. Small glaciers have diminished more than the average (

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Figure 2-1) (Kutuzov & Shahgedanova 2009).

It has been estimated that glaciers in the Syr Darya basin have lost 14% of their total volume over the last 60 years and that 15-40% of the volume will be lost in the coming 40 years (Siegfried et al., 2010). In this report, we will model more precisely the melting rate of glaciers in the Aral Sea Basin.

The modeling work of this project will reveal what will happen to the glaciers and runoff in the coming 40 years. The present records show that small ice caps especially from Tien Shan will disappear and also the glacial melt water will dramatically be reduced (Figure 2-2). It is important to realize that climate warming and melting of glaciers have already been going on especially after 1940. It is evident that there is an accelerating trend in the melting of glaciers.

Figure 2-2. Also geomorphological evidence, marginal moraines indicating bigger size of glacier, shows that glaciers have already receded substantially during the last 150 years (KAZ/KGZ border, Almaty; Google Earth).

2.3. Natural Hazards Accentuated by Climate Change

The recent ice surges, outbursts of glacier-dammed lakes and mudflows have caused large catastrophes in Central Asia. More regular fluctuations of glacier extents may also threaten infrastructures, constructions and property. Glaciers of different geometry, located in different climatic regimes, will react in different ways to a climate change.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

New proglacial lakes may be generated when maybe ice-cored moraines create dams in front of fast receding valley glaciers (Figure 2-3). Such lakes may have extensive amount of water and when the dam collapses, a catastrophic flood may occur. Flood protection interventions are needed in areas identified to be vulnerable to flood damages.

Figure 2-3. The Petrov proglacial lake in Kyrgyzstan has formed as a result of the melting of the glacier. Former 150 years old marginal moraines are damming the lake (Google Earth).

The snowline will raise 200 - 300 meters in average until 2050 (Figure 2-4). Hill slopes which always have been covered by snow will be exposed to erosion. Thawing of permafrost in the higher mountains will make the slopes instable and this will generate landslides and mudflows (Figure 2-5). Such calamities may be dangerous to the settlements and infrastructure downstream. The modeling results do not indicate increasing occurrence of spring-time floods in the future, but due to the increasing mudflows, the impacts of floods may probably be more catastrophic than earlier.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

Figure 2-4. The altitudes of climatic snowline (monthly average T = 0 C) and its rising as a result of climate change in the study area.

Figure 2-5. Rising permafrost altitude will expose new areas for erosion. Therefore, landslides and mudflows become common in the upper parts of the mountain ranges (Northern Pakistan; Google Earth).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part I - Executive Summary & Climate Change

2.4. References

Aizen, V. & E. Aizen (2010). Diagnosis of changes in alpine water storages and land surface degradation in Pamir mountains and Amu Dar’ya River basin. Annual, second year progress report. University of Idaho, USA. 20 pp. Immerzeel, W. W., L. P. H. van Beek, and M. F. P. Bierkens (2010). Climate change will affect the Asian water towers., Science, 328(5984), 1382- 1385, doi:10.1126/science.1183188. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20538947 (Accessed 20 July 2011). Kutuzov, S & M. Shahgedanova (2009): Glacier retreat and climatic variability in the eastern Terskey-Alatoo, inner Tien Shan between the middle of the 19th century and beginning of the 21st century. Global and Planetary Change, 69 2009, p. 59-70.

FCG Finnish Consulting Group Ltd. in association with FINAL REPORT • FutureWater Part 2 • Finnish Meteorological Institute Mountain Hydrology

Asian Development Bank

Water and Adaptation Interventions in Central and West Asia

TA 7532

June 2012

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

TABLE OF CONTENTS

1 Introduction 7

2 Review of Climate Change Impact Studies on Water Resources 9 2.1 Impact of climate change on water resources 9 2.2 Impact of climate change for snow and ice 9 2.3 Glacier changes in Central Asia 10 2.4 Modeling the hydrological response to climate change in high mountain regions 13

3 The Upstream Parts of the Amu and Syr Darya River Basins 18 3.1 Topography 18 3.2 Climate 18 3.3 Glaciers and snow 19 3.4 Land use and land cover 21 3.5 Water management and hydrology 22

4 Methodology 23 4.1 Data sources and preprocessing 23 4.2 Model concepts 25 4.2.1 Model structure 25 4.2.2 Cryospheric processes 27 4.2.3 Rain runoff 31 4.2.4 Baseflow 33 4.2.5 Routing 33 4.3 Calibration 34

5 Current Hydrological Regime 36 5.1 Calibration results 36 5.2 The contribution of glacial and snow melt to river runoff 41

6 Future Hydrological Regime 49 6.1 Climate change scenarios 49 6.2 Future glacier cover development 49 6.3 Future changes in generation and composition of runoff 54 6.3.1 Change in total runoff 54 6.3.2 Changes in snow melt 67 6.3.3 Changes in rain runoff 73 6.3.4 Changes in base flow 79 6.3.5 Changes in runoff composition 84 6.4 Simulated inflow to downstream areas 86

7 Conclusions 98

8 References 100

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

TABLE OF FIGURES

Figure 1-1: Amu Darya and Syr Darya river basins...... 7 Figure 2-1: Changes in glacier mass balance around the world [Kaser et al., 2006]. ... 10 Figure 2-2: The Tien Shan mountains in Kyrgyzstan. Subdivisions of Tien Shan: W, western; N, northern; C, central; I, inner. [Solomina et al., 2004] ...... 11 Figure 2-3: Recession of glacier terminus position for Fedchenko glacier 1933-2006... 12 Figure 2-4: Fedchenko and Murgab meteo station locations in the Pamir mountains. [Khromova et al., 2006] ...... 13 Figure 2-5: Changes in seasonal stream flow assuming 2 °C temperature rise. [Singh and Bengtsson, 2004] ...... 14 Figure 2-6: Anomalies in glacier ice volume ...... 15 Figure 2-7: Total simulated discharge partitioned into rain runoff, glacier runoff, snow runoff and baseflow. [Immerzeel et al., 2011] ...... 15 Figure 2-8: Simulated mean upstream discharge for present (2000-2007) and future climate for the A1B SRES scenario. [Immerzeel et al., 2010a] ...... 16 Figure 3-1: Upstream topography Amu Darya and Syr Darya river basins...... 18 Figure 3-2: Average air temperature 2001-2010 based on downscaled and interpolated station data ...... 19 Figure 3-3: Average annual precipitation 2001-2010 based on the PERSIANN/TRMM dataset ...... 19 Figure 3-4: Area covered by glaciers in Amu Darya and Syr Darya river basins...... 20 Figure 3-5: Areas with glaciers in the Amu Darya and Syr Darya river basins...... 20 Figure 3-6: Land cover in the upstream parts of the Amu Darya and Syr Darya river basins [Defourny et al., 2007]...... 21 Figure 3-7: Land cover type in the upstream parts of the Amu Darya and Syr Darya river basin...... 22 Figure 4-1: General project approach ...... 23 Figure 4-2: DEM with major rivers and HydroSheds catchment and subcatchment boundaries...... 24 Figure 4-3: Correlation between observed and predicted temperature...... 25 Figure 4-4: Spatial distribution of variance of observed and predicted temperature. .. 25 Figure 4-5: Boundaries of upstream Amu Darya and Syr Darya river basins...... 26 Figure 4-6: Schematic model structure AralMountain model...... 27 Figure 4-7: Grid cell area covered with clean-ice glaciers and debris-covered glaciers. 28 Figure 4-8: Schematic representation glacier processes in AralMountain model ...... 29 Figure 4-9: Schematic representation snow processes in AralMountain model ...... 30 Figure 4-10: Schematic representation of rainfall-runoff modeling in AralMountain model ...... 31 Figure 4-11: Runoff factor per grid cell ...... 33 Figure 4-12: Reservoirs used to calibrate the model ...... 34 Figure 5-1: Average annual precipitation 2001-2010 ...... 37 Figure 5-2: Average annual rainfall 2001-2010 ...... 37 Figure 5-3: Average annual snowfall 2001-2010 ...... 38 Figure 5-4: Observed and simulated monthly inflow Nurek reservoir 2001-2010 ...... 38 Figure 5-5: Observed and simulated monthly inflow Toktogul reservoir 2001-2010 .... 39 Figure 5-6: Observed and simulated monthly inflow Andijan reservoir 2001-2010 ...... 39 Figure 5-7: Average total annual runoff per grid cell 2001-2010 ...... 41 Figure 5-8: Average annual glacier melt runoff per grid cell 2001-2010 ...... 42 Figure 5-9: Average annual snow melt runoff per grid cell 2001-2010 ...... 42 Figure 5-10: Average annual rain runoff per grid cell 2001-2010 ...... 43 Figure 5-11: Average annual baseflow runoff per grid cell 2001-2010 ...... 43 Figure 5-12: Average discharge and relative contribution of glacier melt for major streams 2001-2010 ...... 44 Figure 5-13: Average discharge and relative contribution of snow melt for major streams 2001-2010 ...... 44

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-14: Average discharge and relative contribution of rain runoff for major streams 2001-2010 ...... 45 Figure 5-15: Simulated contribution of rain, snow melt, glacier melt and baseflow to inflow Nurek reservoir 2001-2010 ...... 46 Figure 5-16: Simulated contribution of rain, snow melt, glacier melt and baseflow to inflow Toktogul reservoir 2001-2010 ...... 46 Figure 5-17: Simulated contribution of rain, snow melt, glacier melt and baseflow to inflow Andijan reservoir 2001-2010 ...... 47 Figure 6-1: Fractional glacier cover 2010...... 50 Figure 6-2: Fractional glacier cover 2050 for CCCMA GCM...... 51 Figure 6-3: Fractional glacier cover 2050 for CCSM3 GCM...... 51 Figure 6-4: Fractional glacier cover 2050 for CNRM GCM...... 52 Figure 6-5: Fractional glacier cover 2050 for ECHAM GCM...... 52 Figure 6-6: Fractional glacier cover 2050 for MIROC GCM...... 53 Figure 6-7: Fractional glacier cover 2050, mean of 5 above GCM forced model runs. .. 53 Figure 6-8: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Mean of output after forcing model with 5 GCMs...... 55 Figure 6-9: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with MIROC GCM...... 56 Figure 6-10: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with ECHAM GCM...... 57 Figure 6-11: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CNRM GCM...... 58 Figure 6-12: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCSM3 GCM...... 59 Figure 6-13: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCCMA GCM...... 60 Figure 6-14: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Mean of output after forcing model with 5 GCMs...... 61 Figure 6-15: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Output for model forced with MIROC GCM...... 62 Figure 6-16: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Output for model forced with ECHAM GCM...... 63 Figure 6-17: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Output for model forced with CNRM GCM...... 64 Figure 6-18: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Output for model forced with CCSM3 GCM...... 65 Figure 6-19: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Output for model forced with CCCMA GCM...... 66 Figure 6-20: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Mean of output after forcing model with 5 GCMs...... 67 Figure 6-21: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Output for model forced with ECHAM GCM...... 68

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-22: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Output for model forced with CNRM GCM...... 69 Figure 6-23: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Output for model forced with CCSM3 GCM...... 70 Figure 6-24: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Output for model forced with CCCMA GCM...... 71 Figure 6-25: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041- 2050. Output for model forced with MIROC GCM...... 72 Figure 6-26: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Mean of output after forcing model with 5 GCMs...... 73 Figure 6-27: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with MIROC GCM...... 74 Figure 6-28: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with ECHAM GCM...... 75 Figure 6-29: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CNRM GCM...... 76 Figure 6-30: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCSM3 GCM...... 77 Figure 6-31: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCCMA GCM...... 78 Figure 6-32: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Mean of output after forcing model with 5 GCMs...... 79 Figure 6-33: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with MIROC GCM...... 80 Figure 6-34: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with ECHAM GCM...... 81 Figure 6-35: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CNRM GCM...... 82 Figure 6-36: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCSM3 GCM...... 83 Figure 6-37: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCCMA GCM...... 84 Figure 6-38: Projected average annual inflow Toktogul reservoir 2001-2050. Figure shows mean of model output when forced with 5 GCMs and the range of projections when forced by 5 GCMs...... 85 Figure 6-39: Projected average annual inflow Nurek reservoir 2001-2050. Figure shows mean of model output when forced with 5 GCMs and the range of projections when forced by 5 GCMs...... 85 Figure 6-40: Projected average annual inflow Toktogul reservoir 2001-2050 per runoff component. Figure shows mean of model output when forced with 5 GCMs...... 86

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-41: Projected average annual inflow Nurek reservoir 2001-2050 per runoff component. Figure shows mean of model output when forced with 5 GCMs...... 86 Figure 6-42: Upstream catchments used to calculate inflow from upstream to downstream areas. See Table 10 for corresponding catchment names...... 87 Figure 6-43: Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by five GCMs is shown...... 88 Figure 6-44: Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by five GCMs is shown...... 89 Figure 6-45: Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by five GCMs is shown...... 90 Figure 6-46:Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by five GCMs is shown...... 91 Figure 6-47: Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by five GCMs is shown...... 92 Figure 6-48: Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by 5 GCMs is shown...... 93 Figure 6-49: Average change in monthly inflow into downstream areas for Syr Darya basin in 2041-2050. Range for model forced with five GCMs is shown...... 96 Figure 6-50: Average change in monthly inflow into downstream areas for Amu Darya basin in 2041-2050. Range for model forced with five GCMs is shown...... 97

TABLE OF TABLES

Table 1: Land cover in the upstream parts of the Amu Darya and Syr Darya river basin...... 21 Table 2: Estimated runoff factors for different types of land use/land cover...... 32 Table 3: Calibrated parameters in the AralMountain model ...... 36 Table 4: Correlation parameters observed and simulated inflow 2001-2010...... 40 Table 5: Average annual inflow (observed and simulated) 2001-2010 ...... 40 Table 6: Simulated stream flow composition at the major reservoirs...... 47 Table 7: Simulated stream flow composition at the outlet of the upstream hydrological model ...... 48 Table 8: Global circulation models used for future climate projections...... 49 Table 9: Absolute and relative decrease in glacier extent for the five GCM forcings and mean decrease in glacier extent...... 54 Table 10: Catchment names Figure 6-42...... 87 Table 11: Simulated average annual inflow into downstream areas (Mm3) for reference period (2001-2010) and future situations (2021-2030, 2041-2050). Values are mean values for the five GCM outputs...... 94 Table 12: Changes in inflow into downstream areas for future projections (2021-2030, 2041-2050). Ranges indicate maximum and minimum availability change as projected by the five GCMs...... 95 Table 13: Projected average changes in water availability for downstream users Range for five GCMs is shown...... 95

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

ABREVIATIONS AND ACRONYMS

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

1 INTRODUCTION

The hydrological regimes of the two major rivers in the region, the Syr Darya and the Amu Darya, are complex and vulnerable to climate change. The upstream hydrology and the contribution of snow and ice to river runoff are crucial to understand the downstream impact of climate change. This report presents a hydrological modeling study assessing the hydrological properties of the Amu Darya and Syr Darya river basin (Figure 1-1: Amu Darya and Syr Darya river basins.Figure 1-1).

Figure 1-1: Amu Darya and Syr Darya river basins.

Using this model, the importance of snow and ice melt to river runoff can be quantified. With the use of climate change scenarios, the impact of climate change for these rivers in the next decades can be estimated. This report presents the state of the art of what is currently known and unknown as far as climate change impacts on the upstream water resources in the Amu Darya and Syr Darya river basins are concerned. Chapter 2 provides a review of available literature discussing the impact of climate change on water resources in the region. In chapter 3, the area is described in terms of topography, climate and factors which are relevant for the hydrological regime in the river basins. The used model, data sources and model calibration process are 7

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

described in chapter 4. Chapter 5 discusses the current hydrological regime for the river basins as characterized by the model. In chapter 6, the estimated impacts of climate change in the upstream basin for the next decades are described as well as the reliability of the modeling results.

The ultimate objective of this work is to develop national capacity in each of the participating countries (focused on the Kyrgyz Republic, Tajikistan, Kazakhstan, Turkmenistan and Uzbekistan) to use the models to prepare climate impact scenarios and develop adaptation strategies. This will then result in improved national strategies for climate change adaptation.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

2 REVIEW OF CLIMATE CHANGE IMPACT STUDIES ON WATER RESOURCES

2.1 Impact of climate change on water resources Observational records and climate projections provide abundant evidence that freshwater resources are vulnerable and have the potential to be strongly impacted by climate change, with wide-ranging consequences for human societies and ecosystems [Bates et al., 2008]. Observed warming over the last decades has been linked to changes in the hydrological cycle. Examples of changes include changing precipitation patterns, intensity and extremes, changes in snow and ice cover and changes in runoff.

According to the IPCC, climate model simulations for the 21st century are consistent in projecting increased precipitation for high latitudes and parts of the tropics, and decreased precipitation in some subtropical and lower mid- latitude regions [Bates et al., 2008]. Besides a climate change effect on water resources, an increasing population and increasing use of water will put increasing pressure on global water resources: pressures are increasing most rapidly in Africa and parts of southern Asia [Arnell, 1999]. The recent controversy about overestimated melting rates of the Himalayan glaciers, as communicated in the 4th assessment report of the IPCC, shows that the knowledge of high–altitude snow and ice and its response to climate forcing is still very incomplete [Siegfried et al., 2010].

2.2 Impact of climate change for snow and ice Snow and glacial melt are important hydrologic processes in the area surrounding the Tibetan plateau and adjacent mountain belts [Cruz et al., 2007]. Large amounts of water are stored as snow and ice in the mountains; the third-largest ice mass on earth, after the Antarctic and Greenland ice sheets. Changes in temperature and precipitation are expected to have significant effects on the snow and ice storages [Barnett et al., 2005]. The mountain ranges on earth are located in varying climatic zones, ranging from hot to cold and from wet to dry [Viviroli et al., 2011]. Thus every mountainous region on Earth will be impacted by climate change in a different way. The glacier contribution to water resources is minor in monsoon regimes, moderate in most mid-latitude basins and of high importance in very dry basins, like the Aral Sea basin and Tien Shan mountains [Kaser et al., 2010]. The glacier contribution to water resources differs for the catchments of the Amu Darya and Syr Darya. The percentage of glacierized area of the two catchments differs significantly (2% for the Amu Darya vs. 0.15% for the Syr Darya). This could result in different responses to climate change for the two rivers. In general, the Asian high mountains glaciers are show a negative mass balance as seen in Figure 2-1.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 2-1: Changes in glacier mass balance around the world [Kaser et al., 2006].

The impacts of climate change on hydrology and water management of the Central-Asian mountain environment have a complex geographic and climatologic context. It has been estimated that glaciers in the Syr Darya basin have lost 14% of their total volume over the last 60 years and that 15- 40% of the volume will be lost in the coming 40 years [Siegfried et al., 2010]. Over the short/medium term, the warming trend has already increased mean runoff and translated into increased water availability due to significant glacier wastage rates. This may compensate the increasing water demand downstream where climate gets drier and warmer. In the long term water availability may be reduced or the timing may change when the glaciers will be further reduced and the hydrological cycle will be accelerated.

2.3 Glacier changes in Central Asia Multiple studies were done to estimate glacier changes in the Tien Shan mountains. Figure 2-2 shows the Tien Shan area and its traditional subdivision based on its climatic and orographic properties.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 2-2: The Tien Shan mountains in Kyrgyzstan. Subdivisions of Tien Shan: W, western; N, northern; C, central; I, inner. [Solomina et al., 2004]

For the Akshiirak massif in the central Tien Shan and for the Ala Archa glacier basin in the northern Tien Shan glacier changes were estimated using topographic and remote sensing data [Aizen et al., 2007]. Glaciers in the Akshiirak massif in the central Tien Shan lost 12.5% of their surface area between 1943 and 2003. The glacier volume for this area was reduced by approximately 29% for the same period. Glacier retreat was accelerated in the second period of observation (after 1977). However, within this area glacier retreat was not uniform. A few glaciers advanced prior to 1977 while the majority retreated. In the Ala Archa basin in the northern Tien Shan, glacier area decreased by 15.7% between 1963 and 2003. Meteorological observations near the Ala Archa basin in the northern Tien Shan show no trends in observed annual precipitation or summer air temperature for 1913- 2003. However, air temperature increased during spring and autumn, extending the period of glacier ablation and thus favoring glacier retreat. Near the Akshiirak massif in the central Tien Shan no trend in annual precipitation is observed for 1930-2003. Summer air temperature increased for the same period, causing accelerated recession of the glaciers.

Another study conducted at the Akshiirak range in the central Tien Shan plateau estimated the glacier changes between 1943 and 2001 using aerial photographs and ASTER images combined with long-term glaciological and meteorological observations [Khromova et al., 2003]. A small retreat between 1943 and 1977 was observed, followed by accelerated reduction between 1977 and 2001. This was caused by a combination of increased summer and annual air temperatures and decreases in annual precipitation. There has been a strong decrease in the summer/winter precipitation ratio. This ratio change implies reduced accumulation, leading to a lower surface albedo and thus increased summer ablation. On the other hand, climatic warming leads also to warming in deeper layers of the glaciers. The glacier velocity can be

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

accelerated by this process, triggering glacier disintegration. Once a glacier disintegrates, it is more subject to incoming radiation and warm air, which accelerates ice degradation. The authors also state that the role of direct anthropogenic impacts has to be considered. A decrease in surface albedo, through dust deposition originating from gold mining activity, could accelerate summer ablation.

Glacier changes were also studied for the Sokoluk watershed in the northern Tien Shan mountains. The results of this study based on remote sensing and topographic data show a clear trend in glacier retreat between 1963 and 2000 [Niederer et al., 2007]. An overall loss of 28% is observed for this period and degradation accelerated since the 1980’s. This acceleration of degradation was strongest for small glaciers. A general increase in the minimum glacier elevation of 78 meters has been observed over the last three decades, corresponding to one third of the total retreat of minimum glacier elevation since the Little Ice Age maximum. The terminus recession of Fedchenko glacier, the largest glacier in the Pamir mountains is indicated in Figure 2-3. According to the researchers, the decrease in glacier area is related to an increase in annual and summer air temperatures combined with a decrease in summer precipitation in Central Asia.

Figure 2-3: Recession of glacier terminus position for Fedchenko glacier 1933-2006.

For the Saukdara and Zulumart Ranges, located in the high-mountain plateau of the eastern Pamir mountains, glacier retreat was quantified using historical surveys and recent satellite imagery [Khromova et al., 2006]. In the studied area, glaciers have retreated around 10% during the last 30 years. Glacier area has declined and recession of glacier termini is observed. Besides, an increase in debris-covered area and the appearance of new lakes is observed.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

The observed glacier retreat was correlated to two meteorological station records which have very different precipitation cycles despite of their proximity (about 145 km apart) (Figure 2-4). At the ‘Fedchenko’ station 87% of annual precipitation falls during winter while at ‘Murgab’ station winter precipitation accounts for only 27% of the total amount. The glaciers in the Zulumart and Saukdara ranges are located at the boundary of the winter- and summer- dominated precipitation regimes. Glaciers in the more humid part of the Pamir, near ‘Fedchenko’ station should be losing mass as increases in winter precipitation do not compensate for ablation rate increases. In both regions, with contrasting precipitation regimes, the trend in glacier retreat is similar. This may suggest that air temperature is a more important climatic driver for glacier mass balance than precipitation.

Figure 2-4: Fedchenko and Murgab meteo station locations in the Pamir mountains. [Khromova et al., 2006]

2.4 Modeling the hydrological response to climate change in high mountain regions Local scale modeling studies quantifying the hydrological response to climate change were done for multiple catchments in the Himalayan area. For example [Singh and Bengtsson, 2004] concluded for the Satluj river basin that the impact of climate change has more effect on the seasonal scale than on the annual water availability (Figure 2-5). Reduction in melt from the lower part was counteracted by the increase of melt from the upper part of the basin,

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

resulting in a decrease in the magnitude of change in annual melt runoff resulting in reduced summer runoff.

Figure 2-5: Changes in seasonal stream flow assuming 2 °C temperature rise. [Singh and Bengtsson, 2004]

Another study emphasized the regional differences of climatic change impact across the Himalayas [Rees and Collins, 2006]. Two hypothetical catchments were studied: a western catchment in the northeast of Pakistan and an eastern catchment in Nepal. The authors concluded that Himalayan rivers fed by large glaciers descending through considerable elevation range will respond in a broadly similar manner, except that summer snowfall in the east will suppress the rate of initial flow increase, delay peak discharge and postpone eventual disappearance of the ice. Impacts of declining glacier area on river flow will be greater in smaller and more glacierized basins in both the west and east. In the west, where precipitation is scarce, the impact will be important for considerable distances downstream.

[Immerzeel et al., 2011] studied the impacts of climate change for the Langtang catchment in Nepal. A high resolution combined cryospheric hydrological model was developed for the catchment simulating glacier evolution and all major hydrological processes. Parameters related to the glacier modeling are calibrated forcing the model with temperature and precipitation data from 1957-2002 aiming at reproducing the locations of glaciers and permanent snow in 2000, as observed by remote sensing methods. Hydrological parameters were calibrated using 2000-2006 precipitation and temperature data as input and comparing results to observed daily discharges for 2000-2006. The calibrated model is used to estimate glacier changes and discharge changes for 2000-2100 using downscaled data for five different GCM projections. The analysis shows a steady decline in glacier area due to increasing temperature and precipitation (Figure 2-6). The river flow is projected to increase significantly due to the increased ice melt and precipitation (Figure 2-7).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 2-6: Anomalies in glacier ice volume (error bars indicate 1 σ of 5 GCM’s) [Immerzeel et al., 2011]

Figure 2-7: Total simulated discharge partitioned into rain runoff, glacier runoff, snow runoff and baseflow. [Immerzeel et al., 2011]

A large scale hydrological modeling study for five large Asian basins originating in the Himalayan mountains gives estimates for the impact of climate change for these basins [Immerzeel et al., 2010b]. The general conclusion is that Asia’s water towers are threatened by climate change, but the effects of climate change on water availability and food security in Asia differ significantly among basins and cannot be generalized. The study shows that melt water is extremely important in the Indus basin and important for the Brahmaputra basin but plays just a modest role for the Ganges, Yangtze and Yellow rivers. Regional anomalies in glacier response to climate change are observed in the Himalaya region. Glacial expansion can be observed at high altitude, which may be caused by the presence of supra-glacial debris and possibly an increased orographic precipitation. For the Indus and Brahmaputra basin, the effects are likely to be severe because of the large population and high dependency on melt water. On the contrary, in the Yellow River basin climate change may have positive effects, as the dependence on melt water is low and a projected upstream increase in precipitation could enhance water availability. Figure 2-8 shows the predicted flow for the A1B SRES scenario compared to observed flow in 2000-2007.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 2-8: Simulated mean upstream discharge for present (2000-2007) and future climate for the A1B SRES scenario. [Immerzeel et al., 2010a]

Recently, a coupled climate-land ice-hydrological model study was conducted for the Syr Darya catchment [Siegfried et al., 2011]. This model was used to project runoff at the basin level and in the subcatchments of the Naryn and Syr Darya until 2050. These projections were made using the mean climatic change derived from five global circulation models (GCMs), providing a mean change in precipitation and temperature for the Syr Darya region. Preliminary results indicate that the contribution of runoff from glacier melt is small at the basin scale, but can be large for some individual sub-catchments of the Syr Darya. Increased glacier melt due to climate change may help to some extent in compensating for decreasing precipitation and growing water demand in some subcatchments. Considering the uncertainties in climate change projections for Central Asia and the low overall contribution of glacier melt to total runoff in the Syr Darya basin, it is unlikely that climate change induced glacier melt will help to sustain water availability in the short to medium term. Climatic warming has consequences on runoff seasonality due to earlier snow melt. Water stress in unregulated catchments will increase because less water will be available for irrigation in the summer months. Threats from natural hazards, mostly glacier lake outbursts are likely to increase as well.

According to a study conducted by the Eurasian Development Bank, changes in air temperature and precipitation in Central Asia has led to a regression in glacier area [Ibatulin et al., 2009]. For example, the total volume of glaciers in

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

the Pamir Mountains of Tadzhikistan has shrunk by about ten percent. The decline has been particularly dramatic in basins with large glaciers. In recent years, an increase in air temperature has led to more active surging glaciers. Deglaciation led to a decline in runoff despite the increased contribution to runoff from long-lasting ice reserves.

As a general conclusion, it can be stated that the impact of climate change cannot be generalized. The effects of changes in temperature and precipitation can either result in increased water availability or decreased availability of water. This can also differ on different timescales. Regional differences in expected changes in temperature and precipitation as well as the portion of snow and ice covered area in a catchment lead to high variation in effects of climatic change. For the Amu Darya and Syr Darya basins, knowledge on the effects of climate change on water availability is sparse. Besides, the contribution of glacier melt to total runoff differs significantly for both basins, leading to different responses to climate change. Modeling the hydrological and glacial response to climate change for the mountainous upstream parts of the Syr Darya and Amu Darya river basins will provide important insight in future water availability for the downstream regions of the river basins.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

3 THE UPSTREAM PARTS OF THE AMU AND SYR DARYA RIVER BASINS

3.1 Topography The upstream parts of the Amu Darya and Syr Darya river basins are characterized by a high variation in elevation. Figure 3-1 shows main rivers in the upstream parts of the basins and elevation. The elevation ranges from almost 7000 meters above sea level in the southeast to around 170 meters above sea level in the western plains. The Amu Darya and Syr Darya river basins originate in the Tien Shan and Pamir mountain ranges.

Figure 3-1: Upstream topography Amu Darya and Syr Darya river basins.

3.2 Climate The climate in the study area is highly continental with extreme variation in summer and winter temperatures [Ibatulin et al., 2009]. Figure 3-2 shows the average air temperature distribution for 2001-2010. The average air temperature for the region differs from around 20 °C in the western plains to around -20 °C in the high mountains. The climate in the lower areas can be described as arid subtropical. The climate in Central Asian high-mountain regions is typically dry and cold. Within the mountain ranges climate differs over short distances. For example, sharp contrasts in climate between the eastern and western Pamir mountains exist. In the west most precipitation falls during winter, while most precipitation in the east falls during summer [Khromova et al., 2006]. In general, the western mountains get more

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

precipitation compared to the eastern mountains, due to orographic effects, as can be seen in Figure 3-3.

Figure 3-2: Average air temperature 2001-2010 based on downscaled and interpolated station data

Figure 3-3: Average annual precipitation 2001-2010 based on the PERSIANN/TRMM dataset

The data on current and future climate used in this report is generated and analyzed by the Finnish Meteorological institute (FMI), partner in this project. For detailed information on analysis of the climate data see the FMI report on climate change in Central Asia.

3.3 Glaciers and snow Ice and snow melt are important contributors to runoff in the Himalayan area in general. In the upstream Amu Darya basin, 2% of the total area is covered with ice. For the upstream Syr Darya basin the glacial cover is 0.15%. The largest glacier present in the area is the Fedchenko glacier in the Pamir

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

mountains covering 77 km in length and covering 650 km2 [Aizen et al., 2009]. The glacier is the largest in the world outside the polar regions.

Figure 3-4: Area covered by glaciers in Amu Darya and Syr Darya river basins.

In general, a decline in glacial area is observed in the Himalayas, however regional increases in glacial cover are also observed. A glacier’s response to climate change does not only depend on climate, but also on topography and important regulators such as debris cover. West to East and North to South transitions to wetter and warmer climate exist in the Himalayas, resulting in regional differences in glacier response.

Figure 3-5: Areas with glaciers in the Amu Darya and Syr Darya river basins.

Melting water from glaciers in Tajikistan contributes between 10% and 20% to the runoff in large rivers, and in particularly hot and dry years their contribution may rise to 70% [Ibatulin et al., 2009].

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

3.4 Land use and land cover Land use in the Amu Darya and Syr Darya river basins is very variable. The inaccessible mountainous parts at high altitude are characterized by permanent snow and ice and bare areas. At midrange altitudes in the mountains open grasslands dominate while forests are present at low latitudes, especially in the Tien Shan mountains. The Pamir mountains have far less forest land cover. The river valleys downstream are dominated by irrigated croplands, irrigated with water from the Syr Darya and Amu Darya. The downstream areas which are not irrigated are mainly bare areas or areas with sparse vegetation.

Figure 3-6: Land cover in the upstream parts of the Amu Darya and Syr Darya river basins [Defourny et al., 2007].

Table 1: Land cover in the upstream parts of the Amu Darya and Syr Darya river basin. Land cover type Surface area Surface area (absolute, km2) (percentage of total) Bare areas / sparse vegetation 236704 29.6 Grassland 229415 28.7 Croplands 140665 17.6 Croplands / Vegetation 96848 12.1 Permanent snow and ice 78624 9.8 Water bodies 9317 1.2 Forest-Shrubland / Grassland 5011 0.6 Forest 2549 0.3 Artificial areas 125 0.02 21

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

0.63% 0.32% 1.17% 0.02% Bare areas / sparse vegetation 9.84% Grassland 29.62% 12.12% Croplands Croplands / Vegetation Permanent snow and ice 17.60% Water bodies

28.70% Forest-Shrubland / Grassland Forest Artificial areas

Figure 3-7: Land cover type in the upstream parts of the Amu Darya and Syr Darya river basin.

3.5 Water management and hydrology Water from the upstream parts of the Amu Darya and Syr Darya is collected in reservoirs to be used for the generation of hydropower and irrigation. The most upstream reservoirs in the Syr Darya basin are the Toktogul and Andijan reservoirs in Kyrgyzstan and the Charvak reservoir in Uzbekistan. The most upstream reservoirs in the Amu Darya basin are the Nurek reservoir in the Vaksh river in Tadzhikistan and the Gisarak reservoir in Uzbekistan. These reservoirs form the boundary between the upstream basin and the downstream basin. Downstream of these reservoirs the hydrological properties of the rivers don’t correspond to the natural system, which is present upstream of these reservoirs. Upstream of the reservoirs, the influence of human-induced mitigation measures is very small and can be neglected. The upstream system can be approached as a natural system without human interference and this will be studied in the second phase of the project.

Nurek and Toktogul reservoirs have very large catchments with highest observed peak inflows up to 2200 and 1800 m3/s respectively. Andijan and Charvak reservoirs have smaller catchments and peak inflows of 700 to 800 m3/s. Figure 4-12 shows the location of the Nurek, Andijan, Toktogul and Charvak reservoirs.

More information on water management and hydrology and the impact of climate change in the downstream parts of the river basins can be found in a separate report.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

4 METHODOLOGY

4.1 Data sources and preprocessing The study takes a unique approach towards the well-known data-scarcity problems in mountain hydrology, by using a mixture of existing data at various spatial and temporal level. Moreover, data from the public domain is combined with data from Remote Sensing as illustrated in Figure 4-1. The modeling framework is used for (i) integration all this information to enable our understanding of the current situation and (ii) projections of the future using downscaled regional climate information.

Figure 4-1: General project approach

The data used for the model originates from different sources. All data used is available in the online public domain. The topography for the upstream basins is derived from NGA and NASA’s Shuttle Radar Topography Mission (SRTM) dataset. The Digital Elevation Model (DEM) was interpolated from a 15 arc seconds DEM to a 1x1 km cell size grid, to be used for the model (Figure 4-2). Using USGS HydroSheds the catchment boundaries are obtained.1 HydroSHEDS is a mapping product based on SRTM elevation data, providing hydrographic information for regional and global-scale applications. Using HydroSheds, the boundaries of the modeled area are defined to coincide with the catchments boundaries (Figure 4-2). The locations of rivers in the HydroSheds dataset were used to correct the DEM hydrologically. The DEM surface is lowered a bit at the river’s locations to correct for irregularities in the DEM and assure a realistic drainage network.

Data on glacier surface are mainly extracted from the Global Land Ice Measurements from Space (GLIMS) monitoring program.2 Where GLIMS data was insufficient, ice cover data was obtained from the Digital Chart of the World dataset (DCW).3 Hydrological measurements for 2001-2010 used for calibration are obtained from the online Central Asian Water portal.4

1 http://hydrosheds.cr.usgs.gov/ 2 www.glims.org 3 www.naturalearthdata.com 4 www.cawater-info.net 23

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Data on land use and land cover are obtained from the Globcover regional land cover dataset.5 This dataset is made with ENVISAT imagery which resulted in a global land cover map according to UN land cover classification. Locations and outlines of lakes positioned in the upstream basins were obtained from the WWF Global lakes and wetlands database.6

0 200 400 800 Kilometers Figure 4-2: DEM with major rivers and HydroSheds catchment and subcatchment boundaries.

For 2001-2010 climate data was gathered from the national meteorological services of Kazakhstan, Uzbekistan, Turkmenistan, Kyrgyzstan and Tadzhikistan. Daily air temperature observations are available for multiple meteorological stations in the basins. Daily maps covering the whole basins were interpolated for air temperature by kriging; a spatial interpolation method, which gives the best linear unbiased predictors of unobserved values. Besides this method can also take into account external forcing, like altitude which is obtained from the DEM combined with a temperature lapse rate. The programming tool applied in this project is the R language and environment that is created for statistical computing and graphics. More precisely R package called “gstat” is applied here. R is available as Free Software under the terms of the Free Software Foundation's GNU General Public License in source code form and thus the use of this software is free of charge. The spatial variation of temperature can be explained to large extent by surface altitude (Figure 3-2).

The accuracy of the kriging interpolation was checked for each day by calculating the correlation between observed and predicted temperatures.

5 http://ionia1.esrin.esa.int/ 6 https://secure.worldwildlife.org/science/data/item1877.html 24

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 4-3: Correlation between observed and predicted temperature.

Figure 4-4: Spatial distribution of variance of observed and predicted temperature.

The variance of the predicted values was best near observation stations. The variance in the south-east corner of the kriging area was occasionally large indicating the inaccuracy of the data in the mountain area.

Precipitation data were interpolated bilinearly to 0.2° x 0.2° resolution from the PERSIANN 0.25° daily satellite and neural network based precipitation data (http://chrs.web.uci.edu/) [Hsu and Sorooshian, 2009].

More information on the processing of climate data and the generation of climate change scenarios can be found in the FMI report.

4.2 Model concepts

4.2.1 Model structure

The Aral Mountain model is a raster based highly detailed full distributed cryospheric- hydrological model for the upstream part of the Amu Darya and Syr Darya basins. The model is based on commonly accepted standards and the approach has been published in numerous scientific papers. Figure 4 1 shows the areas for the two basins. The model is created in PCRaster environmental modeling software [Karssenberg et al., 2001]. PCRaster is a spatio-temporal environmental modeling language developed at Utrecht University, the Netherlands. The model runs at 1 x 1 km spatial resolution with daily time steps and incorporates all major hydrological processes as well as cryospheric processes.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 4-5: Boundaries of upstream Amu Darya and Syr Darya river basins.

The actual runoff which is calculated for each grid cell consists of four contributing factors. These are: runoff originating from rain, runoff originating from snow melt, runoff originating from glacial melt, and baseflow, as visualized in Figure 4-6. With the daily air temperature and daily precipitation per grid cell as input the model evaluates how much precipitation falls and it is disaggregated into either snow or rain based on the air temperature distribution. The model evaluates the amount of snow and glacier melt or accumulation and which part of snow and glacier melt is directly transformed to runoff and which part refreezes. A part of the rainfall is directly transformed to runoff and a part infiltrates and adds to baseflow. Another part is lost to evapotranspiration. The runoff from all contributing components is routed through the system using the DEM.

The model is calibrated for the period 2001-2010 using air temperature and precipitation data for this period as input and comparing output to observed discharges for the same period. In the next step, climatic data based on climate scenarios is then used as input for the calibrated model, providing estimated river discharges for 2010-2050.

The next paragraphs provide detailed information on the modeling steps for the different cryospheric and hydrological processes in the AralMountain model.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 4-6: Schematic model structure AralMountain model.

4.2.2 Cryospheric processes

The initial glacier cover in the Tien Shan and Pamir mountains is obtained from the GLIMS dataset, replenished with DCW data. With this dataset the glacierized fraction of each grid cell is calculated. Since the model is set up for a 1 x 1 km resolution, the ice cover is described as a fraction varying from 0 (no glacial cover) to 1 (100% glacial cover). In this way, 1 x 1 km grid cells which are partly covered with ice can be simulated. A differentiation is made between clean ice glaciers and debris covered glaciers. This differentiation is made based on elevation and slope. Glaciers at lower altitude tend to have more debris cover because of the cumulative accumulation of debris from higher grounds and glacier parts with a small slope have more debris cover compared to steep-sloped parts of the glacier. In the model, some assumptions were made regarding the occurrence of glaciers and the differentiation between clean ice glaciers and debris covered glaciers. It is assumed that glaciers only occur at elevations above 3100 meters above sea level (Figure 3-4). Debris covered glaciers occur where elevation is between 3100 and 5500 meters and the slope is smaller than 13°. Using these assumptions, the fractions of clean ice and debris covered glaciers where calculated using a DEM with 90 meter spatial resolution to overcome inaccuracy resulting from using the 1 x 1 km DEM used for the modeling. The differentiation between clean ice glaciers and debris covered glaciers is then re-calculated to fractions of the 1 x 1 km grid cells used in the model (Figure 4-7). Summing the fractions of clean ice glacier and clean ice glacier will always result in a total fraction of one.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 4-7: Grid cell area covered with clean-ice glaciers and debris-covered glaciers.

Initial conditions for snow cover were obtained directly from the model. A model run was done simulating several years to develop a balanced snow cover. The snow cover at the end of this model run was used as initial snow cover for further model runs. In the model’s calculations, the amounts of ice and snow are described as millimeters water equivalent.

The modeling of processes involving glaciers is described in a schematic way in Figure 4-8. Melt from clean ice glaciers is defined as the air temperature (if above 0 °C) multiplied by the degree day factor for clean ice, multiplied by the clean ice fraction of the glacier cover and the cell fraction with glacier cover.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 4-8: Schematic representation glacier processes in AralMountain model

For the melt from debris covered glaciers the calculation is similar, although a different degree day factor for debris covered glaciers is specified. Melt rates for debris covered glaciers are lower, since incoming radiation and other heat flows are blocked by the (thick) debris cover.

Degree Day Factors The use of temperature index or degree day models is widespread in cryospheric models to estimate ice and snow melt. In these models an empirical relationship between melt and air temperature based on a frequently observed correlation between the two quantities is assumed [Hock, 2005]. Degree-day models are easier to set up compared to energy-balance models, and only require air temperature, which is mostly available and relatively easy to interpolate.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

The total glacier melt is then calculated by summing the two components from clean ice glacier melt and debris covered glacier melt. A part of glacial melt also refreezes in the glacier when it percolates the ice. A correction for this process is made by adjusting the glacial melt using a glacial runoff factor.

For each cell the model determines if precipitation falls as snow or rain by comparing the actual air temperature to a critical temperature. When air temperature is below or equal to the critical temperature, precipitation will fall as snow. When air temperature is above the critical temperature, precipitation will fall as rain. In the model a differentiation is made between the potential snow melt and the actual snow melt (Figure 4-9).

Figure 4-9: Schematic representation snow processes in AralMountain model

The potential snow melt is defined as the air temperature (if above 0 °C) multiplied by a degree day factor for snow multiplied by the cell fraction covered with snow. The actual snow melt however, is limited by the thickness of the snow pack. No more snow can be melted than the amount of snow which is available at the considered time step. The snow storage is then updated, to be used for the next time step. The updated snow storage is the

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

‘old’ snow storage with the fresh snow added and the actual snow melt subtracted.

The water resulting from snow melt will partially refreeze as it infiltrates the underlying snow pack. The maximum of water that can refreeze is defined by the water storage capacity of the snow pack which depends on the thickness of the snow pack present and the storage capacity of snow (e.g. the total millimeters of melt water that can refreeze per millimeter of snow). The actual amount of water that is stored in the snow pack is defined as the water stored in the snow pack during the previous time step summed by the actual snow melt. Snow melt will become actual snow melt when the amount of snow melt exceeds the water storage capacity of the snow pack.

4.2.3 Rain runoff

The modeling steps for rainfall in the AralMountain model are represented in Figure 4-10. As mentioned in paragraph 4.2.2, precipitation in the model will fall as rain when the air temperature is above a critical temperature.

Figure 4-10: Schematic representation of rainfall-runoff modeling in AralMountain model

Differences in land use and land cover (see Figure 3-6) lead to differences in generated runoff because factors like infiltration and evapotranspiration differ per type of land use and land cover. These differences are taken into account

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

by supplying each type of land use/land cover with a specific runoff factor. A runoff factor value 1 means all water comes to runoff. A runoff factor value 0 means all water is ‘lost’ due to evapotranspiration and infiltration and no water comes to runoff.

Table 2: Estimated runoff factors for different types of land use/land cover.

Type of Land use / land cover Runoff factor Post-flooding or irrigated croplands (or aquatic) 1.0 Rainfed croplands 0.8 Mosaic cropland (50-70%) / vegetation (grassland/shrubland/forest) (20- 0.6 50%) Mosaic cropland (50-70%) / grassland or shrubland (20-50%) 0.6 Mosaic vegetation (grassland/shrubland/forest) (50-70%) / cropland (20- 0.6 50%) Closed (>40%) broadleaved deciduous forest (>5m) 0.8 Closed (>40%) needleleaved evergreen forest (>5m) 0.8 Open (15-40%) needleleaved deciduous or evergreen forest (>5m) 0.8 Closed to open (>15%) mixed broadleaved and needleleaved forest (>5m) 0.8 Mosaic forest or shrubland (50-70%) / grassland (20-50%) 0.7 Mosaic grassland (50-70%) / forest or shrubland (20-50%) 0.7 Closed to open (>15%) broadleaved deciduous shrubland (<5m) 0.8 Closed to open (>15%) herbaceous vegetation (grassland, savannas or 0.6 lichens/mosses) Sparse (<15%) vegetation 0.3 Sparse (<15%) grassland 0.3 Sparse (<15%) shrubland 0.3 Closed to open (>15%) grassland or woody vegetation on regularly flooded 0.3 or waterlogged soil - Fresh, brackish or saline water Artificial surfaces and associated areas (Urban areas >50%) 0.2 Bare areas 0.1 Consolidated bare areas (hardpans, gravels, bare rock, stones, boulders) 0.1 Non-consolidated bare areas (sandy desert) 0.1 Salt hardpans 0.1 Water bodies 1.0 Permanent snow and ice 0.2

Besides land use and land cover, the actual runoff also depends on the hill slope. More runoff is generated on steep slopes because less water can infiltrate the soil and less water can evaporate as the water flows with a higher velocity on the surface.

The slope-dependent runoff factor is calculated as a runoff slope factor multiplied by the sinus of the slope. The land use specific runoff factor and the slope specific runoff factor are averaged resulting in a runoff factor per grid cell ( Figure 4-11).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 4-11: Runoff factor per grid cell

The total amount of rain that comes to runoff per grid cell is calculated as the actual rain in millimeters multiplied by the runoff factor and the fraction of the grid cell which is not covered by glacier or snow.

4.2.4 Baseflow

A module calculating baseflow is incorporated in the model. During periods with low runoff the streams are fed by processes such as sustained ground water flow and/or slow throughflow through the soil from earlier precipitation events. This is referred to as baseflow. The baseflow in the model consists of an initial baseflow summed with a certain recharge. The recharge is calculated for the rain fraction, glacial melt fraction, and the snow melt fraction with a recharge factor describing how much rain, glacial melt and snow melt is contributing to the baseflow.

The amount of baseflow is then calculated as an initial baseflow multiplied by the exponential function to the power of the negative of a recession coefficient. This function simulates the delayed effect of the baseflow. The lag time is determined by the recession coefficient parameter. The amount of recharge is multiplied by one minus the exponential function to the power of the negative of a recession coefficient.

4.2.5 Routing

In the model, the generated runoff is routed through the basin according to a flow direction map based on the DEM. For each sell the local drain direction is defined. The runoff generated per grid cell accumulates with runoff generated in downstream grid cells. Using a linear regression with a regression constant, the time needed for water to flow through the reservoir towards the outflow point is simulated.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

4.3 Calibration Model parameters need to be calibrated in order to simulate past river runoff as accurate as possible and after calibration it can be used for assessing future conditions. The model is calibrated for a ten year period (2001-2010). During this period, three inflow observations per month were reported for four upstream reservoirs: Nurek reservoir for the Amu Darya catchment, and Toktogul, Andijan and Charvak reservoirs for the Syr Darya catchment (Figure 4-12). With these observed inflow data, the AralMountain model is calibrated using the Parameter Estimation software package (PEST). With use of this software, the optimum parameter values were calculated to reproduce the observations for 2001-2010. PEST uses a Gauss-Marquardt-Levenberg algorithm for parameter estimation for nonlinear models. The algorithm runs in an iterative process changing parameter values while trying to constantly reduce the error of simulated values with respect to observed values.

For the calibration phase, the volume and area of glaciers is assumed to be static. No glacier fluctuations are taken into account. Note that glacier fluctuations are taken into account in the future projections with climate change scenarios. Initially, all four reservoirs were used in the calibration process. However, the model performed unsatisfactory for the Charvak reservoir. Simulated inflow was much lower than observed inflow. Why the model doesn’t perform well for this area remains unclear. Possible explanations could be underestimation of precipitation data or wrong observed discharge data. There might also be a high inflow from deep groundwater. The Charvak reservoir was not used for further model calibration.

Figure 4-12: Reservoirs used to calibrate the model

Twelve parameters are used in the model. Most of them are mentioned before in previous paragraphs. An important parameter which is not mentioned before is the correction factor for precipitation. The precipitation input data from PERSIANN and TRMM has a high uncertainty. During the calibration

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

processes a correction factor for precipitation is evaluated to improve model performance. Calibration results are presented in paragraph 5.1.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

5 CURRENT HYDROLOGICAL REGIME

5.1 Calibration results The model is calibrated using the Parameter Estimation Software (PEST) package. In total, twelve parameters are used in the model, of which eleven where determined with the PEST software. Calibrating the model produced the optimum parameter configuration to simulate the observed values for the calibration period 2001-2010 for three reservoirs (Figure 4-12). In table 3, the calibrated parameters are listed. For every parameter, a range of possible values was provided as boundary conditions in PEST. The estimated parameter values for optimum performance of the mode, as obtained by PEST, are listed in the last column. This parameter set is used in the modeled projections for 2011-2050.

Table 3: Calibrated parameters in the AralMountain model

Calibrated parameters Abbreviation used Parameter description Range Value in model Critical temperature for precipitation to fall as TCrit fixed 2 °C snow (°C) Degree Day Factor for clean ice glaciers 7.95 DDFG 5.0-10.0 (mm/°C/day) mm/°C/day Degree Day Factor for debris covered glaciers 3.98 DDFDG 4.0-8.0 (mm/°C/day) mm/°C/day 6.36 DDFS Degree Day Factor for snow (mm/°C/day) 4.0-7.0 mm/°C/day 0.19 SnowSC Storage capacity of the snow pack (mm/mm) 0.05-0.2 mm/mm GlacF Fraction of glacial melt to turn into runoff (-) 0.5-0.9 0.9 0.90- kx Recession coefficient used for routing (-) 0.967 0.99 BaseR Initial daily baseflow (mm) 0.3-0.4 0.3 0.01- RF Recharge factor (-) 0.0988 0.99 0.001- rc Recession constant for baseflow (-) 0.001 0.2 PrecF Correction factor for precipitation (-) 1.0-3.0 1.23 Factor used to estimate slope dependency of RunoffSlope 0.3-0.9 0.9 runoff factor (-)

For the input data, the precipitation data is provided by the PERSIANN dataset. However, the precipitation values for 2006 and 2007 seemed to be significantly overestimated, and the model simulated discharges much higher than observed discharges. For 2006 and 2007, precipitation input data was replaced with TRMM data, leading to better simulation results. Applying the correction factor for precipitation to the input data yields updated input maps for precipitation ( Figure 5-1). Applying the critical temperature to the precipitation input data yields input maps for snowfall and rainfall ( Figure 5-2 and Figure 5-3). These maps show averaged values for the calibration period (2001-2010).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-1: Average annual precipitation 2001-2010

Figure 5-2: Average annual rainfall 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-3: Average annual snowfall 2001-2010

The comparison of the simulated values to the observed values during the calibration period for three reservoirs is shown in Figure 5-4 to Figure 5-6.

Figure 5-4: Observed and simulated monthly inflow Nurek reservoir 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-5: Observed and simulated monthly inflow Toktogul reservoir 2001-2010

Figure 5-6: Observed and simulated monthly inflow Andijan reservoir 2001-2010

The quality of the model performance can be expressed with different criterions for the correlation between the observed values and the simulated values (Table 4).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Table 4: Correlation parameters observed and simulated inflow 2001-2010

Nash-Sutcliffe Pearson’s model correlation Bias efficiency coefficient coefficient Nurek 0.73 -20.6% 0.46 Toktogul 0.82 16.8% 0.56 Andijan 0.74 -17.2% 0.48

For Pearson’s product-moment correlation coefficient, a perfect correlation, where simulated values equal to observed values, would yield a coefficient’s value equal to +1 or -1. When no correlation exists, the coefficient’s value is equal to 0.

The simulation is unbiased when the bias is equal to 0. A positive value for bias indicates overestimation in the simulation, whereas a negative value for bias indicates underestimation in the simulation. The value for bias is given as percentages.

The Nash-Sutcliffe model efficiency coefficient is used to assess the predictive power of hydrological models. The coefficient’s value can range from -∞ to +1. A value of 1 indicates a perfect correlation, where simulated values are equal to observed values. Coefficient value 0 indicates the simulated values are as accurate as the mean of the observed data. For a coefficient value smaller than 0, the mean of the observed data is a better predictor than the model. Another way to provide insight in model performance is to compare observed and simulated average annual inflow Table 5).

Table 5: Average annual inflow (observed and simulated) 2001-2010

Average annual Average annual inflow (observed) inflow (simulated) Mm3 Mm3 Nurek 8726 6932 Toktogul 5857 6840 Andijan 1751 1450

There are a number of uncertainties in the model, causing impediments in predicting the actual runoff. First, there are uncertainties in the temperature and precipitation input data. Temperature data is based on point measurements, interpolated to spatially cover the entire area. This interpolation comes with uncertainties. Besides, precipitation data is based on PERSIANN and TRMM remotely sensed data, which also have uncertainties. The processes described in the model are simplified with respect to what happens in the real world. Assumptions are made in different stages of the modeling process adding to uncertainty.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Overall we conclude that calibration results are very satisfactory considering the complexity and heterogeneity of mountain hydrology, the large areas that have been modeled and the fact that we have used primarily public domain datasets. Moreover, it has been proven that relative model accuracy (= difference between current situation and scenario) is always much higher than relative model accuracy (difference between model output and observations) [Droogers et al., 2008]. This is exactly why this modeling activity is undertaken: compare current climate with future climates.

5.2 The contribution of glacial and snow melt to river runoff With the calibrated model the distribution of runoff generation can be simulated and specified for glacier melt, snow melt, rainfall and baseflow. Figure 5-7 shows the simulated averaged total runoff generation per year per grid cell. The absolute contributions of glacier melt, snow melt rain and baseflow to runoff are shown in Figure 5-8 to Figure 5-11. The values are averaged for the calibration period (2001-2010).

Figure 5-7: Average total annual runoff per grid cell 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-8: Average annual glacier melt runoff per grid cell 2001-2010

Figure 5-9: Average annual snow melt runoff per grid cell 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-10: Average annual rain runoff per grid cell 2001-2010

Figure 5-11: Average annual baseflow runoff per grid cell 2001-2010

For each grid cell, the averaged routed flow during the calibration period can be calculated and the contribution of glacier melt, snow melt, and rain to the total runoff can be specified. Figure 5-12 to Figure 5-14 show the relative contributions of these contributors to the average discharge for major streams in the basins for 2001-2010.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-12: Average discharge and relative contribution of glacier melt for major streams 2001-2010

Figure 5-13: Average discharge and relative contribution of snow melt for major streams 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-14: Average discharge and relative contribution of rain runoff for major streams 2001-2010

The contribution glacial and snow melt to river runoff differs regionally. In rivers in glaciated areas, the contribution of glacial melt is highest. The contribution of glacial melt is lower downstream, where more snow melt and rain runoff contributes to river discharge. The glacier melt is more important for the Amu Darya than for the Syr Darya as can be clearly seen in Figure 5-12. For the total Amu Darya river basin, the contribution of glaciers is much higher compared to the Syr Darya river basin. The contribution of the different contributors can be simulated per time step for each cell in the model. Results of this simulation are displayed for the reservoirs used in the calibration (Figure 5-15 to Figure 5-17).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-15: Simulated contribution of rain, snow melt, glacier melt and baseflow to inflow Nurek reservoir 2001-2010

Figure 5-16: Simulated contribution of rain, snow melt, glacier melt and baseflow to inflow Toktogul reservoir 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-17: Simulated contribution of rain, snow melt, glacier melt and baseflow to inflow Andijan reservoir 2001-2010

For these reservoirs, glacier melt contribution is highest for Nurek reservoir in the Amu Darya basin. Many glaciers are located upstream of this reservoir, contributing to the inflow in the reservoir. For Toktogul reservoir and Andijan reservoir, the contribution of glacial melt is much lower. Snow melt and rain runoff are more important contributors for the inflow into these reservoirs (Table 6). When we consider the stream flow contribution at the outlet of the entire upstream model ( Figure 3-1) the difference between Amu Darya and Syr Darya is striking. For Amu Darya 38% of annual stream flow is glacial melt, whereas snow contributed 26.9%. For the Syr Darya the glacial contribution is only 10.7% of the total river flow, whereas snow contribution to melt is estimated at 35.2%. When we consider the entire river basins of the Amu and Syr Darya then the relative melt water contribution will be even lower.

Table 6: Simulated stream flow composition at the major reservoirs.

Average Contribution Contribution Contribution Contribution inflow of glacier of snow of rain of baseflow (m3/s) melt melt Nurek 527 59.3% 22.4% 6.5% 11.8% Toktogul 520 11.7% 42.2% 27.6% 18.5% Andijan 110 14.8% 43.4% 22.3% 19.6% Charvak 109 22.0% 40.7% 21.1% 16.2%

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Table 7: Simulated stream flow composition at the outlet of the upstream hydrological model Average Contribution Contribution Contribution Contribution outflow of glacier of snow of rain of baseflow (m3/s) melt melt Amu 3549 38.0% 26.9% 16.5% 18.6% Darya Syr 1495 10.7% 35.2% 31.1% 23.1% Darya

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

6 FUTURE HYDROLOGICAL REGIME

6.1 Climate change scenarios Five different Global Circulation Models (GCM’s) are used to estimate the future changes in climate for the Aral Sea Basin between 2010 and 2050 ( Table 8). From the various existing emission scenarios this study uses the A1B GHG emission scenario. This scenario is chosen because it is widely used and recommended by the IPCC. The A1B scenario is considered as the most likely scenario, because it assumes a world of rapid economic growth, a global population that peaks in mid-century and rapid introduction of new and more efficient technologies. The A1B scenario can be seen as an intermediate between the B1 (with the smallest GHG emissions) and A2 (with the largest GHG emissions) scenario. Moreover, A1B is becoming a de-facto standard in assessment studies.

Detailed information on the used climate models and preprocessing is available in the separate report by the Finnish Meteorological Institute (FMI). We used a delta change approach to make the future climate projection at the subcatchment scale. Temperature and precipitation series for the reference period 2001-2010 are repeated four times (2011-2050). The daily projected change (increase or decrease) in temperature and precipitation is added/subtracted to the corresponding day in the reference period.

Table 8: Global circulation models used for future climate projections.

GCM name Developing institute Abbreviation used in report Canadian Centre for Climate CGCM3(T63) CCCMA Modeling and Analysis, Canada Community Climate Community Earth System System Model 3.0 CCSM3 Model (NCAR-CCSM3) Centre National de Recherches CNRM-CM3 CNRM Météorologiques, France Max Planck Institute for ECHAM5/MPI-OM ECHAM Meteorology, Germany Model for Atmosphere and Ocean Interdisciplinary Research Institute, University MIROC Research On Climate of Tokyo (MIROC3.2 HIRES)

6.2 Future glacier cover development As observed in the current hydrological regime, the water supplied by glacier melt has major importance in the river basins (Paragraph 5.2). Therefore it is essential to have a good estimate of future changes in glacier cover to provide good estimates of future inflow into the downstream river basins. For this

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

purpose we developed a glacier development model, which was incorporated in the AralMountain model for 2010-2050. This glacier model was calibrated for 2001-2010 using the observed glacier mass balance in the Pamir mountains [Khromova et al., 2006]. The glacier model estimates glacier surface area and volume change with a monthly time step and for twenty separate glacier size classes between 2010 and 2050. For each model run the results for the 20 different glacier size classes are integrated to derive an overall relative glacier area depletion curve. The depletion and hypsometric curve are then used to estimate the threshold elevation below which glaciers do not persist. This threshold elevation in combination with the elevation distribution within a 1km grid cell is finally used to derive an updated glacier fraction per grid cell. Figure 6-1 to Figure 6-6 show the projected fractional glacier cover in 2050 when the model is forced with the five different GCMs. Figure 6-7 shows the mean of the five outputs.

Figure 6-1: Fractional glacier cover 2010.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-2: Fractional glacier cover 2050 for CCCMA GCM.

Figure 6-3: Fractional glacier cover 2050 for CCSM3 GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-4: Fractional glacier cover 2050 for CNRM GCM.

Figure 6-5: Fractional glacier cover 2050 for ECHAM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-6: Fractional glacier cover 2050 for MIROC GCM.

Figure 6-7: Fractional glacier cover 2050, mean of 5 above GCM forced model runs.

The glacier cover decreases significantly more or less independently of the selected GCM. The decrease until 2050 varies from 46.4% for the CNRM GCM to 59.5% for the MIROC GCM (Table 9). It is very likely that glacier extent in

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

2050 will be just about half of the current extent. Especially small glaciers at low altitudes are affected by the changing climate. This has major impact on the hydrological regime of the Amu Darya and Syr Darya. Contribution of glacier melt to the stream flow changes significantly as described in paragraph 6.3.

Table 9: Absolute and relative decrease in glacier extent for the five GCM forcings and mean decrease in glacier extent.

Glacier extent Glacier Decrease in 2010 (km2) extent 2050 glacier extent (km2) (%) CCCMA 18128.8 9117.6 49.7 CCSM3 18128.8 7395.1 59.2 CNRM 18128.8 9716.1 46.4 ECHAM 18128.8 9017.0 50.3 MIROC 18128.8 7344.0 59.5 Mean 18128.8 8518.0 53.0

6.3 Future changes in generation and composition of runoff

6.3.1 Change in total runoff

As demonstrated in paragraph 5.2 the AralMountain model determines the contribution of glacier melt, snow melt, rain and baseflow to the generated runoff per grid cell. In this paragraph we show how the average annual runoff generation per component changes for 2021-2030 and 2041-2050 with respect to the reference situation (2001-2010). Figure 6-8 shows the mean change in average annual runoff generation per grid cell for 2021-2030 and 2041-2050 with respect to the reference period (2001-2010). Changes are given in millimeters per year. Figure 6-9 to Figure 6-13 show the changes in runoff generation when the model is forced by the five different GCMs, also in millimeters per year.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-8: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Mean of output after forcing model with 5 GCMs.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-9: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with MIROC GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-10: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with ECHAM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-11: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CNRM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-12: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCSM3 GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-13: Change in average annual runoff generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCCMA GCM.

As is obvious from the model output, the generated runoff decreases in general for each of the model outputs forced by the different GCMs. The strongest runoff generation decrease is projected for the CCSM3 scenario. Decreasing runoff is spatially widespread, including the high mountain environment and the lower areas. Increases in runoff are observed for the areas covered with glaciers, since they generate more melt water due to higher temperatures. However, the glaciated area becomes progressively smaller during the period until 2050 reducing the overall generation of glacier melt runoff in the two basins.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-14 shows the mean change in annual average runoff generation from glacier melt for each grid cell for 2021-2030 and 2041-2050 with respect to the reference period (2001- 2010). Changes are given in millimeters per year. Figure 6-15 to Figure 6-19 show the changes in runoff generation from glacier melt when the model is forced by the five different GCMs, also in millimeters per year.

Figure 6-14: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Mean of output after forcing model with 5 GCMs.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-15: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with MIROC GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-16: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with ECHAM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-17: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CNRM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-18: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCSM3 GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-19: Change in average annual runoff generation from glacier melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCCMA GCM.

Runoff generation from glacier melt decreases for the grid cells where the glacierized fraction decreases. The persisting glacierized areas on the other hand, show increasing runoff generation. This is due to increasing temperatures and precipitation, which lead to more melt. Especially the glaciers which persist at relatively low altitude show strong increases in melt water generation. The glacierized area generates more glacier melt runoff in

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

the future, but the glacierized are decreases, leading to an overall reduction in glacier melt water.

6.3.2 Changes in snow melt

Figure 6-20 shows the mean change in average annual runoff generation from snow melt per grid cell for 2021-2030 and 2041-2050 with respect to the reference period (2001-2010). Changes are given in millimeters per year. Figure 6-21 to Figure 6-25 show the changes in runoff generation from snow melt when the model is forced by the five different GCMs, also in millimeters per year.

Figure 6-20: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Mean of output after forcing model with 5 GCMs.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-21: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with ECHAM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-22: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CNRM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-23: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCSM3 GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-24: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCCMA GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-25: Change in average annual runoff generation from snow melt per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with MIROC GCM.

For changes in runoff originating from snow melt a strong correlation with altitude is observed. Due to increases in temperature, the snow line is shifting to higher altitude. Therefore, strongest reductions in the generation of snow melt are occurring in the lower mountain ranges, where less precipitation will fall as snow, and more precipitation will fall as rain. Areas where the glacierized fraction has reduced show an increase in the snow melt in the model, since areas with 100% fractional glacier cover are only contributing to glacier melt and not to snow melt. Snow falling on a glacier is considered as ‘glacier’ in the model.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

6.3.3 Changes in rain runoff

Figure 6-26 shows the mean change in average annual runoff generation from rain per grid cell for 2021-2030 and 2041-2050 with respect to the reference period (2001-2010). Changes are given in millimeters per year. Figure 6-27 to Figure 6-31 show the changes in runoff generation from rain when the model is forced by the five different GCMs, also in millimeters per year.

Figure 6-26: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Mean of output after forcing model with 5 GCMs.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-27: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with MIROC GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-28: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with ECHAM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-29: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CNRM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-30: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCSM3 GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-31: Change in average annual runoff generation from rain per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCCMA GCM.

For most GCMs (except for CCSM3), the overall runoff originating from rainfall increases. There are two reasons for this development. First, the amount of precipitation to fall as rain increases while the amount of precipitation to fall as snow decreases, because the snowline is getting higher, as a result of higher temperatures. Second, the amount of precipitation is increasing for the MIROC, ECHAM, CNRM and CCCMA GCMs. The CCSM3 GCM projects decreasing precipitation, leading to decreased runoff from rain. The grid cells with 100% glaciated fraction don’t generate runoff from rain in the model for the reference situation as well as for the future situations.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

6.3.4 Changes in base flow

Figure 6-32 shows the mean change in average annual base flow generation per grid cell for 2021-2030 and 2041-2050 with respect to the reference period (2001-2010). Changes are given in millimeters per year. Figure 6-33 to Figure 6-37 show the changes in base flow generation when the model is forced by the five different GCMs, also in millimeters per year.

Figure 6-32: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Mean of output after forcing model with 5 GCMs.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-33: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with MIROC GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-34: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with ECHAM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-35: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CNRM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-36: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCSM3 GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-37: Change in average annual base flow generation per grid cell (mm change with respect to reference period) for 2021-2030 and 2041-2050. Output for model forced with CCCMA GCM.

The base flow in the model is related to the total runoff generated by the three other components (glacier melt, snow melt and rain). Therefore, the spatial patterns of changes in base flow generation resemble the spatial patterns of the changes in total runoff generation ( Figure 6-8 to Figure 6-13).

6.3.5 Changes in runoff composition

The relative contribution of melt water to the total runoff differs strongly in space as demonstrated in paragraph 5.2 ( Figure 5-12 to

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 5-14 and Figure 5-4 to Figure 5-6). For the Amu Darya, glacial and snow melt are much more important compared to the Syr Darya. Besides, glacier and snow melt have higher relative contributions in more upstream areas compared to more downstream areas. This also means that the impact of climate change on runoff generation is more severe for the Amu Darya than for the Syr Darya. This is easily illustrated when comparing the development of inflow into Toktogul reservoir in the Syr Darya basin (Figure 6-38) and Nurek reservoir in the Amu Darya basin (Figure 6-39). See Figure 4-12 for the locations of the mentioned reservoirs.

Figure 6-38: Projected average annual inflow Toktogul reservoir 2001-2050. Figure shows mean of model output when forced with 5 GCMs and the range of projections when forced by 5 GCMs.

Figure 6-39: Projected average annual inflow Nurek reservoir 2001-2050. Figure shows mean of model output when forced with 5 GCMs and the range of projections when forced by 5 GCMs.

When looking at the trends for the changes in inflow into the reservoirs a decrease for both reservoirs is observed. However, this decrease is much stronger for Nurek reservoir in the Amu Darya and also more accelerated towards the end of the time interval. By looking at the contributions of the different runoff components (glacier melt, snow melt, rain and baseflow) this can be explained (Figure 6-40, Figure 6-41).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-40: Projected average annual inflow Toktogul reservoir 2001-2050 per runoff component. Figure shows mean of model output when forced with 5 GCMs.

Figure 6-41: Projected average annual inflow Nurek reservoir 2001-2050 per runoff component. Figure shows mean of model output when forced with 5 GCMs.

The glacier melt plays a much larger role for the total runoff for the inflow into Nurek reservoir than for the inflow into Toktogul reservoir. As the contribution of glacier melt decreases dramatically until 2050, the inflow into Nurek reservoir also decreases very significantly. This effect is much smaller for Toktogul reservoir, although the contribution of glacier melt becomes zero, meaning all glaciers in this catchment have disappeared.

6.4 Simulated inflow to downstream areas The runoff generated in the upstream parts of the basins is routed to the downstream parts of the basins. Figure 6-42 shows the division in upstream areas for which the runoff generation is modeled. The runoff flowing into the downstream areas from these areas is modeled and used as input in different locations for the downstream model. The simulated inflow from the mountainous regions into the reservoirs and downstream agricultural areas changes significantly. Especially the inflow from glacier melt decreases significantly in the future. Figure 6-43 to Figure 6-48 show the simulated projected change in inflow at the boundary between the upstream AralMountain model and the downstream WEAP-model for the reference situation (2001-2010) and future situations (2021-2030, 2041-2050). The graphs show the range of projections when the model is forced by the five mentioned GCMs. The AralMountain model output at these locations is used as input for the downstream WEAP-model.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-42: Upstream catchments used to calculate inflow from upstream to downstream areas. See Table 10 for corresponding catchment names.

Table 10: Catchment names Figure 6-42.

Catchment Catchment name no. 1 Toktogul reservoir 2 Andijan reservoir 3 Nurek reservoir 4 Tupalangskoe reservoir 5 Akhangaran reservoir 6 Zaamin reservoir 7 Gissarak reservoir 8 Pachkamar reservoir 9 Kulyab catchment 10 Kurgantube catchment 11 Dushanbe catchment 12 Surkhandarya upstream catchment 13 Surkhandarya downstream catchment 14 Karukum kanal catchment 15 Kashkadarya upstream catchment 16 Kashkadarya downstream catchment 17 Zeravshan Valley catchment 18 Lebap upstream catchment 19 Fergana Valley catchment 20 Syrdaryo, Tashkent, Jizakh catchment 21 South Kazakhstan upstream catchment 22 Charvak reservoir

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

23 Papan reservoir

Figure 6-43: Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by five GCMs is shown.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-44: Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by five GCMs is shown.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-45: Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by five GCMs is shown.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-46:Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by five GCMs is shown.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-47: Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by five GCMs is shown.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-48: Hydrographs for inflow from upstream into downstream areas showing future projections (2021-2030, 2041-2050) compared to reference period (2001-2010). The range of outputs for the model forced by 5 GCMs is shown.

The modeled runoff projections show that the impact of climate change for river runoff is highly variable in the region. Most of the hydrographs show a decrease in runoff. For the highly glacierized Kurgantube catchment, the model projects increased runoff generation for 2021-2030, which turns into a decrease by 2041-2050. As this area is highly glacierized, a temperature increase will lead to initial increases in runoff originating from glacier melt, but ultimately cause a decrease in runoff generation because the glacierized surface decreases.

From the reported changes in water inflow into downstream users (in fact the runoff at the boundary locations between the upstream model and the downstream model) it also becomes clear that the Amu Darya basin is more vulnerable to climate change than the Syr Darya basin, because of the decreasing glacierized area. A typical example is the decreases in runoff for Dushanbe catchment, Gissarak reservoir, Nurek reservoir, Tupalangskoe reservoir, Kashkhadarya upstream and downstream catchments, the Zeravshan Valley catchment, Surkhandarya upstream catchment and Kulyab catchment in Figure 6-43 to Figure 6-48. All of these catchments have glaciers in their catchments. The runoff peaks in early summer for some of them remain quite high, but last much shorter, especially in 2041-2050. The climate change impact is less strong for catchments where runoff is dominated by rain and snow, although also in these areas the climate change impact for runoff generation is significant. See for example the Fergana Valley catchment, Toktogul reservoir, Papan reservoir, and Andijan reservoir. The impact of climate change is less significant for the lower, more downstream parts like

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Surkhandarya downstream catchment, and Karakum kanal catchment. The absolute changes in runoff at the boundary locations is listed in Table 11. Table 12 lists the relative changes in runoff at the boundary locations including the range of projected changes when the model is forced with the five mentioned GCMs.

Table 11: Simulated average annual inflow into downstream areas (Mm3) for reference period (2001-2010) and future situations (2021-2030, 2041-2050). Values are mean values for the five GCM outputs.

Toktogul Andijan Nurek Tupalangsko Akhangaran Zaamin reservoir reservoir reservoir e reservoir reservoir reservoir 2001-2010 16424 3481 16772 2107 362 119 2021-2030 15144 3235 14099 849 333 105 2041-2050 13576 2717 9430 456 328 104

Surkhandary Gissarak Pachkamar Kulyab Kurgantube Dushanbe a upstream reservoir reservoir catchment catchment catchment catchment 2001-2010 353 552 10960 39500 3203 2322 2021-2030 233 480 10402 42245 1614 1142 2041-2050 180 477 6087 38593 1088 737

Surkhandarya Karakum Kashkadarya Kashkadarya Zeravshan Lebap downstream kanal upstream downstream Valley upstream catchment catchment catchment catchment catchment catchment 2001-2010 8718 3309 754 743 6855 4335 2021-2030 6882 2389 458 66 4772 2049 2041-2050 6619 2298 408 51 2918 1924

Syrdaryo, Fergana Tashkent, Charvak Papan Valley Jizakh reservoir reservoir catchment catchment 2001-2010 10212 4947 3464 674 2021-2030 8645 3245 2698 655 2041-2050 7492 3031 2489 562

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Table 12: Changes in inflow into downstream areas for future projections (2021-2030, 2041- 2050). Ranges indicate maximum and minimum availability change as projected by the five GCMs.

Toktogul Andijan Nurek Tupalangsko Akhangaran Zaamin reservoir reservoir reservoir e reservoir reservoir reservoir -13.8 – - -54.1 – - 2021-2030 -6.8 – -9.1% -5.6 – -8.9% -5.8 – -12.0% -8.8 – -17.4% 18.8% 65.6% -16.4 – - -19.1 – - -40.4 – - -76.5 – - 2041-2050 -4.8 – -16.7% -6.3 – -23.4% 18.8% 25.2% 49.1% 81.1%

Surkhandary Gissarak Pachkamar Kulyab Kurgantube Dushanbe a upstream reservoir reservoir catchment catchment catchment catchment -28.5 – - -45.2 – - -45.6 – - 2021-2030 40.6% -10 – -18.5% -3.3 – -8.1% 12.4 – 2.9% 54.4% 56.6% -44.2 – - -38.6 – - -63.8 – - -65.9 – - 2041-2050 56.4% -7.6 – -24.5% 51.3% 3.0 – -9.4% 69.2% 72.1%

Surkhandarya Karukum Kashkadarya Kashkadarya Zeravshan Lebap downstream kanal upstream downstream Valley upstream

catchment catchment catchment catchment catchment catchment 1 -25.6 – - -34.9 – - -90.8 – - -25.1 – - -51.4 – - 2021-2030 -19.5 – -22.6% 31.0% 44.9% 91.5% 36.7% 54.9% -26.0 – - -41.4 – - -92.7 – - -52.2 – - 2041-2050 -20.3 – -28.1% 37.0% 53.3% 94.1% 64.0% -53 – -59.8%

Syrdaryo, Fergana Tashkent, Charvak Papan Valley Jizakh reservoir reservoir catchment catchment -13.7 – - -32.8 – - 2021-2030 17.9% 37.5% -19 – -26.4% -2.2 – -3.6% -35.7 – - -12.3 – - 2041-2050 -23 – -32% 44.7% -25.1 – -34% 21.4%

Summarizing for the two river basins a substantial decrease in water availability for downstream users has been calculated (Table 13). For the Syr Darya basin it is projected that average inflow into downstream areas decreases by 15% for 2021-2030 and 25% for 2041-2050, with respect to the reference situation (2001-2010). For the Amu Darya expected decreases are 13% (2021-2030) and 31% (2041-2050). The results in the table show the range of values for the model when forced with the five mentioned GCMs.

Moreover, changes in water availability are not distributed homogeneously throughout the year. The decrease is strongest in the late summer / start of autumn, when water shortage is already at its peak. In August, September and October water availability decreases by 45% for the Syr Darya and 42%for the Amu Darya in 2050.

Table 13: Projected average changes in water availability for downstream users Range for five GCMs is shown.

Syr Darya Amu Darya

2021-2030 -13% – -17% -11% – -15%

2041-2050 -22% – -28% -26% – -35%

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Note that Syr Darya runoff decreases linearly in time, while the decrease in runoff for the Amu Darya decreases slower until 2021-2030, but is likely to be accelerated towards 2041-2050.

As seen in Figure 6-43 to Figure 6-48, the decrease in runoff is not uniform throughout the year. The decrease in runoff is strongest in the late summer / start of autumn for both river basins, as is visualized in Figure 6-49 and Figure 6-50. In August, September and October strongest runoff decreases are observed with runoff decreasing up to 47-52% for the Syr Darya in September and 36-47%for the Amu Darya in August, September and October in 2041- 2050. The lowest decreases in runoff are observed in spring (March, April, May) for both river basins. Note that the difference between the minimum and maximum prediction based on the five GCMs is higher for the Amu Darya basin.

Figure 6-49: Average change in monthly inflow into downstream areas for Syr Darya basin in 2041-2050. Range for model forced with five GCMs is shown.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Figure 6-50: Average change in monthly inflow into downstream areas for Amu Darya basin in 2041-2050. Range for model forced with five GCMs is shown.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

7 CONCLUSIONS

Climate change might have a big impact on water resources in the Central Asia region, but a rigorous analysis of these impacts is so far missing as the hydrological regimes of the two major rivers in the region (Syr Darya and the Amu Darya) are complex. Only recently, by the advent of advanced computer modeling combined with remotely sensed data and scientific progress, options occur to better understand processes and impact related to climate change.

The work described in this report is a first contribution to a larger study initiated by the Asian Development Bank to better understand and to explore adaptation strategies in the Aral Sea Basin. The ultimate objective of this project is to develop national capacity in each of the participating countries (Kyrgyz Republic, Tajikistan, Kazakhstan, Turkmenistan, Uzbekistan) to use the models, tools, data and results to prepare climate impact scenarios and develop adaptation strategies. This will then result in improved national strategies for climate change adaptation.

This report describes the analysis focusing on the upstream parts of the Amu Darya and Syr Darya river basins. Based on local and public domain datasets and hydro-meteorological observations a spatially distributed glacio- hydrological model has been developed for the upstream parts of the two rivers. This is one of the first models that covers the entire upstream parts of these basins and includes all processes related to glacier and snow melt, rain runoff and base flow.

Some of the key messages resulting from the study are:

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

• Strongest decreases in stream flow are expected for the late summer months (August, September, October), where inflow into downstream areas decreases around 45% for both river basins.

Output of the upstream model as presented in this report will be used as input for a downstream water allocation model to investigate the impact of climate change for water resources in the entire basins until 2050 and to explore possible adaptation measures.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

8 REFERENCES

Aizen, V. B., V. a. Kuzmichenok, A. B. Surazakov, and E. M. Aizen (2007), Glacier changes in the Tien Shan as determined from topographic and remotely sensed data, Global and Planetary Change, 56(3-4), 328-340, doi:10.1016/j.gloplacha.2006.07.016. [online] Available from: http://linkinghub.elsevier.com/retrieve/pii/S0921818106001913 (Accessed 14 September 2011)

Aizen, V. B., P. a. Mayewski, E. M. Aizen, D. R. Joswiak, A. B. Surazakov, S. Kaspari, B. Grigholm, M. Krachler, M. Handley, and A. Finaev (2009), Stable-isotope and trace element time series from Fedchenko glacier (Pamirs) snow/firn cores, Journal of Glaciology, 55(190), 275-291, doi:10.3189/002214309788608787. [online] Available from: http://openurl.ingenta.com/content/xref?genre=article&issn=0022- 1430&volume=55&issue=190&spage=275

Arnell, N. W. (1999), Climate change and global water resources, Global Environmental Change, 9, S31- S49.

Barnett, T. P., J. C. Adam, and D. P. Lettenmaier (2005), Potential impacts of a warming climate on water availability in snow-dominated regions., Nature, 438(7066), 303-309, doi:10.1038/nature04141. [online] Available from: http://www.ncbi.nlm.nih.gov/pubmed/16292301 (Accessed 10 June 2011)

Bates, B. C., Z. W. Kundzewicz, S. Wu, and J. P. Palutikof (2008), Climate Change and Water. Technical Paper of the Intergovernmental Panel on Climate Change, IPCC, Geneva.

Cruz, R. V., H. Harasawa, M. Lal, S. Wu, Y. Anokhin, B. Punsalmaa, Y. Honda, M. Jafari, C. Li, and N. Huu Ninh (2007), Coordinating Lead Authors : Lead Authors : Contributing Authors : Review Editors :, in Asia. Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, pp. 469-506, Cambridge University Press, Cambridge.

Defourny, P., C. Vancutsem, P. Bicheron, C. Brockmann, F. Nino, L. Schouten, and M. Leroy (2007), GLOBCOVER : A 300 m global land cover product for 2005 using ENVISAT MERIS time series, in Proceedings of ISPRS Commission VII Mid-Term Symposium: Remote Sensing: from Pixels to Processes, Enschede (NL) 8-11 May 2006, pp. 8-11.

Droogers, P., A. V. Loon, and W. W. Immerzeel (2008), Quantifying the impact of model inaccuracy in climate change impact assessment studies using an agro-hydrological model, Hydrology and Earth System Sciences, (2004), 669-678.

Hock, R. (2005), Glacier melt: a review of processes and their modeling, Progress in Physical Geography, 29(3), 362-391, doi:10.1191/0309133305pp453ra. [online] Available from: http://ppg.sagepub.com/cgi/doi/10.1191/0309133305pp453ra (Accessed 30 July 2011)

Hsu, K.-lin, and S. Sorooshian (2009), Satellite-Based Precipitation Measurement Using PERSIANN System, in Hydrological Modeling and the Water Cycle, pp. 27-48.

Ibatulin, S., A. Yasinsky, and A. Mironenkov (2009), The Impact of Climate Change on Water Resources in Central Asia. Sector report no. 6.

Immerzeel, W. W., L. P. . Van Beek, and M. F. P. Bierkens (2010a), Climate change will affect the Asian water towers., Science, 328(5984), 1382-5, doi:10.1126/science.1183188.

Immerzeel, W. W., L. P. H. van Beek, M. Konz, A. B. Shrestha, and M. F. P. Bierkens (2011), Hydrological response to climate change in a glacierized catchment in the Himalayas, Climatic Change, doi:10.1007/s10584-011-0143-4. [online] Available from: http://www.springerlink.com/index/10.1007/s10584-011-0143-4 (Accessed 2 August 2011)

100

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology

Immerzeel, W. W., L. P. H. van Beek, and M. F. P. Bierkens (2010b), Climate change will affect the Asian water towers., Science, 328(5984), 1382-1385, doi:10.1126/science.1183188. [online] Available from: http://www.ncbi.nlm.nih.gov/pubmed/20538947 (Accessed 20 July 2011)

Karssenberg, D., P. Burrough, R. Sluiter, and K. de Jong (2001), The PCRASTER software and course materials for teaching numerical modeling in the environmental sciences, Transactions in GIS, 5, 99- 110, doi:10.1111/1467-9671.00070. [online] Available from: http://doi.wiley.com/10.1111/1467- 9671.00070

Kaser, G., J. G. Cogley, M. B. Dyurgerov, M. F. Meier, and a. Ohmura (2006), Mass balance of glaciers and ice caps: Consensus estimates for 1961–2004, Geophysical Research Letters, 33(19), 1-5, doi:10.1029/2006GL027511. [online] Available from: http://www.agu.org/pubs/crossref/2006/2006GL027511.shtml (Accessed 14 July 2011)

Kaser, G., M. Grosshauser, and B. Marzeion (2010), Contribution potential of glaciers to water availability in different climate regimes., Proceedings of the National Academy of Sciences of the United States of America, 107(47), 20223-7, doi:10.1073/pnas.1008162107. [online] Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2996705&tool=pmcentrez&rendertype=abstr act (Accessed 21 June 2011)

Khromova, T. E., M. B. Dyurgerov, and R. G. Barry (2003), Late-twentieth century changes in glacier extent in the Ak-shirak Range, Central Asia, determined from historical data and ASTER imagery, Geophysical Research Letters, 30(16), 3-7, doi:10.1029/2003GL017233. [online] Available from: http://www.agu.org/pubs/crossref/2003/2003GL017233.shtml (Accessed 30 July 2011)

Khromova, T. E., G. B. Osipova, D. G. Tsvetkov, M. B. Dyurgerov, and R. G. Barry (2006), Changes in glacier extent in the eastern Pamir, Central Asia, determined from historical data and ASTER imagery, Remote Sensing of Environment, 102(1-2), 24-32, doi:10.1016/j.rse.2006.01.019. [online] Available from: http://linkinghub.elsevier.com/retrieve/pii/S0034425706000484 (Accessed 30 July 2011)

Niederer, P., V. Bilenko, N. Ershova, H. Hurni, S. Yerokhin, and D. Maselli (2007), Tracing glacier wastage in the Northern Tien Shan (Kyrgyzstan/Central Asia) over the last 40 years, Climatic Change, 86(1-2), 227-234, doi:10.1007/s10584-007-9288-6. [online] Available from: http://www.springerlink.com/index/10.1007/s10584-007-9288-6 (Accessed 30 September 2011)

Rees, H. G., and D. N. Collins (2006), Regional differences in response of flow in glacier-fed Himalayan rivers to climatic warming, Hydrological Processes, 20, 2157- 2169, doi:10.1002/hyp.

Siegfried, T., T. Bernauer, R. Guiennet, S. Sellars, A. W. Robertson, J. Mankin, P. Bauer-Gottwein, and A. Yakovlev (2011), Will climate change exacerbate water stress in Central Asia?, Climatic Change, doi:10.1007/s10584-011-0253-z. [online] Available from: http://www.springerlink.com/index/10.1007/s10584-011-0253-z (Accessed 14 October 2011)

Siegfried, T., S. Sellars, T. Bernauer, R. Guiennet, J. Mankin, A. W. Robertson, and P. B. Gottwein (2010), Coping With International Water Conflict in Central Asia : Implications of Climate Change and Melting Ice in the Syr Darya Catchment, International Studies, 1-36.

Singh, P., and L. Bengtsson (2004), Hydrological sensitivity of a large Himalayan basin to climate change, Hydrological Processes, 18, 2363-2385, doi:10.1002/hyp.1468. [online] Available from: http://doi.wiley.com/10.1002/hyp.1468 (Accessed 5 August 2011)

Solomina, O., R. Barry, and M. Bodnya (2004), The retreat of Tien Shan glaciers (Kyrgyzstan ) since the Little Ice Age estimated from aerial photographs , lichenometric and historical data, Geografiska Annaler, Series A: Physical Geography, 86(2), 205-215, doi:10.1111/j.0435-3676.2004.00225.x. [online] Available from: http://doi.wiley.com/10.1111/j.0435-3676.2004.00225.x

Viviroli, D. et al. (2011), Climate change and mountain water resources: overview and recommendations for research, management and policy, Hydrology and Earth System Sciences, 15(2), 471-504,

101

TA 7532: Water and Adaptation Interventions in Central and West Asia Part II - Mountain Hydrology doi:10.5194/hess-15-471-2011. [online] Available from: http://www.hydrol-earth-syst- sci.net/15/471/2011/ (Accessed 19 July 2011)

102

FCG Finnish Consulting Group Ltd. in association with FINAL REPORT • FutureWater Part 3 • Finnish Meteorological Institute Impact and Adaptation

Asian Development Bank

Water and Adaptation Interventions in Central and West Asia

TA 7532

June 2012

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

TABLE OF CONTENTS

1 Introduction 5

2 Methodology 8 2.1 Water Evaluation And Planning system 8 2.2 Model concepts 8 2.3 Model setup and data sources 9

3 Reference Situation 2001-2010 18 3.1 Population 18 3.2 Agriculture 19 3.3 Reservoirs 21 3.4 Calibration 22 3.5 Water availability and unmet demand 34 3.5.1 Demand and unmet demand Syr Darya basin 34 3.5.2 Demand and unmet demand Amu Darya basin 37 3.5.3 Aral Sea inflow 40

4 Projections 2030 and 2050 41 4.1 Changes in temperature and precipitation 42 4.2 Changes in demand and unmet demand Syr Darya basin 48 4.2.1 Mean of five projections 48 4.2.2 CCSM3 49 4.2.3 CNRM 49 4.2.4 MIROC 50 4.2.5 ECHAM 50 4.2.6 CCCMA 51 4.3 Changes in demand and unmet demand Amu Darya basin 51 4.3.1 Mean of five projections 52 4.3.2 CCSM3 52 4.3.3 CNRM 53 4.3.4 MIROC 53 4.3.5 ECHAM 54 4.3.6 CCCMA 54 4.4 Changes in Aral Sea inflow 55 4.4.1 Mean of five projections 55 4.4.2 CCSM3 56 4.4.3 CNRM 56 4.4.4 MIROC 57 4.4.5 ECHAM 57 4.4.6 CCCMA 57 4.5 Future Aral Sea development 58

5 Adaptation Strategies 62 5.1 Water marginal cost curves 62 5.1.1 Cost curves 62 5.1.2 Measures to close the supply-demand gap 63 5.1.3 Costs of these options 63 5.2 Effectiveness of adaptation measures 64

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

5.3 Water marginal cost curve 66

6 Conclusions 69

7 References 71

TABLE OF FIGURES

Figure 1-1: Amu Darya and Syr Darya river basins...... 5 Figure 2-1: Two-way modeling approach...... 9 Figure 2-2: Division in upstream and downstream basin ...... 10 Figure 2-3: Schematic representation Amu Darya river basin in ARAL-WEAP model. ... 11 Figure 2-4: Schematic representation Syr Darya river basin in ARAL-WEAP model...... 12 Figure 2-5: Subcatchments used in upstream model for input in downstream WEAP- model. See Table 1 for names of the subcatchments, which are also used in Figure 2-3 an Figure 2-4...... 13 Figure 2-6: Geographical visualization of ARAL-WEAP model. AralMountain upstream model area is indicated with blue color. Demand sites are indicated with red dots, catchments are indicated with green dots...... 15 Figure 2-7: Downstream catchments used in WEAP model. See Table 3 for names of the subcatchments, which are also used in Figure 2-3 and Figure 2-4...... 17 Figure 3-1: Population per demand site in 2000. Source: Central Asian Water Info database ...... 18 Figure 3-2: Agricultural area per demand site in 2000. Source: Central Asian Water Info database...... 19 Figure 3-3: Example showing different values for the crop coefficient during the different growing stages of the crop. Source: FAO ...... 20 Figure 3-4: Effective storage capacity for major reservoirs in the Syr Darya river basin. Source: Central Asian Water Info database...... 22 Figure 3-5: Effective storage capacity for major reservoirs in the Amu Darya river basin. Source: Central Asian Water Info database...... 22 Figure 3-6: Observed inflow, outflow and volume data for Toktogul reservoir 2001- 2010. Source: Central Asian Waterinfo Database...... 23 Figure 3-7: Observed inflow, outflow and volume data for Andijan reservoir 2001-2010. Source: Central Asian Waterinfo Database...... 24 Figure 3-8: Observed inflow, outflow and volume data for Charvak reservoir 2001-2010. Source: Central Asian Waterinfo Database...... 24 Figure 3-9: Observed inflow, outflow and volume data for Kayrakkum reservoir 2001- 2010. Source: Central Asian Waterinfo Database...... 25 Figure 3-10: Observed inflow, outflow and volume data for Chardara reservoir 2001- 2010. Source: Central Asian Waterinfo Database...... 25 Figure 3-11: Observed inflow, outflow and volume data for Nurek reservoir 2001-2010. Source: Central Asian Waterinfo Database...... 26 Figure 3-12: Observed inflow, outflow and volume data for Tyuyamuyun reservoir 2001- 2010. Source: Central Asian Waterinfo Database...... 26 Figure 3-13: Observed and simulated in- and outflows Nurek reservoir reference period...... 27 Figure 3-14: Observed and simulated in- and outflows Tyuyamuyun reservoir reference period...... 27 Figure 3-15: Average annual water balance Syr Darya ARAL-WEAP reference period (2001-2010). P catchment is rainfall in the cathment; ETact catchment is the actual evapotranspiration from the natrual landscape and the rainfed crops...... 30 Figure 3-16: Average annual water balance Amu Darya ARAL-WEAP reference period (2001-2010)...... 33 Figure 3-17: Monthly average agricultural demand Syr Darya basin 2001-2010...... 35

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 3-18: Monthly average unmet agricultural demand Syr Darya basin 2001-2010...... 35 Figure 3-19: Monthly average domestic demand Syr Darya basin 2001-2010...... 36 Figure 3-20: Monthly average unmet domestic demand Syr Darya basin 2001-2010. . 36 Figure 3-21: Monthly average agricultural demand Amu Darya basin 2001-2010...... 38 Figure 3-22: Monthly average unmet agricultural demand Amu Darya basin 2001-2010...... 38 Figure 3-23: Monthly average domestic demand Amu Darya basin 2001-2010...... 39 Figure 3-24: Monthly average unmet domestic demand Amu Darya basin 2001-2010. 39 Figure 3-25: Average annual outflow into the Aral Sea 2001-2010...... 40 Figure 4-1: Projected changes in temperature and precipitation for the demand sites in ARAL-WEAP model. Projections show the average and range of 5 GCMs. Projections are for 2021-2030 and 2041-2050...... 48 Figure 4-2: Changes in annual demand and unmet demand Syr Darya basin...... 49 Figure 4-3: Changes in annual demand and unmet demand Syr Darya basin CCSM3 GCM...... 49 Figure 4-4: Changes in annual demand and unmet demand Syr Darya basin CNRM GCM...... 50 Figure 4-5: Changes in annual demand and unmet demand Syr Darya basin MIROC GCM...... 50 Figure 4-6: Changes in annual demand and unmet demand Syr Darya basin ECHAM GCM...... 51 Figure 4-7: Changes in annual demand and unmet demand Syr Darya basin CCCMA GCM...... 51 Figure 4-8: Changes in annual demand and unmet demand Amu Darya basin...... 52 Figure 4-9: Changes in annual demand and unmet demand Amu Darya basin CCSM3 GCM...... 53 Figure 4-10: Changes in annual demand and unmet demand Amu Darya basin CNRM GCM...... 53 Figure 4-11: Changes in annual demand and unmet demand Amu Darya basin MIROC GCM...... 54 Figure 4-12: Changes in annual demand and unmet demand Amu Darya basin ECHAM GCM...... 54 Figure 4-13: Changes in annual demand and unmet demand Amu Darya basin CCCMA GCM...... 55 Figure 4-14: Average annual outflow into Aral Sea. Mean output of model forced by five GCMs...... 55 Figure 4-15: Average annual outflow into Aral Sea for model forced with CCSM3 GCM. 56 Figure 4-16: Average annual outflow into Aral Sea for model forced with CNRM GCM. 56 Figure 4-17: Average annual outflow into Aral Sea for model forced with MIROC GCM. 57 Figure 4-18: Average annual outflow into Aral Sea for model forced with ECHAM GCM. 57 Figure 4-19: Average annual outflow into Aral Sea for model forced with CCCMA GCM. 58 Figure 4-20: Aral Sea shoreline development 1960-2008. Source: www.unimaps.com, based on NASA imagery...... 60 Figure 5-1: Schematic representation of the cost curve...... 62 Figure 5-2: Water marginal cost curve Amu and Syr Darya basin. Note: Cost-axis has been cut off at US$ 0.30. Cost for decreasing domestic demand is 2.00 $/m3...... 66 Figure 5-3: Cumulative water marginal cost curve Amu Darya and Syr Darya basin. .. 68

TABLE OF TABLES

Table 1: Subcatchments upstream parts of the basin (Figure 2-5)...... 14 Table 2: Division of provinces over WEAP demand sites...... 16 Table 3: Catchments used in the downstream model (Figure 2-7) ...... 17 Table 4: Domestic water allocation [Aldaya et al., 2010] ...... 18 Table 5: Average crop coefficients for crops in the study area. Based on [Allen et al., 1998] ...... 20

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Table 6: Crop calendar. Numbers represent fraction of month that the crop is growing in the field. Data based on [Allen et al., 1998] and FAO website...... 21 Table 7: Observed and simulated average annual inflow into the Aral Sea for the reference period 2001-2010 ...... 28 Table 8: Calibrated parameters ARAL-WEAP model Syr Darya basin ...... 28 Table 9: Calibrated parameters ARAL-WEAP model Amu Darya basin ...... 28 Table 10: Global Circulation Models used to force ARAL-WEAP model...... 41 Table 11: Estimating Aral Sea size in 2041-2050 based on historic inflow observations and simulated future inflows using PCRaster and ARAL-WEAP...... 61 Table 12: Annual water demand and unmet demand for Amu and Syr Darya river basins for 2041-2050 for the MIROC GCM climate projection. REF reflects scenario without adaptation measures; A to H differences compared to REF (in Mm3)...... 65 Table 13: Total unmet demand and unmet demand caused by climate change...... 67

ABREVIATIONS AND ACRONYMS

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

1 INTRODUCTION

Figure 1-1: Amu Darya and Syr Darya river basins.

In the Central Asian region, water related issues have been prominent since the break-up of the USSR. Major trans-boundary river basins and management agreements in the Aral Sea Basin include: 1) 1992 Aral Sea Basin Water Allocation and Management; 2) 1993 Aral Sea Basin Program and 3) 1994 Nukus Declaration on Aral Sea Basin Management; 4) 1998 Framework Agreement on Rational Water and Energy Use; 5) 1999 Revised Mandate of the International Fund for Saving the Aral Sea; and 6) 2003- Revised Aral Sea Basin Program, Phase-2. One of the regional environmental initiatives in Central Asia Countries (CACs) is the International Fund for Saving the Aral Sea (IFAS), established in 1994. IFAS, together with its two commissions, the Interstate Commission on Sustainable Development (ICSD) and the Interstate Commission for Water Coordination (ICWC), is charged with mobilizing funds to implement interstate activities on water resources and land 5

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

degradation and other social-economic issues, with financing joint scientific and technical projects, and with participating in international programs and projects directed at the Aral Sea crisis.

UNDP and the Global Water Partnership (2004) have drafted the Integrated Water Resource Management (IWRM) plan that proposed an integrated approach to water management, in which the river basin would be managed holistically, with the participation of water user stakeholders and ensuring environmental sustainability. The regional policy can only be implemented if there is a strong scientific background for investment plans and commitment to international agreements and conventions. It is therefore essential to gain knowledge on the future availability and demand of water resources under of climate change.

The following description of the situation in Central Asia is based on the article ‘Water and Energy Conflict in Central Asia’ by Tobias Siegfried published in Earth Institute’s ‘state of the planet’ blog.

What once was a basin-wide management approach during the Soviet times has become an uncoordinated management situation with conflicting interests for the upstream countries (Kyrgyzstan, Tajikistan and Afghanistan) and the downstream countries (Uzbekistan, Turkmenistan and Kazakhstan). The hydraulic infrastructure is distributed over various independent countries. As a result, the water resources system is not managed collectively and cooperatively. A mixture of regional, national, and interstate institutions now handles allocation decisions, which used to be centrally administered during Soviet times. As a result, water and energy allocation among the various sectors and users is not efficient. Future water resources development in northern Afghanistan will further add fuel to the water and energy conflict in the region.

In short, the upstream / downstream conflict consists of opposed demand patterns for energy and water resources, in space and in time. Kyrgyzstan and Tajikistan need to release water from a number of large reservoirs during the cold months to generate hydropower for heating. There, hydropower provides the cheapest source of energy with generating costs as low as 0.1 cent/kWh. The winter releases frequently cause flooding in the downstream areas. At the same time and in order to have enough hydropower generating capacity during the cold months, these upstream states spend the warmer summer months saving water in those reservoirs.

That is precisely when the downstream countries have the most pressing need for irrigation water where the degradation of agricultural soils and insufficient flows for ecosystems are issues of growing concern. In the region, cotton is an important cash crop, and, at the same time, wheat is considered essential in order to meet national food security goals. Especially for Uzbekistan, considerations of self-sufficiency have become more important in recent times where food grain prices have increased considerably on the world market.

The original idea in Soviet times was to operate the hydro-infrastructure in irrigation mode. The water resources of Central Asia were managed with the aim to maximize crop production. Part of the hydropower produced during irrigation water-releases in spring and summer was conveniently utilized in the

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

downstream for driving lift irrigation and vertical drainage pumps along the 30,000 km of irrigation channels. In return, the upstream areas received energy supplies in the form of gas and coal to cover winter energy demands.

Future climate change poses additional challenges. The discharge in both the Syr Darya and the Amu Darya rivers is driven mainly by snow and glacial melt. The impact of a warming climate on these key hydrological processes is not sufficiently understood and no mitigation and adaptation strategies are in place. Whereas changes in precipitation levels are hard to predict for the future, there is a solid consensus that average global temperatures are rising. As a result, more precipitation will fall as rain in the upstream and the ice volume in the Tien Shan and Pamir mountain ranges will likely shrink. The former will impact the seasonality of the runoff whereas the latter will at least temporarily increase average annual flows. Furthermore, changes in sediment loads may pose additional problems. At this point in time, the impacts are not sufficiently quantified and adaptation and mitigation strategies not in place.

The unfavorable developments in this geopolitically important and fragile region call for urgent attention of the international community. Interdisciplinary research can critically inform decision making in the region for better risk management and the design of mitigation and adaptation strategies1.

This report presents the second part of a two-way modeling study assessing the impact of climate change for future water availability in the Amu Darya and Syr Darya river basins. The parts I and II focused on the impact of climate change for future runoff generation in the mountainous upstream parts of the basins. The projected inflow from the upstream parts into the downstream areas serves as input for a water allocation model developed in the Water Evaluation and Planning system (WEAP) to simulate water demands and resources in the downstream areas. This second part of the modeling study is reported in this document. The effects of future changes in temperature and

1 These sections are partly based on http://blogs.ei.columbia.edu/2009/08/18/water-and-energy- conflict-in-central-asia/

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

precipitation for the future water availability and demand are simulated and the effects of possible adaptation measures are explored.

2 METHODOLOGY

2.1 Water Evaluation And Planning system

WEAP ("Water Evaluation And Planning" system) is a well-known software tool that takes an integrated approach to water resources planning. Allocation of limited water resources between agricultural, municipal and environmental uses now requires the full integration of supply, demand, water quality and ecological considerations. WEAP aims to incorporate these issues into a practical yet robust tool for integrated water resources planning. WEAP is developed by the Stockholm Environment Institute's U.S. Center. WEAP was originally developed for simulating water balances and evaluating water management strategies in the Aral Sea region [Raskin et al., 1992].

A database maintains water demand and supply information to drive a mass balance model on link-node architecture. Simulations calculate water demand, supply, runoff, infiltration, crop requirements, flows, and storage, and pollution generation, treatment, discharge and instream water quality under varying hydrologic and policy scenarios. Policy scenarios evaluate a full range of water development and management options, and takes account of multiple and competing uses of water systems.

WEAP has a user-friendly GIS-based interface with flexible model output as maps, charts and tables. WEAP is available in also Russian and Farsi languages and it is already at use in the Aral Sea Basin. WEAP license is free of charge to non-profit, governmental or academic organization based in a country receiving development bank support (as all the Central Asian countries).2

2.2 Model concepts

The first part of the two way modeling study focused on modeling the impact of climate change for the mountainous upstream parts of the basins (Figure 2-1). This was done using a fully distributed cryospheric-hydrological model simulating all major hydrological and cryospheric processes at 1 km spatial scale with a daily time step. The model was forced with climate change scenarios for five Global Circulation Models (GCMs) to project future runoff generation until 2050. The projected inflow from the upstream parts into the downstream areas serves as input for a water allocation model developed in WEAP simulating water demands and resources in the downstream areas.

In WEAP a database maintains water demand and supply information to drive a water balance model on link-node architecture. Simulations calculate water demand, supply, runoff, infiltration, crop requirements, evapotranspiration, flows, reservoir storage under varying hydrologic and adaptation scenarios.

2 www.weap21.org 8

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Adaptation scenarios evaluate a full range of water development and management options, and take account of multiple and competing uses of water systems.

A water allocation model is developed in WEAP incorporating the agricultural and domestic demand sites, catchments, inflow points from upstream, reservoirs and the connections between them. The effects of future changes in temperature and precipitation for the future water availability and demand are simulated until 2050 and the effects of possible adaptation measures are explored.

The downstream model runs at a monthly time step for three time intervals: for the reference situation (2001-2010) and for two future time interval (2021-2030 and 2041-2050). For each of these time intervals one average year is calculated by averaging the ten years within the interval, to get one representative year. Input from the upstream model is also averaged to the same representative year. The model is calibrated for the reference situation (2001-2010).

Figure 2-1: Two-way modeling approach

2.3 Model setup and data sources

The model is set up for the Amu Darya and Syr Darya river basins. As mentioned, the basin is divided in an upstream part and a downstream part ( Figure 2-2). For the upstream part, the AralMountain model, a combined cryospheric-hydrological model, was developed for this study. For the

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

upstream part, the hydrological situation can be described as a natural situation, where human interference can be neglected. The downstream parts are modeled using the ARAL-WEAP model which is designed in the WEAP-tool. The downstream part includes regions where human interference is significant and comprises all major agricultural areas as well as areas where water is extracted for domestic use in highly populated areas. Figure 2-3 and Figure 2-4 show schematic representations of the model setup.

Division upstream and downstream river basin Upstream basin Downstream basin 0 250 500 1000 km

Figure 2-2: Division in upstream and downstream basin

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 2-3: Schematic representation Amu Darya river basin in ARAL-WEAP model.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 2-4: Schematic representation Syr Darya river basin in ARAL-WEAP model.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

The division of the upstream and the downstream part approximates the division in areas without significant human interference and areas with significant human interference. Partly, this division is well defined where major reservoirs are located in the mountain ranges. Downstream of these locations, the stream flow is human-regulated. In other regions the division in upstream basin and downstream basin is less well defined. For those regions the division is made based on optical analysis of satellite imagery. This boundary approximates the division between the mountain environment and the lower land, extensively used by the human population.

As mentioned before the hydrology in the upstream basin is modeled in the AralMountain model. This is done for the current situation (2001-2010) as well as for future decades (2010-2050). For 2001-2010 the inflow from upstream is daily averaged over the ten years. The generated model output (stream flow) is subsequently used in the ARAL-WEAP model as model input. To obtain stream flow data to be used as input in ARAL-WEAP, the daily generated runoff in the upstream basin is modeled for multiple subcatchments in the upstream basin corresponding to the demand sites and reservoirs used in ARAL-WEAP (Figure 2-5). In the schematic representations of the basins in ARAL-WEAP (Figure 2-3 and Figure 2-4) these inflows from the upstream model are indicated as ‘Inflow AralMountain upstream model’ and ‘Reservoirs’. The figures also indicate in which order these inflows are added to the system. The geographical visualization of the ARAL-WEAP model ( Figure 2-6) shows the geographical positioning of the rivers, demand sites, inflows, catchments, transmission links and return flows as represented schematically in Figure 2-3 and Figure 2-4.

Figure 2-5: Subcatchments used in upstream model for input in downstream WEAP-model. See Table 1 for names of the subcatchments, which are also used in Figure 2-3 an Figure 2-4.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Table 1: Subcatchments upstream parts of the basin (Figure 2-5).

Based on geographical position and data availability, different demand sites were assigned. Each demand site has two components: agricultural demand and domestic demand. Data on reservoir properties, agricultural land use, and demography are taken from the online Central Asian Waterinfo portal.3 Data on land use and populations in this database are arranged at province level for five countries in the Amu Darya and Syr Darya river basins (Uzbekistan, Kazakhstan, Tadzhikistan, Kyrgyzstan and Turkmenistan). No data is available in the database for Afghanistan, although a significant part of the Amu Darya river basin is situated in this country. The used data in the database is updated until the year 2000, which is assumed to be representative for the reference situation (2001-2010).

Since data on agriculture and population numbers are arranged at the province level, the division of demand sites was chosen in a way with close resemblance to the province boundaries. In some cases data of different provinces were combined to form one demand site and in other cases data of one province was divided over multiple demand sites.

3 www.cawater-info.net 14

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 2-6: Geographical visualization of ARAL-WEAP model. AralMountain upstream model area is indicated with blue color. Demand sites are indicated with red dots, catchments are indicated with green dots.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

For example, the demand site labeled ‘Fergana Valley’ is based on all provinces comprising the Fergana Valley. This means, data for the provinces Andijan, Namangan and Fergana provinces in Uzbekistan and data for the Jalalabad and Osh provinces in Kyrgyzstan are combined. An example of a case where data of one province is divided over multiple demand sites is the Kashkhadarya province in Uzbekistan. This province was divided into an upstream demand site and a downstream demand site at the location of the Surkhandarya reservoir, since this is an important feature with high influence on water budgets and therefore needs to be incorporated in the ARAL-WEAP model. The division of population numbers and agricultural surface area over the demand sites is estimated based on satellite imagery. Table 2 shows the translation of provinces to demand sites as used in the model.

Table 2: Division of provinces over WEAP demand sites.

Demand site in WEAP Provinces Dushanbe Rayons of republican subordination (TJK) Andijan (UZB) Jalalabad (KGZ) Fergana Valley Namangan (UZB) Osh (KGZ) Fergana (UZB) Mary (TKM) Karakum desert Akhal (TKM)

Kashkhadarya upstream 20% of Kashkhadarya (UZB)

Kashkhadarya downstream 80% of Kashkhadarya (UZB) Kurgantube 80% of Khatlon (TJK) Kulyab 20% of Khatlon (TJK) Kzylorda Kzylorda (KAZ) Lebap Lebap (TKM) South Kazakhstan South Kazakhstan (KAZ) Surkhandarya upstream 40% of Surkhandarya Surkhandaraya downstream 60% of Surkhandarya Jizakh (UZB) Tashkent (UZB) Syrdarya, Tashkent, Jizakh Syrdarya (UZB) 20% of Sughd (TJK) Khorezm (UZB) Urgenc, Nukus, Aral Sea Karakalpakistan (UZB) Dashoguz (TKM) Bukhara (UZB) Zeravshan Valley Navoiy (UZB) Samarkand (UZB)

Water inflow is also generated in the downstream parts. This is incorporated in the WEAP-model. Catchments are assigned which coincide with the demand sites. For these catchments monthly mean, maximum and minimum temperature and total monthly precipitation are extracted from the 2001-2010 climate data set prepared by FMI. With this data the monthly incoming water (from precipitation) and the water lost by evapotranspiration is calulated. For

this purpose the monthly reference evapotranspiration (ETref) is calculated using the Modified Hargreaves method [Droogers and Allen, 2002]. According to the Modiefied Hargreaves method, the reference evapotranspiration is defined as:

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

0.76 ETref =0.0013 · 0.408RA · (Tavg +17.0) · (TD − 0.0123P)

Where RA is the incoming extraterrestrial radiation in MJm-2d-1, Tavg is the average temperature, TD is the temperature range (Tmax – Tmin) and P is the incoming precipitation. All of these parameters are calculated on a monthly basis from the climate data set. The agricultural area is not taken into account in the calculation of the catchment’s evapotranspiration, since this is modeled for the agricultural demand sites in WEAP separately. Precipitation is calculated for the entire catchment, including the agricultural area. Rainfall- runoff is modelled according to the FAO rainfall-runoff model, which is incorporated in WEAP.

Figure 2-7: Downstream catchments used in WEAP model. See Table 3 for names of the subcatchments, which are also used in Figure 2-3 and Figure 2-4.

Table 3: Catchments used in the downstream model (Figure 2-7) ID Catchment name 1 Kulyab 2 KurganTube 3 Dushanbe 4 Surkhandarya upstream 5 Surkhandarya downstream 6 Karakum kanal 7 Kashkadarya upstream 8 Kashkadarya downstream 9 Zeravshan Valley 10 Lebap 11 Fergana Valley 12 Syrdaryo, Tashkent, Jizakh 13 South Kazakhstan 14 Karakum desert 15 Urgenc, Nukus, AralSea 16 Tyuyamuyn 17 Kzylorda

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

3 REFERENCE SITUATION 2001-2010

3.1 Population

For the reference situation, population figures for the year 2000 are assumed to be representative. The population figures per demand site are presented in Figure 3-1.

Figure 3-1: Population per demand site in 2000. Source: Central Asian Water Info database

Table 4: Domestic water allocation [Aldaya et al., 2010]

Annual Monthly Country domestic water domestic water use (m3/cap) use (m3/cap) Kazakhstan 39 3.25 Kyrgyzstan 63 5.25 Tajikistan 69 5.75 Turkmenistan 74 6.17 Uzbekistan 109 9.08 Average 70.8 5.9

These population numbers are used in ARAL-WEAP to calculate monthly domestic water demand. The annual water use rate per capita in this study is assumed to be 70.8 m3 per capita. This is the average rate for the five countries in the basin (Table 4). The effective domestic consumption is

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

estimated to be 10%, which means 90% of the water allocated for domestic purposes is returned to the system and is available downstream.

3.2 Agriculture

For the reference situation, data on agriculture for the year 2000 are assumed to be representative. Figure 3-2 shows the surface area used for agriculture for each demand site in 2000. Agricultural surface area is sub-categorized for eight crop or land use types (Cotton, forage crops, corn, orchards, cucurbits, grain crops, potatoes). The most important crop in both river basins is cotton. Other important crops are grain crops like wheat. There are large differences between the provinces regarding the types of crops that are grown.

Figure 3-2: Agricultural area per demand site in 2000. Source: Central Asian Water Info database.

To assess the water demand for the agricultural areas it is essential to have a good estimate for the water demand of the different crop types and to have good insights in the crop calendar for the different crops. The crop calendar we use in this study is based on literature, whereas in the most ideal case, data on crop calendar comes from the region directly.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

The potential evapotranspiration (ETpot) is calculated using the reference evapotranspiration (ETref) and the crop coefficient (Kc):

ETpot = Kc * ETref

Based on the availability of water, the actual evapotranspiration is calculated by WEAP.

The appropriate values for the crop coefficients and crop calendar are mainly based on the FAO guidelines for computing crop water requirements [Allen et al., 1998] and information from FAO’s crop water information website4. The crop coefficient differs for the different growth stages of a crop (Figure 3-3). These different crop coefficients are multiplied by the length of each growth stage and then averaged. Because the crops in the available data are often generalized (e.g. Grain crops, Vegetables, Orchards), a most representative crop is chosen.

Figure 3-3: Example showing different values for the crop coefficient during the different growing stages of the crop. Source: FAO

Table 5: Average crop coefficients for crops in the study area. Based on [Allen et al., 1998]

Average crop coefficient (Kc) Cotton 0.78 Forage crops 0.76 Orchards 0.78 Grain crops 0.76 Homestead 0.78 Corn 0.81 Cucurbits 0.75

4 http://www.fao.org/nr/water/cropinfo.html 20

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Potatoes 0.75 Rice 1.00 Sugar beet 0.83 Vegetables 0.88 Vinyards 0.49 Other crops 0.78

Table 6: Crop calendar. Numbers represent fraction of month that the crop is growing in the field. Data based on [Allen et al., 1998] and FAO website.

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Cotton 0 0 0 0.5 1 1 1 1 1 0.5 0 0 Forage crops 0 0 1 1 1 1 1 1 1 1 0 0 Orchards 0 0 0 1 1 1 1 1 1 0 0 0 Grain crops 0 0 0.5 1 1 1 1 0 0 0 0 0 Homestead 0 0 0 0.5 1 1 1 1 0.5 0 0 0 Corn 1 1 1 1 0.33 0 0 0 0 0 0 0.5 Cucurbits 0 0 0 0 0 1 1 1 1 0 0 0 Potatoes 0 0 0 0 1 1 1 1 0 0 0 0 Rice 0 0 0 0 1 1 1 1 1 1 0 0 Sugar beet 1 1 1 1 1 0 0 0 0 0 1 1 Vegetables 0 0 0 1 1 1 1 0.5 0 0 0 0 Vinyards 0 0 0 1 1 1 1 1 1 1 0 0 Other crops 0 0 0 0.5 1 1 1 1 0.5 0 0 0

The efficiency of irrigation for the current situation is estimated to be 90% for upstream agricultural demand sites and 95% for downstream agricultural demand sites. These relatively high efficiencies are the total system ones. So reuse of water is included. Obviously, field efficiencies are much lower.

3.3 Reservoirs

Reservoirs are very important features for water management in the Amu Darya and Syr Darya river basins. The major reservoirs in the two river basins are incorporated in the ARAL-WEAP model (Figure 3-4 and Figure 3-5). The inflow for the reservoirs at the boundary between the upstream model and downstream model is calculated using the upstream AralMountain model. The daily inflow averaged over 2001-2010 is calculated and translated to monthly time steps for use in ARAL-WEAP. For the Syr Darya river basin these reservoirs are Toktogul, Andijan, Charvak, Papan, Akhangaran and Zaamin. For the Amu Darya river basin these reservoirs are Nurek, Tupalangsku, Pachkamar and Gissarak. The inflows in downstream reservoirs, located in between demand sites are calculated within the ARAL- WEAP model.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Reservoir capacity Syr Darya basin 16000

14000

) 12000 3

10000

8000

6000

4000

Effective capacity Effective capacity (Mm 2000

Figure 3-4: Effective storage capacity for major reservoirs in the Syr Darya river basin. Source: Central Asian Water Info database.

Reservoir capacity Amu Darya basin 6000

5000 )

3 4000

3000

2000

1000 Effective capacity (Mm 0

Figure 3-5: Effective storage capacity for major reservoirs in the Amu Darya river basin. Source: Central Asian Water Info database.

3.4 Calibration

The ARAL-WEAP model is calibrated for the reference situation using available data for seven major reservoirs in the basins as well as available data on

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

inflows into the Aral Sea as provided by Central Asian Waterinfo Database. A first estimate on the reliability of these data has been made. In the database, three measurements per month for reservoir inflow, release and storage volume are published. The given observed volumes however, differ significantly from values for volumes which are calculated by adding the observed inflow and subtracting the observed outflow from the initial volume. These differences are shown in Figure 3-6 to Figure 3-10. The cause for this bias between the observed volumes and volumes calculated from the observed in- and outflows is uncertain. Volume losses could be explained to some extent by high evaporation and/or seepage losses and increases in volume could be explained to some extent by inflow of groundwater. Total losses from reservoirs due to evaporation and infiltration in the Amu Darya are estimated to be 14,000 Mm3 per year5.

Figure 3-6: Observed inflow, outflow and volume data for Toktogul reservoir 2001-2010. Source: Central Asian Waterinfo Database.

5 http://www.cawater-info.net/amudarya/losses_e.htm 23

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 3-7: Observed inflow, outflow and volume data for Andijan reservoir 2001-2010. Source: Central Asian Waterinfo Database.

Figure 3-8: Observed inflow, outflow and volume data for Charvak reservoir 2001-2010. Source: Central Asian Waterinfo Database.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 3-9: Observed inflow, outflow and volume data for Kayrakkum reservoir 2001-2010. Source: Central Asian Waterinfo Database.

Figure 3-10: Observed inflow, outflow and volume data for Chardara reservoir 2001-2010. Source: Central Asian Waterinfo Database.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 3-11: Observed inflow, outflow and volume data for Nurek reservoir 2001-2010. Source: Central Asian Waterinfo Database.

Figure 3-12: Observed inflow, outflow and volume data for Tyuyamuyun reservoir 2001- 2010. Source: Central Asian Waterinfo Database.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Despite some questions on the reliability of the observed data as presented in the previous section, these data were used to validate/calibrate the ARAL- WEAP model. Initially, the ARAL-WEAP model performed already satisfactory before any calibration as performed. After some further fine-tuning using the calibration parameters (Table 8, Table 9), the model performed very satisfactory. The parameters are calibrated separately for the Syr Darya basin and the Amu Darya basin.

For each of the two basins the calibration is done for an average year in the reference period (2001-2010). All daily values are averaged for the ten year period to obtain the average year. The model performs very well for the reference period. For the Syr Darya basin, observed reservoir inflow, outflow and volume are available for Toktogul, Andijan, Charvak, Kayrakkum and Chardara reservoirs. For the Amu Darya basin, these data are available for Nurek and Tyuyamuyn reservoirs. .

Figure 3-13: Observed and simulated in- and outflows Nurek reservoir reference period.

Figure 3-14: Observed and simulated in- and outflows Tyuyamuyun reservoir reference period.

When we let the model regulate the water release from the reservoirs based solely on downstream agricultural and domestic demand, the outflow numbers differ from the observed numbers. In reality, other interests next to agricultural and domestic water use, like for example the generation of hydropower, are considered in the water release regime of reservoirs. Since

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

these other interests are not included in the model, certain rules are applied to the reservoirs to mimic the water release regime that was observed for the reference period. Examples of observed and simulated in- and outflows for reservoirs are shown in Figure 3-13 and Figure 3-14

Besides reservoir data, the model is also calibrated using available data for Aral Sea inflow from the Amu Darya and Syr Darya (Table 7). Total observed average annual inflow during the reference period into the Aral Sea is 7082 Mm3 for the Syr Darya and 7309 Mm3 for the Amu Darya. Table 8 and Table 9 list the calibrated parameters used in the ARAL-WEAP model.

Table 7: Observed and simulated average annual inflow into the Aral Sea for the reference period 2001-2010 River Observed Simulated Bias (Mm3) (Mm3) Amu Darya 7309 7419 1.5% Syr Darya 7082 7440 5.0%

Table 8: Calibrated parameters ARAL-WEAP model Syr Darya basin Parameter Value Double cropping factor 1.30 Return flow upstream 10% Return flow downstream 5% Domestic consumed 10% Downstream runoff factor 0.1 Loss from riverbed South Kazakhstan - Kzylorda 15.5%

Table 9: Calibrated parameters ARAL-WEAP model Amu Darya basin Parameter Value Double cropping factor 1.35 Return flow upstream 10% Return flow downstream 5% Domestic consumed 10% Downstream runoff factor 0.1 Loss from riverbed Pyandj-Vaksh-Kerki 1.97% Loss from riverbed Kerki-Tyumayun 10.36% Loss from riverbed downstream Tyumayun 14.23% Loss from riverbed downstream Urgenc-Nukus 13.50%

A double cropping factor was introduced to correctly simulate the use of a plot of land for multiple crops throughout the year. This factor is calibrated for both basins separately. A downstream runoff factor was introduced to produce more realistic runoff generation in the downstream catchments. Since these areas have a very sandy subsoil, the infiltration rates are estimated to be very high. To compensate for this effect, which is not incorporated in the FAO rainfall runoff module in WEAP, we estimate that 90% of generated runoff is lost due to infiltration.

Seepage losses of water into the sandy subsoil is also very important for the actual streams of the Amu Darya and Syr Darya. Besides, water loss due to

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

evaporation is significant given the fact that temperatures get very high in the downstream parts of the basins. Unfortunately, information regarding riverbed losses is very sparse and estimates are not very accurate. Some estimates for riverbed losses for the Amu Darya are reported at the Central Asian Water database6. We used these estimates in our model and made assumptions for loss from the riverbed in the Syr Darya basin.

After calibration the model was run and validated by comparing simulated water demand for irrigated farming in the Amu Darya basin to the reported demand for irrigated farming. The model simulates an average annual demand of 56,672 Mm3 on average for 2001-2010. This correlates excellent to the observed annual demand varying from 56,638 to 58,565 Mm3 for 1997-2010, with a bias of 0.06% to 3.23%.

The average annual modeled water balance for both basins are illustrated in Figure 3-15 and Figure 3-16, showing the calculations regarding inflows, demands, precipitation, evapotranspiration and runoff generation in catchments.

6 http://www.cawater-info.net/amudarya/losses_e.htm 29

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 3-15: Average annual water balance Syr Darya ARAL-WEAP reference period (2001- 2010). P catchment is rainfall in the cathment; ETact catchment is the actual evapotranspiration from the natrual landscape and the rainfed crops.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

To be continued on next page

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 3-16: Average annual water balance Amu Darya ARAL-WEAP reference period (2001- 2010).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

3.5 Water availability and unmet demand

This paragraph list the modeling results for the reference situation. As described in the previous paragraphs, the model runs for the average year of the ten year interval 2001-2010. Modeling output therefore is also representative for the average situation. In years where climate/hydrology/agricultural activity differs from the average year, demand and unmet demand will be different. The model runs at a monthly time-step, providing monthly output. Here we report the agricultural and domestic demand and unmet demand differentiated for the Syr Darya basin and the Amu Darya basin.

3.5.1 Demand and unmet demand Syr Darya basin

In the Syr Darya basin the agricultural demand is about 35 times larger than the domestic demand averaged over the year. During summer months, the agricultural demand can be 95 times larger than the domestic demand. Agricultural demand varies from nearly 0 Mm3 per month in winter to 8600 Mm3 per month in July (Figure 3-17). The largest amounts of water are used in the Fergana Valley and the areas directly downstream of the Fergana Valley.

Unmet agricultural demands occur in the summer months (July, August and September) for all demand sites, when the demand is at its maximum and inflow from the mountains gets lower (Figure 3-18). The water storage in reservoirs is largely depleted by July.

The domestic demand remains constant throughout the year (Figure 3-19). Variations in the diagrams are due to a different number of days in a month. For the Syr Darya basin, the domestic demand is around 110 Mm3 per month.

Although the domestic demand is very small compared to the agricultural demand, unmet demands do occur (Figure 3-20). In the summer months (July, August, September), unmet demands occur simultaneously to the period of unmet agricultural demands. But also in winter (January, February, December) demands are unmet because of the stagnating flow from the upstream mountains.

The unmet demand for the entire Syr Darya basin is around 8.8% on an annual basis.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 3-17: Monthly average agricultural demand Syr Darya basin 2001-2010.

Figure 3-18: Monthly average unmet agricultural demand Syr Darya basin 2001-2010.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 3-19: Monthly average domestic demand Syr Darya basin 2001-2010.

Figure 3-20: Monthly average unmet domestic demand Syr Darya basin 2001-2010.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

3.5.2 Demand and unmet demand Amu Darya basin

The agricultural demand in the Amu Darya basin also has its peak during summer, when the growing season is at its maximum (Figure 3-21). The peak in monthly demand is in June, when agricultural demand is 13200 Mm3. Demands are low during the winter months. Areas consuming the largest amounts of water are the Zeravshan Valley, Karakum desert and the areas close to the Aral Sea.

Unlike the Syr Darya basin, unmet agricultural demands are already occurring in April for the Amu Darya basin, whereas the agricultural unmet demands begin to occur in June in the Syr Darya basin (Figure 3-22). These unmet demands in the early growing season are especially large for the valleys of Amu Darya’s tributaries (Zeravshan Valley, Kashkhadarya Valley). Unmet demands increase until July, when the other demand sites also experience unmet demands. In August, September and October, unmet demands occur again only for the tributary valleys (Zeravshan Valley, Kashkhadarya Valley). From November to March, no unmet agricultural demands are experienced in the Amu Darya basin.

Domestic demand in the Amu Darya basin is about 120 Mm3 per month (Figure 3-23). Largest domestic demands occur in the Zeravshan Valley, in the area near Urgenc and Nukus near the Aral Sea, Karakum desert and around Dushanbe.

The unmet domestic demand shows roughly the same pattern as the unmet agricultural demand (Figure 3-24). Unmet demands occur from April to October and peak in July. Zeravshan Valley and Kashkhadarya experience unmet demands during all of these month, whereas other demand sites only have unmet domestic demands during July. Unmet monthly domestic demand for the entire Amu Darya basin is almost 45 Mm3 at its maximum in July.

The unmet demand in the Amu Darya river basin is 24.8% on annual basis.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 3-21: Monthly average agricultural demand Amu Darya basin 2001-2010.

Figure 3-22: Monthly average unmet agricultural demand Amu Darya basin 2001-2010.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 3-23: Monthly average domestic demand Amu Darya basin 2001-2010.

Figure 3-24: Monthly average unmet domestic demand Amu Darya basin 2001-2010.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

3.5.3 Aral Sea inflow

The total annual inflow into the Aral Sea is 14,859 Mm3 average per year (Figure 3-25). The numbers for the Amu Darya and Syr Darya are both about the same (± 7,400 Mm3). This correlates very well to the observed values for annual Aral Sea inflow (See paragraph 3.4 and Table 7).

Figure 3-25: Average annual outflow into the Aral Sea 2001-2010.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

4 PROJECTIONS 2030 AND 2050

The situation in the future in terms of the availability of water resources and demand is assessed for two time intervals. The first future interval is the timeframe from 2021 to 2030 and the second future interval is the timeframe from 2041 to 2050. This is done for five different climate change scenarios. For five Global Circulation Models (GCM’s) the daily changes in temperature and precipitation are projected for both the upstream parts of the river basins and the downstream parts of the river basins. The impact of climate change in the upstream parts of the river basins is described in the separate report on upstream impacts (Part II).

Table 10: Global Circulation Models used to force ARAL-WEAP model.

Abbreviation GCM name Developing institute used in report Canadian Centre for Climate CGCM3(T63) Modelling and Analysis, CCCMA Canada Community Climate Community Earth System System Model 3.0 CCSM3 Model (NCAR-CCSM3) Centre National de Recherches CNRM-CM3 CNRM Météorologiques, France Max Planck Institute for ECHAM5/MPI-OM ECHAM Meteorology, Germany Model for Atmosphere and Ocean Interdisciplinary Research Institute, University MIROC Research On Climate of Tokyo (MIROC3.2 HIRES)

For the inflow points where output from the upstream model serves as input for the downstream models (Paragraph 2.3) the inflow for an average year within the considered period is used. For the reference situation all daily inflow values for January 1st of 2001-2010 are averaged and this is done for every day in a year to obtain one averaged year for the reference period. The same is done with model output for 2021-2030 and 2041-2050. In this chapter, the impact of climate change for the downstream parts of the Amu and Syr Darya river basins is discussed.

Analyses are performed under the following assumptions:

• First, we assume the population to remain constant at the level observed in the year 2000. • Second, we assume the agricultural area to remain constant at the level observed in the year 2000. This is done to be able to separate effects of

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

climate change from other causes that change the future water demand and unmet demand. • Third, we assume the regime applied to reservoir water release to stay the same as the average during the reference period (2001-2010). In the model, water is sometimes forced to be released from reservoirs, although the water demand downstream is already fulfilled. This is done to mimic the current water release regime. In reality this might be due to other interests than agriculture and domestic use, for example for hydropower generation.

4.1 Changes in temperature and precipitation

For each downstream subcatchment as defined in paragraph 2.3 the average change in temperature and precipitation is extracted from the climate projection dataset at a monthly time scale. The daily temperature and precipitation series for 2011-2050 are constructed by repeating the conditions for the ten year reference period four times and adding or subtracting the projected temperature and precipitation changes. The following range of figures shows how the average daily temperature and average monthly precipitation (summed from daily values) change for 2021-2030 and 2041- 2050 for each of the subcatchments. The figures show the range of projections from the five GCMs for 2021-2030 and 2041-2050. Temperature rises by multiple degrees for all subcatchments according to the projection from all the GCMs. The range of projections is quite small. This increase in temperature will have a substantial impact on the water demand by crops (potential evapotranspiration) and might lead therefore to increasing water shortages and reduced river flows.

The precipitation is likely to decrease by a few millimeters or remain unchanged. However, the range of projections for precipitation changes are quite large as can be seen in the figures. Detailed information can be obtained from the report of the Finnish Meteorological Institute on downscaled climate change scenarios in Central Asia. South Kazakhstan South Kazakhstan

35 60 35 60

30 50 30 50

25 25 40 40 20 20 30 30 15 15 Average (°C) TAverage Average Average (°C) T 20 20 10 10

5 10 5 10

0 0 0 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Syrdaryo, Tashkent, Jizakh Syrdaryo, Tashkent, Jizakh

35 45 35 50

40 45 30 30 35 40 25 25 35 30 30 20 25 20 25 15 20 15 20 Average T (°C) TAverage 15 Average (°C) T 10 10 15 10 10 5 5 5 5 0 0 0 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

Fergana Valley Fergana Valley

35 60 35 60

30 50 30 50

25 25 40 40 20 20 30 30 15 15

Average T (°C) TAverage 20 (°C) TAverage 20 10 10

5 10 5 10

0 0 0 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

Kashkadarya upstream Kashkadarya upstream

35 70 35 70

30 60 30 60

25 50 25 50

20 40 20 40

15 30 15 30 Average (°C) Average T (°C) Average T 10 20 10 20

5 10 5 10

0 0 0 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Zeravshan Valley Zeravshan Valley

35 40 35 40

30 35 30 35 30 30 25 25 25 25 20 20 20 20 15 15

Average T (°C) Average T 15 (°C) Average T 15 10 10 10 10

5 5 5 5

0 0 0 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

Kashkadarya downstream Kashkadarya downstream

40 35 40 35

35 30 35 30 30 30 25 25 25 25 20 20 20 20 15 15

Average T (°C) Average T 15 (°C) Average T 15 10 10 10 10

5 5 5 5

0 0 0 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

KurganTube KurganTube

30 70 35 70

25 60 30 60 25 20 50 50 20 15 40 40 15 10 30 30

Average (°C) Average T (°C) Average T 10 5 20 20 5

0 10 0 10

-5 0 -5 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Surkhandarya upstream Surkhandarya upstream

35 40 35 40

30 35 30 35 30 30 25 25 25 25 20 20 20 20 15 15

Average T (°C) Average T 15 (°C) Average T 15 10 10 10 10

5 5 5 5

0 0 0 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

Lebap Lebap

35 45 35 50

30 40 30 45 35 40 25 25 35 30 20 20 30 25 15 15 25 20 20

Average T (°C) Average T 10 (°C) Average T 10 15 15 5 5 10 10 0 5 0 5 -5 0 -5 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

Dushanbe Dushanbe

35 40 40 40

30 35 35 35 30 30 30 25 25 25 25 20 20 20 20 15

Average (°C) Average T 15 (°C) Average T 15 15 10 10 10 10

5 5 5 5

0 0 0 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Kulyab Kulyab

35 40 35 45

30 35 30 40 35 25 30 25 30 20 25 20 25 15 20 15 20

Average T (°C) Average T 10 15 (°C) Average T 10 15 5 10 5 10

0 5 0 5

-5 0 -5 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

Karakum kanal Karakum kanal

35 60 35 60

30 30 50 50 25 25 40 40 20 20

15 30 15 30

Average T (°C) Average T 10 (°C) Average T 10 20 20 5 5 10 10 0 0

-5 0 -5 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

Surkhandarya downstream Surkhandarya downstream

35 70 35 70

30 30 60 60 25 25 50 50 20 20

15 40 15 40

10 30 10 30

Average (°C) Average T 5 (°C) Average T 5 20 20 0 0 10 10 -5 -5

-10 0 -10 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Karakum desert Karakum desert

35 50 35 50 45 45 30 30 40 40 25 35 25 35 30 30 20 20 25 25 15 15 20 20 Average T (°C) Average T (°C) Average T 10 15 10 15 10 10 5 5 5 5 0 0 0 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

Urgenc, Nukus, Aral Sea Urgenc, Nukus, Aral Sea

35 30 35 35

30 30 25 30 25 25 25 20 20 20

15 15 20 15 10 10 15

Average T (°C) Average T 5 10 (°C) Average T 5 10 0 0 5 5 -5 -5

-10 0 -10 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

Tyuyamuyn Tyuyamuyn

35 25 35 30

30 30 25 20 25 25 20 20 15 20 15 15 15 10

Average (°C) Average T 10 (°C) Average T 10 10 5 5 5 5 0 0

-5 0 -5 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Kzylorda Kzylorda

35 35 35 40 30 30 30 35 25 25 30 20 25 20 15 15 25 20 10 10 20 15 5 5

Average T (°C) Average T (°C) Average T 15 0 10 0 10 -5 -5 5 -10 -10 5 -15 0 -15 0 Average T 2021-2030 Average T 2001-2010 Average T 2041-2050 Average T 2001-2010 Average P 2021-2030 Average P 2001-2010 Average P 2041-2050 Average P 2001-2010

Figure 4-1: Projected changes in temperature and precipitation for the demand sites in ARAL-WEAP model. Projections show the average and range of 5 GCMs. Projections are for 2021-2030 and 2041-2050.

4.2 Changes in demand and unmet demand Syr Darya basin

Changes in demand and unmet demand for 2021-2030 and 2041-2050 are modeled for each of the five GCMs using the upstream PCRaster model and the downstream ARAL-WEAP model. In this paragraph the changes in average annual demand and unmet demand in the Syr Darya basin are presented and compared to the reference situation.

4.2.1 Mean of five projections

The changes in demand and unmet demand are modeled for each of the five GCMs. A mean is calculated from this output. As can be seen in Figure 4-2 the annual demand increases by 3.7% until 2041-2050. The annual unmet demand however increases from 8.8% in 2001-2010 to 34.3% in 2041-2050 for the mean of the five projections. The increasing demand can be explained by the increase in temperature, leading to higher evapotranspiration rates and thus to higher unmet demands. Moreover, since precipitation change is somewhat limited in the downstream areas, the significant increase in unmet demand is caused by the decrease in runoff generation in the upstream mountains (see report on impact of climate change on the upstream part of the Aral Sea).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 4-2: Changes in annual demand and unmet demand Syr Darya basin.

4.2.2 CCSM3

The CCSM3 scenario projects unmet demand above the average of the five GCMs. Changes in annual demand increase by 4.7 % until 2041-2050. Unmet demand increases from 8.8% in 2001-2010 to 39.7% in 2041-2050 (Figure 4-3). This GCM projects high temperature rises, especially in the upstream mountains, leading to less runoff generation. In the downstream regions, higher temperatures increase the evaporation rates of agricultural crops, leading to an increase in water demand.

Figure 4-3: Changes in annual demand and unmet demand Syr Darya basin CCSM3 GCM.

4.2.3 CNRM

Changes in unmet demand are just below average for the CNRM GCM. Projected unmet demand increases from 8.8% in 2001-2010 to 32.2% in 2041-2050 (Figure 4-4). The annual water demand increases by 3.1% for the same period. 49

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 4-4: Changes in annual demand and unmet demand Syr Darya basin CNRM GCM.

4.2.4 MIROC

The MIROC GCM projects changes in demand and unmet demand close to the average of the five GCMs. The model forced by this GCM projects a 3.8% increase in water demand until 2041-2050. The annual unmet demand increases from 8.8% in 2001-2010 to 34.6% in 2041-2050 (Figure 4-5). Because the projection for the MIROC GCM is closest to the average of the five GCMs it is used as representative GCM in the adaptation measures analysis (Chapter 5).

Figure 4-5: Changes in annual demand and unmet demand Syr Darya basin MIROC GCM.

4.2.5 ECHAM

The ECHAM GCM also projects changes in demand and unmet demand which are close to the average of the five GCMs. A 3.6% increase in water demand is projected until 2041-2050. The annual unmet demand increases from 8.8% in the current situation to 33.4% in 2041-2050 (Figure 4-6).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 4-6: Changes in annual demand and unmet demand Syr Darya basin ECHAM GCM.

4.2.6 CCCMA

The CCCMA GCM projects changes in annual demand and unmet demand lower than the average for the five GCMs (Figure 4-7). Annual demand increases by 3.6% until 2041-2050. The annual unmet demand increases from 8.8% in 2001-2010 to 31.6% in 2041-2050.

Figure 4-7: Changes in annual demand and unmet demand Syr Darya basin CCCMA GCM.

4.3 Changes in demand and unmet demand Amu Darya basin

Changes in demand and unmet demand for 2021-2030 and 2041-2050 are modeled for each of the five GCMs. In this paragraph the changes in average annual demand and unmet demand in the Amu Darya basin are presented and compared to the reference situation.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

4.3.1 Mean of five projections

The changes in demand and unmet demand are modeled for each of the five GCMs. A mean is calculated from this output. Figure 4-8 shows how the annual demand increases by 4.4% until 2041-2050. The annual unmet demand however increases from 24.8% in 2001-2010 to 48.6% in 2041-2050 for the mean of the five projections. Like in the Syr Darya basin, the increasing demand can be explained by the projected increase in temperature, leading to higher evapotranspiration rates for agricultural crops and thus to higher unmet demands. The stronger increase in demand for the Amu Darya basin (4.4%) in comparison to the demand increase for the Syr Darya basin (3.8%) is due to the slightly stronger projected increase in temperature in the Amu Darya basin compared to the Syr Darya basin. Moreover, since precipitation does not change much in the downstream areas, the large increase in unmet demand is also caused by the decrease in runoff generation in the upstream mountains.

Figure 4-8: Changes in annual demand and unmet demand Amu Darya basin.

4.3.2 CCSM3

The model forced with the CCSM3 GCM projects a higher increase in unmet demand than the average of the five GCMs (Figure 4-9). Unmet demand increases from 24.8% in 2001-2010 to 54.5% in 2041-2050. The increase in demand is 4.7% until 2041-2050.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 4-9: Changes in annual demand and unmet demand Amu Darya basin CCSM3 GCM.

4.3.3 CNRM

The increase in demand unmet demand for the model forced with the CNRM GCM is slightly below the average of the five GCMs. The demand increases by 3.8% until 2041-2050. Unmet demand increases from 24.8% in the current situation to 46.5% in 2041-2050 (Figure 4-10).

Figure 4-10: Changes in annual demand and unmet demand Amu Darya basin CNRM GCM.

4.3.4 MIROC

For the model forced with the MIROC GCM, a 5.1% increase in demand is projected for 2041-2050. The unmet demand increases from 24.8% in 2001- 2010 to 48.1% in 2041-2050 (Figure 4-11).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 4-11: Changes in annual demand and unmet demand Amu Darya basin MIROC GCM.

4.3.5 ECHAM

For the model forced with the ECHAM GCM, a 4.1% increase in demand is projected for 2041-2050. The unmet demand increases from 24.8% in 2001- 2010 to 48.2% in 2041-2050 (Figure 4-12).

Figure 4-12: Changes in annual demand and unmet demand Amu Darya basin ECHAM GCM.

4.3.6 CCCMA

The CCCMA GCM projects changes in annual demand and unmet demand lower than the average for the five GCMs (Figure 4-7). Annual demand increases by 4.3% until 2041-2050. The annual unmet demand increases from 24.8% in 2001-2010 to 45.8% in 2041-2050.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 4-13: Changes in annual demand and unmet demand Amu Darya basin CCCMA GCM.

4.4 Changes in Aral Sea inflow

In response to lower inflow from upstream to the downstream areas and increasing demand, less water will reach the Aral Sea. The average annual inflow for the Amu Darya and Syr Darya into the Aral Sea is calculated by the model when forced by each of the five GCM projections.

4.4.1 Mean of five projections

The outcomes of the model forced by each of the five GCMs are averaged to obtain the mean output. Average annual outflow into the Aral Sea decreases for both rivers (Figure 4-14). The decrease is strongest for the Amu Darya river. Outflow for the Syr Darya river into the Aral Sea decreases 10.8% until 2041-2050. The outflow from the Amu Darya decreases 38.3% for the same time interval.

Figure 4-14: Average annual outflow into Aral Sea. Mean output of model forced by five GCMs.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

4.4.2 CCSM3

For the model forced with the CCSM3 GCM the average annual inflow into the Aral Sea provided by the Amu Darya decreases by 39.8% until 2041-2050 (Figure 4-15). For the Syr Darya the decrease is 13.1%. The model forced with the CCSM3 GCM projects the largest impact for Aral Sea inflow, compared to the four other GCMs.

Figure 4-15: Average annual outflow into Aral Sea for model forced with CCSM3 GCM.

4.4.3 CNRM

Average annual inflow into the Aral Sea for the model forced with the CNRM GCM is projected to decrease 38.2% for the Amu Darya in 2041-2050 relative to 2001-2010 (Figure 4-16). For the Syr Darya, the projected decrease is 11.6% for 2041-2050 relative to 2001-2010. Both projections are very close to the average of the projections.

Figure 4-16: Average annual outflow into Aral Sea for model forced with CNRM GCM.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

4.4.4 MIROC

The model forced with the MIROC GCM projects the average annual inflow into the Aral Sea from the Amu Darya to decrease by 38.4% until 2041-2050. For the same period, the average annual inflow into the Aral Sea from the Syr Darya decreases by 11.2% (Figure 4-17).

Figure 4-17: Average annual outflow into Aral Sea for model forced with MIROC GCM.

4.4.5 ECHAM

The model forced with the ECHAM GCM projects decreases in Aral Sea inflow which are close to average too. For the Amu Darya, the average annual inflow into the Aral Sea decreases by 38.2%. A decrease of 11.5% is projected for the Syr Darya (Figure 4-18).

Figure 4-18: Average annual outflow into Aral Sea for model forced with ECHAM GCM.

4.4.6 CCCMA

The model forced with the CCCMA GCM projects the lowest decreases in average annual inflow into the Aral Sea compared to the four other

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

projections. For the Amu Darya the average annual inflow into the Aral Sea decreases 37.0%, whereas a decrease of 7.4% in average annual inflow into the Aral Sea is projected for the Syr Darya (Figure 4-19).

Figure 4-19: Average annual outflow into Aral Sea for model forced with CCCMA GCM.

4.5 Future Aral Sea development

Due to decreasing inflows into the Aral Sea from the Amu Darya and Syr Darya rivers, the Aral Sea has been decreasing in size rapidly. Figure 4-20 shows the development of the Aral Sea shoreline for different time steps from 1960 to 2008. In 2005, the northern part of the Aral Sea was dammed with the Kokaral dike, dividing the sea in a northern sea fed by the Syr Darya and a southern sea fed by the Amu Darya.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Figure 4-20: Aral Sea shoreline development 1960-2008. Source: www.unimaps.com, based on NASA imagery.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

A first estimate of Aral Sea dimensions in 2041-2050 was made for the northern and southern Aral Sea. Using the climate data used in the model for 2001-2010 and 2041-2050 it is possible to estimate the annual evaporation from the lake surface and the annual precipitation falling into the lake. The WEAP model calculates the reference evapotranspiration for the area.

Multiplying the reference evapotranspiration (ETref) by an open water coefficient (Kw) gives an estimate of the evaporation from open water:

E = Kw * ETref

A Kw value of 1.10 provides a good estimate of the actual evaporation from open water [Jensen, 2010]. Besides inflow from Amu Darya and Syr Darya rivers into the Aral Sea, precipitation falls on the lake surface. The annual precipitation is also extracted from the climate data set for 2001-2010 and 2041-2050. Table 11 lists the values for lake surface, open water evaporation, precipitation, required inflow to sustain lake size, observed inflow and projected inflow. The evaporation rate and precipitation in 1960 are assumptions, for 2005-2008 and 2041-2050 model climate data is used. Historical lake surface data and inflow observations are obtained from the Central Asian Waterinfo database. Projected inflows are mean of 5 GCM model runs output.

Observed inflow into the Aral Sea in 1960-1970 is much smaller than the inflow required to sustain the lake, resulting in the observed Aral Sea shrinkage. In 2005-2008, the northern Aral Sea is gaining surface area, because of the completion of the Kokaral dike. At the same time, the southern Aral Sea is still shrinking since inflow requirements to sustain the lake’s size are not met.

Projected average annual inflow into the northern Aral Sea is projected to be 6639 Mm3 in 2041-2050, which would be sufficient to maintain a lake with approximately 5158 km2 surface area. In theory, the northern lake could grow, although the lake dimensions are currently limited by the Kokaral dike. The projected average annual inflow for 2041-2050 into the southern Aral Sea is 4576 Mm3, which would be sufficient to maintain a lake with an approximate surface area of 8680 km2. This implies a significant decrease in lake size compared to the 2005-2008 situation.

Table 11: Estimating Aral Sea size in 2041-2050 based on historic inflow observations and simulated future inflows using PCRaster and ARAL-WEAP.

Required Observed Lake Evaporation Precipitatio inflow to inflow Projected surface open water n over lake sustain (Mm3/yr) inflow

(km2) (mm/yr) (mm/yr) lake (Mm3/yr) (Mm3/yr) 1960- 42800 64470 1320 193 72650 1970 North 3000 1409 193 3648 7197 2005- lake 2008 South 14000 1409 193 17025 6254 lake North 5158 1486 199 6639 2041- lake 2050 South 8680 1486 199 4576 lake

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

5 ADAPTATION STRATEGIES

5.1 Water marginal cost curves

5.1.1 Cost curves

The cost-effectiveness of various measures to close the supply-demand gap is be compared in this study by means of the “water-marginal cost curve”, similar to the approach used in a World Bank study for the Middle-East and Northern Africa Water Outlook [Immerzeel et al., 2011]. This cost curve shows the cost and potential of a range of different measures- spanning both productivity improvements and supply expansion – to close the gap. Such a water-marginal cost curve is estimated for the region to assess the total costs to close the supply-demand gap projected under the MIROC scenario in 2041- 2050. The MIROC scenario is used, because it is closest to the average of the five projections used.

Each of these measures is represented as a block on the curve (Figure 5-1). The width of the block represents the amount of incremental water that becomes available from adoption of the measure. The wider a measure, the larger its net impact on water availability. The height of the block represents its unit cost in US$ per m3. The vertical axis measures the financial cost –or savings- per unit of water released by each measure. This is the annualized capital cost, plus the net operating cost compared to business as usual. The unit costs are ordered from the lowest costs to the highest on the cost curve.

Figure 5-1: Schematic representation of the cost curve.

It is important to note that the cost curve’s use is limited to comparing measures’ financial cost and technical potential to close the gap. It does not include or evaluate policies that would be used to enable, incentivize, or enforce the adoption of those measures such as pricing, standards, and behavioral changes. Rather, it provides information on what the cost would be of adopting a set of technical measures, which in turn can be used to inform policy design. Of course, cost is not the only basis on which choices are made, but shedding light on the cost and technical potential of measures allows these to be compared and evaluated in a common context. The cost curve, then, is not prescriptive: it does not represent what the plan for closing the supply- demand gap ought to be. Rather, it should be considered as a tool to help decision-makers understand and compare different options for closing the gap under a given demand scenario. It is therefore important to emphasize that

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

the estimates generated by the cost curve are not explicit predictions, but approximate guides to decision-making.

5.1.2 Measures to close the supply-demand gap

To close the gap between projected future water demand and supply, three core ways of matching water supply and demand are distinguished: • Expanding supply; • Increasing the productivity of existing water use; • Reducing demand by shifting the economy towards less water-intensive activities.

The following potential measures are assessed in this study:

Expanding supply: A: Increased reservoir capacity

Increasing the productivity: B: Improved agricultural practice C: Increased reuse of water in irrigated agriculture D: Increased reuse of water for domestic use

Reducing demand: E: Reduction of irrigated areas F: Reduction of domestic demand G: Deficit irrigation

5.1.3 Costs of these options

The total annual costs for the combined set of measures can be calculated by multiplying the specified deficit by the unit cost of each block required to close the gap. The considered unit cost of each measure is presented below. As there are a large number of measures and a lot of uncertainty about the costs of these measures in the various countries in the future, some crude assumptions have to be made in this study.

A) The costs of expanding reservoir capacity are taken to be 0.04 $/m3 [Immerzeel et al., 2011]. Obviously these costs can vary per region.

B) For improved agricultural practices that increase the productivity of water a unit cost of 0.02 $/m3 is considered [Immerzeel et al., 2011]. There are various kinds of improved agricultural practices, such as drip and sprinkler irrigation, no-till farming and improved drainage, utilization of the best available germplasm or other seed development, optimizing fertilizer use, innovative crop protection technologies and extension services. Costs of such measures vary, but are relatively cheap compared to the water supply measures. Some of the productivity measures can even result in a net cost saving, when operating savings of the measures outweigh annualized capital costs. The 2030 Water Resource Group shows that the majority of the costs of such measures are in the range of 0.02 $/m3 to 0.03 $/m3 [ 2030 Water Resources Group, 2009]. Converting this to costs per hectare (assuming on

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

average 1000 mm of water consumption per hectare) is US$ 200 to US$ 300 per hectare per year.

C) The unit costs of increased reuse of irrigation water are assumed to be 0.04 $/m3 [2030 Water Resources Group, 2009]. These costs are relatively low as it was assumed that this water is only reused for agricultural purposes so that no additional treatment is necessary. The price of 0.04 $/m3 is based on • Reuse of 50 mm = 500 m3 per ha / year • Investment costs of $ 1000 /ha • Annualized capital costs (investment over 10 years) $100 / ha / year; for 500 m3 = 0.02 $/m3 • Annual operational costs (maintenance, pumping) of 0.02 $/m3

D) The unit cost of increased reuse of domestic water depends on the treatment level. According to the 2030 Water Resources Group the unit cost of municipal and industrial waste water reuse is on average 0.30 $/m3 [2030 Water Resources Group, 2009].

E) The unit cost of reduced irrigated areas is assumed to be 0.10 $/m3, as the value of irrigation water ranges usually between 0.05 $/m3 and 0.15 $/m3 [Immerzeel et al., 2011] and foregone benefits can be considered as unit costs. This value is, of course, strongly dependent on the price of agricultural products, which in turn are strongly affected by interventions of governments and trading blocs

F) The unit cost of reduced domestic demand is assumed to be 2.00 $/m3. While drinking water is a necessity of life, its value can be expected to be very high. The other uses of water within households, which make life more comfortable, can be expected to have lower values.

G) The unit cost of deficit irrigation is assumed to be 0.025 $/m3. The average productivity of water in agriculture is around 0.10 $/m3. We assume that reducing the irrigation to 90% of optimum water amounts leads to a reduction in production of 25%. This corresponds to a cost of 0.025 $/m3.

5.2 Effectiveness of adaptation measures

The ARAL-WEAP model for the Amu Darya and Syr Darya basins was used to evaluate the impact of the adaptation measures described in paragraph 5.1.2. The WEAP-model was run for five different climate projections, based on five different Global Circulation Models. The impact of adaptation measures is evaluated for the MIROC GCM, for which the climatic impact for water availability is closest to the mean of the five outputs. No distinction is made between the Amu Darya and Syr Darya basin, the effectiveness of the adaptation measures is evaluated for the total Amu and Syr Darya basin. The effectiveness of the adaptation measures is evaluated for 2041-2050.

Table 12 lists the annual water demand and unmet demand for the entire basin for 2041-2050 with and without adaptation measures.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Table 12: Annual water demand and unmet demand for Amu and Syr Darya river basins for 2041-2050 for the MIROC GCM climate projection. REF reflects scenario without adaptation measures; A to H differences compared to REF (in Mm3).

REF A B C D E F G

DEMAND 100950 0 -14732 -2085 -548 -9821 -274 -19642 Agriculture 98211 0 -14732 -2085 0 -9821 0 -19642 Domestic 2739 0 0 0 -548 0 -274 0 UNMET DEMAND 43133 -3437 -12682 -1871 -203 -8697 -102 -16617 Agriculture 42443 -3388 -12590 -1859 -64 -8636 -32 -16492 Domestic 690 -49 -92 -12 -140 -61 -70 -125

The differences in effectiveness of measures applied to agricultural versus domestic are striking. Since agricultural demands are much higher than domestic demands, adaptation measures applied to agricultural demands are much more effective than adaptation measures applied to domestic demand. Improving for example agricultural practice by 15% reduces the demand by 14732 Mm3 per year, whereas reducing domestic demand by 10% reduces the demand by 274 Mm3; about 54 times less effective.

A: Increased reservoir capacity Increasing the reservoir capacity has limited effect in terms of closing the water gap in the region. Reservoir capacity is large in the region. Increasing the reservoir capacity by 25% will decrease the unmet demand by 3437 Mm3, corresponding to an 8% reduction in unmet demand.

B: Improved agricultural practice Improving agricultural practice reduces the agricultural demand by 15%. The unmet agricultural demand is reduced by 30%. Subsequently unmet domestic demand decreases 13%. In 2041-2050 this reduces the total unmet demand by 12682 Mm3 (29.4% reduction).

C: Increased reuse of water in irrigated agriculture The overall irrigation efficiency, taking into account reuse, is increased from 95% to 97% in the downstream areas and from 90% to 92% in the upstream areas. Increasing the irrigation efficiency in agriculture reduces the agricultural demand by 2.1%. The unmet agricultural demand is reduced by 4.4%. Subsequently unmet domestic demand decreases 1.7%. In 2041-2050 this reduces the unmet demand by 1871 Mm3 (4.3% reduction).

D: Increased reuse of water for domestic use Increasing the reuse of water in domestic use reduces the domestic demand by 20%.The domestic unmet demand is reduced by 20.3%. Subsequently, the agricultural unmet demand decreases by 64 Mm3. In 2041-2050 the total unmet demand is reduced with 203 Mm3 (0.5% reduction).

E: Reduction of irrigated areas Reducing the irrigated areas reduces the agricultural demand by 10%. The agricultural unmet demand decreases by 20.3%. Subsequently unmet domestic demand decreases 8.8%. In 2041-2050 this reduces the unmet demand by 8697 Mm3 (20.2% reduction).

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

F: Reduction of domestic demand Reducing the domestic demand reduces the unmet domestic demand 10.1%. The agricultural unmet demand is reduced by 32 Mm3. Reducing the domestic demand by 10% reduces the total unmet demand 102 Mm3 in 2041-2050 (0.2% reduction).

G: Deficit irrigation Applying deficit irrigation to agricultural areas reduces the agricultural demand with 20%, because evapotranspiration rates decrease strongly. The unmet agricultural demand is reduced by 39%. Subsequently unmet domestic demand decreases 18%. In 2041-2050 this reduces the total unmet demand by 16617 Mm3 (38.5% reduction).

5.3 Water marginal cost curve

The effectiveness of the seven explored adaptation measures in terms of reduction in water shortage is described in the previous section. Another important question is whether these adaptation measures are cost effective. Combining the costs of these adaptation scenarios (paragraph 5.1.3) with the results of the reductions in unmet demand of the measures (paragraph 5.2) results in the water availability cost curve as seen in Figure 5-2.

Figure 5-2: Water marginal cost curve Amu and Syr Darya basin. Note: Cost-axis has been cut off at US$ 0.30. Cost for decreasing domestic demand is 2.00 $/m3.

The water marginal cost curve shows the unit costs of the reductions in unmet demand ordered from the lowest unit cost (0.02 $/m3) to the highest unit cost (2.00 $/m3). The ranking of the adaptation measures (B, G, A, C, E, D, F) reflects the cost-effectiveness of the adaptation measures. The cheapest options are concerning demand reductions in agriculture, and these are also

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

most effective in terms of reducing the unmet demand. Reducing the domestic demand is expensive and reduces the unmet demand insignificantly. The most cost-effective option is the improvement of agricultural practice, followed by the application of deficit irrigation. On the third place comes the increased reuse of water in agriculture and fourth is the reduction of irrigated areas. Least cost-effective are the measures affecting the domestic water use, the increased reuse of domestic water and decreasing the domestic demand.

The unmet demand in 2041-2050 caused by climate change only can be derived from the model results. The total unmet demand for the two basins in the reference period (2001-2010) is 17,800 Mm3/yr (Table 13). This unmet demand rises to 43,133 Mm3/yr, solely in response to the expected climatic changes in the two basins. This means the unmet demand in 2041-2050 which is caused by climate change equals 25,333 Mm3/yr.

Table 13: Total unmet demand and unmet demand caused by climate change.

Amu Syr Darya Total Darya Unmet demand 2001-2010 3,410 14,390 17,800 (Mm3/yr) Total unmet demand 2041-2050 13,846 29,287 43,133 (Mm3/yr) Unmet demand caused by Climate 10,436 14,897 25,333 Change 2041-2050 (Mm3/yr)

Overlaying the projected supply and demand gap (unmet demand) for the Amu and Syr Darya basin in the 2041-2050 representative projection (43,133 Mm3/yr) on the cost curve, it becomes clear that the water gap can be closed completely with the explored adaptation measures (Figure 5-3). Closing the water gap completely would cost about 1,730 million US$ annually (present value). The average unit costs to reach this are 0.040 $/m3. Closing the water gap that is caused by climate change only (25,333 Mm3/yr) would cost about 550 million US$ annually. The average unit costs to reach this are 0.022 $/m3.

It is important to notice that the water gap is calculated for the average climate change projection (MIROC GCM). The water gap is larger for warmer and/or drier climate projections and smaller for cooler and or wetter climate projections.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

Cumulative water marginal cost curve

2500

2000

1500

1000

500 CumulativeCosts (Million$US)

0 0 10000 20000 30000 40000 50000 Additional Water Availability (Mm3/yr)

Total unmet demand 2041-2050 (Mm3/yr) Unmet demand caused by climate change 2041-2050 (Mm3/yr)

Figure 5-3: Cumulative water marginal cost curve Amu Darya and Syr Darya basin.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

6 CONCLUSIONS

Climate change might have a big impact on water resources in the Central Asia region, but a rigorous analysis of these impacts is so far missing as the hydrological regimes of the two major rivers in the region (Syr Darya and the Amu Darya) are complex. Only recently, by the advent of advanced computer modeling combined with remotely sensed data and scientific progress, processes can be better understood and impact analysis related to climate change are more accurate.

The work described in this report in combination with a report on climate change impact for the upstream water resources contribute to a study initiated by the Asian Development Bank to better understand and to explore adaptation strategies in the Aral Sea Basin. The ultimate objective of this project is to develop national capacity in each of the participating countries (Kyrgyz Republic, Tajikistan, Kazakhstan, Turkmenistan, Uzbekistan) to use the models, tools, data and results to prepare climate impact scenarios and develop adaptation strategies. This will then result in improved national strategies for climate change adaptation.

This report describes the analysis focusing on the downstream parts of the Amu Darya and Syr Darya river basins. Based on local and public domain datasets and hydro-meteorological observations a water allocation model has been developed for the downstream parts of the two rivers. The downstream model is coupled to a cryospheric-hydrological model developed for the mountainous upstream parts of the basins. This is one of the first models that covers the entire upstream parts of these basins and includes all processes related to glacier and snow melt, rain runoff and base flow. The downstream water allocation model is used to quantify the impact of climate change for water resources in the two river basins until 2050 and to explore possible adaptation measures.

The key messages resulting from the two-way modeling study are:

• The developed models are able to mimic observed streamflows and water use rates and can be used to explore the impact of climate change on water resources in the region. • There are large differences in the role that melt water plays in runoff generation in the Amu Darya and Syr Darya river basins. Melt water has a higher contribution to runoff in the Amu Darya basin compared to the Syr Darya river basin. • It is very likely that glacier extent in the Pamir and Tien Shan mountain ranges will decrease by 45 to 60% by the year 2050. • The composition of the four components of stream flow (rainfall-runoff, snow melt, glacier melt, base flow) is very likely to change in the future. This will have major impacts on total runoff, especially on seasonal shifts in runoff. The runoff peak will shift from summer to spring and decrease in magnitude. Model output when forced with climate projections generated with five Global Circulation Models shows decreasing runoff generation in

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

the upstream parts of the two basins in 2050. The changes differ strongly spatially. The runoff generation decreases most significantly in upstream areas of glacier retreat. • Total annual runoff into the downstream areas is expected to decrease by 22-28% for the Syr Darya and 26-35% for the Amu Darya by 2050. • Strongest decreases in stream flow are expected for the late summer months (August, September, October), where inflow into downstream areas decreases around 45% for both river basins. • Annual total water demand in the Syr Darya basin increases by 3.0 - 3.9% in 2050. Annual unmet demand increases from 8.8% currently to 31.6 - 39.7% in 2050. • Annual total water demand in the Amu Darya basin increases by 3.8 - 5.0% in 2050. Annual unmet demand increases from 24.8% currently to 45.8 - 54.5% in 2050. • The total extent of the Aral Sea will reduce from about 17,000 km2 currently to 13,800 km2 in 2050. Differences between the North and South Lake are striking. The North Lake has the potential to expand by about 72% in terms of projected inflow (but will be less given spilling to South Lake), while the South Lake will shrink by about 38%. • Most cost-effective adaptation measures are (i) improving agricultural practice, (ii) deficit irrigation, (iii) increasing the reuse of water in agriculture, and (iv) the reduction of irrigated areas. • Costs for closing the entire water gap (43,000 Mm3) are estimated at US$ 1,730 million per year in 2050. • Closing the additional water shortage caused by climate change only (25,000 Mm3) will cost US$ 550 million per year in 2050.

TA 7532: Water and Adaptation Interventions in Central and West Asia Part III - Impact and Adaptation

7 REFERENCES

2030 Water Resources Group (2009), Charting Our Water Future.

Aldaya, M. M., G. Munoz, and A. Y. Hoekstra (2010), Water Footprint of cotton, wheat and rice production in Central Asia, Delft.

Allen, R. G., L. S. Pereira, D. Raes, and M. Smith (1998), Crop evapotranspiration. Guidelines for computing crop water requirements, Rome, Italy.

Droogers, P., and R. G. Allen (2002), Estimating reference evapotranspiration under inaccurate data conditions, Irrigation and Drainage Systems, 16, 33-45.

Immerzeel, W., P. Droogers, W. Terink, J. Hoogeveen, P. Hellegers, M. Bierkens, and R. van Beek (2011), Middle-East and Northern Africa Water Outlook, Wageningen.

Jensen, M. E. (2010), Estimating evaporation from water surfaces, in Proceedings of the CSU/ARS Evapotranspiration Workshop, pp. 1-27, Fort Collins.

Raskin, P., E. Hansen, Z. Zhu, and M. Iwra (1992), Simulation of Water Supply and Demand in the Aral Sea Region, Water International, 17, 55-67.

FCG Finnish Consulting Group Ltd. FINAL REPORT in association with • FutureWater Part 4 • Finnish Meteorological Institute AquaCrop Modeling

Asian Development Bank

Water and Adaptation Interventions in Central and West Asia

TA 7532

Climate Change Impact on the Crop Yields in the Amu Darya and Syr Darya Basins: Preliminary Assessment

November 2012

Table of Contents

1 INTRODUCTION 3

2 METHODS AND DATA 3 2.1 Aral Sea basin 3 2.2 Model selection 3 2.3 AquaCrop model features 5 2.3.1 Theoretical assumptions 6 2.3.2 Climate 7 2.3.3 Crop growth 8 2.3.4 Soil systems 8 2.3.5 Field management 9 2.3.6 Irrigation management 9 2.4 Impact of Climate Change 9 2.5 Crops 12

3 RESULTS AND DISCUSSIONS 13

REFERENCES 17

ABREVIATIONS AND ACRONYMS

ADB Asian Development Bank ET Evapotranspiration ET0/ETref Reference evapotranspiration FAO Food and Agriculture Organization of the United Nations GCM Global Circulation Model KAZ Kazakhstan KGZ Kyrgyzstan MIROC Model for Interdisciplinary Research On Climate, University of Tokyo TAJ Tajikistan TKM Turkmenistan UZB Uzbekistan

1 INTRODUCTION

The Asian Development Bank study “Water and Adaptation Interventions in Central and West Asia” (TA 7532) included a two-way modeling approach assessing the impact of climate change for future water availability in the Amu Darya and Syr Darya river basins. Part I focused on the impact of climate change on future runoff generation in the mountainous upstream parts of the basins. Part II simulated water demands and supplies in the downstream areas. The entire study focused on quantifying the effects of projected changes in temperature and precipitation on the future water availability and demand and potential adaptation measures were explored.

One component not covered in the study was the impact of projected climate change on crop yields. This report explores in which way this could be done by using FAO’s AquaCrop model and demonstrates the use of such an approach for one specific crop in five selected places. This study should therefore not be considered as a full analysis of the impact of climate change on crop yields, but as a first exploration of such an approach.

2 METHODS AND DATA

2.1 Aral Sea basin

The Aral Sea Basin includes the Syr Darya and the Amu Darya river Basins and encompasses five countries: Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, and Uzbekistan. Water resources management in the Central Asia region faces big challenges. The hydrological regimes of the two major rivers in the region are complex and vulnerable to climate change. Water diversions to agricultural, industrial and domestic users have reduced flows in downstream regions, resulting in severe ecological damages.

Future climate change poses additional challenges. The discharge in both the Syr Darya and the Amu Darya rivers is driven mainly by snow and glacial melt. The impact of a warming climate on these key hydrological processes has not sufficiently been understood and only fragmentary mitigation and adaptation strategies are in place. Whereas changes in precipitation levels are hard to predict for the future, there is a solid consensus that average global temperatures are rising. As a result, more precipitation will fall as rain in the upstream and the ice volume in the Tien Shan and Pamir mountain ranges will likely shrink. The former will impact the seasonality of the runoff whereas the latter will at least temporarily increase average annual flows. Furthermore, changes in sediment loads may pose additional problems.

Detailed information on the Aral Sea Basin and its vulnerability to climate change can be found in the associated project reports.

2.2 Model selection

Potential impacts of climate change on world food supply have been estimated in several studies (e.g. Parry et al., 2004). Results show that some regions may improve production, while others suffer yield losses. This could lead to shifts of agricultural production zones around the world. Furthermore, different crops will be affected differently, leading to the need for adaptation of supporting industries and markets. Climate change may alter the competitive position of countries with respect, for example, to exports of agricultural products. This may result from yields increasing as a result of altered climate in one country, whilst being reduced in another. The altered competitive position may not only affect exports, but also regional and farm-level income, and rural employment.

In order to evaluate the effect of climate change on crop production and to assess the impact of potential adaptation strategies models are used frequently (Aerts and Droogers, 2004; Hunink and Droogers, 2011a; 2011b). The use of these models can be summarized as: (i) better understanding of water-food-climate change interactions, and (ii) exploring options to improve agricultural production now and under future climates. Some of the frequently applied agricultural models are: • CropWat • AquaCrop • CropSyst • SWAP/WOFOST • CERES • DSSAT • EPIC

Each of these models is able to simulate crop growth for a range of crops. The main differences between these models are the representation of physical processes and the main focus of the model. Some of the models mentioned are strong in analyzing the impact of fertilizer use, the ability to simulate different crop varieties, farmer practices, etc. However, for this particular project it is required to use models with a strong emphasis on crop-water-climate interactions. The three models that are specifically strong on the relationship between water availability, crop growth and climate change are CropWat, AquaCrop and SWAP/WOFOST. Moreover, these three models are in the public domain, have been applied world-wide frequently, and have a user-friendly interface (Figure 1). Based on previous experiences it was selected to use AquaCrop as it has: • relatively limited data requirements, • a user-friendly interface enabling non-specialist to develop scenarios,

• focus on climate change, CO2, water and crop yields, • developed and supported by FAO, • fast growing group of users world-wide, • flexibility in expanding level of detail.

Figure 1. Typical examples of input screen of AquaCrop: crop development (top) and soil fertility stress (bottom).

2.3 AquaCrop model features

AquaCrop is the FAO crop-model to simulate yield response to water (Steduto et al., 2009; Raes et al., 2009). It is designed to balance simplicity, accuracy and robustness, and is particularly suited to address conditions where water is a key limiting factor in crop production. AquaCrop is a companion tool for a wide range of users and applications including yield prediction under climate change scenarios. AquaCrop is a completely revised version of the successful CropWat model. The main difference between CropWat and AquaCrop is that the latter includes more advanced crop growth routines.

AquaCrop includes the following sub-model components: the soil, with its water balance; the crop, with its development, growth and yield; the atmosphere, with its thermal regime, rainfall, evaporative demand and CO2 concentration; and the management, with its major agronomic practice such as irrigation and fertilization. AquaCrop flowchart is shown in Figure 2.

The particular features that distinguishes AquaCrop from other crop models is its focus on water, the use of ground canopy cover instead of leaf area index, and the use of water productivity values normalized for atmospheric evaporative demand and of carbon dioxide concentration. This enables the model with the extrapolation capacity to diverse locations and seasons, including future climate scenarios. Moreover, although the model is simple, it gives particular attention to the fundamental processes involved in crop productivity and in the responses to water, from a physiological and agronomic background perspective.

Figure 2. Main processes included in AquaCrop.

2.3.1 Theoretical assumptions

The complexity of crop responses to water deficits led to the use of empirical production functions as the most practical option to assess crop yield response to water. Among the empirical function approaches, FAO Irrigation & Drainage Paper nr 33 (Doorenbos and Kassam, 1979) represented an important source to determine the yield response to water of field, vegetable and tree crops, through the following equation:

Eq. 1

where Yx and Ya are the maximum and actual yield, ETx and ETa are the maximum and actual evapotranspiration, and ky is the proportionality factor between relative yield loss and relative reduction in evapotranspiration.

AquaCrop evolves from the previous Doorenbos and Kassam (1979) approach by separating (i) the ET into soil evaporation (E) and crop transpiration (Tr) and (ii) the final yield (Y) into biomass (B) and harvest index (HI). The separation of ET into E and Tr avoids the confounding effect of the non-productive consumptive use of water (E). This is important especially during incomplete ground cover. The separation of Y into B and HI allows the distinction of the basic functional relations between environment and B from those between environment and HI. These relations are in fact fundamentally different and their use avoids the confounding effects of water stress on B and on HI. The changes described led to the following equation at the core of the AquaCrop growth engine:

B = WP · ΣTr Eq. 2

where Tr is the crop transpiration (in mm) and WP is the Water Productivity parameter 2 (kg of biomass per m and per mm of cumulated water transpired over the time period in which the biomass is produced). This step from Eq. 1 to Eq. 2 has a fundamental implication for the robustness of the model due to the conservative behavior of WP (Steduto et al., 2007). It is worth noticing, though, that both equations are different expressions of a water-driven growth-engine in terms of crop modeling design (Steduto, 2003). The other main change from Eq. 1 to AquaCrop is in the time scale used for each one. In the case of Eq. 1, the relationship is used seasonally or for long periods (of the order of months), while in the case of Eq. 2 the relationship is used for daily time steps, a period that is closer to the time scale of crop responses to water deficits.

The main components included in AquaCrop to calculate crop growth are displayed in Figure 3: • Climate • Crop • Soil • Field management • Irrigation management

These five components will be discussed here shortly in the following sections. More details can be found in the AquaCrop documentation (Raes et al., 2009).

Figure 3. Overview of AuqaCrop showing the most relevant components.

2.3.2 Climate

The minimum weather data requirements of AquaCrop include the following five parameters: • daily minimum air temperatures • daily maximum air temperatures • daily rainfall • daily evaporative demand of the atmosphere expressed as reference evapotranspiration (ETo) • mean annual carbon dioxide concentration in the bulk atmosphere

The reference evapotranspiration (ETo or ETref) is the estimation of the evapotranspiration from the "reference surface." The reference surface is a hypothetical grass reference crop closely resembling an extensive surface of green, well-watered grass of uniform height, actively growing and completely shading the ground. In the model the crop evapotransipiration is calculated from the reference evapotranspiration by multiplying the reference evaportranspiration with a crop- specific crop coefficient (Kc value).

The reference evapotranspiration is, in contrast to CropWat, not calculated by AquaCrop itself, but is a required input parameter. This enables the user to apply whatever ETo method based on common practice in a certain region and/or availability of data. From the various options to calculate ETo reference is made to the Penman- Monteith method as described by FAO (Allen et al., 1998; 2006). The same publication makes also reference to the Hargreaves method in case of data shortage. In this application the latter one was used.

AquaCrop calculations are performed always at a daily time-step. However, input is not required at a daily time-step, but can also be provided at 10-daily or monthly intervals. The model itself interpolates these data to daily time steps. The only

exception is the CO2 levels which should be provided at annual time-step and are considered to be constant during the year.

2.3.3 Crop growth

AquaCrop considers five major components and associated dynamic responses which are used to simulate crop growth and yield development: • phenology • aerial canopy • rooting depth • biomass production • harvestable yield

As mentioned earlier, AquaCrop strengths are on the crop responses to water stress. If water is limiting this will have an impact on the following three crop growth processes: • reduction of the canopy expansion rate (typically during initial growth) • acceleration of senescence (typically during completed and late growth) • closure of stomata (typically during completed growth)

Finally, the model has two options for crop growth and development processes: • calendar based: the user has to specify planting/sowing data • thermal based on Growing Degree Days (GDD): the model determines when planting-sowing starts.

2.3.4 Soil systems

AquaCrop is flexible in terms of description of the soil system. Special features: • Up to five horizons • Hydraulic characteristics: o hydraulic conductivity at saturation o volumetric water content at saturation o field capacity o wilting point • Soil fertility can be defined as additional stress on crop growth influenced by: o water productivity parameter o the canopy growth development o maximum canopy cover o rate of decline in green canopy during senescence.

AquaCrop separates soil evaporation (E) from crop transpiration (Tr). The simulation of Tr is based on: • Reference evapotranspiration • Soil moisture content • Rooting depth

Simulation of soil evaporation depends on: • Reference evapotranspiration • Soil moisture content • Mulching • Canopy cover • Partial wetting by localized irrigation • Shading of the ground by the canopy

2.3.5 Field management

Characteristics of general field management can be specified and are reflecting two groups of field management aspects: soil fertility levels and practices that affect the soil water balance. In terms of fertility levels one can select from pre-defined levels (non limiting, near optimal, moderate and poor) or specify parameters obtained from calibration. Field management options influencing the soil water balance that can be specified in AquaCrop are mulching, runoff reduction and soil bunds.

2.3.6 Irrigation management

Simulation of irrigation management is one of the strengths of AquaCrop with the following options: • rainfed-agriculture (no irrigation) • sprinkler irrigation • drip irrigation • surface irrigation by basin • surface irrigation by border • surface irrigation by furrow

Scheduling of irrigation can be simulated as • Fixed timing • Depletion of soil water

Irrigation application amount can be defined as: • Fixed depth • Back to field capacity

2.4 Impact of Climate Change

The impact of climate change on crop production has many components. Some of these components can be assessed with the AquaCrop model, while other components require other approaches. The impact of climate change on crop production that are assessed using the AquaCrop model are in summary:

• Changes in precipitation • Changes in crop water demand due to changes in reference evapotranspiration • Changes in crop base temperature • Changes in crop upper temperature • Changes in irrigation applications

Other impact of climate change that require a more expert view approach are: • Impact on pest and diseases • Impact on weeds

The current study is based on representative areas in the five countries. These are: • Kazakhstan (KAZ): Kyzylorda • Kyrgyzstan (KGZ): Fergana Valley • Tajikistan (TAJ): Khatlon • Turkmenistan (TKM): Akhal • Uzbekistan (UZB): Karakalpakstan

Climate data for this preliminary study have been extracted from the WEAP analysis. However, instead of analyzing five different Global Circulation Models (GCMs) results of the average GCM were used: MIROC3.2 HIRES (Model for Interdisciplinary Research On Climate as developed by the Atmosphere and Ocean Research Institute, University of Tokyo). As emission scenario also the A1B has been used here.

Current and future climate conditions are presented in the following Figures and Tables. Overall it can be concluded that for the five selected locations changes in precipitation are quite limited, but that changes in reference evapotranspiration are substantial.

Figure 4. Monthly climate data for KAZ (2001-2010).

Table 1. Impact of climate change on precipitation and ETref in KAZ (GCM = MIROC). Period Precipitation ETref Precipitation ETref mm/y mm/y % from current 2001-2010 237 2285 2041-2050 251 2425 5.9 6.1

Figure 5. Monthly climate data for KGZ (2001-2010).

Table 2. Impact of climate change on precipitation and ETref in KGZ (GCM = MIROC). Period Precipitation ETref Precipitation ETref mm/y mm/y % from current 2001-2010 273 2619 2041-2050 284 2759 3.9 5.3

Figure 6. Monthly climate data for TAJ (2001-2010).

Table 3. Impact of climate change on precipitation and ETref in TAJ (GCM = MIROC). Period Precipitation ETref Precipitation ETref mm/y mm/y % from current 2001-2010 329 2413 2041-2050 331 2551 0.5 5.7

Figure 7. Monthly climate data for TKM (2001-2010).

Table 4. Impact of climate change on precipitation and ETref in TKM (GCM = MIROC). Period Precipitation ETref Precipitation ETref mm/y mm/y % from current 2001-2010 197 2632 2041-2050 190 2769 -3.3 5.2

Figure 8. Monthly climate data for UZB (2001-2010).

Table 5. Impact of climate change on precipitation and ETref in UZB (GCM = MIROC). Period Precipitation ETref Precipitation ETref mm/y mm/y % from current 2001-2010 193 2561 2041-2050 199 2694 2.7 5.2

2.5 Crops

The current study focusses on the dominant crop in the five countries: wheat. Wheat is grown over vast areas ( Table 6) and yields have shown to increase substantially over the last decade (Figure 9).

Biomass production and yields are calculated by AquaCrop, like almost all other crop growth models, as dry matter. In farm management practice and crop statistics however, yields are always expressed as fresh yields. On average wheat has a dry matter content of 87%, so about 13% moisture is included in the fresh yield. In order to convert AquaCrops results into fresh yields, one has to divide by 0.87. E.g. • 1000 kg dry matter • 1000 / 87% = 1149 kg fresh • 1149 * 13% = 149 kg moist

In this report all yields are presented as fresh yields to ensure proper dissemination amongst stakeholders and policy makers.

The total number of analysis using AquaCrop in this study is 100 years. For each country two simulations of each 10 years have been done: current (2001-2010) and future (2041-2050)

Table 6. Wheat area and yields according to FAOstat for the year 2010. Wheat area in % reflects the area wheat over total arable area in the country. Kazakhstan Kyrgyzstan Tajikistan Turkmenistan Uzbekistan Wheat Area (ha) 13,138,000 375,000 34,2566 850,000 1,420,000 Wheat Area (%) 56 29 46 46 33 Wheat Yield (kg/ha) 965 2246 2369 3172 4624

Figure 9. Wheat yields in the five countries according to FAOstat.

3 RESULTS AND DISCUSSIONS

Results of this preliminary analysis on the impact of climate change on the dominant crop in the Aral Sea basin, wheat, should be considered as indicative rather than fully conclusive. In this study only one representative area in each of the five countries (Kazakhstan, Kyrgyzstan. Tajikistan, Turkmenistan, and Uzbekistan) has been analyzed. It is well known that substantial differences within each of these countries exist in terms of physical setting, farmer practices and socio-economic conditions.

In general it can be concluded that yields are expected to reduce quite substantially. Table 7 presents the results from the AquaCrop simulations for the representative area in each of the five countries for the A1B MIROC projection. Especially for KAZ, TKM and also for UZB severe wheat reductions can be expected. The main reasons for these reductions in yields are an increased water demand and a higher temperature during the growing season. The first will reduce crop yields as more water stress might be observed. The latter has impact on the crop physiology that yield will be reduced if temperatures exceed critical levels.

Interesting is also to explore the annual variation in yields currently and in the future. In Figure 10 to Figure 14 this annual variation in wheat yields is presented as so- called Box-Whisker plots. In these plots are the median, the 25 and 75 quartiles as well as the minimum and maximum yields plotted over a period of 10 years. Again it is clear that median yields are reducing but also that the variation in yields between years is increasing.

The preliminary analysis does not take into consideration any adaptation measures. However several measures could be considered and typical examples will be listed here: • Increase irrigation applications, in case sufficient water is available. • Plant breeding programs to develop drought tolerant and high temperature resistant crop varieties. • Change in cropping patterns. • Improve overall soil fertility conditions. • Improve irrigation practices. • Reduce soil evaporation.

The final conclusion and recommendation from this preliminary scoping study can be summarized as: • Results as presented here are obtained without any validation of the AquaCrop model, so results should be considered as indicative. • It is clear that climate change will have a negative impact on yields in the region. The main reasons are increase in evaporative demand from the atmosphere and temperatures increasing the optimum one for wheat. • Following up analysis should be setup along the following lines: (i) expand to other areas in the countries, (ii) expand to other crops, (iii) improve/validate/calibrate AquaCrop for the specific conditions in the countries, (iv) include an ensemble of GCMs to get a band-width of expected yields, (v) include feedback in terms of water availability from the water resources modeling, and (vi) analyze potential adaptation measures.

Table 7. Impact of climate change on wheat yields based on AquaCrop simulations. Current is average simulated for 2001-2010 and Future for 2041-2050 under the A1B MIROC climate projection. Current Future Difference (ton/ha) (ton/ha) (%) Kazakhstan 0.937 0.425 -55 Kyrgyzstan 2.261 2.071 -8 Tajikistan 2.382 2.071 -13 Turkmenistan 3.153 2.934 -7 Uzbekistan 4.537 3.641 -20

Figure 10. Impact of climate change on wheat yields based on AquaCrop simulations for KAZ. Base is average simulated for 2001-2010 and CC for 2041-2050 under the A1B MIROC climate projection. Median, upper and lower quartiles as well as minimum and maximum yields in 10 years are shown.

Figure 11. Impact of climate change on wheat yields based on AquaCrop simulations for KGZ. Base is average simulated for 2001-2010 and CC for 2041-2050 under the A1B MIROC climate projection. Median, upper and lower quartiles as well as minimum and maximum yields in 10 years are shown.

Figure 12. Impact of climate change on wheat yields based on AquaCrop simulations for TAJ. Base is average simulated for 2001-2010 and CC for 2041-2050 under the A1B MIROC climate projection. Median, upper and lower quartiles as well as minimum and maximum yields in 10 years are shown.

Figure 13. Impact of climate change on wheat yields based on AquaCrop simulations for TKM. Base is average simulated for 2001-2010 and CC for 2041-2050 under the A1B MIROC climate projection. Median, upper and lower quartiles as well as minimum and maximum yields in 10 years are shown.

Figure 14. Impact of climate change on wheat yields based on AquaCrop simulations for UZB. Base is average simulated for 2001-2010 and CC for 2041-2050 under the A1B MIROC climate projection. Median, upper and lower quartiles as well as minimum and maximum yields in 10 years are shown.

REFERENCES

Acevedo, E. P. Silva, H. Silva. Wheat growth and physiology. In: B.C. Curits, S. Rajaram, H.G. Macpherson, Bread wheat. FAO. http://www.fao.org/docrep/006/y4011e/y4011e06.htm Aerts, J. and P. Droogers (Eds). 2004. Climate change in contrasting river basins: Adaptation strategies for water, food, and environment. CABI Books, Londen. 288 p. Afonin, A.N.; S.L. Greene; N.I. Dzyubenko, A.N. Frolov (eds.). 2009. Interactive Agricultural Ecological Atlas of Russia and Neighboring Countries. Economic Plants and their Diseases, Pests and Weeds [DVD]. http://www.agroatlas.ru.http://www.agroatlas.ru/en/80/ Allen, R.G., W.O. Pruitt, J.L. Wright, T.A. Howell, F. Ventura, R. Snyder, D. Itenfisu, P. Steduto, J. Berengena, J. B. Yrisarry, M. Smith, L.S. Pereira, D. Raes, A. Perrier, I. Alves, I. Walter, R. Elliott. 2006. A recommendation on standardized surface resistance for hourly calculation of reference ETo by the FAO56 Penman-Monteith method. Agricultural Water Management 81:1–22. Allen, R.G.; Pereira, L.S.; Raes, D.; Smith, M. 1998. Crop evapotranspiration - Guidelines for computing crop water requirements - FAO Irrigation and drainage paper 56. Amatya, D.M., R.W. Skaggs, J.D. Gregory. 1995. Comparison of methods for estimating REF-ET. Journal of irrigation and drainage engineering, Vol. 121, No. 6. Bazzaz FA, Sombroek WG (1996) Global Climate Change and Agricultural Production: Direct and Indirect Effects of Changing Hydrological, Pedological, and Plant Physiological Processes. Food and Agriculture Organization of the United Nations/Wiley, Chichester, UK/New York Doorenbos, J., Kassam, A.H., 1979. Yield response to water. FAO Irrigation and Drain, Paper 33. Rome, Italy,193 pp. Hunink, J., P. Droogers. 2011a. Climate Change Impact Assessment on Crop Production in Uzbekistan. World Bank Study on Reducing Vulnerability to Climate Change in Europe and Central Asia (ECA) Agricultural Systems. Report FutureWater 106. Hunink, J., P. Droogers. 2011b. Climate Change Impact Assessment on Crop Production in Albania. World Bank Study on Reducing Vulnerability to Climate Change in Europe and Central Asia (ECA) Agricultural Systems. Report FutureWater 105. Maas, E. V. 1990. Crop salt tolerance. In “Agri. Salinity Assessment and Management,” (K. K. Tanji Ed), ASCE Manuals & Reports on Engineering No. 7 1, pp. 2622304, ASCE, New York. Morgounov, A. M. Karabayev, D. Bedoshvili, and H.-J. Braun. 2001. Improving Wheat Production in Central Asia and the Caucasus. In: CIMMYT. Research Highlights of the CIMMYT Wheat Program, 1999-2000. Mexico, D.F. Parry M.L,. Rosenzweig C, Iglesias A, Livermore M, Fischer G, 2004. Effects of climate change on global food production under SRES emissions and socio-economic scenarios. Global Environ. Change. 14, 53–67. Raes, D., P. Steduto, T.C. Hsiao, and E. Fereres. 2009. AquaCrop—The FAO Crop Model to Simulate Yield Response to Water: II. Main Algorithms and Software Description. Agron J: 101: 438–447. Sacks, W.J., D. Deryng, J.A. Foley, and N. Ramankutty. 2010. Crop planting dates: an analysis of global patterns. Global Ecology and Biogeography 19, 607-620. DOI: 10.1111/j.1466-8238.2010.00551.x Steduto, P. 2003. Biomass Water-Productivity. Comparing the Growth-Engines of Crop Models. FAO Expert Consultation on Crop Water Productivity Under Deficient Water Supply, 26 - 28 February 2003, Rome, Italy. Steduto, P. T.C. Hsiao, D. Raes, and E. Fereres. 2009. AquaCrop—The FAO Crop Model to Simulate Yield Response to Water: I. Concepts and Underlying Principles. Agron J: 101: 426–437 Steduto, P., T.C. Hsiao, and E. Fereres. 2007. On the conservative behavior of biomass water productivity. Irrig. Sci. 25:189–207. Tubiello FN, Soussana JF, Howden SM, 2007. Crop and pasture response to climate change. Proceedings of the National Academy of Sciences, USA 2007b;104:19686-19690. Wand, S.J.E., Midgley, G.F., Jones, M.H. and Curtis, P.S. 1999. Responses of wild C4 and C3 grass (Poaceae) species to elevated atmospheric CO2 concentration: a meta-analytic test of current theories and perceptions. Global Change Biology 5: 723-741. You, L., Z. Guo, J. Koo, W. Ojo, K. Sebastian, M.T. Tenorio, S. Wood, U. Wood-Sichra. Spatial Produciton Allocation Model (SPAM) 2000 Version 3 Release 1. http://MapSPAM.info.