Impacts of Land Use Land Cover and Climate Changes on Soil Erosion and Hydrology in Muga Watershed, Abbay River Basin, Ethiopia

DSpace Institution DSpace Repository http://dspace.org

Geography and Environmental Studies Thesis and Dissertations

2021-07-09 IMPACTS OF LAND USE LAND COVER AND CLIMATE CHANGES ON SOIL EROSION AND HYDROLOGY IN MUGA WATERSHED, ABBAY RIVER BASIN, ETHIOPIA

BELAY, TATEK http://ir.bdu.edu.et/handle/123456789/12187 Downloaded from DSpace Repository, DSpace Institution's institutional repository

BAHIR DAR UNIVERSITY FACULTY OF SOCIAL SCIENCE DEPARTMENT OF GEOGRAPHY AND ENVIRONMENTAL STUDIES

IMPACTS OF LAND USE LAND COVER AND CLIMATE CHANGES ON SOIL EROSION AND HYDROLOGY IN MUGA WATERSHED, ABBAY RIVER BASIN, ETHIOPIA

BY TATEK BELAY TEGEGNE

JUNE, 2021 BAHIR DAR, ETHIOPIA

BAHIR DAR UNIVERSITY FACULTY OF SOCIAL SCIENCE DEPARTMENT OF GEOGRAPHY AND ENVIRONMENTAL STUDIES

IMPACTS OF LAND USE LAND COVER AND CLIMATE CHANGES ON SOIL EROSION AND HYDROLOGY IN MUGA WATERSHED, ABBAY RIVER BASIN, ETHIOPIA

A DISSERTATION SUBMITTED TO

THE DEPARTMENT OF GEOGRAPHY AND ENVIRONMENTAL STUDIES, FACULTY OF SOCIAL SCIENCE, IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN GEOGRAPHY AND ENVIRONMENTAL STUDIES (SPECIALIZATION IN ENVIRONMENT AND NATURAL RESOURCE MANAGEMENT)

TATEK BELAY TEGEGNE

ADVISOR: DANIEL AYALEW (PhD, ASSOCIATE PROFESSOR)

JUNE, 2021 BAHIR DAR, ETHIOPIA

DECLARATION

I hereby declare that this PhD dissertation entitled “Impacts of Land Use Land Cover and Climate Changes on Soil Erosion and Hydrology in Muga Watershed, Abbay River Basin, Ethiopia” is my own work and has not been submitted to any other University for the award of any degree. Furthermore, all the sources used in the study are properly acknowledged and cited.

Tatek Belay Tegegne 16/06/2021

Abstract

Land use/ land cover change causes unprecedented changes on the surrounding environment at different spatial and temporal scales. Muga watershed of the Abbay River Basin is characterized by deterioration with mismanagement of natural resources. Thus, this study was aimed at detecting the magnitude and pattern of land use/land cover changes and assessing drivers of changes over the last three decades (1985 to 2017) in the Muga watershed. Synergy of Landsat imageries (1985, 2002, and 2017), household survey, focus group discussion, key informant interview and field observation were used to detect changes and drivers of changes. Land use/land cover changes in the study watershed were detected using digital image analysis techniques, and their socioeconomic and biophysical drivers were analyzed using descriptive statistics. The change detection results exhibited a significant increasing trend of cultivated land and urban areas by 12 and 270%, respectively, from 1985 to 2017. In contrast, grasslands, forest lands, and shrub-bushlands showed a declining trend of about 40, 21 and 12%, respectively. The results from the household survey showed that the expansion of cultivated land, cutting of trees for fuelwood and construction purposes, population growth, land tenure policy and climate variability were the most influential drivers of land use/ land cover changes in the study watershed. Thus, alternative sources of income for youths and landless peasants and integrated watershed management, which has paramount importance in maintaining economic and ecological benefits were suggested to alleviate the adverse effects of LULC changes in Muga watershed.

Soil erosion is one of the major threats in the Ethiopian highlands. In this study, soil erosion in the Muga watershed of the Abbay River Basin under historical and future climate and land use/ land cover (LULC) change was assessed. Future LULC was predicted based on LULC map of 1985, 2002, and 2017. LULC maps of the historical periods were delineated from Landsat images, and future LULC was predicted using the CA-Markov chain model. Precipitation for the future period was projected from six regional circulation models. The RUSLE model was used to estimate the current and future soil erosion rate in Muga watershed. The results of the study show that the average annual rate of soil erosion in the Muga watershed was increased from approximately 15 t ha-1 year-1 in 1985 to 19 t ha-1 year-1 in 2002 and 19.7 t ha-1 year-1 in 2017. If a proper measure against the LULC changes is not taken, the soil loss rate is expected to increase and reach about 20.7 t h-1 yr-1 in 2033. The results of the study also show that in the 2050s, soil erosion is projected to increase by 9.6% and 11.3% under RCP4.5 and RCP8.5, respectively compared with the baseline period. When both LULC and climate changes act together, the mean annual soil loss rate shows a rise of 13.2% and 15.7% in the future under RCP4.5 and RCP8.5, respectively, which is due to synergistic effects. The results of this study can be useful for formulating proper land use planning and investments to mitigate the adverse effect of LULC and climate change on soil loss. The study also demonstrated the importance of an integrated approach, assessing the combined impacts of LULC and climate change, for accurate estimation of soil losses.

LULC and climate change are two main factors affecting watershed hydrology by altering the hydrological response including streamflow, water yield, surface runoff and evapotranspiration. In this study, simulation scenarios were developed using the soil and water assessment tool (SWAT) model to assess the individual and combined effects of LULC and climate change on the hydrological processes in the Muga watershed. The partial least squares regression (PLSR) statistical method was used to assess the effects of changes in individual LULC types on hydrologic components. The results of this study showed that the SWAT model with NSE and PBIAS values of 0.79 and 8.2% for calibration, 0.82 and 9.3% for validation, respectively, performed very well in the simulation of the hydrological process of the watershed. The changes in LULC have also significant impacts on the streamflow, surface runoff, groundwater flow and lateral flow of the study watershed. The results of PLSR analysis showed that crop cultivation has an adverse effect on the water balance such as groundwater flow (GWQ), streamflow (ST), evapotranspiration (ET), and lateral flow (LQ). GWQ, LQ and ST showed a declining trend while ET showed an increasing trend due to the water extraction from the soil and the decrease infiltration capacity of the soil. In contrast, land covered with forest regulates GW, LQ, and ET. The results of the study evidenced the change in water balance of the study watershed due to changes in LULC changes. The results of this study also showed that the combined effects of LULC and climate change on the water balance is relatively higher than the impact of the LULC and climate change alone. Therefore, decision-makers and land use planners should pay attention to LULC and climate change as the main drivers of change in the hydrology of the study watershed. More effort should be focused on the prevention of further deforestation and promotion of afforestation or put other appropriate soil and water conservation in place as a practical management strategy to reduce degradation on the deforested parts of the watershed currently under intensive cultivation.

Keywords: Abbay River basin, Climate change, Climate Scenarios, Hydrology, Land use/ land Cover change, Model, Muga watershed, Soil erosion, Watershed.

ACKNOWLEDGEMENTS I have received various forms of assistance from many people in the course of producing this research. I am glad to use this opportunity to express my indebtedness to all of them. First and foremost is my supervisor Dr. Daniel Ayalew, Bahir Dar University, Ethiopia. His unreserved professional guidance and supervision helped me from the beginning to the end of this study. I am very grateful to him and sincerely appreciate his kindness as well as the confidence you had in me. These were sources of great encouragement for me.

I am grateful to the International Foundation for Science (IFS), which has funded a part of my Ph.D. research work. The fund was very helpful to cover various costs of the study and create a great encouragement for me. I would also like to thank Bahir Dar University and Debre Tabor University for funding this study. My acknowledgments also extended to the Ministry of Water, Irrigation, and Electricity (MoWIE) and the National Meteorological Agency of Ethiopia (NMA), which kindly provided with the daily weather and hydrological data for the study area. I would like to express my gratitude to the Department of Geography and Environmental Studies, Debre Tabor University, for granting me the study leave. I would like to thank the World Climate Research Program for its role to produce the CORDEX dataset and making it available on the Earth System Grid Federation (ESGF) web portals. My thanks and appreciation go to the farmers and development agents and the local administrators of the study watershed, who were involved in the discussion and data collection.

It gives me meaningful pleasure to thank all the staff members and PhD students of the Department of Geography and Environmental Studies, Bahir dar University, for their support and motivation during this study. I extend my deepest gratitude to my friends Tadele M., Asnake W., Sebsibachew A., Hawultu T., Abebaw A., Dr. Belay Z., Yalemwork M., Dagnew T., Mitiku S., Desalew D., Girum G., and Dagnachew M. for their constant encouragement and moral support.

I am very grateful to Liknaw Yirsaw (PhD) and Zewude Tamiru (PhD), language expert and Lecturer at Debre Markos University and Bahir Dar University, respectively, for giving their time to edit this dissertation with the necessary language improvement.

Lastly, my special gratitude goes to my family. In this regard, my beloved parents, Ato Belay Tegegne and W/ro Ajebush Minalu have played immense roles in the development of my career. I would like to extend my heartfelt gratitude to my sisters for their encouragement and emotional support in my life. I feel blessed and honored for having you in my life.

iii

TABLE OF CONTENTS Acknowledgements ...... iii Table of contents ...... iv List of Abbreviations and Acronyms ...... vii List of Figures ...... ix PART I: GENERAL INTRODUCTION OF THE STUDY ...... 1 Chapter 1 ...... 1 1. Introduction ...... 1 1.1. Background ...... 1 1.2. Statement of the problem ...... 3 1.3. Objectives of the Research of the study ...... 6 1.4. Research Questions ...... 6 1.5. Significance of the Study ...... 6 1.6. Scope and limitation of the Study ...... 7 1.7. Structure of the Thesis...... 7 Chapter 2 ...... 9 2. Literature Review...... 9 2.1. Land Use and Land Cover Change: Definition and Causes ...... 9 2.2. Land Use Land Cover Change and Drivers in Ethiopia ...... 11 2.3. Historical and Future Climate Change Scenarios ...... 14 2.4. Impacts of Land Use/ Land Cover and Climate Change on Soil Erosion and Hydrology . 17 2.4.1. Impacts of land use/ land cover change on soil erosion and hydrology ...... 17 2.4.2. Impacts of climate change on soil erosion and hydrology ...... 19 2.4.3. Combined impacts of land use/ land cover and climate change on soil erosion ...... 22 2.4.4. Combined impacts of land use/ land cover and climate change on hydrology ...... 23 2.6. Revised Universal Soil Loss Equation ...... 30 2.7. Soil and Water Assessment Tool (SWAT) ...... 31 2.8. Partial Least Square Regression (PLSR) Model ...... 33 2.9. Conceptual framework of the study ...... 35 Chapter 3 ...... 37 3. Description of the Study Area ...... 37 3.1. Description of the Study Area ...... 37 3.1.1. Location ...... 37 3.1.2. Climate ...... 38 3.1.3. Soils and geology ...... 38 3.1.4. Vegetations ...... 39 3.1.5. Population and socioeconomics ...... 39 Chapter 4 ...... 41 4. Research Methodology ...... 41 4.1. Overview of the Research Philosophy ...... 41

4.2. Research Design and Justifications ...... 42 4.3. Data Sources and Data collection ...... 43 4.3.1. Land use land cover data ...... 43 4.3.2. Climate and hydrological data ...... 43 4.3.3. Other spatial datasets ...... 44 4.3.4. Socioeconomic data ...... 44 4.4. Data Analysis ...... 45 4.4.1. Land use/ land cover change analysis ...... 45 4.4.2. Socioeconomic data analysis ...... 45 4.4.3. Future climate change projection ...... 46 4.4.4. Methods to assess impacts of LULC and climate change on soil erosion ...... 46 4.4.5. Methods to assess impacts of LULC and climate change on hydrological process ...... 46 PART II: RESULTS AND DISCUSSION ...... 47 Chapter 5 ...... 48 5. Land Use and Land Cover Dynamics and Drivers in the Muga Watershed, Abbay River Basin, Ethiopia…………………………………………………………………………………………..48 5.1. Introduction ...... 48 5.2. Material and Methods...... 51 5.2.1. Data sources and processes ...... 51 5.3. Results and Discussion ...... 56 5.3.1. Accuracy assessment ...... 56 5.3.2. Trends of land use land cover change in muga watershed ...... 57 5.3.3. Drivers of land use/ land cover dynamics in muga watershed ...... 63 5.4. Implications of Land Use Land Cover Change in the Muga Watershed ...... 68 5.5. Conclusion ...... 68 Chapter 6 ...... 70 6. Impact of Land Use/ Land Cover and Climate Change on Soil Erosion in Muga Watershed, , Ethiopia ...... 70 6.1. Introduction ...... 70 6.2. Materials and Methods ...... 73 6.2.1. Image processing and preparation of spatial drivers ...... 74 6.2.2. Climate Data ...... 75 6.2.5. Climate projection and bias correction ...... 83 6.2.6. Revised Universal Soil Loss Equation model ...... 86 6.3. Results ...... 93 6.3.1. Land use/ land cover change in Muga watershed (2017-2033) ...... 93 6.3.2. Future climate ...... 94 6.3.3. Impacts of land use/ land cover changes on soil erosion ...... 97 6.3.4. Impacts of climate change on soil erosion ...... 100 6.3.5. Combined impacts of future land use/ land cover and climate change on soil erosion 101

6.4. Discussion ...... 103 6.5. Conclusion ...... 105 Chapter 7 ...... 107 7. Impacts of Land Use/ Land Cover and Climate Change on Hydrology in Muga Watershed, Abbay River basin, Ethiopia ...... 107 7.1. Introduction ...... 107 7.2. Materials and Methods ...... 109 7.2.1. Data inputs ...... 109 7.2.2. The Soil Water Assessment Tool (SWAT) ...... 116 7.2.3. Partial least square regression (PLSR) Model ...... 117 7.3. Results and Discussion ...... 119 7.3.1. SWAT calibration and validation ...... 119 7.3.2. Impacts of land use/ land cover change on Muga watershed hydrology ...... 124 7.3.3. Impacts of individual land-use changes on Muga watershed hydrology ...... 128 7.3.4. Impact of climate change on hydrological components in Muga watershed ...... 135 7.3.5. The combined impacts of land use land cover and climate changes on hydrology in Muga watershed ...... 137 7.4. Uncertainties and limitations ...... 143 7.5. Conclusion ...... 143 PART III: CONCLUSIONS AND RECOMMENDATIONS ...... 145 Chapter 8 ...... 146 8. Conclusions and Recommendations...... 146 8.1. Conclusions ...... 146 8.2. Recommendations ...... 148 References ...... 150 Appendix...... 175

LIST OF ABBREVIATIONS AND ACRONYMS

ANRS Amhara National regional State BCEOM Egis Bceom International C Cropping and land cover factor CA-Markov Cellular -Automata Markov CMIP5 Coupled Model Inter-comparison Project 5 CN Curve Number CSA Central Statistical Agency DEM Digital Elevation Model ERDAS Earth Resource Data Analysis System ET Evaporation and transpiration ETM+ Enhanced Thematic Mapper FAO Food and Agricultural Organization GCM Global climate Model GCPs Ground Control Points GIS Geographic Information System GPS Global Positioning System GWQ Ground Water Runoff Ha Hectare HadCM3 Hadley Center Coupled Model version3 HRUs Hydrological Response Units IDW Inverse Distance Weight IFAD International Fund for Agricultural Development IPCC Intergovernmental Panel on Climate Change K Soil erodibility factor Klocation Kappa for location Kno Kappa for no information Kstandard Kappa for standard LULC Land Use Land Cover MCE Multi-Criteria Evaluation

vii

MEA Millennium Ecosystem Assessment MLC Maximum Likelihood Classification MoFED Ministry of Finance and Economic Development MoWIE Ministry of Water, Irrigation, and Electricity MSS Multi-Spectral Sensor NCAR National Center for Atmospheric Research NMA National Metrological Agency OLI Operational Land Imager P Erosion control practice factor PNSP Productive Safety net Programme Q2 Goodness of prediction 2 Q cum Cumulative goodness of prediction with a given number of factors RCP Representative Concentration Pathways RUSLE Revised Universal Soil Loss Equation SCS Soil Conservation Service SL Slope Length factor SRES Special Report on Emission Scenarios SRTM Shuttle Radar Topography Mission SUFI-2 Sequential Uncertainty Fitting SURQ Surface Runoff SWAT Soil and Water Assessment Tool TM Thematic Mapper UNEP United Nation Environmental Protection UNESCO United Nations Educational, Scientific and Cultural Organization USDA-ARS United States Department of Agriculture-Agricultural Research Service USLE Universal Soil Loss Equation WBISPP Woody Biomass Inventory and Strategic Planning Project WLE Weighted Liner Combination

viii

LIST OF FIGURES Figure 2.1 Conceptual framework on the driving factors of LULC change and associated effect ...... 14 Figure 2.2 Conceptual framework shows the impact of LULC and climate change on soil erosion and hydrology ...... 36 Figure 3.1 Map of the study area…………………………………………………………………………..37 Figure 3.2 Mean monthly rainfall and mean monthly temperature records of the study area ...... 38 Figure 5.1 Flowchart showing the procedures employed to arrive at the final LULC map ...... 53 Figure 5.2 Classified LULC maps for 1985, 2002, and 2017 in the Muga Watershed ...... 58 Figure 5.3 Land use and land covers and their proportionate area in 1985, 2002, and 2017 in the Muga watershed ...... 59 Figure 5.4 Drivers of LULC change in the Muga watershed ...... 66 Figure 6.1 Workflow of the methodology developed in this study ...... 74 Figure 6.2 Standardized biophysical factors of LULC in the study area...... 80 Figure 6.3 Schematic diagram for simulating the future LULC using CA-Markov chain Model ...... 82 Figure 6.4 Soil type (a) and soil erodibility factor map (b) of Muga watershed ...... 88 Figure 6.5 Slope length and steepness factor (LS-factor) map of Muga watershed ...... 90 Figure 6.6 The cover (C) factor map for the year 2017 and 2033 in the Muga watershed ...... 92 Figure 6.7 Conservation practices (P) factor map for the year 2017 and 2033 in Muga watershed ...... 92 Figure 6.8 The 2017 and 2033 LULC of the Muga Watershed...... 94 Figure 6.9 Estimated annual soil loss map of the study area in 1985 (a), 2002 (b), 2017 (c) and 2033 (d)...... 99 Figure 6.10 Soil erosion severity level in 2050 under RCP 4.5 (a) and RCP 8.5 (b)...... 101 Figure 7.1 Daily observed and RCMs simulated (before and after correction) ...... 115 Figure 7.2 Flowchart for LULC and climate change impact assessment using the Soil and Water Assessment Tool (SWAT) and PLSR in Muga watershed...... 119 Figure 7.3 Simulated and observed streamflow for calibration and validation periods ...... 124 Figure 7.4 Water balance of three historical and one future LULC simulated through SWAT using meteorological data from 1985 to 2017 (precipitation equals 1087 mm yr-1) ...... 125 Figure 7.5 Water balance components simulated by SWAT for two different periods: (i) historical climate data (1985-2017) using LULC 2017 and (ii) future climate data (2018) under RCP 4.5 and RCP 8.5 using LULC 2017...... 136 Figure 7.6 Predicted mean annual water balance components for different combinations of climate change and LULC scenarios ...... 142

LIST OF TABLES Table 4.1 List of time series Landsat data used for the study ...... 45 Table 4.2 Dataset sources and types ...... 45 Table 5.1 Land-use types of the study area ...... 54 Table 5.2 Distribution of total and sample respondents ...... 56 Table 5.3 Accuracy assessment results ...... 57 Table 5.4. LULC type in 1985, 2002, and 2017 and rate of change for the periods 1985-2002, 2002-2017, and 1985-2017 in ha and % in Muga watershed ...... 60 Table 5.5. LULC change transition matrices(hectare) for the initial & final reference (1985-2017) ...... 61 Table 5.6. Farmers’ perception and observed LULC changes from 1985-2017 ...... 63 Table 5.7 Drivers of LULC change in the Muga watershed ...... 65 Table 6.1 Dataset sources and types used for RUSLE model ...... 75 Table 6.2 Factors and constraints considered and their weights for predicting LULC conditions in the Muga watershed ...... 79 Table 6.3 Accuracy result of the simulated 2017 LULC ...... 82 Table 6.4 The Regional Climate Models (RCMs) in the CORDEX-Africa used in this study...... 84 Table 6.5 C-factor and P-factor values for the respective LULC classes of Muga watershed, Abbay River basin, Ethiopia ...... 91 Table 6.6 The area and proportion of each land use categories in the Muga watershed ...... 93 Table 6.7. Projected change in rainfall, minimum and maximum temperature of Muga watershed under two RCP climate scenarios for mean of six RCMs for reference and future period ...... 96 Table 6.8 The estimated mean annual soil erosion rate of each LULC type and the entire watershed in 1985, 2002, 2017 and 2033 in the Muga watershed ...... 98 Table 6.9 Annual soil erosion risk classes and area coverage under LULC map of 1985, 2002, 2017 and 2033 in Muga watershed ...... 99 Table 6.10 Annual soil erosion risk classes and area coverage under RCP 4.5 and RCP 8.5 in Muga watershed ...... 100 Table 6.11 The mean annual soil loss (t ha-1year-1) of Muga watershed under current and future LULCand climate ...... 102 Table 7.1. Mean annual minimum and maximum temperature (oc) and rainfall (mm) for the baseline period (1985-2017) and future period (2018-2050) of Muga watershed ...... 110 Table 7.2 The statistical measures of climate variables before bias correction and after bias correction for the Muga watershed ...... 114 Table 7.3 List of parameters with fitted values and global sensitivity results for daily flow ...... 122 Table 7.4. SWAT model evaluation parameters for streamflow in the Muga watershed ...... 124 Table 7.5 Average annual evapotranspiration, surface runoff, groundwater, water yield, and lateral flow change under different LULC scenarios and historical climate data in the Muga watershed ...... 128 Table 7.6 Pairwise Pearson correlation for changes in LULC variables and water components ...... 130 Table 7.7 Summary of the PLSR models for the water balance components in the Muga watershed ..... 132 Table 7.8 Variable importance for the projection (VIP) and partial least square regression (PLSR) weight (w) values for the hydrological components for the Muga watershed ...... 134 Table 7.9 PLSR coefficients presenting the effects of different land use types on hydrological components between 1985 and 2033 (projected) in the Muga watershed ...... 134 Table 7.10. Hydrological components simulated by SWAT under historical climate data sets (1985-2017) and future climate data (2018-2050) under RCP4.5 and 8.5 using LULC 2017 ...... 137 Table 7.11 Predicted mean annual water balance components for different combinations of climate change and LULC scenarios ...... 142

PART I: GENERAL INTRODUCTION OF THE STUDY

Chapter 1 1. Introduction 1.1. Background

Land resources, mainly soil and water, are the basis of agricultural products and services (Zeleke et al., 2006; Hurni et al., 2008). LULC and climate change are the two predominant driving factors that affect soil and water resources (Safriel, 2007). Recently, these threats become among the most important ecological problems worldwide (Leh et al., 2013; Nkonya et al., 2016). Extreme dependence on soil and water resources for agriculture leads to deforestation, overgrazing, climate variability, and the expansion of agriculture to peripheral areas and steep slopes, reducing agricultural productivity (Steenhuis et al., 2013). Consequently, the efforts to increase agricultural yields accelerate the degradation of soil and water resources (Haregeweyn et al., 2015).

Climate change itself has an unequivocal impact on water resources, agriculture, and food systems (Urama & Ozor, 2010; Waithaka et al., 2013; Adhikari et al., 2015). The Horn of Africa is one of the areas in Africa, and it varies significantly in time and space (Souverijns et al., 2016). According to Carty (2017), monsoon rainfalls were very low in the Horn of Africa for the past 60 years. The decline of precipitation in the region during the March-May “long rains” season, leading to drought and famine (Tierney et al., 2015).

In Ethiopia, temperatures have shown a markedly increasing trend over the last 3-5 decades, especially the minimum temperature (Billi, 2015). Based on the meteorological data obtained between 1951 and 2005, the Ethiopian Meteorological Agency report shows that the mean minimum and maximum temperature has increased by 0.13°C/decade and 0.37°c/decade, respectively (NMA, 2007). Furthermore, McSweeney et al. (2008) indicated that the country's mean annual temperature has increased by 0.28 °c/ decade between 1960 and 2006. The rise of temperature has been most rapid in June, August, and September at a rate of 0.32˚C per decade. Unlike temperature, most of the studies in Ethiopia showed that rainfall has no clear trend (positive and negative) (e.g., Mengistu et al., 2014; Alemayehu & Bewket, 2017; Abebe, 2017; Berhane et al., 2020; Admassu et al., 2012; Degefu & Bewket, 2014; Kiros et al., 2016; Tabari et al., 2015).

Like climate change, LULC change is the main component of environmental change in various regions of the world. For example, Sub-Saharan Africa has lost 16% of its forests and 5% of its grasslands in 25 years (Midekisa et al., 2017). In Ethiopia, the conversion of LULC from natural vegetation to farmlands, grazing lands, human settlements, and urban centers is common (e.g., Ariti et al., 2015; Demissie et al., 2017; Kindu et al., 2013; Mengistu et al., 2012; Miheretu & Yimer, 2018; Wubie et al., 2016). However, studies in different parts of the country indicated the heterogeneity of LULC changes in trend, pattern, type, and magnitude (e.g., Zeleke & Hurni, 2001; Bewket, 2002; Mengistu et al., 2012; Teferi et al., 2013; Gebreselassie et al., 2016; F. Demissie et al., 2017). For example, Mengistu et al. (2012) reported a decline in shrub-grassland and riverine trees, while plantation trees, annual crops, and bare/open grasslands were increased in the midwest escarpment of the Ethiopian Rift Valley from 1972 to 2004. On the other hand, Demissie et al. (2017) in their study in Libokemkem district of northwestern Ethiopia, were reported a decline in forests and grasslands over the last four decades, while agricultural land was increased.

LULC change combined with climate change affects soil and water resources of the highland areas of Ethiopian (Mango et al., 2011; Teferi et al., 2013; Setegn et al., 2014; Meshesha et al., 2016), and agricultural productivity (DeFries & Eshleman, 2004; Demessie, 2015). The change in precipitation, temperature and LULC affects water resources in the form of interception, evapotranspiration, runoff, evaporation, and surface infiltration; and soil in the form of erosion soil moisture status; thereby affecting the process of watershed hydrology (Bewket & Sterk, 2005). The problem of soil and water resource degradation is more severe in Ethiopian highlands. This is due to the high population density (about 88% of the Ethiopians live in highland areas), and their dependency on subsistence agriculture (Zeleke, 2010; Demessie, 2015).

To reverse the adverse impacts of climate change and LULC change on soil and water resource from degradation and improve livelihood by enhancing agricultural productivity, various strategies, programs, and policies have been made both at national and local levels. For example, after the 1975 land reform, the government initiated large-scale government conservation programs focused on afforestation (Shiferaw & Holden, 1998), Climate Resilient Green Economy strategy of Ethiopia (FDRE, 2011), the Productive Safety net Programme (PNSP; Gilligan et al., 2009), and the Growth and Transformation Plan (MoFED, 2010). Although the implementation of these strategies and programs has considerable encouraging results, there are still challenges to

achieve sustainable soil and water resource management in Ethiopia (Adimassu & Kessler, 2015; Gebremeskel et al., 2018).

Soil and water conservation, as well as the return of degraded watersheds to protective and productive systems, are essential to improve agriculture productivity and improve the livelihood of a community. This, in turn, requires an understating of the impacts of climate change and LULC change on soil and water resource for soil and water conservation, land use planning, and climate change adaptation interventions. However, in Muga watershed, one of the tributaries of the Abbay River basin originated from Choke Mountain; empirical studies on the impacts of LULC and climate changes on soil erosion and hydrological process at a local level were not conducted to establish context-specific interventions.

1.2. Statement of the problem

Ethiopia is a rich country in terms of biophysical resources such as biodiversity, relatively fertile soil, and water resources (Zeleke et al., 2006). However, soil degradation in the form of soil erosion is the primary and most widespread problem that threatens the sustainable use of agricultural lands in the country. Hence, Ethiopia losses nearly 1.5 million tons of crops annually from the highlands due to soil erosion (Tamene & Vlek, 2008).

LULC change affects the hydrological process (Birhanu et al., 2019) and soil erosion (Anache et al., 2018). Climate change also affects the hydrological process (Bewket & Sterk, 2005; Frankl et al., 2013; Geremew, 2013; Demessie, 2015; Taye et al., 2015), the management of water resource (Kim & Kaluarachchi, 2009; Jury & Funk, 2013; Schmidt & Zemadim, 2013), and soil erosion (Bewket & Sterk, 2005; Gessesse et al., 2009; Mengistu et al., 2015; Fenta et al., 2017). Therefore, it is vital to study the impacts of LULC and climate changes on soil erosion and water resource dynamics to provide useful information on sustainable watershed management and land use policies.

In Ethiopia, several studies were conducted in the Abbay River basin examine the impact of LULC and climate change on soil erosion (e.g., Chimdessa et al., 2019; Bogale, 2020; Hassen & Assen, 2018; Mengistu et al., 2015; Tadesse et al., 2017) and hydrological process (e.g., Bewket & Sterk, 2005; Gebrehiwot et al., 2010; Dile et al., 2013; Tekleab et al., 2013; Roth et al., 2018; Berihun et al., 2019; Gashaw et al., 2018; Woldesenbet et al., 2018; Mengistu et al., 2021; Birhanu et al., 2019).

Some of the previous studies assess the impact of either LULC change (e.g., Steenhuis et al., 2013; Demessie, 2015; Mengistu et al., 2015; Wubie et al., 2016; Haregeweyn et al., 2017; Moges & Bhat, 2017; Tadesse et al., 2017; Aneseyee et al., 2020) or climate change on soil erosion (e.g., Chimdessa et al., 2019; Girmay et al., 2021; Mengistu et al., 2015). Although several studies have been conducted on the individual and combined impacts of LULC and climate changes on soil erosion, there are limited studies on the individual and combined impacts of future LULC and climate change on soil erosion, and the context is not yet well understood in the Abbay River Basin.

In Abbay River basin, water resource is also highly vulnerable to both LULC and climate change (Berihun et al., 2019; Dibaba et al., 2020; Teklay et al., 2021; Woldesenbet et al., 2018), however, most studies focused on the impact of either LULC change (e.g., Andualem & Gebremariam, 2015; Belihu et al., 2020; Bewket & Sterk, 2005; Gashaw et al., 2018; Gebrehiwot et al., 2010; Tekleab et al., 2014; Woldesenbet et al., 2017) or climate change (Abdo et al., 2009; Ayele et al., 2016; Dile et al., 2013; Enyew et al., 2014; Taye et al., 2015; Tekleab et al., 2013) on soil erosion and hydrology (e.g., Mengistu et al., 2015; Moges et al., 2020; Negese, 2021; Yaekob et al., 2020).

Some studies showed that hydrological processes are more affected by climate change than changes in LULC (e.g., Dibaba et al., 2020; Getachew et al., 2021; Mekonnen et al., 2018), and other studies have found that the impacts of LULC change are more significant as compared to the impacts of climate change (Woldesenbet et al., 2018). The results of these studies did not provide conclusive impact of LULC and climate change on hydrological processes. For instance, the increase in surface runoff and total water yield due to LULC change was offset by the decline of the surface runoff and total water yield, due to climate change impacts (Dibaba et al., 2020). On the other hand, the increase in streamflow due to LULC change was amplified by wetting climate

scenarios (Woldesenbet et al., 2018). Very few studies also investigated the combined effect of future LULC and climate change on hydrology (e.g., Getachew et al., 2021; Teklay et al., 2021; Woldesenbet et al., 2018). However, the context is still not well understood at watershed level and the impact of LULC and climate change on soil erosion and hydrology remains a contentious issue and requires further research (Simane et al., 2013; Demessie, 2015). Hence, examine the impact of LULC and climate change on soil erosion and hydrology at a watershed level has a great practical significance (Berihun et al., 2019; Chimdessa et al., 2019; Legesse et al., 2010).

Large-scale irrigation and hydropower projects, including the Grand Ethiopian Renaissance Dam (GERD) is being constructed in the Abbay River basin. Consequentlly, assess LULC dynamics and drivers of the change; impacts of LULC and climate changes on soil erosion and hydrology at watershed level have important implications for the sustainability of large scale and small-scale water resource development projects. Which will be used as a base to choose appropriate climate change adaptation and mitigation measures.

This study was conducted in Muga watershed, which is one of the tributaries of the Abbay River Basin that originates from Choke Mountain. The watershed and its environs are severely affected by land degradation due to the severity of LULC and climate changes in the area. Moreover, small- scale irrigation projects along the Muga River are currently initiated to increase agricultural productivity as well as to improve the livelihood of the community. However, to date, no empirical studies were carried out in the study area concerning the past and future impacts of LULC and climate changes on soil erosion and hydrology. Therefore, this study is crucial and necessary for the sustainability of local and large-scale water resources development projects and management, protection and rehabilitation of the watershed from continued degradation and therefore for socioeconomic development.

1.3. Objectives of the Research of the study

The general objective of the study was to examine the impacts land use/ land cover and climate changes on soil erosion and hydrology in the Muga watershed of the Abbay River Basin, Ethiopia. The specific objectives of the study were to:

 Detect land use/ land cover change and identify the major drivers of the changes in Muga watershed, Abbay River Basin.  Assess the impacts of land use/ land cover and climate changes on soil erosion in the study watershed.  Examine the individual and combined impacts of historical land use/ land cover and climate change on hydrology in Muga watershed.  Examine the individual and combined impacts of future land use/ land cover and climate changes on hydrology in Muga watershed. 1.4. Research Questions  What are the patterns and major drivers of land use/ land cover changes in Muga watershed?  How have historical and projected land use/land cover and climate changes affected soil erosion of Muga watershed?  What impact do historical land use/ land cover and climate changes individually and together on the hydrology of Muga watershed?  What effect do future land use/ land cover and climate changes individually and together on the hydrology of Muga watershed? 1.5. Significance of the Study

Although the Muga watershed is relatively small, it has significant strategic importance for local, regional, and national development. Farmers who have agricultural land along the Muga River started to use water for irrigation during the dry season (Bega season). Furthermore, the local and regional governments also planned to undertake irrigation by constructing irrigation canals along the river to enhance agricultural productivity and the local community livelihoods. Therefore, this study provides valuable scientific insights to various stakeholders (e.g., local government, farmers, policymakers) to formulate and implement appropriate and effective watershed management strategies to minimize the undesirable effects of LULC and climate change on soil erosion and

hydrology. For the improvement of the study watershed relies on proper and timely interventions as well as as permanence of farmers current environmental friendly soil and water management practices. 1.6. Scope and limitation of the Study Muga watershed, part of Ethiopia's Abbay River basin, has a variety of environmental and socioeconomic issues that could be an active area of research. However, this study only considers current and future LULC and climate change conditions as environmental problems. Given the various biophysical and socio-economic impacts of LULC and climate change, this study focuses on trends of LULC change and drivers of the change; historical and future impacts of LULC and climate change scenarios on soil erosion and hydrology. Furthermore, land degradation is a major problem of watersheds located in the Abay basin, where the Muga watershed is located. However, this study is confined to Muga watershed, which has one streamflow gauge and measures the watershed upper parts (an area of about 423 km2). Therefore, this study is limited to 423 km2 of the watershed, which could have a better implication to study the entire area of the watershed for management. Another limitation of the study includes the use of one meteorological station located within the study watershed for analysis, which could have better explained the impacts of climate change on soil erosion and hydrology using the average of more than one station located within the study watershed.

1.7. Structure of the Thesis

This dissertation is organized into eight chapters, consisting of four research outputs from chapter five to seven. A brief description of the contents' outline is presented to simplify understanding the connection between the chapters.

Chapter one presents an overview of the dissertation: which includes the research background, description of the research problem, the research questions, the research objectives, significance and scope, and limitation of the dissertation.

Chapter two presents a literature review of the existing knowledge of the research topic, which includes concepts of LULC change, the drivers of LULC change, climate change, impacts of LULC, and climate change on soil erosion and hydrology, soil and hydrological models.

Chapter three is devoted to a description of the study site by presenting the location, topography, climate, soil, LULC, and socio-economic activity.

Chapter four presents the research methodology, which includes data sources and data collection, and data analysis.

Chapter five presents LULC dynamics and drivers. This chapter is paper 1 (objective 1). It discusses LULC change trends, drivers contributing to the LULC changes, and implications of LULC change in the study area. This was done through household surveys using a questionnaire, FGDs, key informants’ interviews, field observation, and image classification techniques using Landsat imageries.

Chapter six discusses the separate and combined impacts of LULC and climate change on soil erosion in the study watershed. It also highlights the LULC prediction using the CA Markov chain model and climate projection and bias correction. This was achieved through the Revised Universal Soil Loss Equation model.

Chapter seven presents the effects of LULC change on the hydrological process; the hydrological components' response to climate change in the Muga watershed. The hydrological response for LULC and climate change was performed using historical and predicted climate data and calibrated and validated hydrological model.

Chapter eight presents the conclusions and recommendations of the study.

Chapter 2 2. Literature Review 2.1. Land Use and Land Cover Change: Definition and Causes

Land cover and land use are often interrelated. This is logical because the two terms are closely related and to some extent even overlap. According to Mölders (2011), land cover is defined by vegetation, hydrological and geological features, and human-made land surface structures. Land- cover change can range from modifying the landscape character without affecting the overall classifications to the extreme case, where one land-cover type completely replaces another. Lambin et al. (2006) also define land cover in terms of the Earth’s land surface and immediate topographic features that include biota, soil, topography, and surface and groundwater. A similar definition, as defined by Schulze (2000), refers to the biophysical state of the Earth's surface and its immediate subsurfaces, such as cropland, forests, grasslands, settlements, recreation, water bodies, or mining.

Land cover can be changed by natural factors (e.g., long-term climate change or climatic persistence) and human activities through land cover conversion and modification, primarily for agricultural production and development purposes (Mölders, 2011). These conversions and alterations of land cover represent the concept of land use as distinct from land cover. Land use also refers to land cover, but in terms of its socio-economic purpose and intended use. The term land use may vary in nature and intensity depending on its purpose and the land's biophysical characteristics (Verheye, 2009). Lambin et al. (2006) also noted that land use has been defined as the purposes for which humans exploit the land cover. These include settlement, farming, grazing, rangelands, recreation, etc. Changes in land-use at any location may consist of either a shift to a different use or an intensification of the existing one (Mölders, 2011). Therefore, due to people's deliberate role in adapting natural land cover to their benefit, the term land use differs from land cover.

Approximately 40% of the Earth’s land is affected by agriculture, and 85% has anthropogenic influence (Sanderson et al., 2002). Over the past 40 years, land use and land cover changes have significantly impacted the biogeochemical cycle, leading to changes in surface atmospheric energy exchange, carbon and water cycle, soil quality, biodiversity, the ability of biological systems to

meet human needs, and ultimately climate at all scales (Overmars & Verburg, 2005). In addition, land use and land cover changes accelerate soil erosion rates and cause soil degradation and significant pollution of water bodies and aquatic ecosystems (Van Rompaey et al., 2007). There are many studies on land use and land cover change at the regional, national, and local levels.

The most dramatic LULC changes in Africa can be seen due to a reduction of forest cover, often due to agricultural expansion and urbanization. Furthermore, mining in Africa leads to changes in land use and land cover, directly through clearing of forests for mining activities and indirectly through the settlement of workers and the increasing demand for agricultural production in the region (UNEP, 2016). Deforestation is a major problem in Africa. Although the African continent alone accounted for 15.6% of the world’s forests in 2015, between 2010 and 2015 occurred, Africa lost 84% of the world’s forests (MacDicken et al., 2016). This is a large rate of deforestation for a tiny proportion of forest cover. Global population growth, economic growth, and large-scale investment in commercial agriculture are major drivers of deforestation and other land cover changes in Africa (UNEP, 2016).

African population is projected to increase by 115% between 2013 and 2050 (Montanarella et al., 2015). Increasing wealth and shifting dietary preferences to more livestock-based foods increase the need for farmland and put pressure on the food production systems and soil health (Montanarella et al., 2016). Therefore, unsustainable land use and agricultural practices such as the expansion of cropland, increase in the number of livestock, large-scale single cropping, and chemical fertilizers, as well as deforestation and increased use of natural resources, leading to greater loss of biodiversity, soil erosion, and soil fertility loss (Cebecauer & Hofierka, 2008; Montanarella et al., 2015; Borrelli et al., 2017).

Climate change contributes to soil and land degradation; some changes in land use and land cover are inevitable (UNEP, 2016). Such degraded land becomes useless for agriculture or natural vegetation and often turns into a wasteland. Conversely, changes such as deforestation, agricultural expansion, mining, and urbanization play a significant role in climate change. They reduce the sources of natural carbon sinks, causing pollution and greenhouse gas emissions, and accelerate soil degradation.

Like the rest of the world, East Africa is no exception to LULC dynamics. Eastern Africa relies heavily on rain-fed agriculture, making rural livelihoods and food security highly vulnerable to water availability changes. Land is the primary source of livelihood for East Africans, and there has been a steady decline in land acquisition per household. To meet this demand for land, the LULCC in the region has lost its natural forests due to human settlements, urban centers, farmlands, and grasslands (Maitima et al., 2009). Between 1990 and 2015, when forest cover in East Africa decreased by about 1% per year, the human population increased at an average annual rate of 2% (World Bank, 2017). LULC change results in a trade-oﬀs between provisioning of food and ﬁber for human consumption and minimizing adverse effects on other ecosystem services such as water quantity and quality (Mustard et al., 2012).

Food production is dependent on water resources, and therefore any likely impacts of LULC change on water resources have negative implications for food production. Eshleman (2004) found that the increase in water yield associated with forest cover loss depends on many factors. These include deforestation, the degree of forest removal, plant regeneration rates affect ET, climatic conditions, and physical properties of hydrogeology and catchment (Bosch & Hewlett, 1982; Whitehead & Robinson, 1993). The lack of control in the experimental study, which can explain the observed hydrological changes observed through causal mechanisms, was also noted as a factor influencing these changes (Stonestrom et al., 2009).

A study conducted by Maitima et al. (2009) reported that changes in East African land use led to changes in farmlands, grazing lands, human settlements, and urban centers at the expense of natural vegetation. These changes have resulted in deforestation, loss of biodiversity, and land degradation. Land resources in Kenya face major challenges, specifically due to population growth. This puts pressure on land resources and leads to improper land use practices such as intensification of agriculture, expansion of arable land, overgrazing, and tree harvesting for fuelwood. Consequently, environmental degradation ultimately reduces the productivity of natural resources (Miheretu & Yimer, 2018).

2.2. Land Use Land Cover Change and Drivers in Ethiopia

In Ethiopia, numerous studies on LULC change have been conducted and witnessed the massive LULC conversion, e.g., Zeleke & Hurni (2001); Hurni et al. (2005); Bewket (2002); Bewket &

Sterk (2005); Teferi et al. (2013); Geremew (2013); Sewnet (2015); Moges & Bhat (2018); Demissie et al. (2017); Miheretu & Yimer (2018); Wubie et al. (2016); Tadesse et al. (2017); (Hassen & Assen, 2018) in North-western Ethiopia; Ariti et al., (2015); Mengistu et al. (2012); Kindu et al. (2015); Ayele et al. (2016) in the Central Rift Valley of Ethiopia; Betru et al. (2019) in Western Ethiopia; Zewdu et al. (2016) in Southern Ethiopia; Tegene (2002); Tekle & Hedlund (2000); Gessesse & Bewket (2014) in central highlands of Ethiopia; and Tesfaye et al. (2014); Haregeweyn et al. (2012); Haregeweyn et al. (2015); Hishe et al. (2020); Birhane et al. (2019); Gebrelibanos & Assen (2015); Jacob et al. (2016) in Northern Ethiopia. These studies have shown that the direction, pattern, and extent of LULC changes in Ethiopia are heterogeneous.

Recent studies in different parts of Ethiopia have reported the decline of natural vegetation, including forests, shrub-bush lands, and a substantial expansion of cultivated land and settlement areas. For instance, Demissie et al. (2017) in the Libokemkem district; Hurni et al. (2005) in the Dembecha area of Gojjam; Tadesse et al. (2017) in Yezat watershed of northwestern Ethiopia; Teferi et al. (2013) in Jedeb watershed within the Blue Nile; Tegene (2002) in the Derekolli catchment of the South Wollo; Ariti et al. (2015) in the central rift valley of Ethiopia, Dessie & Kleman (2007) in the south-central rift valley of Ethiopia; and Kindu et al. (2015) in the Munesa- Shashemene landscape; Zewdu et al. (2016) in Sago irrigation farm, southern Ethiopia; Miheretu & Yimer (2018) in the Gelana sub-watershed; Tesfaye et al. (2014) in the Gilgel Tekeze catchment of Northern Ethiopia, reported a substantial expansion of cultivated land at the cost of forest, grassland, and shrubland, and this is a major cause of degradation and environmental change which is likely to have an impact on the future. According to Hurni et al. (2010), more than 90 % of the Ethiopian highlands were once cleared, and currently, about 20% of the area is covered by trees and the percentage of forest cover is less than 4%.

On the contrary, some of the previous studies carried out in the country reported improvements in forest cover due to community afforestation, land rehabilitation, and integrated watershed management activities (Bewket, 2002; Bantider et al., 2011; Gebrelibanos & Assen, 2015; Moges & Bhat, 2018). For example, Bantider et al. (2011) in the Eastern Escarpment of Wello; Gebrelibanos & Assen (2015) in the Hirmi watershed of Northern Ethiopia; Haregeweyn et al., (2012) in Enabered watershed of Northern Ethiopia; Bewket, (2002) in the Chemoga watershed of northwestern Ethiopia; Sewnet (2015); and Moges & Bhat (2018) in the Infranz watershed in

Northwestern Ethiopia reported expansion of forest cover at the expense of cropland and grassland. Mengistu et al. (2012) also reported a decline in bush and grasslands, plantation trees, crops, and bare land increased in the midwest escarpment of the Ethiopian rift valley from 1972 to 2004.

Several studies of LULC change conducted in Ethiopia have addressed LULC changes with a variety of drivers, including both proximate and underlying causes, such as population pressures and the associated demand for environmental resources (Tekle & Hedlund, 2000; Bewket, 2002; Tegene, 2002; Mengistu et al., 2012; Gessesse & Bewket, 2014; Kindu et al., 2015; Hailemariam et al., 2016; Wubie et al., 2016; Yalew et al., 2016; Demissie et al., 2017). For example, the studies by Tekle & Hedlund (2000) in Kalu district of Southern Wello of Ethiopia, Tegene (2002) in the Derekolli catchment of South Wollo zone of Ethiopia, and Kindu et al. (2015) in the south-central highlands of Ethiopia, identified expansion of settlement and cutting of trees for fuelwood and charcoal making as proximate causes of LULC changes. The slope and the availability of adequate infrastructures, such as roads and access to markets, were considered as the proximate drivers of the LULC changes in the upper Blue Nile basin (Yalew et al., 2016). Furthermore, population growth, climate variability, land tenure policices, drought, change of government, and poverty were recognized as the main underlining drivers of LULC change (Bewket, 2002; Hurni et al., 2005; Ariti et al., 2015; Zewdu et al., 2016; Hassen & Assen, 2018; Miheretu & Yimer, 2018).

In general, a lot of research has been done on the dynamics of LULC in different parts of the country. Research findings revealed that the extent, magnitude, and drivers of the change are noticeably inconsistent in space and time. Studies have shown that the main drivers of LULC changes across the country are mainly related to population growth, manifested primarily in land pressure for agriculture, even in areas where cultivation is nearly impossible, and socio-economic development.

Biophysical and socioeconomic trigging events

Proximate Causes

Mediating Factors Underlying Drivers Farm Factors Household Factors -Climate change - Size - Demographic composition -Economic factors - Soil quality - Farming background -Policy & institutional factors - Typography - Duration of residence LULC Change -Societal factors - Location - Family labor -Urbanization - Tenure type - Education -Demographic factors - Over cultivation -Income source - Over grazing

 Soil erosion  Change in hydrological process  Soil nutrient depletion  Loss of biodiversity  Decline of land productivity  Food insecurity  Incidence of poverty and migration

 Application of appropriate land use planning  Land use policy consideration  Appropriate conservation and rehabilitation  Investment on land and water resource management

Figure 2.1 Conceptual framework on the driving factors of LULC change and associated effect 2.3. Historical and Future Climate Change Scenarios The Intergovernmental Panel on Climate Change (IPCC) stated that there had been an increase in surface temperature worldwide in the past century. This has happened in two different phases, namely within the 1910s and 1940s (0.35 °c) and more intensely in the 1970s and recently (0.55 °c) (Parry et al., 2007). In the Coupled Model Intercomparison Project Phase 5 (CMIP5), future climate data were obtained in conjunction with the RCP (Representative Concentration Pathways) settings (Taylor et al., 2012). By 2100 four RCP settings with recommended CO2 concentrations reaching 421 ppm (RCP 2.6), 538 ppm (RCP 4.5), 670 ppm (RCP 6.0), and 936 ppm (RCP 8.5) were selected. Each RCP scenario describes a particular discharge trail and consequent radioactive forcing. Future climatic data were obtained, taking into account these parameters (Nazarenko et al., 2015).

Climate is the average weather over many decades in any one location. Climate variability and climate change are often used interchangeably, but they have distinctly different meanings. Climate variability refers to variations in time and spatial scales other than variations in average weather conditions and individual weather events (Parry et al., 2007). This includes the extremes and differences of monthly, seasonal, and annual values from the climatically expected value (temporal mean). Climate variability can cause extreme weather, leading to droughts, floods, and other natural weather events. The underlying trends in the climate can exacerbate these extremes.

The Intergovernmental Panel on Climate Change (IPCC) has defined it as a change in the climate state that can be identified by changes in the mean and variability of its characteristics, which persists for decades or longer. This refers to any change in the climate over time due to natural variability or human activities (Stocker et al., 2013). Another definition set out in the United Nations Framework Convention on Climate Change (UNFCCC) relates to a change of climate, which is directly or indirectly attributed to human activities that change the composition of the global atmosphere in addition to natural climate variability observed over comparable periods (Parry et al., 2007).

In Africa, many parts of the regions experience high interannual variability in precipitation, which has substantial impacts on humans, partly due to the increased reliance on rain-fed agriculture, which in many cases is already limited by a lack of precipitation. Efforts have been made to study how the African climate will change in the future, including several continent-wide studies (Solomon et al., 2007; Giannini et al., 2008). They showed that the models projected a decrease (increase) in precipitation in most subtropical and tropical regions.

Climate has been changing since then. Changes refer to the variability of the long-term trends in the climate or mean changes in temperature and rainfall that persist over an extended period (Field & Barros, 2014). A recent analysis by the intergovernmental panel for climate change indicates that the earth has warmed by approximately 0.6°C ± 0.2°C over the last century with locally and seasonally varying amounts (Parry et al., 2007).

Observed and projected future global annual average temperatures relative to 1986–2005, the observed warming from 1850-1900 to 1986-2005 is 0.61°C. The uneven warming system alters temperature gradients and regional wind patterns and precipitation. Changes in precipitation

patterns and intensity, increased water vapor in the atmosphere, and others have a significant natural variability from year to decade, which masks a long-term trend. According to a report (Stocker et al., 2013), the global surface temperature is estimated at the end of the 21st century and is likely to be over 1.5oc compared to 1850 to 1900 for all RCP scenarios except RCP2.6. It is likely to be greater than 2°c for RCP6.0 and RCP8.5 and more likely not exceed 2°C for RCP4.5. Similarly, a gradual increase in global precipitation is anticipated in the 21st century.

There is currently an increased concern to investigate the impacts of climate change on hydrological processes and changes in water availability and watershed hydrological responses. Increased evaporation with changes in precipitation characteristics affects runoff, frequency, and intensity of floods and droughts, soil moisture, and water supply. Besides, runoff production is influenced by various factors such as the soil surface condition and its vegetative cover (Brown et al., 2005; Zhang et al., 2012).

In Africa, the temperature is projected to rise with global mean temperature changes (Stocker et al., 2013); most of the continent is warming above the worldwide average. James & Washington (2013) have revealed that the ensemble means annual mean temperature change for Africa is 1.1°C at 1°C, 2.3°C at 2°C, 3.4°C at 3°C, and 4.3°C and 4°C in A1B, there is an approximately linear increase. These authors have shown that the precipitation projections exhibit more spatial, seasonal, and inter-model variability than temperature projections.

Ongoma & Chen (2017) conducted a study on the temporal and spatial changes in temperature and precipitation in East Africa between 1951 and 2010 showed the overall decreasing and increasing trends of rainfall and temperature, respectively. Most studies over Eastern Africa predict an increase in temperature of about 3oC. For example, Gebrechorkos et al. (2019) projected 1.5 0C (in the 2020s), 3.1oC (in 2050s), and up to 4.3oC (in 2080s) maximum temperature increase for large parts of Ethiopia, Kenya, and the northern and south-eastern parts of Tanzania; Abera et al. (2018) also reported an increase in maximum temperature by 1oC and 2oC, and in the near-term period and by the end of the century under RCP 4.5, while under RCP8.5, an increase in maximum temperature change is projected by 1oC and 3.5oC, in the near-term period and by the end of the century for central Ethiopia; Gebrechorkos et al. (2019) a maximum and minimum temperature will increase up to 3.7oC and 2.76oC, respectively throughout the 21 century in basins of East

Africa, while precipitation will be higher than during the baseline period throughout the 21 century in large parts of the region.

In Ethiopia, for example, precipitation will increase to more than 40 mm, 84 mm, and 70 mm in the 2020s, 2050s, and 2080s, respectively. Gizaw et al. (2017) report showed an increase of temperature (obtained from an ensemble median of 10 GCMs) in four major river basins of Ethiopia by a 2.3oC and 3.3oC in the 2050s and 2080s, respectively. In contrast, the mean annual precipitation is projected to increase by about 6% and 9% in the 2050s and 2080s, respectively. Gebrechorkos et al. (2019) also reported that the minimum temperature will be warmer (up to 6.1°C) than the reference period in the 2020s, 2050s, and 2080s under the RCPs in the East African region. According to Gabiri et al. (2020), by the end of the 21st century, East Africa's average temperature will rise by 1.7-5.4 °C and precipitation will rise by 5–20%.

In general, water resources will change drastically as temperature and precipitation change over time and space. As hydrological conditions vary regionally and locally, the effects of climate change on the local hydrological processes possibly differ between localities, even under the same climate scenarios. It is primarily at the regional and local scales that policy and technical measures can be taken to avoid or mitigate climate change adverse effects on the natural environment and society.

2.4. Impacts of Land Use/ Land Cover and Climate Change on Soil Erosion and Hydrology 2.4.1. Impacts of land use/ land cover change on soil erosion and hydrology

Land use changes have been caused by various human activities such as urbanization, deforestation and land-use conversion to agricultural land and forest management practices. In Africa and South America, deforestation has increased as more forests are converted to other land uses, mainly agriculture. In Africa, the rate of forest conversion to agriculture and pasture land is high. Between 1990 and 2010, 75 million hectares of forest was converted to agriculture and pasture (FAO, 2010). In East Africa, about 13 million hectares of the original forest was lost over the same 20 years, and the remaining forest is fragmented and continually under threat (FAO, 2010).

Various researchers have investigated the relationship between land use and hydrological processes (Bewket, 2002; Tadele & Förch, 2007; Zhang et al., 2014; Andualem & Gebremariam, 2015). For example, Bewket (2002) points out that land cover change such as forest cover leads to less streamflow and higher evapotranspiration than other land use types such as agriculture, pasture, grassland, and urban areas. Streamflow generation capacity also depends on the type and species of vegetation. On the contrary, Shi et al. (2013) showed that changes in land use (increased forest land) reduce surface water and streamflow and increase actual evapotranspiration in Xixian watershed, China.

In Ethiopia, the dynamic nature of land use is caused by rapid population pressures, poverty, poor agricultural practices, agricultural expansion, and climate change, which significantly impact Ethiopia's water resources (Melesse et al., 2011; Getachew & Melesse, 2012). Deforestation due to overpopulation pressure is the main form of land/ use cover change, which results in the expansion of farmland. On the same occasion, deforestation, which is associated with the use of marginal lands and steep slope with fragile soils and heavy seasonal rains make the highlands of the country vulnerable to soil erosion and can accelerate erosion process during the rainy season (Schmidt & Zemadim, 2013; Tamene et al., 2014). Besides, during the dry season, water scarcity and low water tables cause earlier perennial streams to be intermittent and affect agricultural production (Schmidt & Zemadim, 2013).

Setegn et al. (2010) showed that a quarter of Ethiopia's highland area is highly affected by soil erosion, followed by a large amount of soil being eroded from the area each year. Primarily, the problem of land degradation is severe in the Abbay River Basin, where it is not uncommon for land to be converted into agricultural areas. For example, Haregeweyn et al. (2017) noted that 39% of the total area of the basin is at risk of severe to very severe (>30 t ha-1yr−1). Therefore, the problem is serious because a land conversion has resulted in increased soil erosion, loss of soil fertility, and reduction in dry season flows (Melesse et al., 2011). The problem is closely related to population density, mismanagement of land use, and the lack of proper soil conservation measures (Setegn et al., 2011). Demissie et al. (2004) also studied the impacts of LULC in the Legedad reservoir watershed of Ethiopia. They found that there was a considerable impact on water quantity, erosion, and sediment yield. A study conducted by Getachew & Melesse (2012) showed that LULC change in Angerb watershed (Adjusting the Abbay River Basin) is responsible

for altering the hydrologic response of the watershed, such as quality and quantity of water flowing in a watershed. Tekleab et al. (2014) reported that the monthly variations in streamflow in the Jedeb watershed in the upper Blue Nile Basin increased by 45% in the 1990s and 2000s, while low streamflow decreased by 71%, which is attributed by LULC change.

2.4.2. Impacts of climate change on soil erosion and hydrology

The IPCC Fifth Assessment Report shows that climate change is occurring globally and continentally. There is high confidence in the continuation of anthropogenic climate change, as demonstrated by a wide range of future greenhouse gas emissions scenarios. Besides, there has been a growing awareness that climate change is accelerating due to human activities in recent years.

Potential effects of climate change include severe and irreversible changes in the climate system, which are expected to have severe impacts on basic human interests globally as global warming increases (Edenhofer, 2015). Changes in precipitation intensity and increases in extreme weather events are expected to negatively impact soil and water resources and agricultural production than positive impacts (Oluwatomiwa, 2014).

Although climate change affects soil and water resources both in developed and developing regions of the world, the severity of its influence is most likely region-specific and dependent on the ecosystem, topography, and management (Blanco-Canqui & Lal, 2008). Besides, economic shortage and limited accessibility to effective soil and water conservation measures are expected to exacerbate its effects in low-latitude and developing regions such as Africa and Asia (Blanco- Canqui & Lal, 2008; Elbehri, 2015). This exacerbates the existing imbalance between developed and developing countries (Elbehri, 2015). Developing countries and their economic development, such as crop and livestock production, resource supply, and other components of the agricultural system, may be severely affected by climate change (Case, 2006; Elbehri, 2015).

There is considerable evidence that ecosystems, soil erosion, soil moisture and streamflow changes, biodiversity, and agricultural systems have already responded to climate changes in recent decades by changing precipitation and temperature patterns. Consequently, global warming associated with changes in rainfall and evaporation can affect river flow, water balance, and other parameters such as total water availability, maximum and minimum flow rates, and soil moisture.

This would affect the rate of erosion, sediment concentration in water flow, sediment yields, irrigation, and hydropower production (Demessie, 2015; Oluwatomiwa, 2014).

Since soil detachment and transport by water depending on the amount and intensity of rainfall (Blanco-Canqui & Lal, 2008), temperature and precipitation changes are most likely to increase soil erosion potential in the future. The impact of climate and land cover changes on watershed dynamics has been studied worldwide (Blanco-Canqui & Lal, 2008; Li et al., 2011; Tan et al., 2015; Guo et al., 2016). From a watershed management point of view, the most important impacts are water resource changes and erosion response. These changes will affect the functioning of the watershed's ecosystem services, such as the provision of water and the control of erosion. As a result, increased soil erosion due to the new climate reduces grain and biomass production. The low vegetative cover and biomass inputs can increase erosion, soil degradation, and desertification (Blanco-Canqui & Lal, 2008).

According to Guo et al. (2016) and Tan et al. (2015), the influence of climate change on the annual streamflow is predominant. They noted that the shift in land cover has a moderate impact on the annual streamflow and strongly influences the seasonal stream flows. Ma et al. (2008) found that more than 64% of the average yearly decrease in streamflow in the Shiyang river basin is due to climate variability, mainly the reduction of precipitation. In contrast, Li et al. (2007) investigated that around 87% of the total decrease in mean annual streamflow from 1972 to 1997 in the Wuding River basin due to human activities such as soil and water conservation measures, the remaining 13% is associated with changes in precipitation and possible evaporation.

Although there is considerable uncertainty in the projected global temperature and precipitation patterns, Li et al. (2011) studied that under the conventional tillage, simulated precipitation and temperature change would increase by -2.6 to 17.4%, from 0.6 to 2.6 0C, and by 0.6 to 1.7 0C for the maximum and minimum temperature, respectively, between 2010-2039, compared with the current climate. Which, in turn, predicted a change from -5 to 195% for soil loss and from 10 to 130% for runoff.

In the United States of America, monthly projections for 1950-1999 and 2056-2085 projected by GCMs show that predicted changes in precipitation variance increase runoff by 15%-17% and soil loss by 10 and 19% under conservation and conventional tillage, respectively (Zhang et al., 2004).

They also stated that predicted increases in average temperature increased soil loss by 40% and 19% under conservation and conventional tillage, respectively. Another study was carried out in the western United States and showed no significant annual precipitation changes under the three emission scenarios. At the same time, projected the mean annual runoff and soil loss increased significantly between 1970 and 1996 (Zhang et al., 2012).

In Africa, the Sahel is one of the most vulnerable areas of climate change. The results of limited scientific research have been done to assess and quantify climate change effects on water resources and soil erosion. At the end of the 21st century, research conducted in the Sahel region of morocco, under different climate and land use change scenarios, reveals that the rate of erosion over four years on experimental plots showed spatially contrasted results (Simonneaux et al., 2015). They found out that although heavy rainfall exists in summer and autumn, the annual precipitations would decrease by 10 to14%. But in Africa, particularly in the Sahel and the Nile basin, precipitation changes will be more important than temperature changes: less rainfall causes less vegetation, which leads to overgrazing and less evaporation (van Dam, 2003). Changes in land use changes in Morocco could lead to much larger erosion (up to 250%), while climate change has increased sediment yield by 4.7-10.1% (Simonneaux et al., 2015).

In Ethiopia, various studies have been conducted to assess the effects of climate change on hydrology, e.g., Abdo et al. (2009); in the Lake Tana Basin; Melesse et al. (2011), Kim & Kaluarachchi (2009), and Setegn et al. (2014) in the Upper Blue Nile basin; Legesse et al. (2003) in South Central Ethiopia; Setegn et al. (2011) in the Northern Highlands of Ethiopia. The studies showed the effects of climate change on water resources, ecosystems, forests, soil erosion, irrigation, agricultural productivity, and hydropower development in the Ethiopian highlands.

In Abbay River basin, rainfall and runoff are highly uncertain, but runoff and streamflow may decrease due to higher temperatures while precipitation increases (Melesse et al., 2011). A study carried out in the Gilgel Abay catchment indicates that the modeled impact of climate change on the seasonal and monthly runoff variation will be more significant than the annual variation (Melesse et al., 2011). The study found that an increase in temperature of 2°C and a decrease in rainfall about 20%, more than 33% of the seasonal and annual reduction in runoff is expected in the 2080s. During the main rainy season, runoff is expected to decrease by 11.6% and 10.1% for A2 and B2 scenarios, respectively. Legesse et al. (2003) showed the impact of climate on water

resources at the catchment scale in south-central Ethiopia. They noted that a 10% reduction in precipitation resulted in a 30% decline in the simulated hydrological response of the catchment, while a 1.5 0c increase in air temperature decreases the discharge by 15% decrease.

2.4.3. Combined impacts of land use/ land cover and climate change on soil erosion

Recently, soil erosion has been a global environmental problem that poses a significant threat to the environment and the people. The problem is caused by human activities (e.g., deforestation, inappropriate farming practices, overgrazing, agricultural expansion, etc.) and climate change causing soil nutrient loss and land degradation. Soil erosion has become a pertinent global problem, the continuous decline in natural resources such as forests, grassland, and shrub- bushland, coupled with the effects of climate change.

LULC change is a major cause of soil loss, particularly in sub-Saharan Africa's upland watersheds (Seutloali & Beckedahl, 2015). Rapid changes in land use due to intensive agricultural practices, population pressures, poverty, and the absence of comprehensive land use policies and land tenure systems exacerbate the rates of soil erosion (Bewket, 2002; Hurni et al., 2010; Tadesse et al., 2017; Aneseyee et al., 2020). Soil erosion is also triggered by a combination of topographic factors and climate change and land cover patterns and soil characteristics (Gelagay & Minale, 2016; Lafforgue, 2016). Studies suggest that expansion of cultivated land at the expense of natural vegetation affects the natural environment in general and soil erosion rate in particular (Bantider et al., 2011; Wubie et al., 2016; Tadesse et al., 2017; Miheretu & Yimer, 2018). For example, Tadesse et al. (2017) noted that soil erosion was highest in cultivated lands, while the lowest was under forest land, indicating the importance of soil cover (C-factor of erosion) and land management (P factor). Therefore, continuous expansion of cultivated lands is most likely a major cause of land degradation, which further puts soil fertility at risk.

Over the past several years, researchers have advocated serious concerns about LULC and climate change and their impact on the present and possible future changes upon soils. The changing precipitation patterns due to climate dynamics have undoubtedly increased the risk of soil erosion, surface runoff, and associated environmental consequences. The amount and intensity of rainfall are the main climatic factor that controls soil erosion by water (Renard, 1997; Zhang et al., 2017). An increase in amount, duration, and intensity significantly increase soil detachment and erosion (Vaezi et al., 2017).

Both human activities and natural factors can provoke soil erosion. Natural factors such as precipitation, soil texture, organic matter, terrain slope, and vegetation coverage affect soil erosion (Mengistu et al., 2015; Haregeweyn et al., 2017; Taye et al., 2018; Yesuph & Dagnaw, 2019). Moreover, anthropogenic activities such as urbanization, intensive farming, deforestation, and overgrazing, can aggravate soil erosion (Bewket, 2002; Demessie, 2015; Kindu et al., 2015; Haregeweyn et al., 2015; Wubie et al., 2016; Haregeweyn et al., 2017; Miheretu & Yimer, 2018; Ebabu et al., 2019).

Several studies have quantified the effects of changes in LULC on soil erosion in different parts of the world at large and Ethiopia in particular (Leh et al., 2013; Simonneaux et al., 2015; Borrelli et al., 2017; Haregeweyn et al., 2017; Moges & Bhat, 2017; Tadesse et al., 2017; Zerihun et al., 2018; Ebabu et al., 2019). These studies show that land use and land cover changes are considered to be the major determinants of soil erosion. Climate change also affects soil erosion, mainly caused by changes in the amount and intensity of rainfall and rising temperatures. Therefore, climate change (mainly increased rainfall and temperature) associated with changes in LULC, such as the expansion of cultivated land at the expense of forest, grassland, and shrub-bushland, can exacerbate soil erosion.

2.4.4. Combined impacts of land use/ land cover and climate change on hydrology

Sustainable management of the earth’s surface, including LULC and climate change, is a major environmental problem to be addressed by society (Mustard et al., 2012). Land use changes with climate change are essential determinants of global environmental change and directly and indirectly impact the socio-economic and biophysical environments (Olson et al., 2008; Kim & Kaluarachchi, 2009). This has been evident in several sectors, such as agriculture (Kurukulasuriya & Rosenthal, 2013), health (He & Wu, 2019), and ecosystem and biodiversity (Oliver & Morecroft, 2014). Besides, LULC and climate changes can affect the spatial and temporal changes of watershed hydrology by changing the streamflow, surface runoff, baseflow and evapotranspiration process (Liu et al., 2017; Rahman et al., 2017). More recent studies (e.g., Demessie, 2015; Haregeweyn, et al., 2015; Woldesenbet et al., 2017; Gashaw et al., 2018; Berihun et al., 2019; Betru et al., 2019) showed that human activities in particular the expansion of agricultural land at the expense of forests, grasslands and shrub-bush lands possesses a serious threat to water resources, especially in the face of ubiquitous future climate change.

Pachauri et al. (2014) reported that the global average surface temperature and the frequency of heavy precipitation events are expected to increase in the middle of the 21st century. Temperature and precipitation changes have a direct impact on the amount of evapotranspiration as well as the quality and amount of the runoff. As a result, the local and temporal availability of water resources or the water balance, in general, can change significantly with a temperature change. Besides, land- use change also has a severe impact on water resources, mainly by dividing rainfall amount into interception, evapotranspiration, infiltration, and accumulation of soil moisture, thereby affecting the availability of watershed hydrology (Mishra et al., 2010; Mango et al., 2011).

In recent years, several studies have examined the combined impact of land-use and climate change on hydrologic processes (Serpa et al., 2015; Guo et al., 2016; Liu et al., 2017; Anache et al., 2018; Mekonnen et al., 2018; Aboelnour et al., 2019; Berihun et al., 2019; de Hipt et al., 2019). These studies' results invariably highlight water balance components that can be affected by land-use and climatic change. But, the relative impacts of land use and climate change may vary spatially due to the rate and magnitude of change in climate and land-use. For instance, some studies have found that climate change is having a greater impact on hydrological processes than land-use change (Mekonnen et al., 2018; Aboelnour et al., 2019), while other studies have shown the effects of land-use change are more pronounced (Mwangi et al., 2016; Yin et al., 2017).

The interaction between LULC and climate change is not linear (Jung et al., 2010). For example, some studies show that increased streamflow is caused by land-use changes augmented by wetting climate scenarios (Marhaento et al., 2018). Besides, Sunde et al. (2018) have shown that climate change can significantly offset the increase in runoff due to urbanization. This reveals the importance of considering the combination of LULC and climate change scenarios to understand the effects of environmental changes on water resources.

In Ethiopia, over the last four decades, there have been significant LULC changes and increases of temperature and variability in precipitation, both spatially and over time. Studies conducted on LULC dynamics in different parts of the country (e.g., Teferi et al., 2013; Gebremicael et al., 2019; Wubie et al., 2016; Demissie et al., 2017) revealed that the area coverage of agricultural land has been expanded extensively at the expense of natural forests, shrub-bushlands, and grasslands. LULC changes are one of the main drivers of hydrological changes in watershed areas by altering

interception rates, soil moisture, evapotranspiration (ET), infiltration, and groundwater, leading to changes in surface runoff, streamflow, and flood frequency (Nie et al., 2011; Geremew, 2013; Welde & Gebremariam, 2017; Woldesenbet et al., 2017; Birhanu et al., 2019; Teklay et al., 2019).

The impact of land use and cover changes on the hydrological process is not straightforward but rather complex to ensure any generalization as it is dependent on the scale of the watershed, seasons, climate, and soil conditions (Lambin et al., 2003). Changes in temperature and rainfall can affect water resources' spatial and temporal availability by altering the hydrological cycle (Bekele et al., 2019). According to the World Bank (2020), climate predictions for Ethiopia show a general increase in temperature and precipitation for all months, especially during the rainy season. Changes in both temperature and precipitation show significant regional variability, affecting the hydrology of watersheds (Ayele et al., 2016). Therefore, the effects of climate and land-use changes on watershed hydrology are of very practical importance, especially in a tropical region like Ethiopia, where LULC and climate are changing rapidly (Berihun et al., 2019; Chimdessa et al., 2019).

Climate change is one of the most difficult challenges humans and natural ecosystems face (Stocker et al., 2013). The effects of climate change and adaptation measures are seen as a major current global concern (Field & Barros, 2014). The change in mean and variability of temperature and precipitation affects both the environment and society in many ways (Field & Barros, 2014; Touma et al., 2015). For example, an increase in temperature and precipitation changes can alter the regional water balance and hydrological regime (Bolch et al., 2012). Global rises in surface temperature, variations in precipitation patterns in space and time, and predictable changes in these variations most likely occur over the next century (Kharin et al., 2013).

The Intergovernmental Panel on climate change has defined a series of Representative Concentration Pathways (RCP) scenarios for future climate projections based on the Coupled Model Intercomparison Project (CMIP5) (van Vuuren et al., 2011). They suggest a global average surface temperature rise of more than 2°C by the end of the century compared to the 30-year baseline period from 1986 to 2005. In particular, the average temperature is projected to rise by more than 1°C under a low-emission scenario (RCP 2.6), and over 4°C under an extreme scenario (RCP 8.5) (Knutti & Sedláček, 2013). The effects of climate change are perceived as increasing

over time (McCarthy et al., 2001; Parry et al., 2007; Field & Barros, 2014). Linear trends for 100 years from 1906 to 2005 shows that the average global temperature rises by an average of 0.74 °c (Parry et al., 2007).

The continent of Africa has been identified as vulnerable to climate change due to its vulnerability and low adaptability (Callaway, 2004). The region of East Africa is considered to be one of the most vulnerable regions in Africa. The effects of climate change, such as rising temperatures and unpredictable precipitation intensity and pattern are undeniably evident in this region (Adhikari et al., 2015; McDowell et al., 2016). East African countries are likely to experience adverse impacts from climate change due to their topographical settings (Haile et al., 2019; Ampaire et al., 2020). In East Africa, human and environmental stresses such as LULC changes associated with rapid urbanization and uncontrolled expansion of intensive agricultural production, adversely affects water availability, quality, and other ecosystem services and functions (Adhikari et al., 2015; Souverijns et al., 2016; Gebrechorkos et al., 2019). The negative impacts of changes in the climate change matrix and land use change are also exacerbated by other factors, notably worsening poverty and high population pressures, which are expected to increase food and water demand in the future (Waithaka et al., 2013). If the climate change observed over the last century persists into the future (Stocker et al., 2013), the potential impact on water and soil resources can increase in magnitude, diversity, and severity.

In the Horn of Africa, climate change and variability significantly influence the hydrological cycle and water availability. Therefore, it is essential to assess the impact of climate change on water resources to support building adaption measures to mitigate climate change effects. Many studies have focused on the potential effects of climate change on watershed hydrology, including precipitation, temperature, potential evapotranspiration, streamflow, and soil moisture (Setegn et al., 2011; Dile et al., 2013; Musau et al., 2015). Variations in precipitation have a direct effect on runoff, groundwater storage, frequency and severity of flooding, soil moisture, water supply for irrigation, and hydropower generation (Li et al., 2009; Tshimanga & Hughes, 2012). Higher temperatures increase evapotranspiration, which in turn reduces soil moisture, increases crop water demand, and reduces streamflow (Kusangaya et al., 2014).

2.5. Landuse change modeling

LULC change models are useful for understanding the outcomes of future conditions or future land use direction (Verburg et al., 2004; Yu et al., 2011). They are essential tools to find out and understand the phenomenon of LULC (Marko et al., 2016; Mondal et al., 2017; Jahanishakib et al., 2018). A reliable LULC change model is essential if one wants to know the future status and spatial distribution of LULC in the watershed, based on historical LULC changes (Behera et al., 2012).

Several LULC change models have been developed (e.g., Veldkamp & Lambin, 2001; Parker et al., 2003; Verburg et al., 2004; Koomen & Stillwell, 2007). In recent years, many models have been adapted to model LULC changes. Some of the most common models for LULC changes are statistical models (regression) (Yalew et al., 2016), cellular automata (CA), and Markov chains (Araya & Cabral, 2010; Han et al., 2015; de Oliveira Barros et al., 2018; Munthali et al., 2020), artificial neural networks (Omrani et al., 2017) and multi-agent-based models (Arsanjani, Helbich, Kainz, et al., 2013).

The CA-Markov model is one of the most widely used models due to its reliability and compatibility with many geospatial technologies (Sang et al., 2011; Halmy et al., 2015). The model has been used to predict and simulation of LULC changes in many countries, such as the United States (Myint & Wang, 2006), Libya (Al-sharif & Pradhan, 2014), China (Yang et al., 2012), Brazil (Vick & Bacani, 2019), Portugal (Araya & Cabral, 2010), Egypt (Halmy et al., 2015), Ethiopia (Gidey et al., 2017; Gashaw et al., 2018; Fitawok et al., 2020; Hishe et al., 2020; Kura & Beyene, 2020; Mohamed & Worku, 2020; Yohannes et al., 2020), Bangladesh (Mukhopadhyay et al., 2015), India (Singh et al., 2015).

The CA-Markov chain model is a relatively suitable method for modeling temporal and spatial changes in LULC (Myint & Wang, 2006; Singh et al., 2015; de Oliveira Barros et al., 2018). The CA-Markov chain considers the spatial and temporal components of land-cover dynamics (Houet & Hubert-Moy, 2006). Hence, to demonstrate human interaction with the environment, the Markov Chian (MC) model of LULC change is usually integrated with the cellular automata (CA) model with an agent-based model (Crooks & Heppenstall, 2012). The Markov chain, cellular automata, and cellular automata-Markov model are listed below to predict future land use change models.

2.5.1. Markov chain model

The Markov chain is a dynamic process based on the Markovian random process. It calculates the probability of changes from one state (e.g., forest) to another (e.g., cultivated) at time t2 is predicted from the state the system is in at time t1 (Marko et al., 2016). The Markov chain only calculates the probability of changes but cannot represent a change in an explicit spatial term (Moghadam & Helbich, 2013). Future predictions using Markov chain modeling are usually made by analyzing two qualitative land cover images that were taken at different points in time (Moghadam & Helbich, 2013). As a result of this analysis, three objects are produced. The transition probability matrix stores the probability that each state will change to every other state. The transition area matrix arising from this transition probability matrix (Arsanjani, Helbich, Kainz, et al., 2013; Hyandye & Martz, 2017) stores the expected number of pixels that might change to a predetermined number of units of time (Behera et al., 2012). The conditional probability images are produced to express each pixel's probability belongs to the class specified in the next time step. The Markov chain describes the transitional probability matrix as follows (Equation 1):

ppp11121n

p(p )pppij21222n (1)

pppn1n2nn

The probability of changing from a specific object (ith) to another object (jth) is described as the probability transformation (Pij); n is the type of land cover of the watershed. Pij must meet the following condition (Equation 2):

0p1(i,ij j1,2,3...,n) (2) n i1 p1(i,ij j1,2,3...,n)

Based on the non-after effect of Markovian random processes and probability of Bayes condition, the Markov chain model is obtained by (Equation 3):

ppp(n)(n 1) ij (3)

Where: 푃(푛) is state probability of any times

푃(n-1) is preliminary state probability

2.5.2. Cellular Automata Cellular models use numerous input information that can be used to simulate land cover or land use conversions based on set rules or algorithms that apply synchronization across all spatial units and provides an understanding of the model of the land use change process. During the operation of the Markov chain model, it is clear that it is independent of the states of the cells neighboring the cell under observation (Eastman, 2012). Accordingly, Markov modeling alone is not satisfactory for analyzing the problem simply, as it does not consider the spatial distribution of each class. Although the magnitude of the change can be predicted, the correct direction cannot be obtained (Boerner et al., 1996). Consequently, the cellular automata model is applied considering the spatial nature and the direction of the data (Myint & Wang, 2006).

Proximity to governing change events is a very important geospatial factor (Arsanjani et al., 2013). In a cellular automaton, state changes depend not only on the previous state but also on the state of the cells directly adjacent to the cell in question (Vick & Bacani, 2019). All locally defined transition rules determine the overall performance of the system. A cellular automaton consists of five components. (1) The cell space is the space under a particular cell, (2) the cell state is a representation of the observed spatial variable, (3) the time steps are the discrete steps of time at which the automaton occurs, (4) transition rules are a function that determines the change of the state of each cell taking into consideration its own and its neighboring cell’s states and (5) neighborhood-the spatial distribution of the cells which will be considered when applying the transition rules.

2.5.3. Cellular Automata-Markov The Markov chains-cellular automata model is a combination of two different methods. Markov chains, which is an empirical/statistical model, and cellular automata, a dynamic model that incorporates using the GIS platform (Behera et al., 2012; Arsanjani et al., 2013). The combination of Markov chains and cellular automata (CA-Markov) minimizes the weakness of Markov chains (Moghadam & Helbich, 2013). This is because the combination of the two methods gives a better result than performing a model separately (Arsanjani et al., 2013). Cellular automata represent the spatial-explicit term and location of changes, and Markov chains predict changes quantitatively (Houet & Hubert-Moy, 2006; Behera et al., 2012; Eastman, 2012).

The CA-Markov model is based on the probability that a particular piece of land will change from one state to another (Houet & Hubert-Moy, 2006). These probabilities arise from past changes and are used to predict future changes (Behera et al., 2012). The CA-Markov model can simulate several types of land-use changes (Houet & Hubert-Moy, 2006), which can simulate the transition from one LULC category to another (Behera et al., 2012). The most commonly used models in Ethiopia are the cellular and Markov chain models (Gashaw et al., 2017; Gidey et al., 2017; Hishe et al., 2020). Thus, the CA-Markov chain model integrates with a multi-criteria evaluation to predict future land use of a watershed. The Markov chain model provides the possibility and extent of LU change, while the CA model works with neighborhood interaction and spatial distribution.

2.6. Revised Universal Soil Loss Equation

Soil erosion is a natural phenomenon that is accelerated by human activities such as agriculture, grazing, deforestation, and settlement. The direct effect of soil erosion is the loss of soil organic matter and nutrients from the field, which reduces the soil's depth and fertility. The important factors that influence the rate of soil erosion are the intensity and duration of rainfall, the slope of the land, soil type, LULC, and land management practices. Field experiments and modeling are two approaches that have been used to understand the effect of climate change and LULC change on soil erosion. From these approaches, field experiment is very expensive and difficult, but the latter is relatively easy and economical. The Revised Soil Loss Equation (RUSLE) model (Renard, 1997) is used commonly to estimate the rate of soil erosion. RUSLE is an extension of the Universal Soil Loss Equation (USLE) model by adapting the local conditions' input factors (Renard, 1997). Furthermore, RUSLE is attractive for scenario analysis because of the inclusion of climate and management factors in the equation.

The RUSLE method is has been widely used by previous researchers in Ethiopia (e.g, Bewket & Teferi, 2009; Meshesha et al., 2012; Mengistu et al., 2015; Gelagay & Minale, 2016; Molla & Sisheber, 2017; Tadesse et al., 2017) and other countries (e,g. Angima et al., 2003 in Kenya; Tamene & Le, 2015 in Sub Saharan Africa; Duarte et al., 2016 in Portugal; Ganasri & Ramesh, 2016 in Greece; Fang et al., 2019 in China; Koirala et al., 2019; Kouli et al., 2009 in Nepal; Nyesheja et al., 2019 in Congo; Phinzi & Ngetar, 2019 in South Africa; Rellini et al., 2019 in Italy; Sahli et al., 2019 in Algeria; Mohammed et al., 2020 in Syria; Lin et al., 2020 in Portugal) because of its clear and relatively simple computational inputs.

The universal soil loss equation (USLE) was originally developed by Wischmeier & Smith (1978) and modified by Hurni (1985) for the Ethiopian conditions. Then, the Revised Soil Loss Equation (RUSLE) (Renard, 1997) was modified by Nyssen et al. (2009) to predict the annual amount of soil loss to the Ethiopian conditions. The average annual soil loss (A) due to sheet and rill erosion is calculated by overlaying five raster layers; rainfall erosivity (R) factor, soil erodibility (K) factor, Slope Length and Steepness (SL) factor, and Cover management (C) factor using equation 4.

11 A(tha yr ) R*K*LS*C*P (4) where, A is the average annual soil loss per unit area (t ha-1y-1), R is rainfall erosivity (MJ mm ha- 1 h-1 yr-1), K is soil erodibility (t ha-1 MJ-1 mm-1), LS is the slope length and steepness factor, C is a cover and management factor, and P is the conservation practice factor.

2.7. Soil and Water Assessment Tool (SWAT)

The relationship between rainfall and runoff and thus all processes involved in the hydrological cycle. The Soil Water Assessment Tool (SWAT) is a physically based and semi-distributed hydrological model developed at the USDA-ARS (Arnold et al., 2012). This model was developed to study small-scale to large and complex watersheds with different land use, soil, and slope classes. SWAT can assess the long-term effects of various land management and climate change on biomass production, water quality, and sediments, and agricultural chemical yields with the capability to study large-scale and complex watersheds for longer (Arnold et al., 2012). It simulates a daily hydrological cycle through the water balance of each hydrological response unit (HRUs), which are units with similar soil types, land use, and slope classes (Arnold et al., 2012).

HRUs are the basic units for soil water content, surface runoff, nutrient cycles, sediment yield, crop growth, and simulated management practices. The SWAT model simulates the hydrological cycle at each HRU using the following water balance (Equation 5) (Arnold et al., 2012).

t SWt0day SW(p surf  QE a seep W gw Q (5) i1

Where SWt is the final soil water content (mm), SW0 is the initial soil water content on day i (mm), t is the time (days), Pday is the amount of precipitation on day i (mm), Qsurf is the amount of surface runoff on day i (mm), Ea is the amount of evapotranspiration on day i (mm), Wseep is the amount

of water entering the vadose flow zone from the soil profile on day i (mm), and Qgw is the amount of base flow on day i (mm).

The SWAT model is suitable for understanding the effects of LULC and climate change on soil erosion and streamflow in one or multiple watersheds (Arnold et al., 2012; Koch et al., 2012; Wang et al., 2016). In Ethiopia, the model has been used by some researchers in various watersheds (Betrie et al., 2011; Setegn et al., 2011; Tibebe & Bewket, 2011; Gessesse et al., 2015), and has proved that the SWAT is an effective tool for studying erosion and the hydrological response of a watershed for LULC and climate changes.

Runoff occurs when precipitation on the slope surface exceeds the infiltration rate. Sometimes the precipitation reaches the ground surfaces after saturation, the result in runoff is called saturation overland flow. The SWAT model offers two methods for estimating surface runoff: the Soil Conservation Service (SCS, 1972) Curve Number method and the Green & Ampt (1911) infiltration method. The SCS Curve Number (CN) method, linked to the United States Department of Agriculture (SCS-USDA), is computationally efficient and widely used to predict an approximate amount of surface runoff generation in SWAT. The model predicts the amount of runoff based on land use, soil type, hydro-meteorological, and antecedent moisture condition (Betrie et al., 2011; Fentaw et al., 2018). The SCS CN method computes runoff response for the rainfall events using Equation 6 (Setegn, 2008; Tibebe & Bewket, 2011).

2 (R0.2S)day  3 Qsurf  ,Qinsurf / s (6) (R0.8S)day 

Where Qsurf is the daily surface runoff (mm), Rday is the rainfall depth for the day (mm), and S is the retention parameter (mm). Soil retention parameters derived from curve numbers vary spatially and temporally. The difference depends on soil moisture, soils, land use, land management, and slope with time and space within the watershed (Setegn, 2008; Tibebe & Bewket, 2011). The retention parameter is defined in Equation 7.

1000 S 25.4 10 (7) CN

where S is a drainable volume of soil water per unit area of a saturated thickness (mm/day), CN is the curve number. Further reading material about the SWAT model can be referred from Arnold et al. (1998) and the online resources at http://swat-model.tamu.edu/.

2.7.1. Computation of evapotranspiration Various methods have been suggested to estimate the PET over the past decades. The Soil and Water Assessment Tool (SWAT) provides three methods for estimating potential evapotranspiration (PET). In this study, the SWAT model was implemented using the Hargreaves method to estimate the PET. The Hargreaves-Sami (HS) method is the temperature-based PET estimation method, and it is the most commonly used. The FAO recommends an alternative method for PET estimation, especially for hydrological models when the observed weather data are not available (Hargreaves, 1989). The HS method requires daily minimum and maximum air temperature input data. Thus, the HS method was chosen to estimate the PET in this research. The HS method estimate PET as follows:

HE TTmaxmin ETk*RHSRSamaxmin *(TT)   HT (8) 2

Where ETHS is daily PET in mm/day; Ra is extra-terrestrial radiation in mm/day; Tmax and Tmin are o daily maximum and minimum temperature air temperature in c, respectively; KRS is the empirical radiation adjustment coefficient and the value is set to 0.0023 in this study; HE is empirical Hargreaves exponent and the value is set to 0.5 in this research; and HT is empirical temperature coefficient and the value is set to 17.8 in this study (Hargreaves, 1994).

2.8. Partial Least Square Regression (PLSR) Model

Partial Least Squares Regression (PLSR), developed from the principal component regression, helps to construct models that predict more than one dependent variable (Lorber et al., 1987). This method is used when the number of variables are more than the number of compounds in the data sets and the variables considered for the study are correlated (Abdi, 2010).

PLSR is especially useful when (1) the number of predictor variables is equal to or greater than the number of observations and/ or (2) predictors are highly correlated (i.e., there is strong collinearity) (Khattree & Naik, 2018). These properties make this technique an essential analytical

tool that should be considered when designing scientific studies where the number of sample units collected is small compared to the number of explanatory variables measured. Therefore, for data showing collinearity, the PLSR model is useful for exploring individual variables' effects (Abdi, 2010; Godoy et al., 2014; Khattree & Naik, 2018).

The analysis of hydrological responses using hydrological modeling cannot reveal the contribution of each type of LULC to hydrological changes. Combining the hydrological model and the PLSR may be a viable approach to carefully examine the effect and contribution of each LULC change in hydrological responses to watersheds. The pairwise Pearson correlation, combined with the PLSR model (Abdi, 2010), was used to examine the relationship between individual LULC types and each hydrological component. The PLSR model was used to evaluate the relationship between annual water balance and changes in individual land use class components. PLSR is a robust multivariate regression technique that is appropriate when predictors show multicollinearity (Yan et al., 2013). The most recent studies (e.g. Yan et al., 2013; Woldesenbet et al., 2017; Gashaw et al., 2018; Shawul et al., 2019; Tankpa et al., 2020) used the PLSR method to quantify the contribution of change in individual land use types to water balance components at a watershed level.

An interesting feature of PLSR is that the relationships between predictors and response function can be inferred from the weights and regression coefficients of individual predictors in the most explanatory components (Shi et al., 2013). Therefore, it is possible to determine which land use type most strongly interacts with streamflow. Determining the number of significant components to be kept is usually based on a criterion, including cross-validation. The validity and strength of the model are determined by two main indicators: (1) goodness of fit (R2), that is, the proportion 2 2 of variation in the dependent variable that the model explains; and (2) R cross (cross-validated R ; goodness of prediction), appropriate number in the dependent variable that the model can predict. 2 2 R and R cross are used to obtain the appropriate number of components for each PLSR model (Yan et al., 2013).

2 2 When R is greater than 0.5 and when R cross is greater than 0.0975, the PLSR regression model provides significant and good predictions (Trap et al., 2013). In PLSR modeling, the importance of a prediction for independent and dependent variables is given by the variable importance for the projection (VIP). Terms with larger VIP values (greater than 0.8) were most appropriate to explain

the dependent variable (Shi et al., 2013; Yan et al., 2013). In addition, the regression coefﬁcients of the PLSR models were used to indicate the direction of the relationships between the changes in individual land use types and streamﬂow (Shi et al., 2013; Yan et al., 2013). Thus, it is possible to determine which type of land use interacts most strongly with hydrological components.

PLSR model analysis is performed in statistical software such as SAS (Khattree & Naik, 2018), SPSS; MATLAB (Andersson & Bro, 2000), SAS/STAT (Zhang et al., 2007), XLSTAT software, Excel add-ins statistical tool (Woldesenbet et al., 2017; Shawul et al., 2019), R (Mevik & Wehrens, 2015), SIMCA-P (Shi et al., 2013; Yan et al., 2013), SPSS (Carver & Nash, 2011), and JMP (Cox & Gaudard, 2013).

2.9. Conceptual framework of the study

A conceptual framework serves as a summary of theoretical and empirical literature reviews on the key components of a study, where it integrates the themes to guide the study. LULC and climate change can be expected to continue to change over time in a watershed. LULC change is determined in part by biophysical and socio-economic factors. As these determinants make significant changes in the future, LULC and climate change are likely to change. Changes in LULC and climate also affect soil erosion and the hydrological process of a watershed. Therefore, soil erosion and hydrological process are likely to change in response to a magnitude of LULC and climate changes, as well as the type and effectiveness of soil and water conservation measures. Soil erosion and hydrological processes of a watershed lead to a decline of agricultural productivity, sedimentation of reservoirs and dams. The conceptual framework that shows the impact of LULC and climate change on soil erosion hydrological response is given in Figure 2.2.

Land use/land cover change Climate Change

-Changes in surface conditions -Intensive land use Change in temperature & rainfall -Deforestation/reforestation

Impact assessment

Change in soil erosion & hydrology

Reliable data about the impact of LULC and climate change on soil erosion and hydrology

Figure 2.2 Conceptual framework shows the impact of LULC and climate change on soil erosion and hydrology Note: the relationships represented in broken arrows are not addressed in this study.

Chapter 3 3. Description of the Study Area 3.1. Description of the Study Area 3.1.1. Location The study watershed, Muga, lies within 10° 05´ 00˝ N to 10°43´48˝ N and 37° 49´ 12˝ E to 38° 8´ 56˝ E (Figure 3.1). It is situated in the southeastern part of Mount Choke, which is a headwater area of the Abay river Basin in Ethiopia (Teferi et al., 2010). The study watershed covers a total area of 705 km2, with the gauge of the streamflow being located in the middle of the Muga river. Thus, the upper parts of the watershed (424 km2) drains into the gauge station is the study area considered for this study. The study watershed falls in six districts (Dejen, Enemay, Bebugn, Enarje Enawga, Huletej Enese, and Debay Tilatgin) of East Gojjam Administrative Zone. However, more than 95% of the study watershed lies in the Debay Tilatgin district. The altitude of the study watershed ranges from about 2384 m to 4088 m above mean sea level. The elevation increases from Southeast to North West (Figure 3.1).

Figure 3.1 Map of the study area

3.1.2. Climate

Based on the agro-climatic classification of Ethiopia (Hurni, 1998), the study area is divided into three agro-ecological zones, namely, Weynadega (1500 to 2300 m), Dega (2300 to 3200 m), and Wurch (above 3200m). According to the data obtained from the National Meteorological Agency (NMA) (1985-2017), the average annual rainfall of the study area is about 1086.6 mm. About 85% of the rainfall falls during the rainy season (May to October). Data obtained from NMA also showed that the average monthly minimum and maximum temperatures (1985-2017) were about 9.30C and 23.70C, respectively (Figure 3.2).

350 30.0

300 25.0

250 ) 20.0 200 15.0 150 10.0 Rainfall (mm) Rainfall 100 Temprature (mm Temprature 50 5.0

0 0.0 Jan Feb Mar April May June July Aug Sep Oct Nov Dec Rainfall (mm) Max monthly temprature (0c) Min monthly temprature (0c) Mean monthly temprature (0c) Note: It is based on grided data (records of near by Kuy station and TAMSAT satellite) (1985-2016)

Figure 3.2 Mean monthly rainfall and mean monthly temperature records of the study area

3.1.3. Soils and geology

The Choke mountain's soil types, including the study area, are volcanic in origin, derived from Mio-pliocene shield volcano lavas and at lower elevations, oligocene flood basalts (Kieffer et al., 2004). Under undisturbed conditions, soils are usually deep: the natural depth can extend up to several meters, with the depth of roots in this part of the Ethiopian highlands being up to one meter (Hurni, 1988). As a result, eroded tropical soils are highly susceptible to erosion, and in areas in the western Ethiopian Highlands that have been cultivated with traditional methods, the rate of soil loss can exceed the rate of soil generation by a factor of 4 to 10 (Hurni, 1988), a pattern that is

partly dominated by the traditional ox-drawn farming systems, which has been shown to contribute to rapid erosion in other parts of the highlands of Ethiopia (Nyssen et al., 2000). According to the FAO soil classification system, four major soil types were identified in the study watershed. namely, Eutric Cambisols, Eutric Vertisols, Haplic Luvisols, Haplic Nitisols (BCEOM, 1998). Eutric vertisols are the predominant soil type with an area coverage of 212 km2 in moderately gentle slopes and very deep soils in the study area. The soil organic content in the investigated watershed is between 2.09 and 3.22, and the soil's PH value is between 5.5 and 6.0 (BCEOM, 1998). According to the data obtained from the GIS department of the Ethiopian Ministry of Water and Energy, the geology of the watershed consists of tertiary extrusive and intrusive deposits.

3.1.4. Vegetations

The major forest resources of the Amhara region are Afro-Alpine and Sub-Afro Alpine forests, dry evergreen montane forest and evergreen scrub, Combretum Terminalia woodland, Acacia Commiphora woodland, bamboo forests, and plantations (Wassie, 2017). The study watershed headwater is covered by different types of plant species, including Juniperus pocera, Olea europaea, Hagenia abyssinnica, and Erica arborea, Carissa spinarum, Dovyalis abyssinica, Justicia schimperiana, Dombeya torrida, Eucalyptus tree, Accaica Abyssinica, Cordia Africana, and Olivera Africana.

3.1.5. Population and socioeconomics

According to the national population projection of the year 2014, the population of the woredas (Debay Tilatgin, Enemay, and Dejen) found in and around the study watershed was 548, 621 of whom 49.2% were men and 50.8% were women (CSA, 2013). Debay Tilatgin, Enemay, and Dejen are among the woredas, where the study watershed is located. However, a great portion of the study watershed is found in Debay Tilatgin, according to the national population projection of the year 2014, the population of Debay Tilatgin was 144, 967 (72,194 male and 72,773 female). According to CSA (2013) population projection, the average population density of the study area ranges from 165 to 247 people/ km2.

The livelihood of the local community in the study watershed depends on agriculture. Mixed farming practices (i.e., crop production and livestock rearing) represent the major sources of income for the local community, accounting for approximately 95% of the population’s sources of

income. Due to topographical differences, crops grown in the upper escarpment and lower escarpment are different. The predominant crops grown in the upper escarpment of the watershed are barley (Hordeum vulgare) and potato (Solanum tuberosum), whereas teff (Eragrostis teff), wheat (Triticum vulgare), chickpea (cicer arietinum) and maize (Zea mays) grains are the main cultivated crops in the lower escarpment of the study watershed. Rainfall in the area is characterized by a mono-modal rainy (summer) season, which covers June to September. Thus, crops are harvested once a year in the study area. Similarly, cattle, sheep, equines, and goats are the main livestock resources of the area. They were reared on a free grazing system. They have reared in a free grazing system. Most grazing is free-roaming on the vast communal grazing lands. Farmers often use crop residues as one of the basic feed sources for livestock feed.

Chapter 4 4. Research Methodology

4.1. Overview of the Research Philosophy

For quite a long time, scholars have debated the philosophical foundations or research paradigms to study problems surrounding human beings and nature. Such debates and viewpoints about research paradigms partitioned researchers along two major classes of research in social sciences are often recognized quantitative and qualitative (Jupp, 2006; Walliman, 2017).

Each paradigm bases on particular foundations and applies a specific approach to researching the social world. Bryman (2004) distinguished three significant major characteristics in each that make the point. In quantitative research, i) the orientation uses a deductive approach to test theories; ii) Epistemology, which depends on a positivist approach inherent in the natural sciences and the ontology-objectivist in that social reality is regarded as objective fact. In qualitative research, i) the orientation uses an inductive approach to generate theories; ii) Epistemology, it ignores positivism and basis on individual interpretation of social reality; and the ontology-constructionist, in this case, social reality is viewed as a continually shifting product of perception (Shaw et al., 2010; Williams et al., 2012).

Some scholars contended that mixed methods research has its own philosophical worldview: pragmatism. Consequently, pragmatists trust philosophically in using procedures that work” for a particular research problem under investigation and that one should use different strategies when understanding a research problem (Tashakkori et al., 1998; Tashakkori & Creswell, 2007). The importance of pragmatism as founding for mixed-method opens the door to multiple methods, different worldviews and different assumptions, as well as different forms of data collection and analysis to derive knowledge about the problem in the mixed methods study (Creswell, 2014; Creswell & Clark, 2017). In pragmatism approach, the researchers use different techniques at the same time or one after the other (Mukherjee & Kamarulzaman, 2016)

Accordingly, among these research philosophical orientations, the pragmatism perspective was the general philosophical underpinning of this research. Johnson et al. (2007) also argued that most mixed research methods scholars have supported some version of pragmatism as the most helpful

philosophy to support mixed research methods and rationalize the use of multiple methods in a single study (Creswell & Clark, 2017; Migiro & Magangi, 2011).

4.2. Research Design and Justifications

The choice of methodology depends more on the objectives of the study and the corresponding research questions than the preference of the researcher. According to Johnson & Onwuegbuzie (2004), mixed-method is conceptualized as the class of research where the researcher mixes or combines quantitative and qualitative research techniques, methods, approaches, concepts, or language into a single study.

According to Creswell & Clark (2017), mixed research is a research design with philosophical assumptions as well as methods of inquiry. As a methodology, it involves the philosophical assumptions that guide the direction of the collection and analysis of data and the mixture of qualitative and quantitative approaches in many phases in the research process. As a method, it focuses on collecting, analyzing and mixing both quantitative and qualitative data in a single and series of studies.

In a mixed research approach, the investigator gathers both quantitative and qualitative data and then draws interpretations based on the combined strengths of both sets of data to understand research problems (Creswell, 2014). Therefore, mixed research design was adapted as the general research approach. The quest for triangulation in multiple ways, such as analytical frameworks, data collection from multiple sources, multiple approaches of data analysis and interpretation were considered as rationales to adapt mixed research design approach. In line with this, the central idea of mixed research design is that the combination of both forms of data give a better understanding of a research problem than either quantitative or qualitative data by itself (Creswell, 2014; Creswell & Clark, 2017). In general, using the mixed research methods, qualitative results have given new dimensions to concepts such as drivers of land use land cover change that could not have been captured in quantitative measures/analysis of the concept and vice versa.

The mixed research design has many strands, however, in this study, the concurrent embedded design was used to generate both qualitative and quantitative data concurrently. The concurrent embedded strategy, the direction of presentation must be driven by choices made vis-à-vis how the data interconnect to each other for integration (Creswell, 2014). The study employed mixed

data collection where one data collection method supplied strengths to offset the weaknesses of the other. Such mixing resulted in a more complete understanding of a research problem that results from collecting both quantitative and qualitative data (Creswell, 2014; Creswell & Clark, 2017). Due to the nature of the study and its guiding research questions, more weight was given to the quantitative approach, implying that the quantitative analysis would overweigh the qualitative one. The qualitative information was primarily used to support the quantitative findings to draw valid conclusions.

4.3. Data Sources and Data collection

In this section, a summary of the data sources, software used, and the methodology of the study is presented while the details of data and methodology are given in the respective chapters. The data includes primary and secondary data. Primary and secondary data were collected to achieve the objectives of this study. The collected data includes satellite imagery, DEM, soils, and daily meteorological and river discharge. A summary of data sources and types, methodology and models software used are given below.

4.3.1. Land use land cover data

Landsat Thematic Mapper (TM) of 1985, Landsat Enhanced Thematic Mapper Plus (ETM+) of 2002, and Landsat Operational Landsat Imageries (OLI) of 2017 with path 169 and row 053 images downloaded from the United States Geological Survey (USGS) Earth Explorer (http://earthexplorer.usgs.gov) were used for LULC classification analysis. Furthermore, slope, road, settlement, and elevation maps were used to predict the future LULC change for the study watershed for the year 2033. The LULC maps were used to assess the trend of LULC change; and the impacts of LULC change on soil erosion and hydrological process in Muga watershed.

4.3.2. Climate and hydrological data

Two climate datasets; baseline and modeled climate data, were used in this study. The climate data were used to drive the RUSLE model, future climate scenarios, and hydrological climate impact scenarios. Meteorological data of daily rainfall, maximum and minimum temperature, solar radiation, wind speed, and relative humidity were obtained from the Ethiopian National Meteorological Services Agency. Future climate data from Six Regional Climate Models (RCMs),

namely: HadGEM2-ES, CSIRO-Mk3.6.0, GFDL-ESM2M, CanESM2, MIROC and NorESM1-M forced by two Representative Concentration Pathway (RCPs), RCP4.5 and RCP8.5 were downloaded from Coordinated Regional Downscaling Experiment (CORDEX)-Africa and used to project the future climate change in the study watershed from 2018 to 2050.

Daily streamflow data of gauging station was obtained from the Ethiopian Ministry of Water, Irrigation, and Electricity. The streamflow data were used to calibrate and validate the SWAT hydrological model.

4.3.3. Other spatial datasets

The datasets used in this study comprise the physical characteristics of the watershed, including soil data, Digital Elevation Model (DEM), and population data which were obtained from various sources. The soil map (Scale 1:250,000) of the study watershed that used for the RUSLE and SWAT model was obtained from the Ethiopian Ministry of Water Irrigation and Electricity (MoWIE) hydrology department. A digital elevation model (DEM) of 30 m resolution which was downloaded from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) was used to drive slope, slope length, elevation, and stream network of the study watershed. Besides, ancillary data were utilized during analysis, including topographic maps and thematic layers (towns) that obtained from the Ethiopian Geo-Spatial Information Agency. The road networks in the study watershed were downloaded from OpenStreetMap (https://www.openstreetmap.org/).

4.3.4. Socioeconomic data

Key informant interviews with elders and development agents, household survey, field observations, and focus group discussions, informal discussion with individual farmers were conducted to get detailed information from the local people about the LULC dynamics that occurred in the study watershed and the main drivers of change in the Muga watershed over the last three decades. The population dataset for the study area was obtained from the Ethiopian Central Statistical Agency (CSA) (CSA, 2007, 2013) and woreda agricultural offices of the study watershed. The detail about the types of datasets and their sources are listed in Table 4.1 and Table 4.2.

Table 4.1 List of time series Landsat data used for the study

Satellite and sensor Path/Row Acquisition Date Resolution Sources Landsat 5 TM 169/53 15 April 1985 30 x 30 USGS Landsat 7 ETM+ 169/53 08 March 2002 30 x 30 USGS Landsat 8 (OLI) 169/53 02 January 2017 30 x 30 USGS

Table 4.2 Dataset sources and types No. Name of data Sources 1 DEM (30*30m) ASTER (NASA) 2 Soil data (Scale 1:250,000) MoWIE 3 Land use and land cover (1985, 2002, 2017 & 2033) Landsat image 4 Daily meteorological data of the 7 weather stations in and around the watershed from 1985 to 2017  Precipitation NMA  Maximum & Minimum Temperature  Relative humidity  Wind speed  Solar radiation 5 Daily river discharge data of Muga from 1985 to 2017 MoWIE 6 Road OpenStreetMap 7 Population dataset CSA of Ethiopia 8 Ground Control Points From field using GPS 9 CORDEX data (50*50 km) Earth System Grid Federation (ESGF) https://esgf-node.llnl.gov/projects/esgf-llnl/. 4.4. Data Analysis 4.4.1. Land use/ land cover change analysis

Various techniques are available to extract meaningful information of land use/land cover (LULC) types from remote sensing imagery. In this study, both unsupervised and supervised image classification techniques were used to classify the Landsat imageries. Preprocessing techniques such as layer stacking, color composite, and subsetting areas of interest were done for all images using ERDAS Imagine 2015 software. Then, LULC maps of the historical periods (1985, 2002, and 2017) were delineated from Landsat images, and five classes were identified for the land cover classification: cultivated, grassland, shrub-bushland, forest and urban. The future LULC was predicted with a cellular automata-Markov chain (CA-Markov).

4.4.2. Socioeconomic data analysis

The socioeconomic data collected through household surveys from the local people on LULC dynamics, its drivers, and its impact in the Muga watershed between 1985 and 2017 were analyzed

through simple descriptive statistics (percentage) using SPSS 22. Qualitative information gathered from key informant interviews, observation and FGDs were analyzed through thematic analysis, which were used to triangulate, substitute, and complement quantitative data/ Landsat information.

4.4.3. Future climate change projection

To develop future climate scenarios, grid points of RCMs (daily precipitation and maximum and minimum air temperature data) were extracted using MATLAB software. Then, the output of the RCMs climate data in CORDEX-Africa under emission scenarios RCP4.5 and RCP8.5 were bias- corrected using the linear scaling bias correction method using the baseline data (1985-2018). The bias-corrected data were compared with the baseline climate data to calculate the absolute change in temperature and precipitation. The performance was evaluated by comparing raw data of RCM with bias-corrected RCM data using statistical indicators.

4.4.4. Methods to assess impacts of LULC and climate change on soil erosion

The Revised Universal Soil Loss Equation (RUSLE) was used to assess the impacts of LULC and climate change on soil erosion in Muga watershed. In this study, the values of C and R factors were changed, while the other factors (i.e., soil erodibility, conservation practice factor, and slope length and slope steepness) were kept constant. The inverse distance weighted (IDW) interpolation method was used to estimate the R factor of the baseline and future rainfall data. 4.4.5. Methods to assess impacts of LULC and climate change on hydrological process

In this study, simulation scenarios were developed using the soil and water assessment tool (SWAT) model to assess the individual and combined effects of LULC and climate change on the hydrological processes in the Muga watershed. Besides, Partial Least Squares Regression (PLSR) statistical method was used to assess the effects of changes in individual LULC types on hydrologic components. The SWAT model simulates a daily hydrological cycle through the water balance of each hydrological response unit (HRUs), which are units with similar soil types, land use, and slope classes. The Soil Conservation Service (SCS) Curve Number method and the Hargreaves-Sami (HS) method were used to estimate surface runoff and potential evapotranspiration (PET), respectively.

PART II: RESULTS AND DISCUSSION

Chapter 5 5. Land Use and Land Cover Dynamics and Drivers in the Muga Watershed, Abbay River Basin, Ethiopia Abstract Land use/land cover change causes unprecedented changes on the surrounding environment at different spatial and temporal scales. Muga watershed of the Abbay River Basin is characterized by deterioration with mismanagement of natural resources. Thus, this study was aimed at detecting the magnitude and pattern of land use/land cover changes and assessing drivers of changes over the last three decades (1985 to 2017) in the Muga watershed. Synergy of Landsat imageries (1985, 2002, and 2017), household survey, focus group discussion, key informant interview and field observation were used to detect changes and drivers of changes. Land use/land cover changes in the study watershed were detected using digital image analysis techniques, and their socioeconomic and biophysical drivers were analyzed using descriptive statistics. The change detection results revealed a significant increasing trend of cultivated land and urban areas by 12 and 270%, respectively. In contrast, grasslands, forest lands and shrub-bushlands showed a declining trend of about 40, 21, and 12%, respectively. The socioeconomic data analysis results indicated that the expansion of cultivated land, cutting of trees for fuelwood and construction purposes, population growth, land tenure policy, and climate variability were the most influential drivers of land use/land cover changes in the study watershed. The findings of this study confirmed that unmanaged land cover conversion led to the degradation of natural resources of the study watershed. Thus, alternative sources of income for youths and landless peasants and integrated watershed management, which has paramount importance in maintaining economic and ecological benefits were suggested to alleviate the adverse effects of LULC changes in Muga watershed. Keywords: Drivers, Land use/ land cover change, Muga, remote sensing, Abbay River Basin 5.1. Introduction

LULC of the earth’s land surface has been changing since time immemorial at an accelerating pace (Turner et al., 1994; Ramankutty et al., 2006; Goldewijk & Ramankutty, 2009). The rates and intensities of LULC dynamics are changing dramatically in Africa, where high population growth associated with the overexploitation of natural resources and low productivity of land is prevalent (Eva et al., 2006). For instance, agricultural lands have increased by 57%, whereas the region has

lost 16% of its forests and 5% of its woodlands and grasslands from 1975 to 2000 (Eva et al., 2006). LULC dynamics drivers include cutting trees, charcoal making, poor farming methods, and converting natural vegetation to farmlands. For instance, in the Democratic Republic of Congo, the agricultural expansion through shifting cultivation is the main proximate cause of deforestation (Samndong et al., 2018).

LULC dynamics have important implications for changes in climate change (Kalnay and Cai, 2003), ecosystem service (Tolessa et al., 2017; Gashaw., et al., 2018), soil erosion (Tibebe & Bewket, 2011; Alkharabsheh et al., 2013; Tadesse et al., 2017; Miheretu & Yimer, 2018) and water resources (Getachew & Melesse, 2012; Tekleab et al., 2014; Woldesenbet et al., 2017). In Ethiopia, numerous studies on LULC changes have been conducted and witnessed the massive LULC conversion (e.g., Bewket, 2002; Demissie et al., 2017; Garedew et al., 2009; Gebrelibanos & Assen, 2015; Gebreselassie et al., 2016; Hurni et al., 2005; Kindu et al., 2015; Mengistu et al., 2012; Moges & Bhat, 2018; Teferi et al., 2013; Zeleke & Hurni, 2001). These studies indicated that the direction, pattern, and magnitude of LULC change is heterogeneous across Ethiopia. For instance, Demissie et al. (2017) in the Libokemkem district; Hurni et al. (2005) in the Dembecha area of Gojjam; Teferi et al. (2013) in Jedeb watershed within the Blue Nile; Sewnet (2015) and Moges & Bhat (2018) in the Infranz watershed; Tegene (2002) in the Derekolli catchment of the South Wollo; Ariti et al. (2015) in the central rift valley of Ethiopia, Dessie & Kleman (2007) in the south-central rift valley of Ethiopia; and Kindu et al. (2013) in the Munesa-Shashemene landscape, reported a considerable expansion of cultivated land at the cost of forest, grassland, and shrubland. In contrast, an increase in forest cover in the county was reported (Bewket, 2002; Bantider et al., 2011; Gebrelibanos & Assen, 2015; Moges & Bhat, 2018). For instance, Bantider et al. (2011) in the Eastern Escarpment of Wello; (Gebrelibanos & Assen (2015) in the Hirmi watershed, highlands of Northern Ethiopia; Bewket (2002) in the Chemoga watershed, northwestern Ethiopia; and Moges & Bhat (2018) in the Infranz watershed, Northwestern Ethiopia reported the expansion of forest cover at the expense of cropland and grazing land. Mengistu et al. (2012) also reported a decline in the bush and grassland, whereas plantation trees, crops, and bare lands have increased in the midwest escarpment of the Ethiopian rift valley from 1972 to 2004.

Several LULC change studies conducted in Ethiopia related LULC changes with a variety of drivers, which include both proximate and underlying causes such as demographic pressure and

associated demand on environmental resources (e.g., Tekle & Hedlund, 2000; Tegene, 2002; Bewket, 2002; Demissie et al., 2017; Gessesse & Bewket, 2014; Hailemariam et al., 2016; Kindu et al., 2015; Mengistu et al., 2012). For instance, studies conducted by Tekle & Hedlund (2000) in South Wello of Ethiopia, Tegene (2002) in the Derekolli catchment of South Wollo zone of Ethiopia, and Kindu et al. (2015) in the south-central highlands of Ethiopia, recognized the expansion of settlement and cutting of trees for fuelwood and charcoal making as proximate causes of LULC change. Slope and the availability of adequate infrastructures such as roads and access to markets were considered as the proximate drivers of LULC changes in the Abbay River basin (Yalew et al., 2016). On the other hand, population growth, climate variability, land tenure policy, drought, government change, and poverty were recognized as the major underlining drivers of LULC change (Bewket, 2002; Ariti et al., 2015). Although numerous research initiatives have been conducted on LULC dynamics across the country, the research findings indicated that the extent, magnitude, and drivers of the change are noticeably inconsistent over space and time. The inconsistency of the results would appear to be due to different socio-economic and biophysical related factors.

The Abbay River Basin is one of the most diverse and very important river basins in Ethiopia (Bewket & Teferi, 2009; Demessie, 2015; Hurni et al., 2005; Melesse et al., 2009; Yalew et al., 2016). The basin experiences serious environmental problems, including soil erosion, land degradation, loss of soil fertility, and deforestation (Bewket & Teferi, 2009; Demessie, 2015; Haregeweyn et al., 2017; Mengistu et al., 2012; Steenhuis et al., 2013; Tadesse et al., 2017). Prior studies in the basin assessed the Spatio-temporal aspect of LULC change without giving much attention to its drivers. Besides, analysis of LULC changes utilizing integrated remote sensing and GIS technology with a socio-economic survey at the watershed level are rare in the basin. Designing a development plan with appropriate management interventions requires a proper understanding of local-level LULC dynamics and the driving forces behind them. Thus, information about LULC dynamics in the Muga watershed and its drivers is vital in understanding human-environment interaction dynamics and designing appropriate development plans and environmental management interventions. Therefore, studying LULC status over space and time, assessing its drivers in the study watershed is essential for land use planning and sustainable use and management of watershed resources. For a watershed like Muga, which is undergoing a rapid

LULC change, there is a need for reliable and timely LULC information about resource degradation.

5.2. Material and Methods 5.2.1. Data sources and processes

The time series of the Landsat Thematic Mapper (TM) of 1985, Landsat Enhanced Thematic Mapper Plus (ETM+) of 2002, and Landsat Operational Landsat Imageries (OLI) of 2017 with path 169 and row 053 images were downloaded from the USGS website (http://earthexplorer.usgs.gov) (Table 4.2). For this study, the acquisition years were selected based on a 15-year interval to easily visualize changes in spatiotemporal LULC patterns. Furthermore, Landsat images were selected for years that align with anticipated significant LULC changes in the study area. Accordingly, the 1985 image is indicative of conditions after the collectivization of land resources with the promotion of Agricultural producer cooperatives in Ethiopia including the study watershed during the Dergue regime (Crewett et al., 2008). The year 2002 represents the period aftermath of the Dergue regime, and the year was selected to evaluate the redistribution of land to farmers. Finally, to include recent changes and the study area's current biophysical status, the 2017 Landsat image was used. However, due to Landsat images' accessibility and quality from USGS archives for the study watershed, a certain discrepancy in the time interval (+2 year) was taken into account.

The satellite images were used to prepare the LULC maps of the study watershed. The satellite images were utilized with a spatial resolution of 30 m. Dry season and cloud-free images were used to avoid the effect of seasonal variations (Kindu et al., 2013). The dry season is mostly considered the best period for analysis, as the difference between croplands and the natural environments is more pronounced. Therefore, the acquisition was carried out within January, March, and April. The satellite images were processed using the Earth Resource Data Analysis System (ERDAS) imagine®2015 and ArcGIS®10.4 software.

The satellite imageries have gone through both pre and post-processing. The pre-processing of the images includes layer stacking, mosaic, color composite, and sub-setting. All the acquired satellite images were geometrically and radiometrically corrected before analysis to improve image quality. All the satellite datasets were projected to the Universal Transverse Mercator map projection

system zone 37N and datum of World Geodetic System 84 (WGS84), which ensure consistency between datasets during analysis.

In this study, unsupervised and supervised image classification techniques were used to classify the Landsat images used in the study. The unsupervised classification was initially applied prior to the field survey using the visual interpretation method to differentiate various land use/cover types in the studied watershed. Unsupervised classifications were carried out using Iterative Self- Organizing Data Analysis (ISODATA) clustering algorithm. The LULC maps of the study area were produced using the pixel-based supervised image classification with the maximum likelihood classification algorithm (Congalton & Green, 2019). Supervised image classification is a recommended classification approach to achieve good results when adequate training data are available for the study area (Lillesand et al., 2015; Congalton & Green, 2019). First, about five LULC classes (grassland, cultivation land, shrub-bush land, forest, and urban) were identified from the images. Then, about 300 samples (60 samples per LULC class) were randomly selected from five LULC types as training points whereas 300 samples were used for accuracy assessment based on Lillesand et al. (2015) and Congalton & Green (2019). Besides, high-resolution data (SPOT images and Google Earth), topographic maps, and data from field observations and knowledge of the elderly were used. Generally, the steps of the current LULC change classiﬁcation procedure is presented in Figure 5.1.

Landsat image Ancillary data (Shape 1985, 2002 and Ancillary data file) 2017

Image pre processing 1. Geometric correction 2. Enhancement Unsupervised 3. Layer staking classification 4. Mosaic 5. Subsetting

Field visit, topographic and Preliminary LULC map historical map, Google Earth image, and Spot 5 image

Preparing training points Testing samples

Supervised classification Accuracy assessment Produce training and test files

No Acceptable?

Final LULC Map Yes

Figure 5.1 Flowchart showing the procedures employed to arrive at the final LULC map

As indicated in Table 5.1, five classes of LULC are considered in the image classification. The cultivated areas were included in the same class as it was difficult to distinguish these two classes on the Landsat images. Image classification is based on a sample of the classes; hence, the quality should be checked and quantified (Congalton & Green, 2019). A topographic map of 1984 was obtained from the Ethiopian Geo-Spatial Information Agency use as reference data to check the 1985 Landsat image classifications' accuracy. The 2002 classification was validated using data from the Woody Biomass Inventory and Strategic Planning Project. The knowledge from elder people and Google Earth data sets was also used for 1985 and 2002. Field visits and Google Earth data sets were used to collect samples for the reference year 2017. The ground control points used for the accuracy assessment were independent of those used as training samples.

Table 5.1 Land-use types of the study area

S. N Class name Description 1 Cultivated Areas used for farmlands, and the scattered rural settlements that are closely associated with the cultivated fields. These were combined into one class as it was difficult to detect the dispersed rural settlements as a detached LULC class where fragmented cultivated land exists around settlements. 2 Grassland Land predominately covered with grasses, and land units are allocated as a source of animal feed. 3 Forest Forest areas covered with relatively tall and dense trees formed nearly closed canopies around religious sites and at the upper escarpment of the watershed. This class also included plantation trees. 4 Shrub- Shrub-bushland dominated area with small isolated trees always with a lower range bushland of grass. 5 Urban Areas with all types of artificial surfaces including residential areas (emerging rural towns, villages) occupied by living houses and constructions.

The accuracy assessment was done by comparing the classes' areal extent in the classified images relative to the reference data set using error matrixes, which is used to define the quality of the map derived from the data. Accordingly, overall accuracy, producers and users accuracies, and Kappa statistics were computed from the error matrix for each year (Lillesand et al., 2004). The Kappa coefficient is calculated as follows (Equation 9):

푟 푟 푁 ∑𝑖=1 푋𝑖𝑖 − ∑𝑖=1(푋𝑖+ ∗ 푋+𝑖) 퐾ℎ푎푡 = 2 푟 (9) 푁 − ∑𝑖=1(푋𝑖+ ∗ 푋+𝑖)

푟 Where, Khat= kappa coefficient; N is the total number of values; 푁 ∑𝑖=1 푋𝑖𝑖 is observed accuracy; 푟 and ∑𝑖=1(푋𝑖+ ∗ 푋+𝑖) is chance accuracy.

The change detection is carried out by considering the LULC image on a pixel-by-pixel basis (Lillesand et al., 2015). A change that was extracted from the study observation years is presented in change metrics to get “where” and “from-to” information of change direction for the reference time interval. The changes calculated for each LULC class as the percentage of the difference in area between a final year, t2, and an initial year, t1. The percentage of change for each LULC class is calculated by using the following equation (Kindu et al., 2013; Hassen & Assen, 2018).

AAt2 t1 C  *100 (10) A Where C the rate of percent change, At1 is the areat1 of one type of land use in t1 time, At2 is the area of the same type of land use in t2 time.

In this study, focus group discussions, survey questionnaires, field observations, and key informant interviews with individual farmers and development agents were used to collect socioeconomic data. The socioeconomic data were used to gain detailed information from the local people regarding the trends and important driving factors of LULC changes in the study area during the study period. The survey questionnaire was obtained households, which was administrated to a sample of household heads obtained from the total household heads using a multi-stage sampling technique. First, the study watershed was stratified into three agro-ecologically representative kebeles, following Hurni's classification (1998). Second, a simple random sampling technique was used to select a kebele from each of the study area's traditional agro-ecological zones. The selected kebeles were; Debet, Kuyzuria, and Enekoy, representing Weynadega, Dega, and Wurch, respectively. Thirdly, the sample households were selected using simple random sampling procedures taken from each kebele. The total household population was listed, and the sample size calculated using Equation 11 (Kothari, 2004).

푧2 ∗ 푝 ∗ 푞 ∗ 푁 푛 = (11) 푒2 ∗ (푁 − 1) + 푧2 ∗ 푝 ∗ 푞 where; n is the samples size when the population is less than 10,000; N is sample frame (population); e is acceptance error, i.e, 0.05 (since the estimate should be within 5% of true value); z is size 1.96 i.e. sampling error ( as per the table of the area under the normal curve for the given confidence level of 95%); p is 0.5, p-value being the proportion of defectiveness in the universe (q=1-p). Based on the formula, the sample size/ number of respondents were 355 as the total population size was 4172. The number of samples in each kebele was proportional to the population's size, which was 115, 93, and 147 from Weynadega, Dega, and Wurch, respectively (Table 5.2). From the sample respondents, 5.63% were females, and 94.36% were males.

To complete the survey with qualitative data, interviews with key informants, focus group discussions (FGD), and field observations were conducted to investigate the LULC change trends and driving factors in the Muga watershed. Key informant interviews were undertaken with 15 farmers and three kebele development agents, and natural resource experts to obtain additional information regarding LULC changes and drivers of the change. The key informants were selected based on their position, experience, or responsibilities and who can provide information about

local facts, attitudes, and beliefs. They include elderly farmers, natural resource experts or development agents, and community leaders.

The focus group discussions were conducted with a group of individuals who have knowledge, experience, or interest in the issue or who were affected by the topic being discussed. They include elderly farmers, vulnerable groups (such as women), youth, local community leaders, and natural resource experts or development agents. All FGD participants were not included in the in-depth interview. The interview was conducted with an open-ended questionnaire at the DA office, churches, and farm fields. Detailed information about landscape changes and drivers of change over the last 32 years (1985-2017) was collected. Transect walks and informal discussions were held with farmers to understand the watershed major observed problems better. A total of 3 FGDs were carried out in three kebeles selected from the upper, middle, and lower parts of the watershed. Each focus group has ten members, and the facilitator leads the discussion. Discussions were held using a checklist as a memory guide and to assist in record keeping. The checklist covers the extent, trend, and drivers of LULC change in the study site's past and present situation. Besides, checklists cover challenges to prevent and reduce the drivers of ecosystem change. Key informants and focus group discussions were recorded, and brief notes were also taken.

Table 5.2 Distribution of total and sample respondents Woreda Selected rural kebele Agroecology Total household heads Sample household heads Administration Male Female Total Male Female Total Enekoy Wurch 1689 39 1728 144 3 147 Kuyzuria Dega 989 104 1093 84 9 93 Debet Weynadega 1259 92 1351 107 8 115 Total - - 3937 235 4172 335 20 355 5.3. Results and Discussion 5.3.1. Accuracy assessment

Accuracy assessment is used to determine the correctness of the classified image. In this study, the accuracy assessment is performed for 1985, 2002, and 2017 classified images using ground control points collected using a topographic map, Woody Biomass Inventory and Strategic Planning Project data, knowledge from the elderly, field visits, and Google earth data sets. The overall accuracy of the three classified images is shown in Table 5.3. The classification's overall accuracy for the years 1985, 2002, and 2017 is 85.7%, 87.9%, and 89%, respectively, with kappa values of 0.82, 0.86, and 0.86. The accuracy statistics and kappa coefficient values presented in Table 5.3

are well above the recommended values (Congalton & Green, 2019). These demonstrated a successful classification of the Landsat images of 1985, 2002, and 2017. Table 5.3 Accuracy assessment results

1985 2002 2017 Land cover class User Producer User Producer User Producer accuracy accuracy accuracy accuracy accuracy accuracy Cultivated 82.8 80.0 89.3 83.3 87.9 85.0 Grassland 80.7 83.3 80.6 88.5 81.8 90.0 Forest 90.0 90.0 90.7 89.0 96.1 89.1 Shrub-bushland 82.1 83.7 84.5 89.0 82.8 87.3 Urban 97.4 95.0 97.8 90.0 100.0 94.0 Overall accuracy 85.7 85.7 87.9 87.9 88.9 88.9 Kappa Statistics 0.82 0.82 0.86 0.86 0.86 0.86

5.3.2.Trends of land use land cover change in muga watershed

The produced LULC maps of the Muga watershed for the three reference years (1985, 2002, and 2017) are presented in Figure 5.2. The trend analysis made for two consecutive periods, 1985- 2002 and 2002-2017, showed spatiotemporal changes in LULC classes.

Figure 5.2 Classified LULC maps for 1985, 2002, and 2017 in the Muga Watershed As can be seen from Table 5.4, cultivated land has remained the dominant land use type in the study watershed. The LULC map of 1985 showed 66.1% of the watershed area was under cultivation, followed by grassland (19.1%) and shrub-bushland (11.2%). The proportions of forest and the urban areas were 3.0% and 0.6%, respectively. The 2017 LULC map showed that 74% of the watershed area was covered by cultivated land followed by grassland (11.4%) and shrub-bushland (9.8%). The area under forest and urban was estimated at 2.4 % for each land use class. Over 32 years, cultivated land and urban areas showed an increasing trend of 0.4% and 8.5% -1 year , respectively. While grasslands, shrub-bushlands, forest areas showed declining trends of 1.3%, 0.4%, and 0.7% year-1, respectively.

Comparing the two study periods indicated that cultivated land increased consistently at a rate of 6.2% in the first period (1985-2002) and 5.4% in the second period (2002-2017), and at a rate of 102.3 and 107.2 ha year-1, respectively. Similarly, urban areas increased by 2.2% and 0.4% per year during the first period (1985 and 2002) and the second period (2002 and 2017) of the study, respectively. As indicated in Table 5.4, urban areas showed significant increases with higher amounts than cultivated lands during the study period (1985-2017). On the other hand, a significant decrease in the watershed area under grassland was observed (1985: 19.1%; 2017:11.4%), shrub- bushland (1985:11.2%; 2017:9.8%), and forest (1985:3%; 2017:2.4%) (Figure 5.3 and Table 5.4). Although shrub-bushlands and forest lands showed a declining trend in the study period, these LULC classes showed an increasing trend of 3.3% and 0.6% year-1 in the second period (2002- 2017), respectively. In contrast, a higher rate of reduction in grassland was observed in the second period. The area's increasing tendency under shrub-bushland and forest cover in the second period was likely due to improved land use and management practices occurring (such as rehabilitation/ area closure and reforestation) by the local community.

32000 28000 1985 24000 2002 2017 20000

LULC 16000 12000 8000 4000 0

Land use/ land cover

Figure 5.3 Land use and land covers and their proportionate area in 1985, 2002, and 2017 in the Muga watershed

Table 5.4. LULC type in 1985, 2002, and 2017 and rate of change for the periods 1985-2002, 2002-2017, and 1985-2017 in ha and % in Muga watershed

1985 2002 2017 1985-2002 2002-2017 1985-2017 LULC Types Area (ha) % Area (ha) % Area (ha) % Area (ha) % Area (ha) % Area (ha) % Cultivation 28092.2 66.1 29830.6 70.2 31438.4 74.0 1738.4 6.2 1607.9 5.4 3346.2 11.9 Grass land 8103.8 19.1 7467.1 17.6 4862.7 11.0 -636.7 -7.9 -2604.4 -35.0 -3241.1 -40.0 Shrub-bushland 4742.6 11.2 3422.3 8.1 4170.5 9.8 -1320.3 -28.0 748.2 21.9 -572.1 -12.1 Forest 1282.8 3.0 859.3 2.0 1008.7 2.4 -423.5 -33.0 149.4 17.4 -274.1 -21.4 Urban 274.1 0.6 916.1 2.2 1015.1 2.4 642.0 234 99.0 10.8 741.0 270.3

The change matrix analysis presented in Table 5.5 shows the LULC shift from one class to the other. The highest proportion of transition occurred among cultivated, grassland and shrub- bushland LULC types over the study period. For instance, between 1985 and 2017, about 7495.2 ha of cultivated land was mainly due to the shift from grassland (5138.5 ha) and shrub-bushland (2356.7 ha). Likewise, about 2068.7 ha of shrub-bushland was mainly due to the shift from grassland (1575.9 ha) and Forest areas (492.7 ha). The area under different LULC classes that remained unchanged was estimated as 24427.2 ha, 604.2 ha and 1732.1 ha in the cultivated, forest, and shrub-bushland, respectively, during 1985 and 2017. Similarly, about 18.2 ha and 1077.2 ha of urban and grassland remained unchanged. Therefore, it could be concluded that the largest transition was observed from grassland to cultivation, from grassland to shrub-bushland, and from shrub-bushlands to cultivation.

The highest net gain was revealed in cultivation (3954.6 ha) and urban (673 ha) while the highest net loss was observed in grassland (-3648.7 ha) followed by shrub-bushland (-665.4 ha) and forest (-275.3 ha). The net change is the difference between the total gain and the total loss in a specific LULC class. Negative values of net change indicate net loss and positive values indicate net gain.

Table 5.5. LULC change transition matrices(hectare) for the initial & final reference (1985-2017)

Land use Land cover 2017 Urban Cultivation Forest Grassland Shrub-bushland Total Loss Urban 18.2 156.3 0.0 22.9 0.1 197.5 179.3 Cultivation 594.7 24427.2 18.8 3017.5 225.5 28283.7 3856.5 Forest 3.2 159.5 604.2 18.2 492.7 1277.8 673.6 Grassland 220.0 5138.5 33.1 1077.2 1575.9 8044.7 6967.6 Shrub-bushland 34.5 2356.7 346.4 222.1 1732.1 4691.7 2959.6 Total 2017 870.5 32238.2 1002.5 4357.8 4026.4 27858.8a Gain 852.3 7811.0 398.3 3280.7 2294.3 Loss 179.3 3856.5 673.6 6929.4 2959.6 Net change 673.0 3954.6 -275.3 -3648.7 -665.4 a Is the sum of diagonal areas in hectare and indicates the total area which remained unchanged b Net change = gain-loss

LULC change is unavoidable as it is derived from social and economic development (Wu et al., 2008). In this regard, recent LULC change studies conducted in Ethiopia reported substantial and rapid LULC changes (Tadesse et al., 2017; Miheretu & Yimer, 2018). This study also revealed that tremendous LULC changes, particularly in urban, forest, and grassland areas were observed

in the last three decades. Similar results were reported by prior LULC change studies in the highlands of Ethiopia. For instance, in the central highlands of Ethiopia (Gessesse & Bewket, 2014), in the Hirmi watershed of Northern Ethiopia (Gebrelibanos & Assen, 2015), in the Libokemikem district of South Gonder Zone (Demissie et al., 2017), and Gelda catchment of the Lake Tana (Hassen & Assen, 2018), reported the expansion cultivated land at the expense of forest and grassland. Expansion of cultivated land often comes with many environmental costs as it is being practiced without proper land management.

Direct and indirect driving factors have caused LULC changes in the Muga watershed. Among the direct drivers of LULC change, expansion of farmland, cutting of trees for fuelwood and construction purpose, expansion of road access and public institutions, settlement and urbanization, and lack of alternative sources of income were the main proximate drivers in the Muga watershed. This result is concurrent with previous studies in Ethiopia (Bewket, 2002; Yalew et al., 2016; Demissie et al., 2017; Miheretu & Yimer, 2018; Kindu et al., 2015). The key informants and focus group discussants indicated that an abundant livestock population compared to the limited amount of available grazing land created high pressure on grassland, thereby, caused overgrazing. Overgrazing, expansion of croplands, and settlement in the study area, grassland has declined over time. This is in line with previous studies (Mengistu et al., 2012; Gessesse & Bewket, 2014; Kindu et al., 2015; Wubie et al., 2016), which confirmed the expansion of agriculture was the main driver of LULC changes. In contrast, Tegene (2002) concluded that the expansion of farmland was not one of the primary drivers for changes in LULC change in the Derekolli catchment area of the South Wollo zone of the Amhara region.

Among the direct factors, infrastructure development such as road access and market center is one of the drivers of LULC changes in the study watershed. In contrast, Kindu et al. (2015) indicated that road access expansion was not responsible for LULC changes in the Munessa-Shashemene landscape.

According to Tekleab et al. (2013) and Mengistu et al. (2014), the upper Blue Nile basin temperature increased during the last three decades. Climate variability could lead to land conversion to different land uses. This study showed that climate variability, which was recognized by 65.2% of the sample households, was one of the drivers of LULC changes in the study

watershed. Gessesse & Bewket (2014) also identified climate as a major driver of LULC changes in the central highlands of Ethiopia. It was understood from key informants and FGDs that in earlier times, teff was grown well at altitudes below 2400 m. However, teff is currently growing at higher altitudes up to 2900 m. This is most likely due to the changing of local climatic conditions, mainly the upper escarpment of the watershed becoming warm.

The results obtained from both key informant interviews and focus group discussions confirmed the decline of crop yields due to the high cost and limited access to agricultural inputs as contributors to the expansion of cropland which compensated for yield reduction. The interview results with elders revealed that the area under shrub-bushland and grassland decreased due to an illegal expansion of cultivated land in adjoining shrub-bushland areas. According to the evidence obtained during focus group discussions, the declining trend of grasslands in the study period will result in a scarcity of grazing land and animal fodder, which will likely lead to declining livestock production. Consequently, the expansion of areas under cultivation and the decline of areas under forest and grassland may eventually result in soil erosion and changes in the hydrological regime of the Muga watershed. Table 5.6, shows that 82% of the respondents claimed that both the area under forests and grasslands have decreased in the last three decades (between 1985 and 2017). In contrast, about 78.5% and 67% of the respondents indicated that the area under cultivation and urban has increased.

Table 5.6. Farmers’ perception and observed LULC changes from 1985-2017. Numbers indicate the percentage that perceived the observed changes per LULC type

Perceived 1985-2017 change Cultivation Forest Grassland Shrub-bushland Urban Increased 66.7 12.1 13.6 48.5 78.5 No change 7.6 6.1 4.5 15.2 4.6 Decreased 16.9 81.8 81.8 36.4 16.9 5.3.3. Drivers of land use/ land cover dynamics in muga watershed

The respondents identified 12 factors (proximate and underlying drivers) as important drivers contributing to LULC changes in the study watershed (Table 5.7). Expansion of farmland, settlement, and urbanization, lack of alternative sources of income, expansion of road access and public institutions, distribution of communal lands to landless youths and returnee soldiers, small- scale irrigation, land-related policies, cutting of tree for fuelwood, and construction purposes were

the top eight drivers of LULC changes in the study area. Similar results were also revealed from key informant interviews and FGDs in which expansion of cultivated land, cutting of tree for fuelwood and construction purpose, expansion of road access and public institutions, and agricultural expansion were identified as the main causes of LULC in the study area over the last three decades. Expansion of farmland, settlement, urbanization and lack of alternative sources of income are the top three proximate drivers of LULC changes in the Muga watershed during the study period. According to the household survey analysis, about 82% of the respondents perceived that the lack of alternative sources of income apart from agriculture contributed to LULC changes in the study watershed. As a result, young and landless peasants, who sustain their livelihood by making charcoal and cutting trees for cultivation land, were among the major causes for the decline in forest resources in the study watershed.

Focus group discussants noted that farmers have been practicing subsistence farming with inappropriate land management practices and very low additional farm inputs. Hence, such poor land management practices led to a decline in crop productivity in the study watershed. Irrespective of their age and sex, all focus group discussants noted that the inhabitants' per capita income in the study watershed is declining. Focus group discussants and key informant interviewees indicated that the abundant livestock population in the study watershed compared to the available grazing land created high pressure on grassland, thereby overgrazing. Thus, the proportion of grasslands showed a declining trend over time due to overgrazing, expansion of cropland, and settlement. The participants in FGD also indicated that public institutional development such as roads and schools were among the drivers of the LULC changes in the study watershed.

This study also revealed that the land-related policies were among the identified driving factors of LULC change in the study area, as indicated in Table 5.7. The focus group discussion participants stated that lack of law enforcement belongs to farmers' illegal encroachment onto shrub-bushlands, forest and grazing lands has contributed to LULC changes in the study area. Moreover, the periodic redistribution of land to farmers is among policy-related drivers of LULC changes in the study watershed.

According to FGD participants and selected households, extraction of wood for fuel and construction purposes is also one of the main factors contributing to changes in LULC in the study watershed. Obviously, in areas like the Muga watershed, where most of its inhabitants depend on

wood for fuel and other purposes, there is an ever-increasing high demand for fuelwood and construction purposes. This is confirmed by 83% of the respondents, who used fuelwood as the main energy source for cooking. Only 17% of the respondents used an alternative source of energy such as cattle dung and crop residues for cooking. Thus, it can be inferred that the decline in forest resources in the study watershed is partly due to an ever-increasing wood extraction.

Although the demand for fuelwood and construction purposes coupled with poverty and shortage of land forced farmers to cut natural forests, a slight increase in plantation forest was observed in the study watershed between the 2002 and 2017 study period. This slight increase may be attributed to the government-supported reforestation initiatives implemented in the area since the 1990s. Table 5.7 Drivers of LULC change in the Muga watershed

% of respondents Drivers of LULC change yes No Expansion of farmland 97.0% 3.0% Settlement and urbanization 87.9% 12.1% Land related polices 71.0% 29.0% Distribution of communal lands to landless youths and returnee soldiers 78.8% 21.2% Cutting of tree for fuelwood and construction purpose 70.0% 30.0% Expansion of small trade and enterprise 55.0% 45.0% Expansion of small irrigation schemes. 77.3% 22.7% Soil erosion and/or soil fertility decline 75.0% 25.0% Financial capital of rural farmers (poverty) 65.0% 35.0% Expansion of infrastructure such as roads, health centers, and schools, etc. 81.8% 18.2% Climate variability 65.2% 34.8% Lack of alternative sources of income 82.0% 18.0%

Yes No Absence of alternative sources of income Climate variability Expansion of infrustructures such as road,… Financial capital of rural farmers (poverty) Soil erosion and/or soil fertility decline Expansion of small irrigation schemes. Expansion of small trade and enterprise Cutting of tree for fuel wood and… Distribution of communal lands to landless… Drivers of LULC chnage Drivers of LULC Land tenure system. Expansion of settlement and urbanization Increasing demand for farming. 0.0% 15.0% 30.0% 45.0% 60.0% 75.0% 90.0% Response in percent

Figure 5.4 Drivers of LULC change in the Muga watershed

As it is shown in Table 5.7 and Figure 5.4, expansion of farmland, expansion of settlement, poverty, land use-related policy, lack of alternative sources of income, and high cost of agricultural inputs for production were the major underlying drivers of LULC change. According to the Central Statistical Agency (CSA) of Ethiopia reports, the population of the Debay Tilatgin district, where more than 90% of the study area is located, increased from 99,840 in 1994 (CSA, 1999) to 142,124 in 2017 (CSA, 2013). This revealed that the study watershed had experienced rapid population growth, which subsequently impacts LULC of the area due to the increasing demand for food, settlement, fuel and construction wood. The expansion of cultivated land and urban areas at the expense of grasslands and forest areas is a typical example. Thus, population pressure was one of the major drivers of LULC change in the study watershed. Similar results were reported by earlier studies in the Ethiopian highlands (Bewket, 2002; Wubie et al., 2016; Yalew et al., 2016).

During the key informant interview, key informants indicated that population growth was an important factor of LULC changes in the Muga watershed. The survey results also showed that the household's average landholding size was 0.93 ha (3.72 timad) with an average family size of five per household, less than 1 ha. Similar studies in different parts of Ethiopia also showed the scarcity of cultivated land (Mengistu et al., 2012; Kindu et al., 2015; Wubie et al., 2016). It was understood

from key informants and household survey results that the increasing population has contributed to the scarcity of farmland. This will likely result in the conversion of forests and pastures to agricultural land and contribute to land degradation in the study area. According to Malthus (1959), a growing rural population increases the demand for agriculture to feed the ever-increasing population, leading to the expansion of agricultural land into areas of marginal land and resulting in land fragmentation and decreased productivity. These are pathways to poverty and environmental degradation. However, Boserup (1988) considered that arable land becomes scarce as a population grows, necessitating people to intensify agricultural production. From this perspective, agricultural change is driven primarily by the local population's changing consumption needs resulting from population growth. The finding of this study is in line with previous studies conducted in Ethiopia. Similar studies conducted in Ethiopia showed that population growth increases the demand for further farmland and urban areas, and this will ultimately lead to land degradation (Bewket, 2002; Mengistu et al., 2012; Ariti et al., 2015; Gebrelibanos & Assen, 2015; Ayele et al., 2016; Wubie et al., 2016; Gashaw et al., 2017; Tadesse et al., 2017; Mekonnen et al., 2018).

Land use related policies are also key underlying factors in explaining LULC change. In this study, over 70% of the respondents considered land use-related policies as drivers of LULC changes in the study area. This is also supported by an interview with key informants responsible for managing the land resources noted that the utilization of land resources and land management investment decisions in the study area are connected. In line with this study, it has been argued that land tenure and land resource-related policies of the government had a major influence on LULC changes (Mengistu et al., 2012; Kindu et al., 2015).

The other underlying driver of LULC change identified was the lack of alternative sources of income. Sample respondents have recognized that a lack of alternative sources of income apart from on-farm activities contributed to LULC changes. The key informants and focus group discussants revealed that young and landless peasants have less opportunity for off-farm activities. This is most likely the reason for the expansion of cultivated lands at the expense of forest and grazing lands, as one of the opportunities for generating incomes and coping strategies during the decline or failure of yield production. In support of this, Gebrelibanos & Assen (2015) indicated

that a lack of alternative sources of income was an important underlying driver of the observed LULC changes.

5.4. Implications of Land Use Land Cover Change in the Muga Watershed

LULC change is necessary and essential for economic development and social progress (Wu et al., 2008) but often comes with substantial environmental costs. The rapid expansion of cultivated lands and urban areas at the expense of forest, grassland, and shrub-bushland have a severe impact on land resources of the Muga watershed. The conversion of grasslands and forests to cultivation and urban development also affected the hydrological cycle and increased soil erosion and runoff. This leads to land degradation, which may affect the livelihood of the local community. Proper management of grassland, forest, and shrub-bushland has immense importance for the sustainability of the study watershed resources.

5.5. Conclusion

In this study, the LULC change of the Muga watershed over the last three decades was analyzed. This study revealed that a significant part of the study area has been changed during the study period. Cultivated and urban areas showed an increasing trend of 12 and 270%, respectively, while grasslands, shrub-bushland and forest lands showed a declining trend of 40%, 12% and 21%, respectively. This result indicates that the conversion of grassland, forest and shrub-bushlands to cultivation and urban areas may have far-reaching environmental problems such as land degradation (e.g., soil erosion and soil fertility loss) and change in the local hydrological regime. Lack of alternative sources of income further intensified the vulnerability of the Muga watershed for land degradation. Thus, the rate of soil erosion caused by land conversion is anticipated to increased in the study area unless land management actions are undertaken. It was identified that various proximate and underlying factors driven LULC change in the Muga watershed. Expansion of farmland, settlement, and urbanization, lack of alternative sources of income, expansion of road access and public institutions, distribution of communal lands to landless youths and returnee soldiers, expansion of small-scale irrigation, land-related policies, cutting of tree for fuelwood and construction activities were among the major factors of LULC change in the study area. A major modification of LULC is anticipated unless an appropriate and integrated approach in implementing policies and strategies related to land resource management interventions are undertaken. Hence, this study suggested integrated watershed management, including proper land

use planning and management interventions with the local community active participation, to prevent undesirable LULC changes in the watershed. The interventions may include: creating alternative sources of income for young and landless peasants in rural areas, creating awareness among the farmers, limiting cultivated land expansion through intensification of agricultural production using improved land management practices and affordable agricultural inputs.

Chapter 6 6. Impact of Land Use/ Land Cover and Climate Change on Soil Erosion in Muga Watershed, Abbay River Basin, Ethiopia Abstract Soil erosion is one of the major threats in the Ethiopian highlands. In this study, soil erosion in the Muga watershed of the Abbay River Basin under historical and future climate and land use/ land cover (LULC) change was assessed. Future LULC was predicted based on LULC map of 1985, 2002, and 2017. LULC maps of the historical periods were delineated from Landsat images, and future LULC was predicted using the CA-Markov chain model. Precipitation for the future period was projected from six regional circulation models. The RUSLE model was used to estimate the current and future soil erosion rate in Muga watershed. The results of the study show that the average annual rate of soil erosion in the Muga watershed was increased from approximately 15 t ha-1 year-1 in 1985 to 19 t ha-1 year-1 in 2002 and 19.7 t ha-1 year-1 in 2017. If a proper measure against the LULC changes is not taken, the soil loss rate is expected to increase and reach about 20.7 t h-1 yr-1 in 2033. The results of the study also show that in the 2050s, soil erosion is projected to increase by 9.6% and 11.3% under RCP4.5 and RCP8.5, respectively compared with the baseline period. When both LULC and climate changes act together, the mean annual soil loss rate shows a rise of 13.2% and 15.7% in the future under RCP4.5 and RCP8.5, respectively, which is due to synergistic effects. The results of this study can be useful for formulating proper land use planning and investments to mitigate the adverse effect of LULC and climate change on soil loss. The study also demonstrated the importance of an integrated approach, assessing the combined impacts of LULC and climate change, for accurate estimation of soil losses.

Keywords: Land use land cover, Climate change, Impact, Soil erosion, Cellular Automata- Markov

6.1. Introduction

Soil is an indispensable resource, however, it has been affected by human beings since the beginning of agriculture (Amundson et al., 2015). LULC and climate are expected to change in the future as a result of human activities (Field & Barros, 2014) and expected to influence soil (Mengistu et al., 2015; de Hipt et al., 2019; Anache et al., 2018; Berberoglu et al., 2020; Hu &

Gao, 2020). Soil erosion has become a significant threat to terrestrial ecosystems, and it denotes a vital environmental risk (Sun et al., 2013).

Abbay river basin is one of the most diversified and noteworthy river basins in Ethiopia (Bewket & Teferi, 2009; Melesse et al., 2009; Demessie, 2015; Yalew et al., 2016; Gelete et al., 2019). Currently, the river basin faces severe environmental challenges such as soil erosion, land degradation, loss of soil fertility, and deforestation (Steenhuis et al., 2013; Demessie, 2015; Mengistu et al., 2015; Haregeweyn et al., 2017; Tadesse et al., 2017; Wubie et al., 2016; Yesuph & Dagnaw, 2019).

In Ethiopia, soil erosion has been recognized as a serious environmental problem (Tamene & Vlek, 2008; Haregeweyn, Tsunekawa, et al., 2015; Haregeweyn et al., 2017). The loss of top fertile soil by water erosion creates severe limitations to sustainable agricultural land use, which leads to reduced soil productivity and food insecurity (Hurni et al., 2015).

Soil erosion jeopardizes the sustainability of agriculture and leads to siltation of streams, lakes, dams, and reservoirs (Haregeweyn et al., 2017; Yaekob et al., 2020) and downstream ecosystems impacts (Fenta et al., 2016). As a result, it can have serious implications for irrigation agriculture and related investments in different parts of the country. Thus, soil erosion caused by water is a severe problem of the country both in its onsite (e.g. productivity loss, soil fertility loss) and off- site effects (e.g. siltation) (Demessie, 2015; Wagena et al., 2016; Berihun et al., 2019; Chimdessa et al., 2019; Yaekob et al., 2020; Aneseyee et al., 2020; Belihu et al., 2020). The risk varies across different areas, depending on the watershed landscape, local climatic conditions, soil characteristics, land use, and soil and water conservation and management practices. Therefore, it is a potential threat to the national food supply.

Numerous studies in Ethiopia (e.g., Mengistu et al., 2015; Taye et al., 2018; Tadesse et al., 2017; Haregeweyn et al., 2017; Gelagay & Minale, 2016; Molla & Sisheber, 2017; Hassen & Assen, 2018; Miheretu & Yimer, 2018; Anache et al., 2018; Ebabu et al., 2019; Berihun et al., 2019; Kidane et al., 2019; Aneseyee et al., 2020) reported that changes in LULC and climate could significantly affect the intensity of soil erosion at various spatiotemporal scales. For instance, Mengistu et al. (2015), reported the influence of temperature and precipitation on biomass

production and soil organic carbon, thereby increasing soil erosion in the Abay river basin. An increase in precipitation levels and intensity will lead to intensified erosion (Field & Barros, 2014).

Similarly, studies on the soil erosion caused by LULC and climate change were conducted in the Abbay river basin of Ethiopia (e.g., Steenhuis et al., 2013; Demessie, 2015; Mengistu et al., 2015; Wubie et al., 2016; Haregeweyn et al., 2017; Moges & Bhat, 2017; Tadesse et al., 2017; Aneseyee et al., 2020). For instance, a study by Moges & Bhat (2017) in Rib watershed, and Tadesse et al. (2017) in Yezat watershed of the Abbay River basin, Ethiopia, showed that change in LULC of an area led to soil erosion. Mengistu et al. (2015) in the Abay river basin reported that climate change leads to soil erosion. These studies indicate that LULC change and changing precipitation patterns have an impact on soil erosion. Thus, to better understand the effects of LULC and climate change on soil erosion at a watershed level, a holistic and multidisciplinary approach is required. Modeling future soil loss rates due to LULC and climate change is an essential step (Mullan et al., 2012).

The effects of change in LULC and climate on soil erosion can be investigated by associating field measurements of soil erosion variables with different LULC classes and climate data (Adugna & Abegaz, 2016; Ebabu et al., 2019). However, measurement of soil erosion is often not possible over the required temporal and spatial scale. Thus, field studies have to be complemented by soil erosion model simulation (Giri et al., 2015; Woznicki et al., 2016). Furthermore, results from soil erosion models have been used to predict impacts of LULC and climate change on soil erosion and as a scientific basis for soil erosion control and management at different scales (Addis & Klik, 2015; Serpa et al., 2015; Woldesenbet et al., 2018; Anache et al., 2018).

Some researchers have developed and used different LULC change models depending on their study objectives and backgrounds. For instance, Araya & Cabral (2010) in Setúbal and Sesimbra, Portugal; de Oliveira Barros et al. (2018) in Montes Claros, Brazil; Omrani et al. (2017) in Luxembourg; Arsanjani et al. (2013) in Tehran; (Xie et al., 2007) in China. Currently, the most widely used models in LULC change monitoring and prediction in Ethiopia are cellular and Markov chain models in Ethiopia (e.g., Gidey et al., 2017; Gashaw et al., 2018; Hishe et al., 2020; Kura & Beyene, 2020; Fitawok et al., 2020; Mohamed & Worku, 2020), and these models are verified in the Ethiopian context. Accordingly, the incorporation of GIS, CA-Markov model,

climate model, and soil erosion models have been used to assess the effects of change in LULC and climate on soil erosion.

Studies including Maeda et al. (2010) in Kenya; Plangoen et al. (2013) in Thailand; Routschek et al. (2014) in Saxony, German; Ferreira et al. (2015) in the south of Portugal; Mengistu et al. (2015) in the Abbay River Basin of Ethiopia; and Perović et al. (2019) in Vranjska Valley of Serbia were conducted to evaluate the potential impacts of climate change on soil erosion using soil erosion models with different scenarios. On the other hand, Sharma et al. (2011) in India; Plangoen et al. (2013) in Thailand; Ferreira et al. (2015) in the south of Portugal; de Hipt et al. (2019) in Burkina Faso, applied soil erosion model to simulate the potential effects of LULC change on soil erosion. The combined and separate effect of the current LULC and climate change on soil erosion were evaluated by several studies (e.g., Ferreira et al., 2015; Mengistu et al., 2015; de Hipt et al., 2019; Perović et al., 2019). Most of the studies focused on the separate effects of LULC and climate changes on soil erosion in different parts of the world. Although some researches were conducted on the combined and individual impacts of climate and land-use changes on soil erosion, the context is not yet well understood in the Abbay River Basin. In addition, the environmental impact of future LULC and climate are still contentious issues and unresolved problems and require further research (Simane et al., 2013; Demessie, 2015). Hence, predicting the impact of future LULC and climate changes on soil erosion is very important for land use planning and watershed management issues.

Therefore, this study aims to analyze the impact of LULC and climate change on soil erosion in the Muga watershed, Abbay River Basin, Ethiopia. This study will help the natural resource experts and land-use planners to understand the effect of future LULC and climate change in the Muga watershed. The results will also help land use planners to improve future land use planning to reduce soil erosion in the study watershed and watersheds with similar settings. Moreover, applying the CA-Markov chain model, climate model, and RUSLE model at the watershed level is the main contribution of this study in Ethiopia.

6.2. Materials and Methods

In this study, LULC from Landsat imageries, historical and future climate model outputs, historical climate datasets are used as model input. Thus, future land cover expansion and climate change

scenarios are also considered for their potential impacts on soil erosion in the Muga watershed. An integrated method of erosion and climate model and LULC datasets were used to achieve this. The procedure of this study is summarized in Figure 6.1. A detailed description of the datasets and methodology are provided in the subsections.

Historical & future Mean annual rainfall Soil data DEM Landsat images (1985, 2002 & 2017) CORDEX rainfall data data (1985-2017)

CSIRO-Mk3.6.0 Image classification GFDL-ESM2M Soil Type Flow Slope (%) HadGEM2-ES Organic matter direction Land use/Land cover Map of Extract point content of 1985, 2002 & 2017 CanESM2 data & bias different soils types Flow NorESM1-M correction accumulation Change analysis MIROC Soil Erodibility (K) factor Topographic Future prediction of (LS) factor land cover (2033) Future projection Interpolation (2018-2050) Conservation practice (P) factor Cropping and Land Rainfall Erosivity Cover (C) factor (R) factor

RUSLE Model

A (t ha-1 yr -1) = R*K*LS*C*P Soil loss estimation

Figure 6.1 Workflow of the methodology developed in this study 6.2.1. Image processing and preparation of spatial drivers

The LULC status of the study watershed for the years 1985, 2002, and 2017 was studied in the previous part of this study. In this study, the CA-Markov chain model was used to simulate and predict the future land cover map of 2033. Different sets of spatial and non-spatial data were used as input to the CA Markov model. The simulation and prediction processes were performed using LULC maps of 1985, 2002 and 2017, roads and towns, distance to a river, slope, and elevation maps. The LULC maps of 1985, 2002, and 2017 were mainly used to generate transitional matrix using the Markov chain process. Besides maps of a slope, road, settlement and elevations maps were used to create transitional potential maps. The datasets were used in combination to predict future change in the LULC using the CA Markov chain model. Potential land-use change drivers were identified through literature reviews (Bewket & Teferi, 2009; Teferi et al., 2013; Gidey et

al., 2017; Hishe et al., 2020) and field interviews with farmers, local farming experts, regional land bureau officials, and through spatial correlations.

A digital elevation model (DEM) of 30 m resolution from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) was used to prepare the watershed's slope, elevation, and stream network of the study watershed. The watershed road network was also downloaded from OpenStreetMap (https://www.openstreetmap.org/), and its geometric consistency was verified in QGIS software. The population dataset for the study area was obtained from the Ethiopian Central Statistical Agency (CSA). Consequently, these data sets were produced using ArcGIS software and then exported to IDRISI Selva to run the potential transition maps. The list and sources of datasets used to analyze future land use map and RUSLE model in the Muga watershed is presented in Table 6.1.

Table 6.1 Dataset sources and types used for RUSLE model

No. Name of data Sources 1 Digital Elevation Model (DEM) (30 m) USGS 2 Soil data of the watershed MoWIE 3 Daily precipitation data from 1985 to 2017 National meteorological agency of Ethiopia 4 Road OpenStreetMap 6.2.2. Climate Data Two climate datasets; baseline and modeled climate data, were used in this study. The baseline climate data were used to estimate the baseline soil loss and validate the predicted climate scenarios. The rainfall data were compiled from seven meteorological stations operating within and around the study watershed from 1985 to 2017; the data obtained from the National Metrological Agency (EMA) of Ethiopia.

Climate models were used to quantify the relative change in the current and future climate, often used as an input to the soil erosion models. This study used six regional climate models (RCMs) with the driving model ICHEC-EC-EARTH under CORDEX-Africa (Fick & Hijmans, 2017; Shamir et al., 2015). CORDEX-Africa provides projected climate outputs at a relatively higher spatial resolution (50 km). The models were: HadGEM2-ES, CSIRO-Mk3.6.0, GFDL-ESM2M, CanESM2, MIROC, and NorESM1-M, because these models are appropriate based on earlier research conducted in other parts of Ethiopia (Alemseged & Tom, 2015; Teklesadik et al., 2017; Worku et al., 2018).

The representative concentration pathways (RCPs) scenarios RCP4.5 and RCP8.5 were considered for this study to drive the RUSLE model (Alemseged & Tom, 2015). Policymakers usually focus on events occurring in the 2050s compared to the events in the far future (Weber, 2006). Thus, the authors select climate change data from 2018 to 2050 rather than the end of the century.

The CORDEX grid points were extracted using MATLAB software. It is recognized that climate model output data contain systematic errors and cannot be used immediately in soil and hydrological simulations (Christensen et al., 2008; Teutschbein & Seibert, 2010, 2012; Chen et al., 2016). Thus, the RCM climate data outputs in CORDEX-Africa under emission scenarios RCP4.5 and RCP8.5 were bias-corrected for this study. 6.2.3. Land use land cover prediction using CA Markov chain model

Several LULC change models were developed (e.g., Veldkamp & Lambin, 2001; Parker et al., 2003; Verburg et al., 2004; Koomen & Stillwell, 2007). Some of the most widely used LULC models include statistical models (regression) (Yalew et al., 2016), Cellular automata (CA), and Markov chains (Araya & Cabral, 2010; Han et al., 2015; de Oliveira Barros et al., 2018; Munthali et al., 2020), evolutionary models (neural networks) (Omrani et al., 2017), and multi-agent-based models (Xie et al., 2007; Arsanjani, Helbich, & de Noronha Vaz, 2013). These approaches are often combined to create an integrated model that determines the probabilities of LULC changes. Frequently, the choice of models depends on the study's objective and the level of complexity required (Nainggolan et al., 2012).

The Markov Chian (MC) model of LULC change has been widely used to predict LU changes (Sang et al., 2011) and usually integrated with a cellular automata (CA) model with an agent-based model to express the human interaction on the landscape (Xie et al., 2007; Ralha et al., 2013). This is due to its reliability and compatibility with many geospatial technologies (Halmy et al., 2015). The reliability and compatibilities of the CA and Markov chain model was checked in the Ethiopian context by Gidey et al. (2017); Gashaw et al. (2018); and Hishe et al. (2020). Thus, this study used the CA-Markov chain model integrated with the multi-criteria evaluation to predict LULC of 2033 in the Muga watershed. The Markov chain model provides the probability and extent of LULC change, whereas the CA model operates on neighborhood interactions and spatial distribution.

The complex interrelationships between the physical and human factors lead to the conversion from one land use type to another (Eastman, 2012). Thus, with the CA-Markov model, a multi- criteria evaluation (MCE) technique was used to support land allocation decision process using different land use criteria (Al-sharif & Pradhan, 2014; Deng et al., 2015). The MCE is useful for studying the land use suitability for the possible conversion of certain land use to another, revealing each criterion's influence and importance (Yalew et al., 2016; Gidey et al., 2017). 6.2.3.1. Preparation of suitability maps

LULC can be driven by a multitude of socio-economic and biophysical factors. The future change in LULC can be determined by the inherent change and external factors/ spatial variables such as the proximity to towns, distance from a river and urban, elevation, slope, and areas suitable for each change in each class. A slope was considered as a constraint for cultivation and urban areas because steep slopes prohibit both cultivation and urban expansion.

An integrated evaluation procedure was used to generate potential transition maps based on biophysical and socioeconomic indicators. In this regard, more than five biophysical and socioeconomic variables including elevation, slope, distance to towns, and distance to roads were considered. The selected transition potential maps are in different units. Therefore, the maps were converted into a uniform measurement scale through standardization techniques for weighted overlay analysis (WOA) (Reshmidevi et al., 2009; Zabihi et al., 2015). Once all the maps were standardized, the weight for each criterion was calculated using the analytical hierarchy process (AHP) (Yalew et al., 2016; Gidey et al., 2017).

Consequently, specific weights were assigned to each factor (Hishe et al., 2020) and used to compute suitable maps in IDRISI software. The relative weights for a group of factors were defined based on the authors' knowledge and experience about the studied landscape, a review of the scientific literature, and the farmers' and local experts' opinions. The highest weight value is the most influential factor, while the lowest is a less important LULC change factor. After determining each criterion layer's relative importance through a pairwise comparison matrix, these values were entered using IDRISI software to produce associated weights and consistency ratio value.

Table 6.2 shows inputs to the pair-wise comparison for the AHP analysis to determine weights. The weights produced from the AHP procedure using the inputs in Table 6.2 are between 0 and 1,

where 0 denotes the no probability and 1 the high probability. The consistency ratio of the pairwise comparisons for the computation of criteria weights is shown in Table 6.2, which is an acceptable range (Eastman, 2012). Consequently, a weighted overlay analysis was performed using IDRISI software. The final standardized factors are depicted in Figure 6.2.

Table 6.2 Factors and constraints considered and their weights for predicting LULC conditions in the Muga watershed

Land use types Factors Factor Consistency Constraint and classes weight considered Suitable areas for conversion to Cultivation 0.420 Proximity to developed land 0.269 Distance to rivers 0.109 Elevation 0.201 Suitable areas for conversion to Shrub-bushland 0.328 Proximate to developed land 0.396 Elevation 0.206 Suitable areas for conversion to Grassland 0.110 Proximate to developed land 0.582 Elevation 0.309 Forest Suitable areas for conversion to Forest 0.425 Slope (None) Proximate to developed land 0.213 Elevation 0.080 Suitable areas for conversion to urban 0.160 0.01 Proximate to developed land 0.166 Distance to urban 0.417 Distance to roads 0.160 Elevation 0.090

Figure 6.2 Standardized biophysical factors of LULC in the study area. (a) Elevation, (b) proximity to a stream, (c) slope, (d) proximity to road, (e) proximity to urban, (f) proximity to shrub-bushland, g proximity to cultivation, (h) proximity to grassland, (i) proximity to forest.

The 1985-2002 transition matrix and the 2002 transition potential maps were integrated to simulate LULC map of 2017. The data were integrated using the CA model to simulate LULC of 2017. The same procedure was used to predict the LULC of 2033. The transition matrix from 2002-2017 and transition potential map of 2017 were integrated to run the prediction for 2033.

6.2.4. Validation of land use land cover prediction model

Model calibration and validation are essential in predicting future decadal changes where no datasets are available for predicted data accuracy (Srivastava et al., 2014; Singh et al., 2015). To validate the model, the actual LULC map of the year 2017 was compared with the 2017 map simulated by CA-Markov and based on the kappa statistics and a comparison of each simulated LULC class with the real class has been used. According to the CA-Markov model's required preparation, a calibration map for 2017 was prepared. The number of repetitions in the model was set equal to the number of years between the reference map and the map predicted by the model (15 years).

The actual 2017 LULC map was then used as a reference map to compare with the 2017 simulated LULC map results. In this method, LULC classes of the simulated map were checked against LULC classes of the corresponding reference LUC map of 2017. The accuracy level of the simulated and reference LULC classes of 2017 was computed using the following formula;

2  Accuracy(%)100((pp )*100ij (12) Where, Pi: reference LULC 2017; Pj: simulated LULC 2017

Accordingly, the analysis result has shown consistency between the reference and the simulated LULC maps of 2017 (Table 3).

Furthermore, an attempt was also made to examine its accuracy using the kappa coefficient. The Kappa variations used to validate the CA-Markov model for LULC change predictions in this study were computing the kappa coefficient using a validated tool in IDRIS Selva software. Kappa statistics are used for testing accuracy are the traditional Kappa (Kstandard) or Kappa for no information/ability (Kno), Kappa for location (Klocation), and quantity of correct cells (Kappa for quantity). Kappa variations have been strongly recommended and widely used to validate LULC change predictions (Pontius Jr, 2002; Singh et al., 2015).

The Kappa coefficient for quality and location was computed. The statistics show that Kno is 0.8803, Klocation is 0.8270, KlocationStata is 0.8270, and Kstandard is 0.8141 (overall). The k values are greater than 0.8, which shows the model accuracy to predict future LULC change. In other words, the kappa coefficient result indicates the model's ability to specify grid cell level location of future change (Mondal et al., 2016). The result indicates that K values of 0.80 and above are considered strong and reasonable to make plausible future projections (Yang et al., 2014; Gidey et al., 2017). The LULC scenario was prepared for 2033, accounting for the implications of future land use on soil erosion due to the variation in C factor. Figure 6.3 shows the process for simulating the future LULC using CA-Markov chain Model. Table 3 Accuracy result of the simulated 2017 LULC

LULC type LULC referece (2017) LULC simulated (2017) Difference Difference (%) Accuracy (%) Cultivation 314.38 320.89 6.51 2.07 97.93 Grass land 48.63 44.51 -4.12 -8.46 91.54 Shrub-bushland 41.71 38.71 -3.00 -7.18 92.82 Forest 10.09 11.87 1.79 17.72 82.28

LULC 1985, 2002 & 2017

Markov chain

Conditional probability Slope (%) MCE (Weight factor) MCE Transition probability Proximity to road (m) AHP weight derivation Transition area file Proximity to river (m)

Proximity to bush/shrub land (m) Physical & socio Fuzzy set membership CA-Markov economic factors function (0-255) Proximity to grassland (m) Baseline LULC 2017

Proximity to forest (m) Transition area file

Proximity to built-up area (m) Transition suitability image

Proximity to cultivated land (m) Transition Simulation suitability images

Simulation 2017

Model analysis Simulations (2033) LULC Validation (Actual and simulated LULC 2017)

Figure 6.3 Schematic diagram for simulating the future LULC using CA-Markov chain Model

6.2.5. Climate projection and bias correction 6.2.5.1. Climate projection Climate scenarios have served as an essential tool in climate change research in the past and will likely continue to do so in the future. There are four greenhouse gas scenarios recently published in the fifth IPCC Assessment Report (AR5), called the representative concentration pathways named based on their possible range of radiative forcing values like (RCPs) 2.5, 4.5, 6, and 8.5. In this study, the researcher has decided to select two climate change scenario data (Representative Concentration pathways; RCP 4.5 and 8.5) that cover the entire range of radiative forcing from the newly available Coupled Model Intercomparison Project Phase 5 (CMIP5). The output of the RCM ensemble of the CORDEX for African domain projection was used as an input to the hydrological model.

RCP 4.5 scenarios are medium-term emissions scenarios at the stabilization level of about 4.5 2 w/m (about 650 ppm CO2 equivalent) not exceeding this value by the year 2100 (Meinshausen et al., 2011; van Vuuren et al., 2011), supposing that all countries around the world undertake emission mitigation policies (Thomson et al., 2011). RCP 8.5 scenarios are the worst-case scenario in terms of greenhouse gas emissions, with no clear climate policy. The RCP 8.5 scenario shows 2 a radiative forcing pathway leading to 8.5 w/m (greater than 1370 ppm CO2-equivalent) in 2100 (van Vuuren et al., 2011). Climate scenario data sets consist of a historical run (1976– 2005) and projection (2018–2050) with a spatial resolution of 0.44° based on the emission scenarios of RCP 4.5 and RCP 8.5. In this study, climate data up to 2050 (2018-2050) was used rather than considering the end of the century, as small and large dams (with a lifespan of 30 to 50 years) in the Abbay River Basin of Ethiopia, including the study watershed has been studied, designed, and started to implement for irrigation and hydropower purposes, which will be affected by a changing climate in the short period of time; and policymakers often focus on what is happening sooner than what is happening in the distant future (Weber, 2006).

The CMIP5 data for two specific periods (historical and future periods) were downloaded from https://esgf-node.llnl.gov/projects/esgf-llnl/. The grid points of the regional climate scenarios (daily precipitation and maximum and minimum air temperature data) were extracted using MATLAB software. The software was also used to convert NetCDF files into text format, taking into account the spatial references of a particular weather station. Seven stations were used to

extract the downloaded data; Bichena, Debre Work, Felege Birhan, Kuy, Rob Gebeya, Yetemn, and Yetnora. The models were: HadGEM2-ES, CSIRO-Mk3.6.0, GFDL-ESM2M, CanESM2, MIROC, and NorESM1-M, as these models are appropriate based on earlier research conducted in other parts of Ethiopia (Alemseged & Tom, 2015; Teklesadik et al., 2017; Worku et al., 2018) (Table 6.4). In this study, the six models' ensemble mean was computed and used as input for the RUSLE model.

Table 6.4 The Regional Climate Models (RCMs) in the CORDEX-Africa used in this study

Institution RCM name Country CSIRO-Commonwealth Scientific and Industrial Research Organization CSIRO-Mk3.6.0 Australia NOAA GFDL: Geophysical Fluid Dynamics Laboratory GFDL-ESM2M USA MOHC: Met Office Hadley Centre HadGEM2-ES United Kingdom CCCma: Canadian Centre for Climate Modelling and Analysis CanESM2 Canada NCC: The Norwegian Climate Centre NorESM1-M Norway MIROC: Developed by the Japanese research community MIROC Japan 6.2.5.2. Bias correction of regional climate models simulation

Given that the relatively low spatial resolution of GCMs and RCMs outputs systematic errors such as estimation of climate variables (over or under), inaccurate estimates of seasonal variations of precipitation (Christensen et al., 2008), and more wet-less days compared to observed (Ines & Hansen, 2006). As a result, the data from RCPs were bias-corrected to prevent over-or-under estimation and to ensure a realistic representation of the future climate. Because, climate model output cannot be used as direct input data for hydrological simulations (Christensen et al., 2008; Teutschbein & Seibert, 2010, 2012; Chen et al., 2016). The model performance in simulating the observed precipitation and maximum and minimum temperature were assessed, and those with unacceptable time series-based-metrics results were bias-corrected. The historical RCMs and observed data were used for bias correction. Therefore, RCM outputs (precipitation) are typically adjusted to eliminate any bias (Maraun, 2012; Teutschbein & Seibert, 2012).

Several bias correction techniques were developed to correct the bias, ranging from very simple to more complex methods, such as quantile mapping (QM), general quantile mapping (GQM), power transformation (PT), and linear scaling (LS) (Teutschbein & Seibert, 2012). They can be classified according to the degree of their complexity and simpler methods such as scaling factors and more sophisticated techniques such as probability mapping. Although the RCM climate variables' bias

correction significantly improves hydrological modeling, there is a major drawback. All bias correction methods are based on the assumption of static model errors (Teutschbein & Seibert, 2012). This means that the correlation algorithm for current climate conditions and its parameters are considered valid even under the changing climate conditions.

According to Teutschbein & Seibert (2012) and Sisay et al. (2017), linear scaling is a simple technique and provides better efficiency in correcting the ensemble RCMs climatological biases. In this study, the linear scaling bias correction method was applied to adjust the raw ensemble of climate scenarios (rainfall) output simulation data.

The linear scaling bias correction method was selected following a review of the study by Teutschbein & Seibert (2010), which evaluated five bias correction methods for precipitation and showed that linear scaling is suitable for precipitation. The linear scaling method uses a correction factor for each month based on long-term historical RCM data and the observed data. Correction factors were applied for future climate scenarios of the RCP 4.5 and RCP 8.5 emissions. The bias correction was used to estimate and remove the bias correction in future RCMs output. The linear scaling bias correction method (V.1.0) Microsoft Excel file described by Shrestha (2015) was used in this study to adjust the climate model’s average value to observations appropriately. Rainfall is corrected with a multiplier, which is the ratio between the observed monthly average and the data simulated for the historical run (Equations (13) and (14)) (Teutschbein & Seibert, 2012; Luo et al., 2018).

* Phis (d) Phis (d).  m (P obs (d))/  m (P his (d)) (13)

* Psim (d) Psim (d).  m (P obs (d))/  m (P his (d)) (14) where, d=daily, µm=long term monthly means, *=bias corrected, his=RCM simulated 1985-2017, sim=RCM simulated 2018-2050, obs=observed 1985-2017

To verify the bias correction method and the improvement obtained after bias correction, the average daily observed and simulated data (before and after bias correction) were again compared with the observed data by model evaluation statistical methods such as the correction coefficient (r) and root mean square error (RMSE). Statistical results of the bias correction show that the bias correction significantly improves the simulated data as the RMSE values decrease, the SD values

closer to observed data. The correlation coefficient between the observed and simulated data was significantly improved from 0.74 to 0.76 for rainfall data. It indicates that the bias-corrected values better represent the patterns of precipitation over the study watershed. This result showed that the bias correction result is acceptable and consistent with previous studies' results (Gebre & Ludwig, 2015; Sisay et al., 2017). The two RCP scenarios (RCP RCP 4.5 and RCP 8.5) were considered to estimate the rainfall erosivity factor.

6.2.6. Revised Universal Soil Loss Equation model

In this study, the RUSLE model for erosion modeling was used. RUSLE is the revised form of the USLE model (Wischmeier & Smith, 1978), with substantial improvements in predicting the yearly amount of soil loss; and it revised by Nyssen et al. (2009) to the Ethiopian conditions. The annual soil loss (A) of the study watershed was calculated by overlaying five raster layers using Equation 15:

AR*K*LS*C*P (15)

Where, A is the average annual soil loss per unit area (t ha-1y-1), R is rainfall erosivity (MJ mm ha- 1 h-1 yr-1), K is soil erodibility (t ha-1 MJ-1 mm-1), LS is the slope length and steepness factor, C is a cover and management factor, and P is the conservation practice factor. The procedures and techniques used for these factors are presented as follows.

Rainfall erosivity factor (R-factor) indicates the input, which represents the effect of rainfall intensity on soil erosion and requires in-depth, unceasing precipitation data for its calculation (Renard & Freimund, 1994; Angima et al., 2003). The rainfall erosivity is calculated as the total storm energy multiplied by the maximum 30 minutes intensity (Renard, 1997). However, such data are not available for the study area, limiting the R factor's spatiotemporal application. Empirical relationships between measured rainfall amount and R-factor values are widely used as an alternative approach (Hurni, 1985; Nigussie et al., 2014; Mengistu et al., 2015). This method is widely used by previous researchers in Ethiopia and other countries (e.g., Bewket & Teferi, 2009; Mengistu et al., 2015; Tamene & Le, 2015; Duarte et al., 2016; Gelagay & Minale, 2016; Haregeweyn et al., 2017; Molla & Sisheber, 2017).

In this study, the annual rainfall erosivity was calculated according to the method of Hurni (1985) to the Ethiopian highlands as Equation 16;

R8.12(0.562*P)  (16)

Where, R is the rainfall erosivity factor (in MJ mm ha-1 h-1 year-1), and P is the mean annual rainfall (mm).

After accounting the R factor of the baseline (1985-2017) and future period (2018-2050) using equation 16, the R factor raster was created using the inverse distance weighted (IDW) interpolation method in the Geostatistical Analysis extension of ArcGIS 10.4 software.

Soil erodibility factor (K-factor) represents soil's vulnerability to erosion (Wischmeier & Smith, 1978). According to the soil difference, the K value of an area is also different depends on the parameters of soil texture, organic matter content, and permeability (Renard, 1997). The soil erodibility was determined from the available soil data of the Abbay River Basin (scale 1:250,000), which was obtained from the Ministry of Water and Energy. In this study, the nomograph method to calculated the K value was used as proposed by Wischmeier et al. (1971).

6 1.14 K(1.292)[2.1*10fact  f(12p ) 0.0325(S2) om 0.025struc (f3)](17)  perm

in whichfp (100 ppsilt ) clay

Where, fp is the particle size parameter (unitless), pom is the percent organic matter (unitless), sstruc is the soil structure index (unitless), fperm is the profile-permeability class factor (unitless), psilt is the percent silt (unitless) and pclay is the percent clay (unitless).

In Equation 17 the factor (1.292) is needed to convert K-factor from the imperial to the international system units (i.e., SI metric units) (Streile et al., 1996). The soil structure index, Sstruc, is 4 for blocky, platy, or massive soil, 3 for medium or coarse granular soil, 2 for fine granular soil, and 1 for very fine granular soil (Mengistu et al., 2015). The profile-permeability class factor, fperm, is 1 for very slow infiltration, 2 for slow infiltration, 3 for slow to moderate infiltration, 4 for moderate infiltration, 5 for moderate to rapid infiltration, and 6 for rapid infiltration (Streile et al., 1996). Generally, Equation 17 can help capture relative differences in erodibility between soil types and help appropriate the resistance to erosion of different soils under

consideration (Ganasri & Ramesh, 2016). The soil type and soil erodibility map of the watershed is presented in Figure 6.4.

Figure 6.4 Soil type (a) and soil erodibility factor map (b) of Muga watershed

Topographic factor (LS-factor) refers to the effect of topography on soil erosion. The slope length (L) and slope gradient (S) factors are joined in a single index, LS-factor, to describe the topographic factor for soil loss. The slope length refers to the distance from the point of origin of overland flow to the point where either the slope gradient declines enough in which sedimentation starts or the runoff water enters a well-defined channel (Renard, 1997). The LS factor (Figure 6.5) represents the ratio of soil loss per unit area on-site to the corresponding loss from a 22.13 m long experimental plot with a 9% slope (Renard, 1997). Higher slope lengths may increase the overland flow and lead to more land surface soil erosion (Moore & Burch, 1986a). Moreover, higher slope gradients also promote the runoff rate and bring out more soil erosion.

In this study, the L factor was computed following Equation (18) proposed by Moore & Burch (1986a, 1986b). The algorithm (Equation (19)) recommended by Desmet & Govers (1996) was used to calculate the S factor. It is based on flow accumulation and slope steepness. The slope

steepness (S) factor was calculated for high (>9%) and low slope land (<9%) from the slope angle. Finally, the L and S factor were multiplied to derive the LS factor for the study watershed using the Spatial Analysis Tool Map Algebra Calculator in the ArcGIS 10.4 environment and the spatial variability of the slope length steepness factors (Renard, 1997).

  m   L,   22.1  where  (18)  sin     0.0896 m,    0.8  1 3 sin0.56  

10.8*θ+0.03, whereslopegradient <9% S= (19) 16.8*sinθ-0.5, whereslopegradient ³9% where, L and S indicate slope length and steepness factor (dimensionless); λ is the slope length (in meter), cell size is the size of the grid cell ( for this study 30 m); m is an adjustable slope length exponent to β; while β represents the rill to interrill erosion ratio; and  is the slope angle (in degrees).

The study area's topography is steep with non-uniform terrain on the upper parts of the watershed, whereas the area's lower escarpment is gentle and uniform. The combined steeper and longer slopes resulted in larger cumulative runoff volumes with high velocity and erosive power.

Figure 6.5 Slope length and steepness factor (LS-factor) map of Muga watershed

Cover management factor (C-factor) represents the effect of vegetation and the cover on the amount of soil erosion (Bewket & Teferi, 2009; Haregeweyn et al., 2017). The individual values of C vary between 0 for a completely non-erodible condition, and 1 implies conditions more erodible than those normally experienced under unit plot conditions, which can occur for conditions with very extensive tillage, and C is strongly related to land use.

In Ethiopia, satellite-derived images are a good proxy for land cover on relatively large basins and watershed levels and were applied in the Abbay River basin of Ethiopia (Bewket & Teferi, 2009; Mengistu et al., 2015; Molla & Sisheber, 2017; Taye et al., 2018). In this study, C values were determined based on the most recent (2017) and predicted future (2033) land use data, as suggested in the literature (Table 6.5).

Conservation practice (P-factor) reflects the effects of support practices, such as terracing and contour tillage, to reduce the rate of soil erosion (Renard, 1997). The higher the P factor, the less effectively the practice facilities deposition to take place close to the source. It is primarily used to assess the effects of conservation measures implemented on soil loss (Mengistu et al., 2015). Our

field visits indicated that soil and water conservation activities are not widely practiced in the study area. However, farmers commonly use stone bunds and contour farming to protect the soil from erosion. Still, they are poorly maintained as implementation was carried out in a top-down approach, and there was no map of conserved areas in the sub-watershed. In areas where contour plowing is widely practiced, Hurni (1985) suggested using P values of 0.9 and 0.8 for agricultural and non-agricultural lands, respectively. Thus, a similar method was used to assign P values to the study area for the baseline period (2017) and future scenario (2033).

Finally, the annual soil loss was estimated on a cell-by-cell basis of multiplying the five RUSLE factors using Equation (1). As Landsat images and ASTER-DEM used in this study had 30 m spatial resolution, all the raster maps were resampled to 30 × 30 m cell size and re-projected to UTM Zone 37° N, WGS 1984 datum. Table 6.5, Figure 6.6, and Figure 6.7 shows the C values and the P values derived from LULC information for the baseline and future period.

Table 6.5 C-factor and P-factor values for the respective LULC classes of Muga watershed, Abbay river basin, Ethiopia Land-use and land C factor P factor cover classes Grassland 0.05 (Hurni, 1985; Bewket & Teferi, 0.8 (Hurni, 1985; Adugna et al., 2015) 2009) Cultivated 0.15 (Hurni, 1985; Adugna et al., 0.9 (Bewket & Teferi, 2009; Adugna et 2015) al., 2015; Taye et al., 2018) Shrub-bush land 0.02(Mengistu et al., 2015) 0.8 (Hurni, 1985) Forest 0.05 (Molla & Sisheber, 2017) 0.8 (Hurni, 1985) Urban 0.05 (Moges & Bhat, 2017) 0.8 (Hurni, 1985)

Figure 6.6 The cover (C) factor map for the year 2017 and 2033 in the Muga watershed

Figure 6.7 Conservation practices (P) factor map for the year 2017 and 2033 in Muga watershed

6.3. Results 6.3.1. Land use/ land cover change in Muga watershed (2017-2033)

The produced LULC maps of Muga watershed for the three reference years (1985, 2002, and 2017) are presented in Figure 6.8. The trend analysis made for the two consecutive periods, 1985-2002 and 2002-2017, showed spatiotemporal changes in LULC classes. The magnitude and extent of LULC change for the study watershed between 1985 to 2017 were obtained from Belay & Mengistu (2019). The LULC change between 2017 and 2033 were calculated (Table 6.6). As shown in LULC map of 2033, cultivated land remains a dominant land use type, which accounts for about 76.8% of the study watershed. From 2017 to 2033, the areal coverage of grassland decreased by 35.4%, while urban, shrub-bushland, forest area, and cultivation area would increase by 2.9%, 10.4%, 6.8% 3.8%, respectively. It should be noted that the decline of grassland area from 2017 to 2033 probably due to the conversion of grasslands to cultivated land and shrub- bushland, as shown in Table 6.6.

Table 6.6 The area and proportion of each land use categories in the Muga watershed

Reference (2017) Future (2033) Change (2017-2033) Area (Ha) % Area (Ha) % (%) Grassland 4862.7 11.4 3143.2 7.4 -35.4 Cultivated 31438.4 74.0 32624.7 76.8 3.8 Urban 1008.7 2.4 1038.3 2.4 2.9 Forest 1015.1 2.4 1084.2 2.6 6.8 Shrub-bushland 4170.5 9.8 4605.0 10.8 10.4 Total area 42495.4 100.0 42495.4 100.0

Figure 6.8 The 2017 (A) and 2033 (B) LULC of the Muga Watershed. 6.3.2. Future climate

Projected changes in annual precipitation and the maximum and minimum temperature for the RCP4.5 and RCP8.5 climate change scenarios of the six CMIP5 RCMs for the 2050s relative to the reference period 1985-2017 are shown in Table 7. The mean daily maximum (minimum) temperature of Muga watershed are projected to increase by about 0.67°C (1.24°C) under the RCP4.5 climate scenario and by 0.5°C (1.59°C) under RCP8.5 climate scenarios in the future period relative to the reference period, respectively. As expected, projected changes in minimum temperature are higher than projected changes in maximum temperature.

In the study watershed, compared with 1985-2017 (1086.6mm), the mean for RCMs ensemble (2018-2050) showed an increase in mean annual rainfall of 1272.9 mm (RCP4.5) and 1307.0 mm (RCP 8.5). The findings showed that the rainfall erosivity is expected to increase from 602.6 MJ mm ha-1 h-1 yr-1 (1985-2017) to 707.26 and 726.4 MJ mm ha-1 h-1 yr-1 for the period 2018-2050, under RCP 4.5 and RCP8.5, respectively. It is expected to increase by 17.15% and 20.27% under

RCP 4.5 and RCP 8.5, respectively. This result agrees with previous studies (Conway & Schipper, 2011; Kassie et al., 2014; Abera et al., 2018; Fentaw et al., 2018) that predicted increasing precipitation in other parts of Ethiopia.

Table 7. Projected change in rainfall, minimum and maximum temperature of Muga watershed under two RCP climate scenarios for mean of six RCMs for reference and future period

Future Relative change in Temperature Historial RCP 4.5 (2018-2050) RCP 8.5 (2018-2050) Relative change in (°C) rainfall (%) RCP 4.5 RCP 8.5 RCP 4.5 RCP 8.5 Tmin Tmax Tmin Tmax 1087 9.48 23.69 1273 10.7 24.36 1307 11.07 24.19 17.15 20.28 1.24 0.67 1.59 0.5

6.3.3. Impacts of land use/ land cover changes on soil erosion

LULC change in the area may increase or decrease soil erosion. The LULC dynamics of the Muga watershed from 1985 to 2017 was studied by Belay & Mengistu (2019). In this study, to assess the impact of LULC change on soil erosion, the values of C and R factors were changed, while the other factors (i.e., soil erodibility, conservation practice factor, and slope length and slope steepness) kept constant.

The results of the study show that the average annual rate of soil erosion in the Muga watershed was increased from approximately 15 t ha-1 year-1 in 1985 to 19 t ha-1 year-1 in 2002, and 19.7 t ha- 1 year-1 in 2017 (Table 6.8). The study results also indicate that if the LULC changes are not managed, the soil loss rate is expected to continue in 2033, which is expected to reach 20.7 t h-1 yr-1 in 2033. Areas with soil erosion rates over 15 t ha-1 yr-1 during the study years were widely distributed on the upper and steep areas of the watershed, while those with an erosion rate of less than 10 t ha-1 yr-1 was mainly concentrated in the gentle areas of the lower and upper part of the watershed (Figure 6.9). Increasing some types of LULC, such as cultivable land in steep areas and irrigated agriculture, will accelerate soil erosion by reducing soil cover in the future.

The validation and consistency of the model output was compared with the quantitative outputs of previous experimental observations and similar empirical studies conducted in Ethiopia, mainly in the northwestern highlands. Besides, selected field observations were carried out. In supporting this process, the color printed model output soil erosion severity map was taken in the field to check the reality on the ground. Consequently, the estimated rate of soil loss and the spatial patterns are generally realistic as compared with the findings in the field and from the results of previous studies (Hurni, 1983b, 1983a, 1985; Bewket & Teferi, 2009; Mengistu et al., 2015)

The relative contribution of each LULC type to soil erosion was assessed in the study watershed, and the result showed noticeable differences among LULC types in 1985, 2002, 2017, and 2033. The highest mean annual soil erosion rate was predicted on cultivated land (e.g., 18 t ha-1 year-1 in 1985, 19 t ha-1 year-1 in 2002, and 21.7 t ha-1 year-1 2017).

The area under cultivation is expected to increase by 2033, the average annual rate of soil erosion is expected to be higher than 1985, 2002, and 2017. The causes of the higher soil loss estimated from cultivated land may be related to the encroachments of sloping plowing in mountainous areas of the watershed, as the expense of expansion of urban areas into cultivation. The results of this study suggest that the rate of soil loss in shrub-bushland areas will be higher than in grassland and urban in the future (2033), which may be due to the scanty vegetation and the steepness of the area where the shrub-bushland area is located. This indicates that in the year 2033, the area of shrub- bushland is expected to have less capability to protect the soil from erosion than in the year 2017. Thus, future LULU changes under the business-as-usual scenario, with increasing LULC changes in the study area (i.e., expansion of cultivated land into steep slope) contributing to more soil erosion, which is anticipated to increase further with higher precipitation. It is estimated that in 2033 there will be higher rates of soil erosion in cultivated, followed by grassland and shrub- bushland, which are the key areas to prevent soil erosion in the Muga watershed in the future.

Table 6.8 The estimated mean annual soil erosion rate of each LULC type and the entire watershed in 1985, 2002, 2017 and 2033 in the Muga watershed

Mean annual soil erosion rate (t ha-1 yr-1) Entire Year Grassland Cultivated land Urban Forest Shrub-bushland watershed 1985 8.0 18.0 11.0 3.0 7.5 15.0 2002 12.0 19.0 12.0 8.5 12.0 19.0 2017 14.2 21.7 10.4 11.2 11.9 19.7 2033 13.6 25.0 6.5 9.1 13.0 20.7

As shown in Table 6.9, the percentage areas under low to medium-level soil erosion were decreased between 1985 and 2033. In contrast, the portion of areas under high and very high soil erosion classes was increased during the same periods, but this change is relatively low. The increase in the areas with high and very high soil erosion rates between 1985 and 2033 may be due to the conversion of some parts of the watershed with low and moderate erosion classes into the range of high and very high erosion. Between 1985 and 2017, there was an upward trend in the area occupied by severe and very severe erosion classes, while the severe erosion category is expected to decline in 2033. It was also found that the area classified as very severe erosion is uniform in 2017 and 2033.

Figure 6.9 Estimated annual soil loss map of the study area in 1985 (a), 2002 (b), 2017 (c) and 2033 (d). Table 6.9 Annual soil erosion risk classes and area coverage under LULC map of 1985, 2002, 2017 and 2033 in Muga watershed

Study period Soil erosion intensity class 1985 2002 2017 2033 and rate of soil loss Area (ha) % Area (ha) % Area (ha) % Area (ha) %

Low (0-5 t ha-1 yr-1) 19138.9 45.0 18343.9 43.2 17679.0 41.6 17412.5 41.0 -1 -1 Moderate (5-20 t ha yr ) -1 -1 High (20-50 t ha yr ) -1 -1 Very high (50-100 t ha yr ) -1 -1 Severe (100-150 t ha yr ) Very severe (>150 t ha-1 yr-1) 260.5 0.6 330.5 0.8 413.2 1.0 406.4 1.0

6.3.4. Impacts of climate change on soil erosion

Rainfall amount and rainfall erosivity under RCP 4.5 and RCP 8.5 for the study area showed an increasing trend, which is anticipated to affect soil loss negatively. Increasing rainfall will lead to an increase in annual rainfall erosivity, which increases the annual rate of soil loss. Thus, the average annual rate of soil erosion from the watershed by modifying climatic data the in this study was analyzed during the study period to assess the relationship between soil erosion and climate change in 2017 and 2050.

In this study, the researchers seek to simulate the future rate of soil erosion due to changes in climatic conditions that may increase the risk of soil and land degradation in the Muga watershed, which can also affect agricultural productivity and livelihoods of the local community. The results are shown in Table 6.10 and Figure 6.10. The results show that the average annual rate of soil loss in the Muga watershed in 2050 under RCP 4.5 and RCP 8.5 will be 21.6 t ha -1 year-1 and 22.2 t ha -1 year-1, respectively, which is equivalent to an annual soil loss of 917,900.6 t year-1 and 960,396.0 t year-1, respectively. Therefore, the rate of soil erosion is predicted to increase by 9.7% and 12.7% under RCP 4.5 and RCP 8.5, respectively, compared to the reference period. This means that the mean annual soil erosion rate in the Muga watershed showed an increasing trend in the future compared to the reference period due to climate change. Table 6.10 Annual soil erosion risk classes and area coverage under RCP 4.5 and RCP 8.5 in Muga watershed

Soil severity categories and rate LULC 2017 & LULC 2017 & of soil erosion 2050 (RCP 4.5) 2050 (RCP 8.5) Area (ha) % Area (ha) % Low (0-5 t ha-1 yr-1) 15904.1 37.4 15526.2 36.5 Moderate (5-20 t ha-1 yr-1) 13677.6 32.2 13864.7 32.6 High (20-50 t ha-1 yr-1) 7041.0 16.6 7097.8 16.7 Very high (50-100 t ha-1 yr-1) 4273.5 10.1 4346.7 10.2 Severe (100-150 t ha-1 yr-1) 1083.5 2.5 1136.0 2.7 Very severe (>150 t ha-1 yr-1) 515.9 1.2 524.2 1.2 Total area 42495.4 100.0 42495.4 100.0 Recognizing the soil loss response helps to know the rate of soil erosion changes with climate change, which helps to identify priority areas for implementing soil management measures in the study watershed. Increased soil erosion due to climate changes can affect the watershed's local ecosystems and can cause hydrological changes in streams originating from the watershed.

100

Comparing LULC and climate change, as shown in Table 6.11, Climate changes are expected to have a higher impact on soil erosion than LULC change in the future. Figure 6.10 shows a map of soil erosion in the Muga watershed in 2050 by averaging six RCMs under RCP 4.5 and RCP 8.5. The figures show that high soil losses are concentrated on the watershed's upper escarpment where cultivation and other human activities are carried out on the steep slopes of Choke Mountain.

Figure 6.10 Soil erosion severity level in 2050 under RCP 4.5 (a) and RCP 8.5 (b). 6.3.5. Combined impacts of future land use/ land cover and climate change on soil erosion

In addition to assessing the individual impacts of future LULC and climate change, the combined and synergistic effects of these changes on soil erosion was also evaluated in the Muga watershed. By replacing the two factors of rainfall erosivity (R) and cover management factor (C), and keeping all other factors constant in the RUSLE model, soil erosion was predicted using future LULC and climate change scenarios: 2033 LULC map and climate scenarios under RCP 4.5 and RCP 8.5 in 2050. The magnitudes and rates of mean annual loss of soil under future LULC and climate change scenarios are presented in Table 6.11.

The results of this study demonstrated that the highest soil erosion rate for the study watershed is predicted when LULC change is combined with climate change under the RCP 8.5 scenario, which is estimated about 22.8 t ha-1year-1. In contrast, the average annual rate of soil loss under RCP 4.5

101

is estimated about 21.6 t ha-1year-1. Therefore, changes in LULC under RCP 4.5 and RCP8.5 scenarios showed a slight variation in annual soil loss rates. It should also be noted that the ranges of relative change in average annual soil loss rates for the future period due to climate change is larger than LULC change. Table 6.11 shows that soil erosion in the Muga watershed appears to be more sensitive to future climate changes than future LULC changes. The results of this study showed that the magnitude of the soil loss rates is expected to increase in the future due to the synergy effects of LULC and climate changes. As it is shown in Table 6.11, the combined effects of LULC and climate change on soil loss rate is expected to increase in the future.

The results of this study showed that LULC changes are expected to exacerbate the rate of soil loss by 5%, while climate change is also predicted to increase the rate of soil loss by approximately 9.7% and 12.7% under RCP 4.5 and RCP 8.5, respectively. When LULC and climate change act together, the average annual soil loss rate increases to 13.2% and 15.7% under RCP 4.5 and RCP 8.5, respectively, which is higher than the individual effects of LULC and climate change. Thus, the combination of modeled LULC and climate change is expected to have a substantial impact on soil erosion in the future. Increasing average annual soil erosion rates caused by LULC and climate change will significantly impact land use planning and implementation in the Muga watershed. Besides, as the LULC types change, adverse effects may be intensified. Thus, there should be appropriate land management strategies that consider future LULC and climate change.

Table 6.11 The mean annual soil loss (t ha-1year-1) of Muga watershed under current and future LULCand climate

Model input (a) (b) (c) (d) (e) (f) LULC 2017 2033 2033 2033 2017 2017 Climate 1985-2017 1985-2017 RCP 4.5 RCP 8.5 RCP 4.5 RCP 8.5 (2018-2050) (2018-2050) (2018-2050) (2018-2050) Mean annual soil erosion 19.7 20.7 22.3 22.8 21.6 22.2 (t ha-1 yr-1) Note: (a) Baseline period (LULC in 2017 and baseline climate), (b) future LULC (2033) and baseline climate, (c) Combined impacts of future LULC (2033) and climate under RCP 4.5, (d) Combined impacts of future LULC and climate under RCP 8.5, (e) only impacts of future climate under RCP 4.5, (f) only impacts of climate under RCP 8.5

102

6.4. Discussion

Soil erosion prevention patterns varied considerably depending on location and time. Globally, as environmental change accelerates, more frequent and intense changes in LULC associated with a higher frequency of extreme climate events will increase soil degradation and make the ecosystem less resilient to natural disturbance (Cramer et al., 2018; Guerra et al., 2020). The results of the study show a spatial and temporal variability in soil erosion rate due to climate dynamics and changes in LULC. According to Zhou et al. (2008), a soil vegetation cover of more than 78% greatly reduce erosion by water. The results of this study show that less than 5% of the study watershed is covered with forest; hence, the study watershed is sensitive to soil erosion. Furthermore, the expansion of cropland at the expense of forest and shrubland reduced the protection and soil organic matter of the soil and exposed it to the impact of climate change thereby accelerated soil erosion rate (Wijitkosum, 2012).

The results of future climate projection showed that the mean annual precipitation in the watershed is expected to increase by 17.15% and 20.27% under RCP 4.5 and RCP 8.5, respectively, compared to the historical period of 1985-2017. This result agrees with previous studies (e.g., Conway & Schipper, 2011; Kassie et al., 2014; Abera et al., 2018; Fentaw et al., 2018) that predicted the increase in precipitation in the other parts of Ethiopia.

Muga watershed also showed significant dynamics in LULC from 1985 to 2017. The results of the study show that LULC changes played a significant role in increasing the soil erosion rate in the study watershed. This finding coincides with previous research conducted in the Abbay River Basin (e.g., Gessesse et al., 2015; Gelagay & Minale, 2016; Moges & Bhat, 2017; Tadesse et al., 2017), who reported that LULC change was a responsible factor for the increased soil erosion in their respective study areas.

Table 6.8 shows the estimated soil loss in 1985, 2002, 2018, and 2033. The estimated average soil loss was increased due to LULC dynamics that occurred between 1985 and 2017. This shows that LULC has a significant impact on soil loss by water erosion. The results of this study show that the average annual rate of soil loss in the Muga watershed (19.0 t ha-1 yr-1 in 2002, 19.7 t ha-1 yr-1 in 2017, and 20.7 t ha-1 yr-1 in 2033) was substantially higher than the average soil loss of Abbay

103

River Basin (16 t h-1 yr-1) reported by Mengistu et al. (2015), and Fenta et al. (2021) who reported 16.5 t ha-1 yr-1 average soil loss rate for Ethiopia.

This study also shows that about 25% of the study watershed has a soil loss rate of 20 t ha-1 yr-1 and above, which is higher than the tolerable soil loss limits estimated for Ethiopia. Tolerable soil loss rates suggested for Ethiopia is 12 t ha-1 yr-1 (Hurni, 1983b). According to Khosrokhani & Pradhan (2014), the rate of soil formation in tropical areas is generally slow, and soil loss of more than 1 t ha-1 yr-1 is regarded as irreversible soil erosion. However, soil loss of 1 t ha-1 yr-1 or less is considered as an acceptable soil erosion rate. The predicted average soil loss for the study year exceeds the tolerable soil erosion rates of 12 t ha-1 yr-1 and the estimated soil formation rate of 2 t ha-1 yr-1 for Ethiopia (Hurni, 1983b), which can affect soil productivity.

In the presents study, the average annual rate of soil loss varies with the type of LULC types, with the main contributor is being cultivated land (about 18 t ha-1 year-1 in 1985, 19 t ha-1 year-1 in 2002, and 21.7 t ha-1 year-1 2017), followed by grasslands and shrub-bushlands (Table 6.8). The lowest soil loss rates were observed in forest and urban areas. This is due to vegetation cover and the corresponding low C-factor values (Zhou et al., 2008). The highest rates of soil loss on cultivated land indicated that land conversion from natural vegetation (e.g., forest and shrub-bushland) to cultivated land would exacerbate land degradation due to soil erosion.

Similarly, Fenta et al. (2021) estimated an average annual soil loss rate of 36.4 t ha–1 yr–1 from cultivated land for Ethiopia's highlands. This is twice the overall average rate of soil loss (16.5 t ha–1 yr –1), they reported for the whole highlands of Ethiopia. Contrary to our research results, Taye et al. (2015) reported the highest soil loss rate of 38.7 t ha-1 yr-1 from grassland compared with 7.2 t ha-1 yr-1 from cropland.

Studies showed that rainfall is one of the most sensitive factors for soil erosion (Zhang & Nearing, 2005; Her et al., 2019; Berberoglu et al., 2020; Borrelli et al., 2020; Ciampalini et al., 2020; Eekhout & De Vente, 2020). This study results show an increase in the future period's annual rainfall erosivity compared to the baseline. The RCP 8.5 scenario is expected to have the highest amount of erosivity factor and will have the highest erosion rate, followed by RCP 4.5 in 2018- 2050. The rate of soil erosion under RCP 4.5 and RCP 8.5 is expected to increase by 9.7% and 12.7%, respectively, compared to the baseline period.

104

As shown in Figure 6.9, the areas with moderate and very high-intensity soil erosion rate are mainly occupied in the study watershed ranges from 19.7 t ha -1 year-1 under LULC 2017 and climate of 1985-2017 to 22.3 t ha -1 year-1 under LU 2033 and RCP4.5 and 22.8 t ha -1 year-1 under LU 2033 and RCP 8.5, respectively.

The result of this study is consistent with the results in the Abbay River Basin (e.g., Bewket & Teferi, 2009; Mengistu et al., 2015). However, the rate of soil erosion in the study area is lower than the value previously reported by Yesuph & Dagnaw (2019) for Beshillo catchment of the Abbay River basin, which is 37 t ha -1 year-1. According to Kouli et al. (2009), soil loss rates above 10 t ha -1 year-1 will not reverse for 50 to 100 years. Considering this threshold, the total area of the study watershed where the risk of soil erosion exceeds the soil loss tolerance is expected to be 19,348 hectares and 21,057 hectares in the 2050s under RCP 4.5 and RCP 8.5, respectively.

The future scenario shows that LULC combined with climate change substantially increases the average soil erosion by 2070s globally (Borrelli et al., 2020). The results of this study also showed that when projected land use is combined with simulated climate change, the mean annual soil erosion rate is expected to increase by 13.2% under LULC map of 2033 and RCP 4.5 and 15.7% under LULC map of 2033 and RCP 8.5 compared with the baseline.

6.5. Conclusion

The study showed the effects of impacts of land use land cover and climate changes on soil erosion using an integrated approach of CA Markov chain, climate and soil erosion models. The outcomes indicate that soil erosion rate in the Muga watershed has shown an increasing trend from 19.7 t h- 1 yr-1 in 2017 to 20.7 t h-1 yr-1 in 2033 due to LULC change. Furthermore, it is concluded that by the 2050s, the rainfall erosivity factor may increase which can lead to more soil erosion rate. Hence, the soil loss rate in Muga watershed is projected to increase to 22.0 t ha -1 year-1 and 22.8 t ha -1 year-1, under RCP 4.5 and RCP 8.5 scenarios, respectively, due to higher erosive power of the future intense rainfall. When the combined effect of LULC and climate change considered, the average annual soil loss rate was increased by 13.2% and 15.7% under RCP 4.5 and RCP 8.5, respectively, which is much higher than the individual effects of LULC and climate change. The change in soil erosion rate showed a spatial variation, hence the upper escarpment of the watershed which consists of steep slopes is highly vulnerable to soil erosion hazard.

105

The outcome of this investigation would be imperative in the decision and implementation of appropriate soil and water conservation methods and local climate adaption strategies within the study area and for deducing the changes in the future. Likewise, the results obtained from this study can be used by local and regional government agencies, developers and policymakers to diminish the rate of soil loss in the study watershed. Integrated use of CA Markov chain, climate, and soil erosion models have demonstrated to provide relevant information about the impact of LULC and climate change in soil erosion at a watershed level.

106

Chapter 7 7. Impacts of Land Use/ Land Cover and Climate Change on Hydrology in Muga Watershed, Abbay River Basin, Ethiopia Abstract

LULC and climate change are two main factors affecting watershed hydrology by altering the hydrological response including streamflow, water yield, surface runoff and evapotranspiration. In this study, simulation scenarios were developed using the soil and water assessment tool (SWAT) model to assess the individual and combined effects of LULC and climate change on the hydrological processes in the Muga watershed. The partial least squares regression (PLSR) statistical method was used to assess the effects of changes in individual LULC types on hydrologic components. The results of this study showed that the SWAT model with NSE and PBIAS values of 0.79 and 8.2% for calibration, 0.82 and 9.3% for validation, respectively, performed very well in the simulation of the hydrological process of the watershed. Future projections of maximum and minimum temperature and precipitation under RCP4.5 and RCP8.5 scenarios were used as input in the calibrated SWAT model. In Muga watershed, LULC change have an impact on streamflow, surface runoff, groundwater flow and lateral flow. However, the expected ET and water yield response to LULC change is low. The results of the PLSR analysis revealed that cultivation has an adverse effect on groundwater flow (GWQ), streamflow (ST), ET and lateral flow (LQ). In contrast, land covered with forest have regulated GW, LQ, and ET. Furthermore, the results of this study showed a strong negative correlation between LULC change and SURQ and WYLQ, while a positive correlation with GW, ST, ET and LQ. Therefore, the results of this study indicated that the change in LULC led to a change in the water balance. The combined effect of LULC and climate change on water balance is relatively higher than the impact of LULC change scenario alone, which is accompanied by an increase of rainfall under RCP4.5 and RCP8.5. This study will help decision-makers and planners to plan appropriate mitigation strategies accordingly.

Keywords: Climate change, Hydrology, Land use/ land cover, PLSR, Soil Water Analysis Tool 7.1. Introduction The potential hydrological impacts of LULC and climate change could lead to unforeseen future water resource crises (Talib & Randhir, 2017). Hence, analysis of the spatiotemporal LULC and

107

climate changes play important roles in water resources management (Mango et al., 2011; Field & Barros, 2014; Afzal & Ragab, 2020; Parry et al., 2007). Climate change has multiple effects on the hydrological cycle, which affects water resources (Swain et al., 2020). Several drivers of change, such as population growth, climate variability and land use policy have contributed to significant changes in land cover and land use, which are affecting the hydrological system at both watershed, sub-basin and regional level (Bewket, 2003; Dagnachew Legesse et al., 2003; Birhanu et al., 2019).

Water resources are influenced by different land uses such as cultivated land and urban expansion, the decline in forest and grassland. In addition, climate change has worsened the effects of population growth, poverty and rapid urbanization on the depletion of natural resources (Yesuf et al., 2008; Gebre & Ludwig, 2015). Climate change (i.e. temperature and precipitation) is likely to continue to change in the 21st century (Gizaw et al., 2017), which could lead to water scarcity. A change in LULC also changes the characteristics of surface runoff, groundwater recharge, and evapotranspiration.

The effects of LULC and climate change on water availability are exacerbated by a rapidly growing population, putting pressure on the water resource availability, and adversely affecting the agricultural sector and food supplies of nations (Kurukulasuriya & Rosenthal, 2013). Therefore, to adopt and implement adequate management of water resources to ensure the sustainable production of water resources, it is important to assess the hydrological response of a watershed to LULC and climate changes. Moreover, in order to successfully manage land and water resources and mitigate climate change, it is important to assess the contribution of LULC and climate changes to water resource change.

The general purpose of this study was to analyze the individual and combined effect of LULC and climate change on hydrology using LULC and climate change prediction including hydrological models in the Muga watershed. The specific objectives of the study were to examine the contribution of individual LULC class on hydrological components; assess the individual and combined impacts of LULC and climate change on hydrological components.

108

7.2. Materials and Methods 7.2.1. Data inputs

In this study, an integrated method CA-Markov chain model, hydrological and climate model were used. LULC and climate change scenarios are therefore considered to assess the impacts of LULC and climate change on hydrology in the Muga watershed. The spatial and non-spatial data of the study watershed published in Belay & Mengistu (2019), as well as the data used in chapter six, were used in this chapter. In this study, it was necessary to prepare a set of spatially distributed data to run the SWAT model. Spatial and temporal data, including DEM, soil, land use, meteorological and streamflow data were used as input for the SWAT model. A detailed description of data used as an input for the SWAT model is provided below.

7.2.1.1. Hydro-climatic data

The hydrological processes of the watershed were simulated based on combined hydro-climatic and watershed information. The meteorological data required for daily calculation of hydrological processes in the SWAT model were long term daily rainfall, maximum and minimum temperature, solar radiation, wind speed and relative humidity and the data were obtained from the Ethiopian National Meteorological Services Agency (NMSA).

Most meteorological stations located within and close to the study watershed have a long period of recorded continuous daily climate data. These weather stations are; Yetmen, Kuy, Yetnora, Bichena, Felege Birhan, Debre Work and Rob-Gebeya. Only two stations Yetnora and Debre Work have all the climatic variables of daily maximum and minimum temperature, precipitation, wind speed, sunshine hours and relative humidity while the remaining five stations have only temperature and precipitation data. Kuy meteorological station is located within the study watershed and contains only precipitation and temperature data. In this study, Yetnora meteorological station was chosen and used to calculate the parameters to fill gaps in meteorological data. Inputs of WGNMaker 4.1 include daily precipitation, daily highest and lowest temperature, and daily maximum precipitation in half an hour. In this study, the dew points required for SWAT were then calculated for the stations using precipitation and temperature (minimum and maximum) data. The geographical location of meteorological stations, and their

109

mean annual rainfall, minimum and maximum temperature for the baseline period (1985-2017) and future period (2018-2050) of Muga watershed is given in Table 7.1.

Table 7.1. Mean annual minimum and maximum temperature (oc) and rainfall (mm) for the baseline period (1985-2017) and future period (2018-2050) of Muga watershed

Station Coordinate Baseline period (1985-2017) Future climate (2018-2050) location RCP 4.5 RCP 8.5 Lat Lon Tmin Tmax Rainfall Tmin Tmax Rainfall Bichena 10.4 38.2 10.3 24.6 967.5 13.8 27.1 1307.0 14.1 27.2 1392.4 Debre Work 10.7 38.2 10.4 24.3 956.3 10.9 24.0 1329.7 11.2 24.0 1430.2 Felege Birhan 10.7 38.1 10.3 23.8 1193.3 11.7 25.3 1263.8 11.8 25.4 1309.1 Kuy 10.5 38.0 9.3 23.7 1058.7 9.4 22.2 1275.5 9.8 22.2 1277.1 Rob Gebeya 10.6 37.8 8.3 23.2 1253.8 9.4 22.2 1275.5 9.8 22.2 1277.1 Yetemn 10.3 38.2 10.2 24.0 1096.6 13.8 27.1 1307.0 14.1 27.2 1392.4 Yetnora 10.3 38.1 10.2 23.4 1034.8 12.5 22.2 1073.5 12.8 25.3 1109.0

Various methods were proposed over the last decades to estimate potential evapotranspiration (PET). In this study, Hargreaves method was used for PET estimation as input for the SWAT model. The Hargreaves-Sami (HS) method is a temperature-based PET estimation method, it is the most commonly used method and recommended by FAO as an alternative method when observed meteorological data are not available (Allen et al., 1998), especially in the research fields of hydrological models (Hargreaves, 1989). The HS method requires the daily minimum and maximum air temperatures as input. Thus, in this study, the HS method was chosen to estimate the PET.

As explained in chapter six, in this study, future climate data up to 2050 (2018-2050) was used rather than considering the end of the century.

Future climate data from six Regional Climate Models (RCMs), forced by two Representative Concentration Pathway (RCPs), RCP4.5 and RCP8.5 were downloaded from Coordinated Regional Downscaling Experiment (CORDEX)-Africa and used to project the future climate change in the study watershed from 2018 to 2050. The future climate data used as input for the SWAT model were daily rainfall, maximum and minimum temperatures.

Climate variables such as solar radiation, relative humidity, and wind speed for the period from 2018-2050 were generated by the stochastic weather generator (WXGEN) (Sharpley & Williams, 1990). This stochastic weather generator is often used for studies on climate change (Ficklin et al., 2009; Cousino et al., 2015; Fan & Shibata, 2015). In this study, future meteorological data (i.e.,

110

solar radiation, relative humidity, and wind speed) were individually estimated using the WXGEN weather generator model in SWAT based on historical statistics of the observed data climate data. Therefore, stationarity with solar radiation, relative humidity and wind speed was assumed.

Hydrological data of daily river discharge for the Muga watershed (Muga near Dejen) used for calibration and validation of the SWAT model was obtained from the Hydrology Department of the Ethiopian Ministry of Water, Irrigation and Electricity (MoWIE). The missing data values of streamflow were filled by using statistically similar day averages (long-term averages) regarding the previous years where there were existing data values (Perry & Hollis, 2005).

7.2.1.2. Land use land cover data

LULC data is an essential input parameter for the hydrological model. LULC change is the most important factor influencing the hydrological process of watersheds and determining the watershed characteristics (Welde & Gebremariam, 2017; Woldesenbet et al., 2018). LULC maps of 30 m spatial resolution for 1985, 2002, and 2017 are published in Belay & Mengistu (2019). LULC maps were an important component of the SWAT model. Therefore, the classified (1985, 2002, and 2017) and predicted LULC (2033) maps were used separately to show the hydrological response to changes in LULC changes in the study watershed. The SWAT model requires the conversion of the LULC types into four-digit SWAT codes. The land use codes were assigned by the SWAT database, namely AGRL, FRST, SHRB, GRAS and URBN were assigned to indicate areas of cultivation, forest, shrub-bushland, grassland, and urban respectively. The CA-Markov chain model integrated with the multi-criteria evaluation was used to predict LULC map of 2033 for the study watershed. The detailed procedures of the model is well presented in the previous chapter of this study.

7.2.1.3. Soil and geological information

The soil physical and chemical properties parameters required by SWAT, such as texture, available water content, hydraulic conductivity, bulk density, and organic carbon content for different layers of each soil type are determining factors of runoff. According to the FAO classification, there are five types of soil in the study watershed as it is described in the earlier part of this study. The soil map (Scale 1:250,000) of the study watershed used for the SWAT model was obtained from the Ethiopian Ministry of Water Irrigation and Electricity (MoWIE) hydrology department. To

111

integrate the sol map with the SWAT model, a soil database with physical and chemical properties of soils was prepared for each soil layer and added to the SWAT soil database.

7.2.1.4. Digital elevation model (DEM)

The spatial data used in the SWAT model for this study was a Digital Elevation Model (DEM) data. The DEM (30 m resolution) of the study watershed was obtained from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), which was used to derive the hydrologic parameters such as flow direction and flow accumulation, watershed delineation, sub- basin definition, and HRUs setup in the ArcSWAT interface. The DEM was also used to drive the topographic parameters of the study watershed such as terrain slope, elevation, slope length and stream network. The stream network was validated through ground-truthing during fieldwork. Based on the topographic characteristics, the SWAT model identified 13 sub-watersheds.

7.2.1.5. Climate projection and bias correction

The detailed procedures about the selection of climate scenarios, climate models and methods of bias correction for rainfall and temperature minimum and maximum used in this study are available in chapter six of this study.

In this study, the linear scaling bias correction method was selected for temperature minimum and maximum following a review published in Teutschbein & Seibert (2010) who evaluated four bias correction techniques for temperature. The results of this study showed that linear scaling is suitable for both precipitation and temperature. The linear scaling method uses a correction factor for each month based on long-term historical RCM data and the observed data. The correction factor was applied for future climate scenarios of the RCP4.5 and RCP8.5 emissions. Bias correction was used to estimate and remove the bias correction in future RCMs output. In this study, the linear scaling bias correction method (V.1.0) Microsoft Excel file described by Shrestha (2015) was used to adjust the climate model average value to the observed values.

The temperature is corrected with an additive term, which is the difference between the monthly observation and the simulation data for the historical run given in Equations (20) and (21) below (Teutschbein & Seibert, 2012; Luo et al., 2018). The bias equations used for bias correction are shown below.

112

* T(d)T(d)(T(d))(T(d))his his   mobs mhis  (20)

* T(d)T(d)(T(d))(T(d))sim sim   mobs mhis  (21) where, d=daily, µm=long term monthly means, *=bias corrected, his=RCM simulated 1985-2017, sim=RCM simulated 2018-2050, obs=observed 1985-2017

For temperature (Equations (19) and (20)), the subscripts refer to a similar meaning as that of the equations in the rainfall equations. The bias correction was applied separately at each meteorological station for the baseline and future simulations.

Figure 7.1a shows the average observed and simulated rainfall and temperature data of the Muga watershed for the reference period (1985-2017). In the simulations from 1985-2017, an overestimation of the RCMs daily rainfall occurred. In this sense, the bias correction effectively adjusts the RCMs simulation of daily rainfall bias very well except for a slight overestimation. Other studies have shown that bias correction can correct climate model simulations (Teutschbein & Seibert, 2012; Fang et al., 2014; Manish Shrestha et al., 2017). Also, the RCM model is characterized by a steady underestimation of the maximum temperature (Tmax) under RCP 4.5 (Figure 7.1b), while the ensemble model of the RCM overestimates the minimum temperature (Tmin) under RCP 4.5 (Figure 7.1c). The RCM overestimation of Tmin and the underestimation of Tmax in RCM were effectively corrected for the observed daily Tmax and Tmin by bias correction. As a result, the cold bias and warm bias of the maximum and minimum temperatures improved significantly.

Figure 7.1 shows the bias-corrected statistical results under RCP 4.5 and the results showed that the bias correction significantly improves the simulated data as the RMSE values decrease and the SD values closer to observed data. As indicated in Table 7.2 the correlation between observation and simulation data has been significantly improved. It means that the correlation values of uncorrected rainfall data are generally lower than their bias-corrected counterparts, indicating that the bias-corrected values better represent the patterns of temperature over the study watershed. The results of this study showed that the bias correction result is acceptable and consistent with previous research results as reported by Gebre & Ludwig (2015), Sisay et al. (2017). The two RCP scenarios (RCP RCP 4.5 and RCP 8.5) were considered for hydrological modeling.

113

Table 7.2 The statistical measures of climate variables before bias correction and after bias correction for the Muga watershed

Before bias After bias Variables Statistics Observed correction correction Rainfall (RCP4.5) Mean (mm/day) 3.48 2.92 Correlation 2.5 0.74 0.76 RMSE (mm/day) 4.21 3.74 Temperature Mean (oC) 11.78 10.7 minimum (RCP 4.5) Correlation 0.65 0.79 RMSE (oC) 10.5 2.2 1.07 Temperature max Mean (oC) 27.28 24.3 Maximum (RCP 4.5) Correlation 26.4 0.79 0.81 RMSE (oC) 1.46 1.27 Rainfall (RCP8.5) Mean (mm/day) 3.4 2.62 Correlation 2.7 0.76 0.81 RMSE (mm/day) 4.1 3.4 Temperature Mean (oC) 11.78 11.07 minimum (RCP 8.5) Correlation 11.2 0.69 0.78 RMSE (oC) 2.8 1.2 Temperature Mean (oC) 27.28 24.2 Maximum (RCP 8.5) Correlation 26.3 0.73 0.75 RMSE (oC) 1.75 1.23

114

Figure 7.1 Daily observed and RCMs simulated (before and after correction) rainfall (a), maximum temperature (b), and minimum temperature (c) from the study watershed during 1985-2018. Raw and corrected represent model simulations before and after bias correction under RCP 4.5

In this study, the nearest point from the real rain gauging and meteorological stations were selected to calibrate the SWAT model for the Muga watershed, as recommended by Cousino et al. (2015). After the bias correction was made, the monthly water balance component simulation of the reference period (1985-2017) was compared to the next period (2018-2050) to evaluate the effects of climate change on the hydrological process.

115

7.2.2. The Soil Water Assessment Tool (SWAT)

The Soil Water Assessment Tool (SWAT) is a physically based and semi-distributed hydrological model developed at the USDA-ARS (Arnold et al., 2012). The model is developed to study small- scale and complex watersheds with varying land use, soil, and slope classes. The SWAT can evaluate the long-term impacts of different land management and climate variability/ change on biomass production, water quality, and sediment and agricultural chemical yields with the capability to study large-scale and complex watersheds for longer (Arnold et al., 2012). It simulates the hydrological cycle daily through water balance for each hydrological response unit (HRU), which are units with similar soil types, land use, and slope classes (Arnold et al., 2012).

HRUs are the basic units for soil water content, surface runoff, nutrient cycles, sediment yield, crop growth and simulated management practices. The model simulates the hydrological cycle at each HRU using the following water balance equation (Equation 22) (Arnold et al., 2012).

t SWSW(pQEWQt0daysurfaseepgw  (22) i1

Parameters of the weather generator were calculated by WGNMaker 4.1. As it is mentioned above Yetnora weather station was selected to fill gaps in meteorological data. WGNMaker 4.1 includes daily precipitation, daily highest and lowest temperature, and maximum daily precipitation in half an hour. Due to lack of maximum daily precipitation in half an hour, the data were estimated by using daily precipitation divided by 48 approximately.

The SWAT model is suitable to understand the effects of climate variability and LULC change on soil erosion and streamflow on single or multiple watersheds (Arnold et al., 2012; Koch et al., 2012; Wang et al., 2016). Some researchers have used the model across different watersheds in Ethiopia (e.g., Betrie et al., 2011; Setegn et al., 2011; Tibebe & Bewket, 2011; Gessesse et al.,

116

2015). They proved that the SWAT is a useful tool to examine the erosion and streamflow response of a watershed to LULC and climate change.

The SWAT model provides two methods for estimating runoff: The Soil Conservation Service (SCS, 1972) Curve-Number method and the Green & Ampt (1911) infiltration method. The Curve- Number method, which is linked to the United States Department of Agriculture (SCS – USDA), is the standard method used by SWAT. In this study, the watershed runoff response for the rainfall events was simulated using the SCS runoff equation (Schneiderman et al., 2007) to support the determination of runoff mechanism using Equation 23.

(PI) 2 Q  a (23) (PI)sa

Where, Ia is the rainfall lost as initial abstraction; S: is the maximum retention storage of the soil in mm; P-Ia is the effective rainfall in mm. The initial abstraction accounts for all water losses due to interception, evaporation, depression, storage, surface detention, evaporation and infiltration before runoff begins. The maximum retention storage conceptually represents a potential amount of water retained in the soil.

The simulated runoff from Equation 23 using the developed Se value was plotted against the measured runoff, and both R2 and Nash Sutcliff Efficiency (NSE), Nash & Sutcliffe (1970) were used to indicate how well the plot of observed versus simulated data fits the 1:1 line. Using the SCS-CN method, it is possible to estimate the surface runoff blade, considering precipitation data and parameters that characterize the watershed (Betrie et al., 2011; Fentaw et al., 2018). The parameters include spatial topography, Digital Elevation Model (DEM), land use, soil, and hydro- meteorological data. The LULC and climate change impact assessment procedure using the Soil and Water Assessment Tool (SWAT) is summarized in Figure 7.2.

7.2.3. Partial least square regression (PLSR) Model

Analyzing hydrological responses and change using hydrological simulation cannot reveal each LULC type's contribution to hydrological change. The combined use of the hydrologic model and PLSR could be a viable approach to scrutinize the impact and contribution of each LULC change on the catchment hydrological responses. The pair-wise Pearson correlation combined with the

117

PLSR model (Abdi, 2010) was applied to investigate the relationship between individual LULC types and each hydrological component. The PLSR model was used to evaluate the relationship between annual water balance components and particular land use classes.

PLSR is a robust multivariate regression method that is appropriate when the predictors exhibit multicollinearity (Yan et al., 2013). The change in the Land use maps of 1985, 2002, 2017, and 2033 and the mean annual water balance components simulated based on each LULC map were used in this study. These are used to quantify the contribution of changes in individual LULC types to water balance components at a watershed level. In this research, the independent variables are the changes in the five LULC types (cultivation, forest, shrub-bush land, grassland, and urban), while the dependent variables are hydrological components.

One of the exciting features of PLSR is that the relationship between the predictors and the response function can be inferred from the weights and regression coefficients of individual predictors in the most explanatory components (Shi et al., 2013). Thus, it is possible to determine which LULC types most strongly interact with streamflow. The determination of the number of significant components is usually based on a criterion that involves cross-validation. Two main indices assess model validity and strength: (1) goodness of fit (R2), i.e., the proportion of variation 2 2 in the dependent variable that the model explains; and (2) R cross (cross-validated R ; goodness of prediction), i.e., the proportion of variance in the dependent variable that the model can predict.

A PLSR regression model provides significant and good predictions when R2 is greater than 0.5 2 and when R cross is greater than 0.097 (Trap et al., 2013). In PLSR modeling, the importance of a predictor for both the independent and the dependent variables is given by the variable importance for the projection (VIP). The terms having large VIP values (greater than 0.8) were the most relevant for explaining the dependent variable (Shi et al., 2013; Yan et al., 2013). Besides, the regression coefﬁcients of the PLSR models were used to show the direction of the relationships between the changes in individual LULC types and streamﬂow (Shi et al., 2013; Yan et al., 2013). It is thus possible to determine which LULC types most strongly interact with the hydrological components. Detailed information on the application of PLSR can be obtained from (Abdi, 2010) and Yan et al. (2013), and thus only a brief description of this technique is present here. The PLSR model analysis was performed using R software.

118

Climate change analysis Land use Water balance (past & future) analysis analysis

Spatial data Land use in 9 Land use ArcSWAT GCM Climate Data 2002 & 2017 driving factors Soil RCP 4.5 and 8.5 Temperature Climate Setup model Land use CA- (Tmax & Tmin) & run Extract future RCM Markov chain Model Land use RCP 4.5 and RCP 8.5 Initial parameters Precipitation DEM for calibration Future Land Bias correction use Type SWAT CUP

Sensitivity No Prediction future Land use in 1985, Analysis climate data 2002, 2017, and 2033 Observed Calibration flow data Climate data (past yes & future) SWAT2012 PLSR Validation Simulations

Impact of individual LULC LULC change SWAT 2012 class on hydrology impacts Simulation

Conditions: 1) Only LU change 2) Only climate change 3) Both change

Change Analysis of water balance

Figure 7.2 Flowchart for LULC and climate change impact assessment using the Soil and Water Assessment Tool (SWAT) and PLSR in Muga watershed. 7.3. Results and Discussion 7.3.1. SWAT calibration and validation Successful application of hydrological models relies heavily on the parameters' calibration and sensitivity analysis (Abbaspour et al., 2018). Calibration and validation processes are used efficiently with only observed data. Calibration of the SWAT hydrological process requires an accurate simulation of the streamflow in this regard. Discharge data collected from the Ethiopian Ministry of Water, Irrigation, and Electricity was used to calibrate and validate the SWAT model.

The SWAT model was calibrated over 21 years (1988–2009), and the years from 2010 to 2015 were used for model validation. The model simulations were divided into a warm-up, calibration, and validation. Three years (1985-1987) were used for the initialization of the model start-up (warm-up period) to minimize the impact of initial conditions on model simulations.

119

Analysis of the model parameters' sensitivity can identify the parameters that influence the model simulation results. Attention to the calibration of those parameters reduces model uncertainty and improves the reliability of the model simulations. In this study, an automatic parameter estimation procedure, the Sequential Uncertainty Fitting version 2 (SUFI-2) in the SWAT-CUP (SWAT Calibration and Uncertainty Procedures) (Abbaspour et al., 2018), was used to analyze the sensitivity of the parameter and to calibrate the model parameters. The observed streamflow data at the gauging station of Muga watershed (Figure 3.1) was used in this SWAT calibration analysis. The Sequential Uncertainty Fitting (SUFI-2) algorithm stands out due to its capability to account for all sources of uncertainty within the parameter ranges (Abbaspour et al., 2018). SUFI-2 is widely used (Setegn, 2008; Gebremicael et al., 2013; Dile et al., 2016) due to the relatively fewer required number of runs to achieve an acceptable result. Calibration is an attempt to better parameterize the model for a given set of local conditions and then reduce the uncertainty of the prediction.

Hydrological models often have at least some parameters that cannot be determined directly by field measurements. Therefore, identifying highly sensitive parameters can improve the performance of the calibration process. The highly sensitive model parameters were selected in the calibration procedure by referring to the previous literature related to SWAT calibration (Setegn, 2008; Gebremicael et al., 2013; Dile et al., 2016) and the results of initial sensitivity analysis. The most appropriate parameter range of the model was determined, and the uncertainty analysis was analyzed by reviewing the range results for each parameter. Consequently, the number of parameters in the calibration procedure decreased by eliminating the non-sensitive parameters.

A global sensitivity analysis for selected model parameters was implemented. T-stat and P-Value were used to assess the significance of the relative sensitivity of the parameters. The former index measures a parameter's sensitivity; the higher the absolute value, the more sensitive the parameter will be. The latter index determines the importance of sensitivity. If it is close to 0, the parameter is more sensitive. After calibration was completed, the parameters were adjusted manually for runoff simulation in the model, based on the optimal parameter set of the SWAT model. Observed runoff and simulated runoff were compared to evaluate model performance. Global sensitivity

120

results for the watershed showed that parameters with more negative and positive T-test values and less than 5% of P-values were highly sensitive to streamflow parameters.

In this study, thirteen parameters were initially selected. Finally, seven sensitive parameters with the most sensitivity values were selected. The set of calibrated sensitive parameters of the SWAT model and the adjusted values obtained and are presented in Table 7.3.

The results showed that the SCS runoff curve number (CN2) was more sensitive to streamflow, which is consistent with several other studies (Setegn, 2008; Gebremicael et al., 2013; Taffese & Zemadim, 2013; Dile et al., 2016; Melaku et al., 2018; Worqlul et al., 2018). Besides, a threshold depth of water in the shallow aquifer required for return flow to occur (GWQMN), Baseflow alpha- factor (ALPHA_BF), Surface runoff lag time (SURLAG), Soil evaporation compensation factor (ESCO), Groundwater delay (GW_DELAY), Deep aquifer and percolation fraction (RCHRG_DP) were found to be the most sensitive parameters for streamflow simulations. The ranking of sensitive parameters was determined by the absolute values of T-stat and P-value. The T-stat value measures the size of the difference in terms of variability in the sample data, whilst the P-value represents the probability of observing a test statistic at least as large as the one calculated with the assumption of a hypothesis is correct. A parameter with the highest absolute t-stat value and minimum P-value compared to the other parameters was considered highly sensitive (Abbaspour et al., 2018).

121

Table 7.3 List of parameters with fitted values and global sensitivity results for daily flow

Parameter name Description Parameter range Best sim Rank t-stat P-value A_CN2.mgt SCS runoff curve number 11.23-14.53 11.56 1 -23.62 0.00 V_ESCO.hru Soil evaporation compensation factor -0.09-0.12 0.02 2 10.46 0.00 V_ALPHA-BF.gw Base flow alpha factor or recession constant (days) 0.32-0.51 0.42 3 8.01 0.00 R__RCHRG_DP.gw Deep aquifer and percolation fraction -0.25-0.12 -0.02 4 -3.21 0.01 A__GWQMN.gw Threshold depth of water in the shallow aquifer 3621.29-3810.43 3693.16 5 2.63 0.02 required for return flow to occur (mm) R__GW_DELAY.gw Groundwater delay (days) 0.05-0.64 0.07 6 -2.19 0.04 V__SURLAG.bsn Surface runoff lag coefficient -0.07-0.29 0.01 7 2.18 0.05

122

The ability of the model to predict streamflow should be evaluated using statistical measures to determine the quality and reliability of SWAT hydrological predictions by comparing the simulated values with those observed (Abbaspour et al., 2018). SUFI2 offers many options as model performance indicators for calibration and validation periods. Two commonly used precision statistical measures; The Nash–Sutcliffe efficiency (NSE) (Nash & Sutcliffe, 1970) and the Percentage of Bias (PBIAS), were used to evaluate the performance of the model in this study. They indicate the similarity between the simulated runoff and observed runoff. NSE and PBIAS are calculated as follows:

n 2 i1 QQ NSE1obsisimi (24) n 2 i1 QQ  obsimean n i1 QQobsisimi PBIAS1n X100 (25) i1 Qobsi

Where NSE is the Nash-Sutcliffe coefficient, PBIAS is the percentage of deviation, Qobs is the observed, Qsim is simulated, and Qmean is mean streamflow.

The statistical coefficient for performance evaluation shows that the simulation streamflow has a good agreement with the observed, resulting in NSE of 0.79 and 0.82 for calibration and validation, respectively, and percent bias (PBIAS) of 8.2% and 9.3%, for calibration and validation, respectively (Table 7.4). Thus, the SWAT model has a “very good” performance for the calibration and validation periods, as NSE > 0.5 and PBIAS < + 25% (Moriasi et al., 2007).

A comparison of the simulated monthly streamflow versus the observations is shown in Figure 7.3, and the statistical evaluation measures are listed in Table 7.4. The hydrograph shows a very good agreement between the two in the watershed. The model overestimated some peak flows during the rainy season and underestimated some low flows during the dry season. Therefore, the SWAT hydrological model successfully simulates the monthly streamflow with reasonable accuracy.

123

Figure 7.3 Simulated and observed streamflow for calibration and validation periods Table 7.4. SWAT model evaluation parameters for streamflow in the Muga watershed Constituent Evaluation parameters Calibration period (1988-2010) Validation period (2010-2015) NSE 0.79 0.82 PBIAS 8.2% 9.3%

To investigate the impact of LULC and climate change on hydrological components, this analysis considers three conditions: (a) the individual impact of LULC changes on hydrological components, (b) the individual impact of climate change on water balance components, and (c) the combinations of the impact of both on the hydrological components. The simulations were carried out using land uses of 1985, 2002, 20117, and 2033 and two periods of climate change (1985-2017 and 2018-2050).

7.3.2. Impacts of land use/ land cover change on Muga watershed hydrology

Changes in LULC due to human activities such as deforestation, loss of vegetation and increased settlement are the major contributors to changes in the hydrological response of a watershed (Welde & Gebremariam, 2017; Woldesenbet et al., 2017; Gashaw et al., 2018; Berihun et al., 2019). Four different periods of land use (1985, 2002, 2017, and 2033) were used individually in the SWAT model to investigate the impact of LULC changes on the water balance components of the Muga watershed. The calibrated and validated SWAT model was simulated using the four LULC maps, but all the other SWAT inputs were kept the same in all simulations. Figure 7.4 shows the impacts of past and future LULC on the proportion of hydrological components.

124

Figure 7.4 Water balance of three historical and one future LULC simulated through SWAT using meteorological data from 1985 to 2017 (precipitation equals 1087 mm yr-1)

The results of this study showed that although changes in LULC lead to increased surface runoff and water yields, a considerable reduction was also observed in streamflow, evapotranspiration and lateral flow. The mean annual values of water balance components (i.e., mean annual surface runoff (SURQ), water yield, lateral flow (LQ), groundwater flow (GWQ), and evapotranspiration (ET)) simulated by SWAT under different LULC are given in Table 7.5. This study showed that the mean monthly streamflow for the rainy season was increased from 133.2 mm in 1985 to 147.0 mm in 2002, 146.8 mm in 2017, and it is expected to increase to 149.6 mm in 2033, while the average monthly streamflow for the dry season has decreased from 71.1 mm in 1985 to 43.5 mm in 2002, 48.6 mm in 2017 and, expected 49.0 mm in 2033.

The average annual streamflow for the LULC 1985, 2002, 2017 and 2033 scenarios is 203.3, 190.5, 195.4 and 198.6 mm, respectively. This revealed that the average annual streamflow decreased significantly during the study period. The decline in mean annual streamflow is due to expansion of cultivated and urban areas as well as loss of forests, shrub-bush and grasslands. On the other hand, surface runoff increased from 190.4 mm in 1985 to 204.6 mm and 220.4 mm in 2002 and

125

2017, respectively, and expected to rise to 219.5 mm in 2033. These indicate that increased in cultivated and urban areas at the expense of forests and shrub-bush lands may be the reason for the increased rate of surface runoff. The results of this study are similar to the previous studies conducted by Gashaw et al. (2018) and Berihun et al. (2019). They reported that surface runoff was increased while streamflow was declined due to rapid deforestation of natural forests, shrub- bushlands and grassland as well as expansion of farmland and urban land.

In this study, groundwater flow was decreased by -4.36% and -9.21% in 2002 and 2017, respectively, and is expected to decrease by 10% in 2033 compared to 1985. Similarly, the average annual ET was decreased from 592.4 mm in 1985 to 580.1 mm in 2002, 583.3 mm in 2017, and expected to 586.8 mm in 2033, which is 12.3 mm lower in 2002, 9.1 mm lower in 2017, and 5.6 mm lower in 2033 compared to 1985. A possible reason for the decline in the average annual ET is the decline in the area of forests and shrub-bushlands. Similarly, there was a negative trend in the mean annual lateral flow over the Muga watershed, declining by 3.9% in 2002, 8.5% in 2017, and 7.0% in 2033 compared to 1985. According Ayivi & Jha (2018), a slight decrease in groundwater flow and lateral flow can be attributed to low soil infiltration and high surface runoff.

The results of this study showed that the mean annual water yield was increased slightly in 2002 and 2017 in the study watershed as compared to 1985. In 2002 it increased by 2.7 mm and in 2017 by 5.6 mm, while water yield in the study watershed is expected to increase by 3.8 mm in 2033. According to the results of this study, the decline of vegetation cover predominantly forest induces to an increase in surface runoff and further increases water yield. According to Woldesenbet et al. (2017) and Gashaw et al. (2018), the decline of forest cover leads to higher water yield. Aylward (2005) also reported that water yield was increased with a decline of forest cover, and water yield would increase if forests were cleared.

The results of this study showed that between 1985 and 2017, some parts of forest, grassland and shrub-bushlands were converted to cultivation and urban, and are expected to continue in the future, which causes a considerable increase in surface runoff and water yield in Muga watershed. The results of this study show an agreement with previous studies (Bewket & Sterk, 2005; Shrestha et al., 2018), who reported that the conversion of forested areas to cultivated areas contributed to an increase in surface runoff. The decline of vegetation cover, creates less canopy interception,

126

resulting in more rainfall reaching the soil surface (also known as direct rainfall), increasing the potential of infiltration and runoff from the land (Rogger et al., 2017).

The increase in surface runoff can be attributed to the decrease in infiltration (Benegas et al., 2014). In addition, soil infiltration capacity is reduced in areas where deforestation has occurred and areas of increased settlement. The reduction of soil water infiltration, in turn, results in a decrease in groundwater flow. This means that forests absorb water through leaves and roots and promote the infiltration of rainwater into groundwater. The increasing demand of the growing population for the use of groundwater can also reduce groundwater by increasing the abstraction of groundwater.

This study showed that the mean annual water yield increased slightly by 0.6% in 2002 and by 1.2% in 2017 compared to the baseline year, which is most likely due to the gradual increase in the cultivated land. The increase in surface runoff was due to the decline of forests and shrub- bushlands, leading to higher water yield. The average annual water yield in the future under the LULC change scenario is expected to be almost the same as the baseline period, and it will likely increase by only 0.8%.

Expansion of residential areas together with increased agricultural activities, reduction in vegetation cover and forest areas leads to greater areas of impervious surfaces and soil compaction, which reduces the potential for soil infiltration and possibly increases surface runoff (Woldesenbet et al., 2018). Table 7.5 shows water balance components under different LULC scenarios and historical climate data (1985-2018) in the Muga watershed. In general, if the incoming water through overland flow is more than the infiltration water, the excess water can cause an increase in surface runoff. As a result, water yield, a combination of surface runoff, groundwater flow, and lateral flow, has also increased with the cultivated and urban areas.

The results of this study showed an agreement with other studies (e.g., Bewket & Sterk, 2005; Gebrehiwot et al., 2010; Welde & Gebremariam, 2017; Gashaw et al., 2018; Birhanu et al., 2019; Shawul et al., 2019; Woldesenbet et al., 2018) which reported an increase in surface runoff and water yield. For example, Bewket & Sterk (2005) in Chemoga watershed, upper Blue Nile basin of Ethiopia concluded that forests have the effect of reducing surface runoff and increasing the infiltration and evapotranspiration, while the expansion of agriculture reduces infiltration. Gashaw et al. (2018) also showed that a decrease in groundwater flow and an increase in surface runoff in

127

Andasa watershed in the Upper Blue Nile basin between 1985-2015 were associated with changes in LULC of the watershed. According to Shawul et al. (2019), between 1974 and 2014, an increase in surface runoff and decreased groundwater flow in the upper Awash river basin were associated with LULC change of the basin. Contrary, when agricultural land is plowed, compaction of the lower soil horizons occurs, causing the reduction of infiltration capacity and ultimately more runoff (Jin et al., 2008). In Muga watershed, a rapid change in LULC was observed between 1985 and 2017 due to different proximate and underlying drivers. As a result, the scarcity of land and the need to cultivate more land encouraged the community to cultivate on steep slopes. Cultivation of steep slopes increases, in turn, increased surface runoff and reduces infiltration, which affects the availability of water resources in Muga watershed.

Table 7.5 Average annual evapotranspiration, surface runoff, groundwater, water yield, and lateral flow change under different LULC scenarios and historical climate data (1985-2018) in the Muga watershed

P STQ (m3/s) ET SURQ GWQ LQ WYLD Year mm mm % Wet Dry Mm % mm % mm % mm % mm % 1985 1087 203.3 133.2 71.1 592.4 190.4 208.5 60.8 459.7 2002 1087 190.5 -7.2 147.0 43.5 580.1 -2.08 204.6 7.46 199.4 -4.36 58.4 -3.9 462.4 0.59 2017 1087 195.4 -5.8 146.8 48.6 583.3 -1.54 220.4 15.76 189.3 -9.21 55.6 -8.55 465.3 1.22 2033 1087 198.6 -3.3 149.6 49.0 586.8 -0.95 219.5 15.28 187.5 -10.0 56.5 -7.07 463.5 0.83

The trend in surface runoff and water yield was inconsistent during the study periods, however these showed an increasing trend in 2002, 2017 and 2033 as compared to 1985. On the other hand, a decline of mean annual streamflow, evapotranspiration and lateral flow was observed as compared to the reference year. The study analysis revealed that the area of forest cover has declined, historically. Therefore, forest loss (deforestation) will accelerate surface runoff and water yield while slowing evapotranspiration over the Muga watershed in 2002, 2017, and 2033 compared to LULC in 1985. The results of this study showed a similar result with previous studies (e.g., Chemura et al., 2020; Ridwansyah et al., 2020), which revealed that the shift of high evapotranspiration LULC to low evapotranspiration leads to an increment of surface runoff. The reduction in evapotranspiration in 2002, 2017, and 2033 compared to 1985 may be due to the loss of vegetation cover (forests, grasses, and shrub-bush). As a result, it supports surface runoff and water yields.

7.3.3. Impacts of individual land-use changes on Muga watershed hydrology

Investigating hydrological response to LULC change using hydrological modeling may not reveal the contribution of each LULC type to the hydrological response. The combination of hydrological

128

model and PLSR may be a viable method to study the effect and contribution of each LULC change to the hydrological response of the watershed.

The pairwise Pearson correlation combined with PLSR model was carried out to assess the relationship between individual LULC types (cultivated, grassland, shrub-bushlands, forests, and urban areas) and hydrologic components (i.e., surface runoff, groundwater recharge, lateral flow, water yield, and evapotranspiration) (Abdi, 2010). The relationship between each type of LULC and hydrological component is computed using the pair-wise Pearson correlation, while the contribution of their change to the hydrological component can be quantified using the PLSR model, as shown in previous studies (e.g., Nie et al., 2011; Shi et al., 2013; Yan et al., 2013; Woldesenbet et al., 2017). Pearson correlation analysis was performed at the watershed level to analyze the linear correlation between LULC (dependent variables) and hydrological components (explanatory variables). This approach is vital in determining whether the observed LULC change was large enough to induce changes in the hydrological components.

In this study, almost all land use types showed a strong correlation with individual hydrological components in 1985, 2002, 2017 and 2033. For instance, changes in cultivated land had a positive correlation with annual surface runoff (SURQ) and total water yield but a negative correlation with groundwater recharge (GWQ), Streamflow (STQ), ET, and lateral flow (LQ). In contrast, changes in forest area had a significant positive correlation with ET at a significant level of 1% and positively correlated with groundwater and lateral flow, but a negative correlation with surface runoff and total water yield. Furthermore, grassland changes had a strong negative correlation with surface runoff and water yield, while its correlation with groundwater, STQ, ET, and lateral flow was positive. The expansion of urban and cultivated land accompanied by a decrease of forest, grassland, and shrub-bushland cover during the study period increased total water yield and surface runoff while groundwater flow and evapotranspiration were declined. The pairwise relationships of the five LULC types and hydrologic components are presented in Table 7.6.

129

Table 7.6 Pairwise Pearson correlation for changes in LULC variables and water components between LULC in 1985, 2002, 2017, and 2033

*&** Significant value at p=0.05 & 0.01, respectively, GL: Grassland; CL: Cultivated land; SHRUB: Shrub-bushland; STQ: Streamflow; WetSURQ: Surface runoff (mm); LQ: Lateral flow (mm); GWQ: Groundwater flow (mm); WYLD: Water Yield (mm); ET: Evaporation transpiration (mm)

The correlation circle plot provides a graphical display of the X and Y variables using the two PLSR component correlation coefficients on the t1 and t2 axes. It presents an overview of the relationship between the model variables in monoplot form. In the correlation graph, the circle represents the regression coefficient (R² = 1.0), the length of the line to the circle represents how well the variables are represented. While the short line that appears near the center of the circle within R² < 0.2 is not well represented by the variables. The angle between the vectors estimates the correlation between the variables and the cosine of the angle between these lines gives the correlation coefficient. A small angle indicates that the variables are positively correlated, while a 90-degree angle indicates that the variables are not correlated, and the angle close to 180 degrees indicates that the variables are negatively correlated (Tenenhaus et al., 2005).

The monoplot exhibits that the change in cultivated land affected surface runoff, and the area of cultivated land is inversely correlated with the groundwater flow, lateral flow and ET. Interestingly, the correlation map showed that forest and shrub bushland were strongly correlated with ET, while grassland was highly correlated with lateral flow variation. As the urban area

130

increases, there will be more impermeable soils, which will increase the surface runoff, thus increasing the watershed total water yield. Therefore, a strong correlation between the urban area and water yield was obtained. In addition, it can be inferred from the monoplot that the correlation between x or predictor variables is that cultivated land is negatively associated with grassland and forest.

The results of the PLSR model for the annual water balance components (STQ, ET, SURQ, GWQ, LQ, and WYLD) in the Muga watershed are given in Table 7.7, and a summary of the PLSR weights, variable importance of the projected values (VIP) and percent of explained variability and cumulative percentage of explained variability in the response variables is given in Table 7.8. The PLSR model automatically generated two model components (Comp1 and Comp2). The variable importance for the projection (VIP) of each explanatory variable for two components allows identifying the most explanatory variables that contribute the most to the models. The correlation circle graph indicates the importance of individual land use classes on the variation of values of the watershed's hydrological components; however, the VIP values can provide a more comprehensive expression of the relative importance of predictors. Values VIP > 1 and/ or VIP > 0.8 based on Wold (1995) criteria indicate that the predictor variable is considered to be significantly important to the corresponding dependent variable.

The two PLSR components are generally sufficient for a limited number of products, and the choice of two components allows convenient graphical analysis (Tenenhaus et al., 2005). The cumulative explained variations in Y and cumulative Q2 that corresponds to the correlations between the explanatory (X) variable and dependent (Y) variables with the components were above 0.75, which indicated their good predictive ability of the model. The first PLSR component explained 97.7% of the annual, wet, and dry streamflow variance, 90.4% of the variance in the ET, 81.6% in the SURQ, 81.4% in the GWQ, 82.0% in the LQ, and 82.2.0% in the WYLD. The addition of second components explained about 2.2% to 17.8% of the variance. This implies that the addition of components to the second PLSR models did not substantially increase the percentage of variance that was explained (Table 7.7). As linear combinations of original variables that define scores, PLSR weights can describe the quantitative relationship between predictors and results (Abdi, 2010).

131

Table 7.7 Summary of the PLSR models for the water balance components in the Muga watershed

Variables R2 MSE RMSE Q2 Comp Explained Cum explained Q2 cum variation in Y (%) variation in Y (%) Annual, wet 0.99 0.18 0.13 0.99 1 97.7 97.7 0.96 and dry STQ 2 2.2 99.9 0.99 ET 0.98 0.22 0.47 0.95 1 90.4 90.4 0.95 2 9.4 99.8 0.99 SURQ 0.94 2.91 2.0 0.86 1 81.6 81.6 0.86 2 17.2 98.8 0.99 GWQ 0.98 1.45 1.20 0.77 1 81.4 80.1 0.77 2 17.8 99.2 0.99 LQ 0.88 0.46 0.68 0.85 1 82.0 82.0 0.73 2 16.3 98.3 0.99 WYLD 0.79 0.85 0.92 0.78 1 82.2 76.0 0.77 2 16.5 98.5 0.99 R2 indicates explained variation in response Q2 indicates the predictive ability of the model

Although the PLSR weight (Table 7.8) indicates the importance of individual changes in the LULC classes for the water balance components, a more convenient and comprehensive expression of the relative importance of the predictors can be obtained by examining their VIP Values. The results of the relative importance of predictors of the model showed that grassland (VIP=1.07), cultivated land (VIP=1.069), and urban land (VIP=1.03) were the most significant contributors for the variables of hydrological components. Forestland and shrub-bushland were also important contributors affecting water balance components with the VIP values greater than 0.8 but less than 1 (Table 7.8). Furthermore, the PLSR weight (w) value gives a more precise and accurate idea of the direction of influence that LULC class has on the corresponding water balance components. The squares of w values more than 0.3, i.e. (w2 > 0.3), indicate that the PLSR components are mainly weighted on the corresponding variables. The bold values shown in Table 7.8 show that the PLSR component is considered to be more important for the response variables.

For streamflow (annual, dry, and wet), for example, the highest VIP =1.179 was obtained by forest followed by urban and shrub-bushland, with VIP value of VIP = 1.144 and 0.996, respectively. The w value indicates the direction of the inﬂuence of the LC class on the corresponding water balance components. For SURQ and GWQ, the highest VIP values are obtained for the urban land (VIP = 1.245), followed by cultivated land (VIP = 1.233) and grassland (VIP = 1.132). The change in cultivation land area has a positive regression coefficient for SURQ and a negative regression

132

coefficient for GWQ (0.339 and -0.351, respectively) (Table 7.9). The amount of SURQ is inversely proportional to the forest area (w = -0.104) while it is directly related to change in cultivated land (w = 0.339) and urban area (w = 0.297). The highest VIP value for LQ was obtained for cultivated land (VIP =1.267; w = -0.317); for urban areas (VIP = 1.217; w = -0.294) and for grasslands (VIP = 1.191; w =0.305). Similarly, the expansion of cultivated land had a greater impact on GWQ (VIP =1.267; w = -0.351); for urban (VIP = 1.217; w = -0.297) and for grassland (VIP = 1.191; w =0.354). Change in cultivation land area has a positive regression coefficient for WYLD and a negative regression coefficient for ET (0.280 and -0.067, respectively). In contrast, changes in forest and shrub-bushland area had a negative regression coefficient for WYLD (-0.167 and -0.065, respectively) and a positive regression coefficient for ET (0.407 and 0.411, respectively).

In this study, a significant correlation of each LULC class with most hydrological components indicates that almost all land use is very important for the change. This result supports the general idea that increasing the amount of one hydrologic component increases some other components while reducing the number of other components.

133

Table 7.8 Variable importance for the projection (VIP) and partial least square regression (PLSR) weight (w) values for the hydrological components for the Muga watershed

PLSR Annual, wet, and dry STQ SURQ and WYLD GWQ and LQ ET parameters VIP w*1 w*2 VIP w*1 w*2 VIP w*1 w*2 VIP w*1 w*2 GL 0.918 +0.264 -0.557 1.132 -0.508 -0.424 1.191 0.536 0.402 0.981 0.093 -0.620 CL 0.991 0.389 0.418 1.233 0.556 0.251 1.267 -0.572 -0.226 0.855 -0.246 0.505 Urban 1.144 +0.542 0.094 1.245 0.562 -0.043 1.217 -0.550 0.066 1.059 -0.482 0.191 Forest 1.179 -0.549 0.394 0.932 -0.321 0.547 0.951 0.281 -0.567 1.346 0.617 0.288 SHRUB 0.996 -0.428 0.604 0.896 -0.116 0.679 0.868 0.065 -0.687 1.247 0.564 0.532 Bold-faced values are greater than the threshold value of VIP > 1, and the square of w > 0.3 indicates that the predictor variables are more important to the response variables; (+ or −) = direction of regression parameters.

Table 7.9 PLSR coefficients presenting the effects of different land use types on hydrological components between 1985 and 2033 (projected) in the Muga watershed

PLSR Response PLSR predictors Model variables Grassland Cultivated Urban Forest Shrub-bushland Annual STQ -0.094 -0.021 -0.223 0.420 0.445 Wet STQ -0.231 0.280 0.318 -0.250 -0.151 Dry STQ 0.105 -0.188 -0.303 0.346 0.295 SURQ -0.335 0.339 0.297 -0.104 0.028 GWQ 0.354 -0.351 -0.297 0.081 -0.057 LQ 0.305 -0.317 -0.294 0.127 0.007 WYLD -0.255 0.280 0.287 -0.167 -0.065 PLSR 4 ET -0.041 -0.067 -0.249 0.407 0.411 Note: The positive and the negative signs revealed the position of influence

134

The direction of the change in components of water balance with LULC change in this study is comparable to previous hydrological studies conducted in the Abbay River Basin (Bewket & Sterk, 2005; Welde & Gebremariam, 2017; Gashaw et al., 2018; Berihun et al., 2019), which showed substantial variation in the hydrology of the watershed related to LULC change. In this study, the tendency for future evapotranspiration is lower than the year 2017, which may be due to the conversion of agricultural land and forests into residential areas in the future. On the other hand, total water yield showed an increasing trend between 1985 and 2017 and is expected to increase in 2033. The results of this study also showed that in the case of future LULC, the surface runoff is predicted to increase by approximately 16.9% compared to the year 2017 due to a conversation of 5298 hectares of grassland, forest, and shrub-bushland into cultivated lands and urban areas (Table 6.6).

7.3.4. Impact of climate change on hydrological components in Muga watershed

In this study, climate projections were used in the hydrological model to examine the effect of future climate change on hydrological components (surface flow, groundwater flow, water yield, lateral flow and evapotranspiration). Thus, water balance components simulated by SWAT model for the baseline period (1985-2017) were compared with the results for the next period (2018- 2050) under RCP4.5 and RCP8.5 to examine the effects of climate change. The effect of the historical climate and two trajectories (RCP 4.5 and RCP8.5) on water balance with bias correction for the period (2018-2050) using the recent LULC of 2017 on water balance components is shown in Figure 7.5.

135

Figure 7.5 Water balance components simulated by SWAT for two different periods: (i) historical climate data (1985-2017) using LULC 2017 and (ii) future climate data (2018) under RCP 4.5 and RCP 8.5 using LULC 2017.

Climate change causes evapotranspiration changes due to increasing radiative forcing, which ultimately changes the water cycle system (Field & Barros, 2014). The results of this study showed that climatic parameters (i.e., rainfall, minimum and maximum temperature) are expected to increase in the future, which is reflected a change of water balance components specifically, mean annual surface runoff, groundwater flow, streamflow, lateral flow and water yield for the period consists of the years 2018-2050 under RCP4.5 and RCP8.5, as compared with historical climate datasets (1985-2017). The baseline period consists of the years 1985-2017, surface runoff was approximately 217.4 mm and in the future climate scenario, which is expected to increase to 230.2 mm and 231.5 mm yr-1 under RCP 4.5 and RCP 8.5, respectively. The results of this study is inline with the results of the study conducted by Chanapathi & Thatikonda (2020).

In contrast, the result of the study by Dibaba et al. (2020) in the Finchaa catchment, upper Blue Nile basin of Ethiopia revealed that surface runoff decreased by 7.33% and 12.32% under RCP 4.5 and RCP 8.5, respectively, during the years 2021-2050 as compared to the baseline period consists of the years 1986-2015. As a result, the increase in surface runoff directly relates to erosion and sedimentation and affects erosion and sedimentation. Therefore, the transport capacity of the surface runoff determines the amount of material subject to erosion. Abebe & Gebremariam (2019) reported that reduced surface runoff significantly reduced sediment yield.

The results of this study indicate that average annual rainfall is expected to increase in the 2050s, causing an increase in the amount of streamflow discharge. Similarly, future temperature is also expected to increase under the RCP 4.5 and 8.5 emission scenarios, and that tends to increase ET under simulation LULC 2017 and RCP 4.5 and LULC 2017 and RCP 8.5 compared to historical climate data sets. ET is expected to rise by 10.4% (61.0 mm) under simulation LULC 2017 and RCP 4.5 and by 11.3% (66.3 mm) under simulation LULC 2017 and RCP 8.5 compared to historical climate data. Therefore, there is a distinct increase in the prediction of ET given by the climate simulation scenario.

136

The intensity of the effects of climate change on hydrological components considered in this study (i.e., water yield, groundwater recharge, lateral flow and ET) will increase in the future under both emission scenarios, which is due to an increase in precipitation and temperature in the 2050s within the study watershed. In this study, the expected increase in most hydrologic components is more tremendous in the RCP 8.5 simulation scenario than in RCP 4.5. For example, the average projected average annual streamflow will increase by 13.8% and 21.3% under RCP 4.5 and RCP8.5 scenarios, respectively, from the historical baseline period. The average changes in groundwater recharge, surface runoff, lateral flow and water yield will be 20.1%, 5.9%, 10.4%, 2.5%, and 11.6%, respectively, according to RCP 4.5 with 2017 LULC, compared to the reference period consists of the years 1985-2017. For the same period and the 2017 LULC map, the corresponding increases for RCP 8.5 are 21.4%, 6.5%, 11.3%, 7.0%, and 12.9%. Changes in average annual groundwater recharge, lateral flow and water yield indicate the effect of trajectories (RCPs) on these hydrological components.

The result of this study is consistent with Dile et al. (2013) and Dibaba et al. (2020), they report that climate change is expected to change hydrological elements and alter the hydrological cycle behavior. Table 7.10 shows the number of hydrological components under historical LULC 2017 and climate change periods (1985-2017) and (2018-2050). As shown in Table 7.10, the increasing trend of the study area hydrological components is accompanied by a significant increase in rainfall for both RCP 4.5 and RCP 8.5.

Table 7.10. Hydrological components simulated by SWAT under historical climate data sets (1985-2017) and future climate data (2018-2050) under RCP4.5 and 8.5 using LULC 2017

STQ ET SURQ GW LQ WYLQ Periods mm ∆% Mm ∆% mm ∆% mm ∆% mm ∆% mm ∆% LULC 2017 & CC 1985-2017 193.4 586.3 217.4 190.3 56.8 463.3 LULC 2017 & RCP 4.5 220.1 13.8 647.3 10.4 230.2 5.9 228.6 20.1 58.2 2.5 517.0 11.6 LULC 2017 & RCP 8.5 232.2 21.3 652.6 11.3 231.5 6.5 231.0 21.4 60.8 7.0 523.3 12.9 Note: CC: Climate change; LULC: Land use land cover in 2017

7.3.5. The combined impacts of land use land cover and climate changes on hydrology in Muga watershed The water balance of a basin or watershed is an important component of water resource management, which is affected by many natural and human factors. Changes in these factors can cause an imbalance in the environment, leading to water shortages or flood hazards. Understanding

137

the effects of LULC and climate change on water resource is significant for water utilization and distribution and have important implications for ecological and socioeconomic activities (Woldesenbet et al., 2018; Ghodichore et al., 2019). Both climate and hydrological models are used to evaluate the combined effects of LULC and climate changes on water resources (e.g., Gadissa et al., 2019; Chanapathi & Thatikonda, 2020; Getahun et al., 2020).

The simulated water balance components under future land use and climate datasets (2018–2050) under RCP 4.5 and 8.5 scenarios were compared with a simulation scenario of LULC 2033 and climate dataset (1985-2017). To examine the effects of climate change on water balance components, LULC remained constant based on 2017 conditions. On the other hand, variations of LULC were considered in SWAT model to compare the individual and combined effects of LULC and climate change. As discussed above, rainfall and temperature variations have a significant impact on elements of water balance such as the increment of evapotranspiration, surface runoff, groundwater flow, lateral flow, water yield, etc.

The distinct and combined impact of LULC and climate change on water balance components of the study watershed in the future are shown in Table 7.11 and Figure 7.6. The results showed that both LULC and climate change have contributions to the modification of hydrological components in the Muga watershed, but the individual effect of climate played a greater role than LULC. For example, the mean annual ET of the study area is expected to increase in the future compared to the simulation of LULC 2017 & CC 1985-2017 due to the combined effect of LULC and climate change. The results of this study showed that the smallest and highest increase in ET is expected to be under simulation LULC 2033 and RCP 4.5 scenario (10.6%) and LULC 2033 and RCP 8.5 scenario (11.3%), respectively. The results of this study showed that when the combined effects of future LULC and climate change on ET are compared with individual effects of future LULC change scenario (LULC 2033 and CC 1985-2017), changes in temperature and rainfall will have a higher impact on future ET than LULC change.

The direction of change in the average annual ET under the combined LULC and climate change scenario is like changes found when scenarios of LULC and climate change are considered individually. This implies that the combined effect of LULC and climate change is similar to the distinct effect of LULC and climate change, as the effects of the two drivers generally indicate the same direction in ET. However, ET is slightly higher in future climate change scenarios than

138

simulations under the combined impacts of LULC and climate change scenarios. This research indicates a similar considerable increment in ET due to land use and climate change scenarios. This means that the individual and combined effects of LULC and climate change are expected to have almost similar effects on the Muga watershed ET compared to simulation under LULC 2017 and CC 1985-2017.

In this study, the individual and combined effects of LULC and climate change on surface runoff were examined. As shown in Table 7.11, the result showed that the average annual surface runoff under simulation LULC 2033 and RCP 8.5 scenario is expected to increase as compared with simulation under LULC 2017 and CC 1985-2017; it is expected to increase by 14.2%. Surface runoff is expected to increase by 10.4% and 9.8% under simulations LULC 2017 and RCP 8.5 and LULC 2033 and RCP 4.5 scenarios, respectively. Although surface runoff is expected to increase in the future, this may happen mainly due to future climate rather than future LULC. Additionally, future LULC changes, followed by less vegetation cover and settlement expansion, will increase surface runoff and water yield. Hence, the increment of surface runoff and water yield in the study watershed require special attention, as they may cause massive erosion and sedimentation in the future.

The combined effects of LULC and climate change on groundwater flow is also expected to increase by 20.8% and 22.2% under LULC 2033 and RCP 4.5 and LULC 2033 and RCP 8.5 scenarios, respectively compared to simulation under LULC 2017 & CC 1985-2018. The result of this study showed that groundwater flow is mainly impacted by climate change, whereby the range of groundwater flow under climate change scenario is greater than LULC change scenario alone. This study showed a slight effect of LULC alone on groundwater flow, whereby a higher groundwater flow is found in simulations under LULC 2017 and CC RCP 4.5, LULC 2017 and CC RCP 8.5, LULC 2033 and RCP 4.5 and LULC 2033 and CC RCP 8.5 scenarios. However, the extent of groundwater flow change under different LULC scenarios was lower than expected. The difference in the extent between the impact of LULC and climate change are consistent with the study conducted by Pan et al. (2017), Marhaento et al. (2018), and Yan et al. (2016), they found that the intensity of change in groundwater flow due to climate change would be much higher than the effects of LULC change.

139

The combined effects of both climate change and LULC change show an increase in water yields in the study area in the future as compared to simulation under LULC 2017 and CC 1985-2017, but there are some differences among the scenarios. The variations in the mean annual total water yield is pronounced for climate change, where the average annual water yield is expected to increase by about 13.8% and 17.2% under scenarios of LULC 2017 and RCP 4.5 and LULC 2017 and RCP 8.5, respectively, which is approximately 2.0% and 5.0% more than under the LULC change scenario alone. The reason can be explained by the increase in precipitation under RCP 4.5 and RCP 8.5. The abundance of water on the soil surface due to excess rainfall makes the water infiltration less likely. Furthermore, surface runoff and base flow increase with an increase in infiltration capacity. An increase in surface runoff is expected as surface water does not infiltrate completely.

In this study, the lateral flow response to the combined scenario of future land use and climate change scenario is also assessed. The simulation results show that under simulations LULC 2033 and RCP 4.5 and LULC 2033 and RCP 8.5 scenarios, the average lateral flow is expected to increase by 3.5% and 9.5%, respectively, as compared to simulation under LULC 2017 and CC 1985-2017.

The combined effect of LUCC and climate change was also evaluated by comparing the values of streamflow discharge under future climate and land-use scenarios with simulation values under LULC 2017 and CC 1985-2017. The result showed that the combination of future LULC and climate change scenario considerably affects streamflow discharge compared to LULC 2017 and historical climate datasets. This study revealed that streamflow is likely to be highest under simulation LULC 2033 and RCP 8.5 compared to simulation under LULC 2017 and CC 1985- 2017. Future rainfall increased by about 20.2% under RCP 8.5, resulting in a corresponding increase of 40.6 mm (21.0%) in streamflow discharge. Similarly, in the context of LULC 2033 and RCP 4.5 scenario, which showed a 17.2% increase in precipitation as compared to historical period, streamflow is expected to increase by 36.4 mm (18.8%). The change in streamflow is about 4.4% and 6.3% higher than the LULC scenario only. This revealed that the combined effect of future LULC and climate change outperforms the impact of the LULC scenario alone, which is accompanied by the intensification of rainfalls under RCP 4.5 and RCP 8.5, but the contribution of LULC change to streamflow change is still substantial. Therefore, the rise of rainfall in the

140

future due to climate change scenarios is expected to promote average annual streamflow discharge and water resources availability in the Muga watershed. Accordingly, increasing streamflow for future time is beneficial for the local and regional water resources infrastructure planning and operation in the study area as well as downstream areas of the Abbay River Basin of Ethiopia.

141

Figure 7.6 Predicted mean annual water balance components for different combinations of climate change and LULC scenarios Table 7.11 Predicted mean annual water balance components for different combinations of climate change and LULC scenarios

Scenarios STQ ET SURQ GW LQ WYLQ mm ∆% mm ∆% mm ∆% mm ∆% mm ∆% mm ∆% LULC 2017 & CC 1985-2017 193.4 586.3 217.4 190.3 56.8 463.3 LULC 2033 & CC 1985-2017 220.1 13.8 647.3 10.4 230.2 5.9 185.6 -2.5 55.2 -2.8 517 1.7 LULC 2017 & RCP 4.5 232.2 21.3 652.6 11.3 231.5 6.5 231 21.4 60.8 7 523.3 12.9 LULC 2017 & RCP 8.5 236 22 652.3 11.3 240 10.4 234.2 23.1 65.8 15.8 540 16.6 LULC 2033 & RCP 4.5 229.8 18.8 648.4 10.6 238.7 9.8 229.8 20.8 58.8 3.5 527.3 13.8 LULC 2033 & RCP 8.5 234 21 650.8 11 248.3 14.2 232.5 22.2 62.2 9.5 543 17.2 Note: LULC: Land use/land cover map; RCP 4.5 & RCP 8.5 (2018-2050)

142

7.4. Uncertainties and limitations

The SWAT model is an effective and useful tool to assess the impacts of LULC and climate change on hydrology, but some limitations and uncertainties exist in the model. Hydrological component of a watershed may be influenced by resolution of DEM, soil parameters, LULC and HRU delineation (Gassman et al., 2007; Glavan & Pintar, 2012). In this study, model parameters after calibration and bias correction were used to simulate future water balance components of Muga watershed. However, future LULC and climate change could change the parameters (Vaze et al., 2010). Consequently, limitations might reduce the accuracy of the model results. Despite the acknowledged limitations and uncertainties, the results of this study are relevant, reliable and valid under current scenarios of land use and climate change according to the standards provided by Moriasi et al. (2007). Therefore, although a precise comparison of LULC and climate change effects are problematic, it is possible to identify and/ or assess the individual and combined impacts of LULC and climate change on the hydrology of a watershed.

7.5. Conclusion

Land use/ land cover and climate change affect hydrological process of a watershed. Therefore, assessing the individual and combined impacts of LULC and climate change on hydrology at the watershed level has paramount importance for policymakers and land managers to develop more effective water resource management and climate adaptation strategies for a watershed. Four period LULC and two climate change scenarios were considered to assess the individual and combined impact of LULC and climate change on hydrology.

The results of this study showed that LULC change had impact on streamflow, surface runoff, groundwater flow and lateral flow, and the impact on ET and water yields was small. Future LULC change is expected to have the same impact on hydrology as the historical LULC did. Cultivated land affected GWQ, STQ, ET, and LQ negatively, but it influenced SURQ and WYLQ positively. Forest areas regulated GW, LQ, ET, SURQ and WYLQ. Grassland had a strong negative correlation with SURQ and WYLQ, its correlation with GW, ST, ET, and LQ was positive.

The impact of climate change on hydrology was also examined. Hence, by the 2050s, surface runoff, groundwater flow, streamflow, lateral flow and water yield in Muga watershed are

143

projected to increase under RCP 4.5 and RCP 8.5 scenarios compared to the reference climate datasets (1985-2017). The results of this study showed that the impact of future LULC and climate change on hydrology is relatively higher in the study watershed as compared to simulation under LULC 2017 and CC 1985-2017. The highest impact is expected to be under simulation LULC 2033 and RCP 8.5.

The combined effects of LULC and climate change on streamflow was estimated. Hence, streamflow is projected to increase to 40.6 mm (21.0%) and 36.4 mm (18.8%), under LULC 2033 and RCP 8.5 and LULC 2033 and RCP 4.5 scenarios, respectively, as compared with the historical period. The change in streamflow discharge is about 4.4% and 6.3% higher than LULC change scenario only. The combined effect of LULC and climate change on hydrological components is relatively higher than the impact of LULC change scenario alone, which is accompanied by the intensification of rainfalls under RCP 4.5 and RCP 8.5; however, the contribution of LULC change is still substantial.

Integrated use of CA Markov chain, climate, hydrological models and statistical tools have proven to provide relevant information about the impact of LULC and climate change in hydrology at a watershed level. The results of this study can contribute to the selection and implementation of appropriate climate adaption strategies and soil and water conservation techniques to the study watershed. Furthermore, it is beneficial for planning and operation of local and regional water resource infrastructure in the study watershed. However, examine the impacts of LULC and climate change on hydrology contains uncertainties, hence, the results of this study would be used to show the individual and the combined impacts of future land use and climate change on hydrology.

144

PART III: CONCLUSIONS AND RECOMMENDATIONS

145

Chapter 8 8. Conclusions and Recommendations

8.1. Conclusions

Lack of up-to-date and reliable information is a major obstacle to develop sustainable soil and water resource management strategies at a watershed level. Hence, the objective of this study was to detect the magnitude and pattern of LULC change with the drivers of changes over the last three decades (1985-2017) and investigate the impacts of LULC and climate change on soil erosion and hydrology of Muga watershed, Abbay River Basin of Ethiopia. The main findings of this study are summarized as follows:

The results of the study indicated that cultivated land and urban areas showed increasing trends while grasslands, forests, and shrub-bushlands declined over the last three decades. The expansion of cultivated land, cutting of trees for fuelwood and construction purposes, population growth, land tenure policies, and climate variability were the main drivers of LULC changes in the Muga watershed. If proper management measures are not implemented, the increasing trend of cultivated land and built up-areas at the cost of forests, shrubs and grasslands are expected to continue in the future.

The changes in land use and land cover had increased soil erosion, and highest amount of soil losses were recorded from cultivated lands. The mean annual rate of soil erosion was increased from about 15 t ha-1 year-1 in 1985 to 19 t ha-1 year-1 in 2002 and 19.7 t ha-1 year-1 in 2017. The study also revealed that the rate of soil loss is expected to continue unless LULC trends are managed, which is expected to increase to 20.7 t h-1 yr-1 in 2033. Areas with erosion rates over 15 t ha-1 yr-1 prevailed in the watershed upper and steeper areas in 2002 and 2017, while the erosion rates below 10 t ha-1 yr-1 were mainly concentrated in the gentle and downstream areas of the watershed. Similarly, the rate of soil erosion is expected to increase in 2033 in the upper parts of the watershed. The results pointed out that the highest average soil erosion rate on cultivated land was high (for example, 18 t ha-1 year-1 in 1985, 19 t ha-1 year-1 in 2002 and 21.7 t ha-1 year-1 2017).

By 2033s, the average annual rate of soil erosion in the cultivated land is expected to be 25 t ha-1 year-1 followed by grassland (mean of 13.6 t ha-1 year-1), shrub-bushland (13 t ha-1 year-1), forest (9.1 t ha-1 year-1) and urban (5.7 t ha-1 year-1). The result of this study also indicated that climate

146

change causes an increase in the average annual soil loss rate. Hence, by 2050s, the soil loss rate in Muga watershed is expected to increase to 21.6 t ha -1 year-1 and 22.2 t ha -1 year-1, under LULC 2017 and RCP 4.5 and LULC 2017 and RCP 8.5, respectively. When future LULC and climate change act together, the average annual soil loss rate is projected to increase by 13.2% and 15.7% under LULC 2033 and RCP 4.5 and LULC 2033 and RCP 8.5, respectively, which is higher than the individual effects of LULC and climate change.

The results of this study showed that LULC changes over the period 1985-2017 are recognized to have major impacts on hydrology. In Muga watersheds, cultivated land affected groundwater flow, streamflow, evapotranspiration and lateral flow negatively, but it positively influenced surface runoff and water yield. Forest areas regulated groundwater flow, lateral flow, ET, surface runoff and water yield. Grassland had a strong negative correlation with surface runoff and water yield, its correlation with groundwater flow, streamflow, ET and lateral flow was positive. Future LULC is also expected to have the same effect on hydrology similar to historical LULC. The expected increases of wet season streamflow during the rainy season can cause flooding, and a decrease in dry season flow will affect future irrigation practices.

The impacts of climate change on the hydrology of Muga watershed were also investigated in this study. The results of this study indicated that precipitation are expected to increase by 17.2% and 20.2% in the 2050s under RCP 4.5 and RCP 8.5, respectively. This leads to an increase in surface runoff, groundwater flow, streamflow, lateral flow and water yield by 2050s compared to historical climate datasets (1985-2017). The combined effects of future LULC and climate change on the hydrology of Muga watershed are expected to be relatively higher than historical period. The highest impact is expected to be under simulation LULC 2033 and RCP 8.5. under LULC 2033 and RCP 8.5 and LULC 2033 and RCP 4.5 scenarios, streamflow is projected to increase by 40.6 mm (21.0%) and 36.4 mm (18.8%), respectively, as compared with the historical period. The combined effect of LULC and climate change on hydrology is relatively higher than the impact of LULC change scenario alone, which is accompanied by the intensification of rainfalls under RCP 4.5 and RCP 8.5. However, the contribution of LULC change is still substantial.

147

In general, the results of this study will enhance to understand the individual and combined impacts of LULC and climate change on soil erosion and hydrology in Muga watershed, as well as to formulate more effective and sustainable land use planning, soil and water resource management strategies in the study watershed.

Integrated use of CA Markov chain, climate, hydrological models and statistical tools have proven to provide up-to-date information on the impact of LULC and climate change on soil erosion and hydrology at a watershed level, either individually or in combination. This would benefit stakeholders and decision makers to make better choices for planning and management of land and water resource. Since climate change has become the main driver of soil erosion and hydrologic change, watershed managers need to prioritize and pay more attention to climate change adaptation and mitigation strategies to soil and water resource degradation in Muga watershed. The approach used in this study can be applied to other watersheds with similar geographical conditions to study the effects of LULC and climate change on soil erosion and hydrology, either individually or in a combination.

8.2. Recommendations

This study provided insights into drivers and trends of LULC dynamics, and impacts of LULC and climate change on soil erosion and hydrology in the Muga watershed. Based on the findings of the study, the following recommendations are forwarded.

 Considerable attention to income diversification strategies (e.g. weaving, handicrafts, grain and livestock trade and selling local food and drinks, etc.) is required instead of solely sticking to on-farm income alone.  It is important to prevent or take measures against further expansion of croplands into forest, vegetation and very steep slopes, which prevent undesirable LULC changes and thereby reduce its negative impact on soil erosion and hydrology in the study area. This will be supported through intensification of agricultural production using improved land management practices and agricultural inputs.  Afforestation, reafforestation and implementation of appropriate soil and water conservation by considering the biophysical, land use, slope, and topography of the watershed are highly necessary to address problems of deforestation and soil erosion in the study watershed.

148

 Appropriate and integrated approaches to land use planning, land management and climate change need to be designed and implemented to prevent the expected impact of LULC and climate change on soil erosion and hydrology.  The involvement of different stakeholders (i.e., development agents, natural resource experts, local communities, government and NGOs) is required to develop and implement intervention measures to prevent deforestation and further expansions of crop cultivation to reduce the impact of LULC and climate change on soil erosion and hydrology in the study watershed.  Climate change adaptation and mitigation strategies, including crop diversification, irrigation and community-based integrated watershed management, need to be implemented to reduce the adverse effects of LULC and climate change on soil erosion and hydrology.  Even though future LULC and climate data should have the same year for comparison, the prediction for this study was based on LULC 2033 and climate data sets (2018-2050), as there may be more uncertainty for an extended year of LULC prediction. Therefore, this issue should be considered in future studies.

149

References Abbaspour, K., Vaghefi, S. A., & Srinivasan, R. (2018). A guideline for successful calibration and uncertainty analysis for soil and water assessment: A review of papers from the 2016 International SWAT Conference. Multidisciplinary Digital Publishing Institute. Abdi, H. (2010). Partial least squares regression and projection on latent structure regression (PLS Regression). Wiley Interdisciplinary Reviews: Computational Statistics, 2(1), 97–106. Abdo, K. S., Fiseha, B. M., Rientjes, T. H. M., Gieske, A. S. M., & Haile, A. T. (2009). Assessment of climate change impacts on the hydrology of Gilgel Abay catchment in Lake Tana basin, Ethiopia. Hydrological Processes: An International Journal, 23(26), 3661–3669. Abebe, G. (2017). Long-term climate data description in Ethiopia. Data in Brief, 14, 371–392. Abebe, T., & Gebremariam, B. (2019). Modeling runoff and sediment yield of Kesem dam watershed, Awash basin, Ethiopia. SN Applied Sciences, 1(5), 446. https://doi.org/10.1007/s42452-019-0347- 1 Abera, K., Crespo, O., Seid, J., & Mequanent, F. (2018). Simulating the impact of climate change on maize production in Ethiopia, East Africa. Environmental Systems Research, 7(1), 4. https://doi.org/10.1186/s40068-018-0107-z Aboelnour, M., Gitau, M. W., & Engel, B. A. (2019). Hydrologic response in an urban watershed as affected by climate and land-use change. Water, 11(8), 1603. Addis, H. K., & Klik, A. (2015). Predicting the spatial distribution of soil erodibility factor using USLE nomograph in an agricultural watershed, Ethiopia. International Soil and Water Conservation Research, 3(4), 282–290. Adhikari, U., Nejadhashemi, A. P., & Woznicki, S. A. (2015). Climate change and eastern Africa: A review of impact on major crops. Food and Energy Security, 4(2), 110–132. Adimassu, Z., & Kessler, A. (2015). Impact of the productive safety net program on farmers’ investments in sustainable land management in the Central Rift Valley of Ethiopia. Environmental Development, 16, 54–62. Admassu, H., Getinet, M., & Thomas, T. S. (2012). East African Agriculture and Climate Change: A Comprehensive Analysis-Ethiopia. Adugna, A., & Abegaz, A. (2016). Effects of land use changes on the dynamics of selected soil properties in northeast Wellega, Ethiopia. Soil, 2(1), 63–70. Adugna, A., Abegaz, A., & Cerdà, A. (2015). Soil erosion assessment and control in Northeast Wollega, Ethiopia. Solid Earth Discussions, 7(4), 3511–3540. https://doi.org/10.5194/sed-7-3511-2015 Afzal, M., & Ragab, R. (2020). Impact of the Future Climate and Land Use Changes on the Hydrology and Water Resources in South East England, UK. American Journal of Water Resources, 20(8), 218– 231. https://doi.org/10.12691/ajwr-8-5-2 Alemayehu, A., & Bewket, W. (2017). Local spatiotemporal variability and trends in rainfall and temperature in the central highlands of Ethiopia. Geografiska Annaler: Series A, Physical Geography, 99(2), 85–101. Alemseged, T. H., & Tom, R. (2015). Evaluation of regional climate model simulations of rainfall over the Upper Blue Nile basin. Atmospheric Research, 161, 57–64. Alkharabsheh, M. M., Alexandridis, T. K., Bilas, G., Misopolinos, N., & Silleos, N. (2013). Impact of land cover change on soil erosion hazard in northern Jordan using remote sensing and GIS. Procedia Environmental Sciences, 19, 912–921. 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. Fao, Rome, 300(9), D05109. Al-sharif, A. A., & Pradhan, B. (2014). Monitoring and predicting land use change in Tripoli Metropolitan City using an integrated Markov chain and cellular automata models in GIS. Arabian Journal of Geosciences, 7(10), 4291–4301.

150

Ampaire, E. L., Acosta, M., Huyer, S., Kigonya, R., Muchunguzi, P., Muna, R., & Jassogne, L. (2020). Gender in climate change, agriculture, and natural resource policies: Insights from East Africa. Climatic Change, 158(1), 43–60. Amundson, R., Berhe, A. A., Hopmans, J. W., Olson, C., Sztein, A. E., & Sparks, D. L. (2015). Soil and human security in the 21st century. Science, 348(6235), 1261071–1261075. https://doi.org/10.1126/science.1261071 Anache, J. A., Flanagan, D. C., Srivastava, A., & Wendland, E. C. (2018). Land use and climate change impacts on runoff and soil erosion at the hillslope scale in the Brazilian Cerrado. Science of the Total Environment, 622, 140–151. Andersson, C. A., & Bro, R. (2000). The N-way toolbox for MATLAB. Chemometrics and Intelligent Laboratory Systems, 52(1), 1–4. Andualem, T. G., & Gebremariam, B. (2015). Impact of land use land cover change on stream flow and sediment yield: A case study of Gilgel Abay watershed, Lake Tana sub-basin, Ethiopia. Int. J. Technol. Enhanc. Merg. Eng. Res, 3, 28–42. Aneseyee, A. B., Elias, E., Soromessa, T., & Feyisa, G. L. (2020). Land use/land cover change effect on soil erosion and sediment delivery in the Winike watershed, Omo Gibe Basin, Ethiopia. Science of The Total Environment, 728, 138776. https://doi.org/10.1016/j.scitotenv.2020.138776 Angima, S. D., Stott, D. E., O’neill, M. K., Ong, C. K., & Weesies, G. A. (2003). Soil erosion prediction using RUSLE for central Kenyan highland conditions. Agriculture, Ecosystems & Environment, 97(1–3), 295–308. Araya, Y. H., & Cabral, P. (2010). Analysis and modeling of urban land cover change in Setúbal and Sesimbra, Portugal. Remote Sensing, 2(6), 1549–1563. Ariti, A. T., van Vliet, J., & Verburg, P. H. (2015). Land-use and land-cover changes in the Central Rift Valley of Ethiopia: Assessment of perception and adaptation of stakeholders. Applied Geography, 65, 28–37. Arnold, J. G., Kiniry, J. R., Srinivasan, R., Williams, J. R., Haney, E. B., & Neitsch, S. L. (2012). Soil and water assessment tool input/output file documentation: Version 2012. Texas Water Resources Institute TR 436. Arsanjani, J. J., Helbich, M., & de Noronha Vaz, E. (2013). Spatiotemporal simulation of urban growth patterns using agent-based modeling: The case of Tehran. Cities, 32, 33–42. Arsanjani, J. J., Helbich, M., Kainz, W., & Boloorani, A. D. (2013). Integration of logistic regression, Markov chain and cellular automata models to simulate urban expansion. International Journal of Applied Earth Observation and Geoinformation, 21, 265–275. Ayele, G. T., Demessie, S. S., Mengistu, K. T., Tilahun, S. A., & Melesse, A. M. (2016). Multitemporal land use/land cover change detection for the Batena Watershed, Rift Valley Lakes Basin, Ethiopia. In Landscape Dynamics, Soils and Hydrological Processes in Varied Climates (pp. 51–72). Springer. Ayele, H. S., Li, M.-H., & Tung, C.-P. (2016). Assessing Climate Change Impact on Gilgel Abbay and Gumara Watershed Hydrology, the Upper Blue Nile Basin, Ethiopia. Terrestrial, Atmospheric & Oceanic Sciences, 27(6). Ayivi, F., & Jha, M. K. (2018). Estimation of water balance and water yield in the Reedy Fork-Buffalo Creek Watershed in North Carolina using SWAT. International Soil and Water Conservation Research, 6(3), 203–213. https://doi.org/10.1016/j.iswcr.2018.03.007 Aylward, B. (2005). Land use, hydrological function and economic valuation. Forests, Water and People in the Humid Tropics: Past, Present and Future Hydrological Research for Integrated Land and Water Management, 99–120. Bantider, A., Hurni, H., & Zeleke, G. (2011). Responses of rural households to the impacts of population and land-use changes along the Eastern Escarpment of Wello, Ethiopia. Norsk Geografisk Tidsskrift-Norwegian Journal of Geography, 65(1), 42–53. BCEOM. (1998). Abbay river basin integrated development master plan project, phase 2, data collection and site-investigation survey and analysis, section II, sectoral studies: Vol. XIV.

151

Behera, M., Borate, S. N., Panda, S. N., Behera, P. R., & Roy, P. S. (2012). Modelling and analyzing the watershed dynamics using Cellular Automata (CA)–Markov model–A geo-information based approach. Journal of Earth System Science, 121(4), 1011–1024. Bekele, D., Alamirew, T., Kebede, A., Zeleke, G., & Melesse, A. M. (2019). Modeling climate change impact on the Hydrology of Keleta watershed in the Awash River basin, Ethiopia. Environmental Modeling & Assessment, 24(1), 95–107. Belay, T., & Mengistu, D. (2019). Land use and land cover dynamics and drivers in the Muga watershed, Upper Blue Nile basin, Ethiopia. Remote Sensing Applications: Society and Environment, 15, 100249. Belihu, M., Tekleab, S., Abate, B., & Bewket, W. (2020). Hydrologic response to land use land cover change in the upper Gidabo watershed, Rift Valley Lakes Basin, Ethiopia. HydroResearch. Benegas, L., Ilstedt, U., Roupsard, O., Jones, J., & Malmer, A. (2014). Effects of trees on infiltrability and preferential flow in two contrasting agroecosystems in Central America. Agriculture, Ecosystems & Environment, 183, 185–196. https://doi.org/10.1016/j.agee.2013.10.027 Berberoglu, S., Cilek, A., Kirkby, M., Irvine, B., & Donmez, C. (2020). Spatial and temporal evaluation of soil erosion in Turkey under climate change scenarios using the Pan-European Soil Erosion Risk Assessment (PESERA) model. Environmental Monitoring and Assessment, 192(8), 1–22. Berhane, A., Hadgu, G., Worku, W., & Abrha, B. (2020). Trends in extreme temperature and rainfall indices in the semi-arid areas of Western Tigray, Ethiopia. Environmental Systems Research, 9(1), 1–20. Berihun, M. L., Tsunekawa, A., Haregeweyn, N., Meshesha, D. T., Adgo, E., Tsubo, M., Masunaga, T., Fenta, A. A., Sultan, D., Yibeltal, M., & Ebabu, K. (2019). Hydrological responses to land use/land cover change and climate variability in contrasting agro-ecological environments of the Upper Blue Nile basin, Ethiopia. Science of The Total Environment, 689, 347–365. https://doi.org/10.1016/j.scitotenv.2019.06.338 Betrie, G. D., Mohamed, Y. A., Griensven, A. van, & Srinivasan, R. (2011). Sediment management modelling in the Blue Nile Basin using SWAT model. Hydrology and Earth System Sciences, 15(3), 807–818. https://doi.org/10.5194/hess-15-807-2011 Betru, T., Tolera, M., Sahle, K., & Kassa, H. (2019). Trends and drivers of land use/land cover change in Western Ethiopia. Applied Geography, 104, 83–93. Bewket, W. (2002). Land cover dynamics since the 1950s in Chemoga watershed, Blue Nile basin, Ethiopia. Mountain Research and Development, 22(3), 263–269. Bewket, W. (2003). Towards integrated watershed management in highland Ethiopia: The Chemoga watershed case study. Bewket, W., & Sterk, G. (2005). Dynamics in land cover and its effect on stream flow in the Chemoga watershed, Blue Nile basin, Ethiopia. Hydrological Processes: An International Journal, 19(2), 445–458. Bewket, W., & Teferi, E. (2009). Assessment of soil erosion hazard and prioritization for treatment at the watershed level: Case study in the Chemoga watershed, Blue Nile basin, Ethiopia. Land Degradation & Development, 20(6), 609–622. Billi, P. (2015). Landscapes and landforms of Ethiopia. Springer. Birhane, E., Ashfare, H., Fenta, A. A., Hishe, H., Gebremedhin, M. A., & Solomon, N. (2019). Land use land cover changes along topographic gradients in Hugumburda national forest priority area, Northern Ethiopia. Remote Sensing Applications: Society and Environment, 13, 61–68. Birhanu, A., Masih, I., van der Zaag, P., Nyssen, J., & Cai, X. (2019). Impacts of land use and land cover changes on hydrology of the Gumara catchment, Ethiopia. Physics and Chemistry of the Earth, Parts A/B/C, 112, 165–174. Blanco-Canqui, H., & Lal, R. (2008). Principles of soil conservation and management. Springer Science & Business Media. Boerner, R. E., DeMers, M. N., Simpson, J. W., Artigas, F. J., Silva, A., & Berns, L. A. (1996). Markov models of inertia and dynamism on two contiguous Ohio landscapes. Geographical Analysis, 28(1), 56–66.

152

Bogale, A. (2020). Review, impact of land use/cover change on soil erosion in the Lake Tana Basin, Upper Blue Nile, Ethiopia. Applied Water Science, 10(12), 235. https://doi.org/10.1007/s13201-020- 01325-w Bolch, T., Kulkarni, A., Kääb, A., Huggel, C., Paul, F., Cogley, J. G., Frey, H., Kargel, J. S., Fujita, K., & Scheel, M. (2012). The state and fate of Himalayan glaciers. Science, 336(6079), 310–314. Borrelli, P., Robinson, D. A., Fleischer, L. R., Lugato, E., Ballabio, C., Alewell, C., Meusburger, K., Modugno, S., Schütt, B., & Ferro, V. (2017). An assessment of the global impact of 21st century land use change on soil erosion. Nature Communications, 8(1), 1–13. Borrelli, P., Robinson, D. A., Panagos, P., Lugato, E., Yang, J. E., Alewell, C., Wuepper, D., Montanarella, L., & Ballabio, C. (2020). Land use and climate change impacts on global soil erosion by water (2015-2070). Proceedings of the National Academy of Sciences, 117(36), 21994–22001. https://doi.org/10.1073/pnas.2001403117 Bosch, J. M., & Hewlett, J. D. (1982). A review of catchment experiments to determine the effect of vegetation changes on water yield and evapotranspiration. Journal of Hydrology, 55(1–4), 3–23. Boserup, E. (1988). Environment, Population and Technology. The Ends of the Earth-Perspectives on Modern Environmental History, Cambridge University Press, Cambridge, 23–38. Brown, A. E., Zhang, L., McMahon, T. A., Western, A. W., & Vertessy, R. A. (2005). A review of paired catchment studies for determining changes in water yield resulting from alterations in vegetation. Journal of Hydrology, 310(1–4), 28–61. Bryman, M. A. H. A. (2004). Handbook of data analysis. Callaway, J. M. (2004). Adaptation benefits and costs: Are they important in the global policy picture and how can we estimate them? Global Environ. Change - Human Policy Dimens, 14, 273–282. https://doi.org/10.1016/j.gloenvcha.2004.04.002 Carty, T. (2017). A climate in crisis: How climate change is making drought and humanitarian disaster worse in East Africa. Carver, R. H., & Nash, J. G. (2011). Doing data analysis with SPSS: Version 18.0. Cengage Learning. Case, M. (2006). Climate change impacts on East Africa: A review of scientific literature. Cebecauer, T., & Hofierka, J. (2008). The consequences of land-cover changes on soil erosion distribution in Slovakia. Geomorphology, 98(3–4), 187–198. Chanapathi, T., & Thatikonda, S. (2020). Investigating the impact of climate and land-use land cover changes on hydrological predictions over the Krishna river basin under present and future scenarios. Science of The Total Environment, 721, 137736. https://doi.org/10.1016/j.scitotenv.2020.137736 Chemura, A., Rwasoka, D., Mutanga, O., Dube, T., & Mushore, T. (2020). The impact of land-use/land cover changes on water balance of the heterogeneous Buzi sub-catchment, Zimbabwe. Remote Sensing Applications: Society and Environment, 18, 100292. Chen, J., Brissette, F. P., & Lucas-Picher, P. (2016). Transferability of optimally-selected climate models in the quantification of climate change impacts on hydrology. Climate Dynamics, 47(9–10), 3359– 3372. Chimdessa, K., Quraishi, S., Kebede, A., & Alamirew, T. (2019). Effect of Land Use Land Cover and Climate Change on River Flow and Soil Loss in Didessa River Basin, South West Blue Nile, Ethiopia. Hydrology, 6(1), 2. Christensen, J. H., Boberg, F., Christensen, O. B., & Lucas-Picher, P. (2008). On the need for bias correction of regional climate change projections of temperature and precipitation. Geophysical Research Letters, 35(20). Ciampalini, R., Constantine, J. A., Walker-Springett, K. J., Hales, T. C., Ormerod, S. J., & Hall, I. R. (2020). Modelling soil erosion responses to climate change in three catchments of Great Britain. Science of The Total Environment, 749, 141657. https://doi.org/10.1016/j.scitotenv.2020.141657 Congalton, R. G., & Green, K. (2019). Assessing the accuracy of remotely sensed data: Principles and practices. CRC press.

153

Conway, D., & Schipper, E. L. F. (2011). Adaptation to climate change in Africa: Challenges and opportunities identified from Ethiopia. Global Environmental Change, 21(1), 227–237. https://doi.org/10.1016/j.gloenvcha.2010.07.013 Cousino, L. K., Becker, R. H., & Zmijewski, K. A. (2015). Modeling the effects of climate change on water, sediment, and nutrient yields from the Maumee River watershed. Journal of Hydrology: Regional Studies, 4, 762–775. Cox, I., & Gaudard, M. (2013). Discovering partial least squares with JMP. SAS Institute. Cramer, W., Guiot, J., Fader, M., Garrabou, J., Gattuso, J.-P., Iglesias, A., Lange, M. A., Lionello, P., Llasat, M. C., Paz, S., Peñuelas, J., Snoussi, M., Toreti, A., Tsimplis, M. N., & Xoplaki, E. (2018). Climate change and interconnected risks to sustainable development in the Mediterranean. Nature Climate Change, 8(11), 972–980. https://doi.org/10.1038/s41558-018-0299-2 Creswell, J. W. (2014). A concise introduction to mixed methods research. SAGE publications. Creswell, J. W., & Clark, V. L. P. (2017). Designing and conducting mixed methods research. Sage publications. Crewett, W., Bogale, A., & Korf, B. (2008). Land tenure in Ethiopia: Continuity and change, shifting rulers, and the quest for state control. Crooks, A. T., & Heppenstall, A. J. (2012). Introduction to Agent-Based Modelling. In A. J. Heppenstall, A. T. Crooks, L. M. See, & M. Batty (Eds.), Agent-Based Models of Geographical Systems (pp. 85–105). Springer Netherlands. https://doi.org/10.1007/978-90-481-8927-4_5 CSA. (1999). The 1994 population and housing census of Ethiopia. Central Statistical Authority Addis Ababa. CSA. (2007). The 2007 population and housing census of Ethiopia. Central Statistical Authority Addis Ababa. CSA. (2013). Population projection of Ethiopia for all regions at wereda level from 2014–2017. CSA Addis Ababa. de Hipt, F. O., Diekkrüger, B., Steup, G., Yira, Y., Hoffmann, T., Rode, M., & Näschen, K. (2019). Modeling the effect of land use and climate change on water resources and soil erosion in a tropical West African catch-ment (Dano, Burkina Faso) using SHETRAN. Science of The Total Environment, 653, 431–445. de Oliveira Barros, K., Ribeiro, C. A. A. S., Marcatti, G. E., Lorenzon, A. S., de Castro, N. L. M., Domingues, G. F., de Carvalho, J. R., & dos Santos, A. R. (2018). Markov chains and cellular automata to predict environments subject to desertification. Journal of Environmental Management, 225, 160–167. DeFries, R., & Eshleman, K. N. (2004). Land-use change and hydrologic processes: A major focus for the future. Hydrological Processes, 18(11), 2183–2186. Degefu, M. A., & Bewket, W. (2014). Variability and trends in rainfall amount and extreme event indices in the Omo-Ghibe River Basin, Ethiopia. Regional Environmental Change, 14(2), 799–810. Demessie, E. T. (2015). Soil hydrological impacts and climatic controls of land use and land cover changes in the Upper Blue Nile (Abay) basin. CRC Press. Demissie, F., Yeshitila, K., Kindu, M., & Schneider, T. (2017). Land use/Land cover changes and their causes in Libokemkem District of South Gonder, Ethiopia. Remote Sensing Applications: Society and Environment, 8, 224–230. Demissie, T., Storm, D. E., & White, M. (2004). A GIS-based watershed modeling and water resources use optimization under dynamic changing land use/land cover conditions. 2004 ASAE Annual Meeting, 1. Deng, Z., Zhang, X., Li, D., & Pan, G. (2015). Simulation of land use/land cover change and its effects on the hydrological characteristics of the upper reaches of the Hanjiang Basin. Environmental Earth Sciences, 73(3), 1119–1132. Desmet, P. J. J., & Govers, G. (1996). A GIS procedure for automatically calculating the USLE LS factor on topographically complex landscape units. Journal of Soil and Water Conservation, 51(5), 427– 433.

154

Dessie, G., & Kleman, J. (2007). Pattern and magnitude of deforestation in the South Central Rift Valley Region of Ethiopia. Mountain Research and Development, 27(2), 162–168. Dibaba, W. T., Demissie, T. A., & Miegel, K. (2020). Watershed Hydrological Response to Combined Land Use/Land Cover and Climate Change in Highland Ethiopia: Finchaa Catchment. Water, 12(6), 1801. Dile, Y. T., Berndtsson, R., & Setegn, S. (2013). Hydrological response to climate change for gilgel abay river, in the lake tana basin-upper blue Nile basin of Ethiopia. PloS One, 8(10), e79296. Dile, Y. T., Daggupati, P., George, C., Srinivasan, R., & Arnold, J. (2016). Introducing a new open source GIS user interface for the SWAT model. Environmental Modelling & Software, 85, 129–138. Duarte, L., Teodoro, A. C., Gonçalves, J. A., Soares, D., & Cunha, M. (2016). Assessing soil erosion risk using RUSLE through a GIS open source desktop and web application. Environmental Monitoring and Assessment, 188(6), 351. Eastman, J. R. (2012). IDRISI Selva manual, version 17. Clark University, Worcester, 322. Ebabu, K., Tsunekawa, A., Haregeweyn, N., Adgo, E., Meshesha, D. T., Aklog, D., Masunaga, T., Tsubo, M., Sultan, D., & Fenta, A. A. (2019). Effects of land use and sustainable land management practices on runoff and soil loss in the Upper Blue Nile basin, Ethiopia. Science of the Total Environment, 648, 1462–1475. Edenhofer, O. (2015). Climate change 2014: Mitigation of climate change (Vol. 3). Cambridge University Press. Eekhout, J. P., & De Vente, J. (2020). How soil erosion model conceptualization affects soil loss projections under climate change. Progress in Physical Geography: Earth and Environment, 44(2), 212–232. https://doi.org/10.1177/0309133319871937 Elbehri, A. (2015). Climate change and food systems: Global assessments and implications for food security and trade. Food and Agriculture Organization of the United Nations (FAO). Enyew, B. D., Van Lanen, H. A. J., & Van Loon, A. F. (2014). Assessment of the impact of climate change on hydrological drought in Lake Tana catchment, Blue Nile basin, Ethiopia. J. Geol. Geosci, 3, 174. Eshleman, K. N. (2004). Hydrological consequences of land use change: A review of the state-of-science. GMS, 153, 13–29. Eva, H. D., Brink, A., & Simonetti, D. (2006). Monitoring land cover dynamics in sub-Saharan Africa. Institute for Environmental and Sustainability, Tech. Rep. EUR, 22498. Fan, M., & Shibata, H. (2015). Simulation of watershed hydrology and stream water quality under land use and climate change scenarios in Teshio River watershed, northern Japan. Ecological Indicators, 50, 79–89. Fang, G. H., Yang, J., Chen, Y. N., & Zammit, C. (2014). Comparing bias correction methods in downscaling meteorological variables for hydrologic impact study in an arid area in China. Hydrology and Earth System Sciences Discussions, 11(11), 12659–12696. Fang, G., Yuan, T., Zhang, Y., Wen, X., & Lin, R. (2019). Integrated study on soil erosion using RUSLE and GIS in Yangtze River Basin of Jiangsu Province (China). Arabian Journal of Geosciences, 12(5), 173. FAO. (2010). Global forest resources assessment 2010: Main report. Food and Agriculture Organization of the United Nations. FDRE. (2011). Ethiopia’s Climate-Resilient Green Economy: Green Economy Strategy. Federal Democratic Republic of Ethiopia Addis Ababa, Ethiopia. Fenta, A. A., Tsunekawa, A., Haregeweyn, N., Tsubo, M., Yasuda, H., Kawai, T., Ebabu, K., Berihun, M. L., Belay, A. S., & Sultan, D. (2021). Agroecology-based soil erosion assessment for better conservation planning in Ethiopian river basins. Environmental Research, 110786. Fenta, A. A., Yasuda, H., Shimizu, K., & Haregeweyn, N. (2017). Response of streamflow to climate variability and changes in human activities in the semiarid highlands of northern Ethiopia. Regional Environmental Change, 17(4), 1229–1240.

155

Fenta, A. A., Yasuda, H., Shimizu, K., Haregeweyn, N., & Negussie, A. (2016). Dynamics of soil erosion as influenced by watershed management practices: A case study of the Agula watershed in the semi-arid highlands of northern Ethiopia. Environmental Management, 58(5), 889–905. Fentaw, F., Hailu, D., Nigussie, A., & Melesse, A. M. (2018). Climate Change Impact on the Hydrology of Tekeze Basin, Ethiopia: Projection of Rainfall-Runoff for Future Water Resources Planning. Water Conservation Science and Engineering, 3(4), 267–278. https://doi.org/10.1007/s41101-018-0057- 3 Ferreira, V., Panagopoulos, T., Cakula, A., Andrade, R., & Arvela, A. (2015). Predicting soil erosion after land use changes for irrigating agriculture in a large reservoir of southern Portugal. Agriculture and Agricultural Science Procedia, 4, 40–49. Fick, S. E., & Hijmans, R. J. (2017). WorldClim 2: New 1-km spatial resolution climate surfaces for global land areas. International Journal of Climatology, 37(12), 4302–4315. Ficklin, D. L., Luo, Y., Luedeling, E., & Zhang, M. (2009). Climate change sensitivity assessment of a highly agricultural watershed using SWAT. Journal of Hydrology, 374(1–2), 16–29. Field, C. B., & Barros, V. R. (2014). Climate change 2014–Impacts, adaptation and vulnerability: Regional aspects. Cambridge University Press. Fitawok, M. B., Derudder, B., Minale, A. S., Van Passel, S., Adgo, E., & Nyssen, J. (2020). Modeling the Impact of Urbanization on Land-Use Change in Bahir Dar City, Ethiopia: An Integrated Cellular Automata–Markov Chain Approach. Land, 9(4), 115. Frankl, A., Zwertvaegher, A., Poesen, J., & Nyssen, J. (2013). Transferring Google Earth observations to GIS-software: Example from gully erosion study. International Journal of Digital Earth, 6(2), 196–201. Gabiri, G., Diekkrüger, B., Näschen, K., Leemhuis, C., van der Linden, R., Majaliwa, J.-G. M., & Obando, J. A. (2020). Impact of Climate and Land Use/Land Cover Change on the Water Resources of a Tropical Inland Valley Catchment in Uganda, East Africa. Climate, 8(7), 83. https://doi.org/10.3390/cli8070083 Gadissa, T., Nyadawa, M., Behulu, F., & Mutua, B. (2019). Assessment of catchment water resources availability under projected climate change scenarios and increased demand in Central Rift Valley Basin. In Extreme Hydrology and Climate Variability (pp. 151–163). Elsevier. Ganasri, B. P., & Ramesh, H. (2016). Assessment of soil erosion by RUSLE model using remote sensing and GIS-A case study of Nethravathi Basin. Geoscience Frontiers, 7(6), 953–961. Garedew, E., Sandewall, M., Söderberg, U., & Campbell, B. M. (2009). Land-use and land-cover dynamics in the central rift valley of Ethiopia. Environmental Management, 44(4), 683–694. Gashaw, T., Tulu, T., Argaw, M., & Worqlul, A. W. (2017). Evaluation and prediction of land use/land cover changes in the Andassa watershed, Blue Nile Basin, Ethiopia. Environmental Systems Research, 6(1), 17. Gashaw, T., Tulu, T., Argaw, M., & Worqlul, A. W. (2018). Modeling the hydrological impacts of land use/land cover changes in the Andassa watershed, Blue Nile Basin, Ethiopia. Science of the Total Environment, 619, 1394–1408. Gashaw, T., Tulu, T., Argaw, M., Worqlul, A. W., Tolessa, T., & Kindu, M. (2018). Estimating the impacts of land use/land cover changes on Ecosystem Service Values: The case of the Andassa watershed in the Upper Blue Nile basin of Ethiopia. Ecosystem Services, 31, 219–228. Gassman, P. W., Reyes, M. R., Green, C. H., & Arnold, J. G. (2007). The soil and water assessment tool: Historical development, applications, and future research directions. Transactions of the ASABE, 50(4), 1211–1250. Gebre, S. L., & Ludwig, F. (2015). Hydrological response to climate change of the upper blue Nile River Basin: Based on IPCC fifth assessment report (AR5). Journal of Climatology & Weather Forecasting, 3(01), 1–15. Gebrechorkos, S. H., Bernhofer, C., & Hülsmann, S. (2019). Impacts of projected change in climate on water balance in basins of East Africa. Science of The Total Environment, 682, 160–170. https://doi.org/10.1016/j.scitotenv.2019.05.053

156

Gebrechorkos, S. H., Hülsmann, S., & Bernhofer, C. (2019). Regional climate projections for impact assessment studies in East Africa. Environmental Research Letters, 14(4), 044031. Gebrehiwot, S. G., Taye, A., & Bishop, K. (2010). Forest cover and stream flow in a headwater of the Blue Nile: Complementing observational data analysis with community perception. Ambio, 39(4), 284– 294. Gebrelibanos, T., & Assen, M. (2015). Land use/land cover dynamics and their driving forces in the Hirmi watershed and its adjacent agroecosystem, highlands of Northern Ethiopia. J Land Use Sci 10 (1): 81–94. Gebremeskel, G., Gebremicael, T. G., & Girmay, A. (2018). Economic and environmental rehabilitation through soil and water conservation, the case of Tigray in northern Ethiopia. Journal of Arid Environments, 151, 113–124. https://doi.org/10.1016/j.jaridenv.2017.12.002 Gebremicael, T. G., Mohamed, Y. A., Betrie, G. D., Van der Zaag, P., & Teferi, E. (2013). Trend analysis of runoff and sediment fluxes in the Upper Blue Nile basin: A combined analysis of statistical tests, physically-based models and landuse maps. Journal of Hydrology, 482, 57–68. Gebremicael, T. G., Mohamed, Y. A., & Van der Zaag, P. (2019). Attributing the hydrological impact of different land use types and their long-term dynamics through combining parsimonious hydrological modelling, alteration analysis and PLSR analysis. Science of the Total Environment, 660, 1155–1167. Gebreselassie, S., Kirui, O. K., & Mirzabaev, A. (2016). Economics of land degradation and improvement in Ethiopia. In Economics of land degradation and improvement–a global assessment for sustainable development (pp. 401–430). Springer, Cham. Gelagay, H. S., & Minale, A. S. (2016). Soil loss estimation using GIS and Remote sensing techniques: A case of Koga watershed, Northwestern Ethiopia. International Soil and Water Conservation Research, 4(2), 126–136. Gelete, G., Gokcekus, H., & Gichamo, T. (2019). Impact of climate change on the hydrology of Blue Nile basin, Ethiopia: A review. Journal of Water and Climate Change. Geremew, A. A. (2013). Assessing the impacts of land use and land cover change on hydrology of watershed: A case study on Gigel-Abbay Watershed, Lake Tana Basin, Ethiopia [PhD Thesis]. Gessesse, B., & Bewket, W. (2014). Drivers and implications of land use and land cover change in the central highlands of Ethiopia: Evidence from remote sensing and socio-demographic data integration. Ethiopian Journal of the Social Sciences and Humanities, 10(2), 1–23. Gessesse, B., Bewket, W., & Bräuning, A. (2015). Model-based characterization and monitoring of runoff and soil erosion in response to land use/land cover changes in the Modjo watershed, Ethiopia. Land Degradation & Development, 26(7), 711–724. Gessesse, D., Gizaw, A. K., & Hurni, H. (2009). Assessment of soil erosion and soil conservation practices in Angereb watershed, Ethiopia: Technological and land user context. Biophysical and Socio- Economic Frame Conditions for the Sustainable Management of Natural Resources: Book of Abstracts, 52. Getachew, B., Manjunatha, B. R., & Bhat, H. G. (2021). Modeling projected impacts of climate and land use/land cover changes on hydrological responses in the Lake Tana Basin, upper Blue Nile River Basin, Ethiopia. Journal of Hydrology, 595, 125974. Getachew, H. E., & Melesse, A. M. (2012). The impact of land use change on the hydrology of the Angereb Watershed, Ethiopia. International Journal of Water Sciences, 1(4). Getahun, Y. S., Li, M.-H., & Chen, P.-Y. (2020). Assessing Impact of Climate Change on Hydrology of Melka Kuntrie Subbasin, Ethiopia with Ar4 and Ar5 Projections. Water, 12(5), 1308. Ghodichore, N., Dhanya, C. T., & Hendricks-Franssen, H.-J. (2019). Assessing the individual and combined impacts of land use land cover changes (LULC) and climate change on hydrological processes over India. Geophysical Research Abstracts, 21. Giannini, A., Biasutti, M., Held, I. M., & Sobel, A. H. (2008). A global perspective on African climate. Climatic Change, 90(4), 359–383. https://doi.org/10.1007/s10584-008-9396-y

157

Gidey, E., Dikinya, O., Sebego, R., Segosebe, E., & Zenebe, A. (2017). Cellular automata and Markov Chain (CA_Markov) model-based predictions of future land use and land cover scenarios (2015– 2033) in Raya, northern Ethiopia. Modeling Earth Systems and Environment, 3(4), 1245–1262. Gilligan, D. O., Hoddinott, J., & Taffesse, A. S. (2009). The impact of Ethiopia’s Productive Safety Net Programme and its linkages. The Journal of Development Studies, 45(10), 1684–1706. Giri, S., Nejadhashemi, A. P., Zhang, Z., & Woznicki, S. A. (2015). Integrating statistical and hydrological models to identify implementation sites for agricultural conservation practices. Environmental Modelling & Software, 72, 327–340. Girmay, G., Moges, A., & Muluneh, A. (2021). Assessment of Current and Future Climate Change Impact on Soil Loss Rate of Agewmariam Watershed, Northern Ethiopia. Air, Soil and Water Research, 14, 1178622121995847. Gizaw, M. S., Biftu, G. F., Gan, T. Y., Moges, S. A., & Koivusalo, H. (2017). Potential impact of climate change on streamflow of major Ethiopian rivers. Climatic Change, 143(3), 371–383. https://doi.org/10.1007/s10584-017-2021-1 Glavan, M., & Pintar, M. (2012). Strengths, weaknesses, opportunities and threats of catchment modelling with Soil and Water Assessment Tool (SWAT) model. Water Resources Management and Modeling, 39–64. Godoy, J. L., Vega, J. R., & Marchetti, J. L. (2014). Relationships between PCA and PLS-regression. Chemometrics and Intelligent Laboratory Systems, 130, 182–191. Goldewijk, K. K., & Ramankutty, N. (2009). Land use changes during the past 300 years. Land-Use, Land Cover and Soil Sciences, 1, 147–168. Green, W. H., & Ampt, G. A. (1911). The flow of air and water through soils. J. Agric. Sci, 4, 1–24. Guerra, C. A., Rosa, I. M. D., Valentini, E., Wolf, F., Filipponi, F., Karger, D. N., Nguyen Xuan, A., Mathieu, J., Lavelle, P., & Eisenhauer, N. (2020). Global vulnerability of soil ecosystems to erosion. Landscape Ecology, 35(4), 823–842. https://doi.org/10.1007/s10980-020-00984-z Guo, J., Su, X., Singh, V. P., & Jin, J. (2016). Impacts of climate and land use/cover change on streamflow using SWAT and a separation method for the Xiying River Basin in northwestern China. Water, 8(5), 192. Haile, G. G., Tang, Q., Sun, S., Huang, Z., Zhang, X., & Liu, X. (2019). Droughts in East Africa: Causes, impacts and resilience. Earth-Science Reviews, 193, 146–161. Hailemariam, S. N., Soromessa, T., & Teketay, D. (2016). Land use and land cover change in the bale mountain eco-region of Ethiopia during 1985 to 2015. Land, 5(4), 41. Halmy, M. W. A., Gessler, P. E., Hicke, J. A., & Salem, B. B. (2015). Land use/land cover change detection and prediction in the north-western coastal desert of Egypt using Markov-CA. Applied Geography, 63, 101–112. Han, H., Yang, C., & Song, J. (2015). Scenario simulation and the prediction of land use and land cover change in Beijing, China. Sustainability, 7(4), 4260–4279. Haregeweyn, N., Berhe, A., Tsunekawa, A., Tsubo, M., & Meshesha, D. T. (2012). Integrated watershed management as an effective approach to curb land degradation: A case study of the Enabered watershed in northern Ethiopia. Environmental Management, 50(6), 1219–1233. Haregeweyn, N., Tesfaye, S., Tsunekawa, A., Tsubo, M., Meshesha, D. T., Adgo, E., & Elias, A. (2015). Dynamics of land use and land cover and its effects on hydrologic responses: Case study of the Gilgel Tekeze catchment in the highlands of Northern Ethiopia. Environmental Monitoring and Assessment, 187(1), 4090. Haregeweyn, N., Tsunekawa, A., Nyssen, J., Poesen, J., Tsubo, M., Tsegaye Meshesha, D., Schütt, B., Adgo, E., & Tegegne, F. (2015). Soil erosion and conservation in Ethiopia: A review. Progress in Physical Geography, 39(6), 750–774. Haregeweyn, N., Tsunekawa, A., Poesen, J., Tsubo, M., Meshesha, D. T., Fenta, A., Nyssen, J., & Adgo, E. (2017). Comprehensive assessment of soil erosion risk for better land use planning in river basins: Case study of the Upper Blue Nile River. Science of the Total Environment, 574, 95–108.

158

Hargreaves, G. H. (1989). Accuracy of estimated reference crop evapotranspiration. Journal of Irrigation and Drainage Engineering, 115(6), 1000–1007. Hargreaves, G. H. (1994). Defining and using reference evapotranspiration. Journal of Irrigation and Drainage Engineering, 120(6), 1132–1139. Hassen, E. E., & Assen, M. (2018). Land use/cover dynamics and its drivers in Gelda catchment, Lake Tana watershed, Ethiopia. Environmental Systems Research, 6(1), 4. He, S., & Wu, J. (2019). Relationships of groundwater quality and associated health risks with land use/land cover patterns: A case study in a loess area, Northwest China. Human and Ecological Risk Assessment: An International Journal, 25(1–2), 354–373. https://doi.org/10.1080/10807039.2019.1570463 Her, Y., Yoo, S.-H., Cho, J., Hwang, S., Jeong, J., & Seong, C. (2019). Uncertainty in hydrological analysis of climate change: Multi-parameter vs. multi-GCM ensemble predictions. Scientific Reports, 9(1), 4974. https://doi.org/10.1038/s41598-019-41334-7 Hishe, S., Bewket, W., Nyssen, J., & Lyimo, J. (2020). Analysing past land use land cover change and CA- Markov-based future modelling in the Middle Suluh Valley, Northern Ethiopia. Geocarto International, 35(3), 225–255. Houet, T., & Hubert-Moy, L. (2006). Modeling and projecting land-use and land-cover changes with Cellular Automaton in considering landscape trajectories. Hu, Y., & Gao, M. (2020). Evaluations of water yield and soil erosion in the Shaanxi-Gansu Loess Plateau under different land use and climate change scenarios. Environmental Development, 34, 100488. Hurni, H. (1983a). Soil erosion and soil formation in agricultural ecosystems: Ethiopia and Northern Thailand. Mountain Research and Development, 131–142. Hurni, H. (1983b). Soil Formation Rates in Ethiopia: Ethiopian Highland Reclamation Study. Addis Ababa: SCRP. Hurni, H. (1988). Degradation and Conservation of the Resources in the Ethiopian Highlands. Mountain Research and Development, 8(2/3), 123–130. https://doi.org/10.2307/3673438 Hurni, H. (1998). Agroecological belts of Ethiopia: Explanatory notes on three maps at a scale of 1: 1,000,000. Soil Conservation Research Program of Ethiopia, Addis Ababa, Ethiopia, 31. Hurni, H. (1985). Erosion-productivity-conservation systems in Ethiopia. Conservation and Productivity: Proceedings, 654–674. https://doi.org/10.7892/boris.77547 Hurni, H, Abate, S. M., Bantider, A., Debele, B., Ludi, E., Portner, B., Yitaferu, B., & Zeleke, G. (2010). Land degradation and sustainable land management in the Highlands of Ethiopia. https://doi.org/10.7892/BORIS.5959 Hurni, H, Herweg, K., Portner, B., & Liniger, H. (2008). Soil erosion and conservation in global agriculture. In Land use and soil resources (pp. 41–71). Springer. Hurni, H, Tato, K., & Zeleke, G. (2005). The implications of changes in population, land use, and land management for surface runoff in the upper Nile basin area of Ethiopia. Mountain Research and Development, 25(2), 147–154. Hurni, Hans. (1985). Erosion-productivity conservation systems in Ethiopia. IV International Conference on Soil Conservation November 3-9, 1985 Maracay Venezuela, 654–674. Hurni, K., Zeleke, G., Kassie, M., Tegegne, B., Kassawmar, T., Teferi, E., Moges, A., Tadesse, D., Ahmed, M., & Degu, Y. (2015). Soil degradation and sustainable land management in the rainfed agricultural areas of Ethiopia: An assessment of the economic implications. Report for the Economics of Land Degradation Initiative. Hyandye, C., & Martz, L. W. (2017). A Markovian and cellular automata land-use change predictive model of the Usangu Catchment. International Journal of Remote Sensing, 38(1), 64–81. Ines, A. V. M., & Hansen, J. W. (2006). Bias correction of daily GCM rainfall for crop simulation studies. Agricultural and Forest Meteorology, 138(1), 44–53. https://doi.org/10.1016/j.agrformet.2006.03.009

159

Jacob, M., Romeyns, L., Frankl, A., Asfaha, T., Beeckman, H., & Nyssen, J. (2016). Land use and cover dynamics since 1964 in the Afro-Alpine vegetation belt: Lib Amba Mountain in north Ethiopia. Land Degradation & Development, 27(3), 641–653. Jahanishakib, F., Mirkarimi, S. H., Salmanmahiny, A., & Poodat, F. (2018). Land use change modeling through scenario-based cellular automata Markov: Improving spatial forecasting. Environmental Monitoring and Assessment, 190(6), 332. James, R., & Washington, R. (2013). Changes in African temperature and precipitation associated with degrees of global warming. Climatic Change, 117(4), 859–872. Jin, K., Cornelis, W. M., Gabriels, D., Schiettecatte, W., De Neve, S., Lu, J., Buysse, T., Wu, H., Cai, D., Jin, J., & Harmann, R. (2008). Soil management effects on runoff and soil loss from field rainfall simulation. CATENA, 75(2), 191–199. https://doi.org/10.1016/j.catena.2008.06.002 Johnson, R. B., & Onwuegbuzie, A. J. (2004). Mixed methods research: A research paradigm whose time has come. Educational Researcher, 33(7), 14–26. Johnson, R. B., Onwuegbuzie, A. J., & Turner, L. A. (2007). Toward a definition of mixed methods research. Journal of Mixed Methods Research, 1(2), 112–133. Jung, I. W., Chang, H., & Moradkhani, H. (2010). Quantifying uncertainty in urban flooding analysis caused by the combined effect of climate and land use change scenarios. Hydrology and Earth System Science, 15, 617–633. Jupp, V. (2006). The Sage dictionary of social research methods. Sage. Jury, M. R., & Funk, C. (2013). Climatic trends over Ethiopia: Regional signals and drivers. International Journal of Climatology, 33(8), 1924–1935. Kassie, B. T., Rötter, R. P., Hengsdijk, H., Asseng, S., Ittersum, M. K. V., Kahiluoto, H., & Keulen, H. V. (2014). Climate variability and change in the Central Rift Valley of Ethiopia: Challenges for rainfed crop production. The Journal of Agricultural Science, 152(1), 58–74. https://doi.org/10.1017/S0021859612000986 Kharin, V. V., Zwiers, F. W., Zhang, X., & Wehner, M. (2013). Changes in temperature and precipitation extremes in the CMIP5 ensemble. Climatic Change, 119(2), 345–357. Khattree, R., & Naik, D. N. (2018). Applied multivariate statistics with SAS software. SAS Institute Inc. Khosrokhani, M., & Pradhan, B. (2014). Spatio-temporal assessment of soil erosion at Kuala Lumpur metropolitan city using remote sensing data and GIS. Geomatics, Natural Hazards and Risk, 5(3), 252–270. Kidane, M., Tolessa, T., Bezie, A., Kessete, N., & Endrias, M. (2019). Evaluating the impacts of climate and land use/land cover (LU/LC) dynamics on the Hydrological Responses of the Upper Blue Nile in the Central Highlands of Ethiopia. Spatial Information Research, 27(2), 151–167. https://doi.org/10.1007/s41324-018-0222-y Kieffer, B., Arndt, N., Lapierre, H., Bastien, F., Bosch, D., Pecher, A., Yirgu, G., Ayalew, D., Weis, D., Jerram, D. A., Keller, F., & Meugniot, C. (2004). Flood and Shield Basalts from Ethiopia: Magmas from the African Superswell. Journal of Petrology, 45(4), 793–834. https://doi.org/10.1093/petrology/egg112 Kim, U., & Kaluarachchi, J. J. (2009). Climate Change Impacts on Water Resources in the Upper Blue Nile River Basin, Ethiopia. JAWRA Journal of the American Water Resources Association, 45(6), 1361– 1378. Kindu, M., Schneider, T., Teketay, D., & Knoke, T. (2013). Land use/land cover change analysis using object-based classification approach in Munessa-Shashemene landscape of the Ethiopian highlands. Remote Sensing, 5(5), 2411–2435. Kindu, M., Schneider, T., Teketay, D., & Knoke, T. (2015). Drivers of land use/land cover changes in Munessa-Shashemene landscape of the south-central highlands of Ethiopia. Environmental Monitoring and Assessment, 187(7), 452. Kiros, G., Shetty, A., & Nandagiri, L. (2016). Analysis of variability and trends in rainfall over northern Ethiopia. Arabian Journal of Geosciences, 9(6), 451.

160

Knutti, R., & Sedláček, J. (2013). Robustness and uncertainties in the new CMIP5 climate model projections. Nature Climate Change, 3(4), 369–373. Koch, F. J., Van Griensven, A., Uhlenbrook, S., Tekleab, S., & Teferi, E. (2012). The Effects of land use change on hydrological responses in the choke mountain range (Ethiopia)-a new approach addressing land use dynamics in the model SWAT. Koirala, P., Thakuri, S., Joshi, S., & Chauhan, R. (2019). Estimation of soil erosion in Nepal using a RUSLE modeling and geospatial tool. Geosciences, 9(4), 147. Koomen, E., & Stillwell, J. (2007). Modelling land-use change. In Modelling land-use change (pp. 1–22). Springer, Dordrecht. Kothari, C. R. (2004). Research methodology: Methods and techniques. New Age International. Kouli, M., Soupios, P., & Vallianatos, F. (2009). Soil erosion prediction using the Revised Universal Soil Loss Equation (RUSLE) in a GIS framework, Chania, Northwestern Crete, Greece. Environmental Geology, 57(3), 483–497. https://doi.org/10.1007/s00254-008-1318-9 Kura, A. L., & Beyene, D. L. (2020). Cellular automata Markov chain model based deforestation modelling in the pastoral and agro-pastoral areas of southern Ethiopia. Remote Sensing Applications: Society and Environment, 100321. Kurukulasuriya, P., & Rosenthal, S. (2013). Climate Change and Agriculture: A Review of Impacts and Adaptations. World Bank, Washington, DC. Kusangaya, S., Warburton, M. L., Archer van Garderen, E., & Jewitt, G. P. W. (2014). Impacts of climate change on water resources in southern Africa: A review. Physics and Chemistry of the Earth, Parts A/B/C, 67–69, 47–54. https://doi.org/10.1016/j.pce.2013.09.014 Lafforgue, M. (2016). Impact of forests on water quality. Forest and the Water Cycle: Quantity, Quality, Management. Newcastle upon Tyne (UK): Cambridge Scholar Publishing, 204–265. Lambin, E. F., Geist, H., & Rindfuss, R. R. (2006). Introduction: Local processes with global impacts. In Land-use and land-cover change (pp. 1–8). Springer. Legesse, D., Abiye, T. A., Vallet-Coulomb, C., & Abate, H. (2010). Streamflow sensitivity to climate and land cover changes: Meki River, Ethiopia. Hydrology and Earth System Sciences, 14(11), 2277– 2287. Legesse, Dagnachew, Vallet-Coulomb, C., & Gasse, F. (2003). Hydrological response of a catchment to climate and land use changes in Tropical Africa: Case study South Central Ethiopia. Journal of Hydrology, 275(1–2), 67–85. Leh, M., Bajwa, S., & Chaubey, I. (2013). Impact of land use change on erosion risk: An integrated remote sensing, geographic information system and modeling methodology. Land Degradation & Development, 24(5), 409–421. Li, L.-J., Zhang, L., Wang, H., Wang, J., Yang, J.-W., Jiang, D.-J., Li, J.-Y., & Qin, D.-Y. (2007). Assessing the impact of climate variability and human activities on streamflow from the Wuding River basin in China. Hydrological Processes: An International Journal, 21(25), 3485–3491. Li, M.-H., Tien, W., & Tung, C.-P. (2009). Assessing the impact of climate change on the land hydrology in Taiwan. Paddy and Water Environment, 7(4), 283. https://doi.org/10.1007/s10333-009-0175-9 Li, Z., Liu, W.-Z., Zhang, X.-C., & Zheng, F.-L. (2011). Assessing the site-specific impacts of climate change on hydrology, soil erosion and crop yields in the Loess Plateau of China. Climatic Change, 105(1–2), 223–242. Lillesand, T., Kiefer, R., & Chipman, J. (2004). Remote sensing and image interpretation (5th ed.). John Wiley & Sons. Lillesand, T., Kiefer, R. W., & Chipman, J. (2015). Remote sensing and image interpretation. John Wiley & Sons. Lin, J., Guan, Q., Tian, J., Wang, Q., Tan, Z., Li, Z., & Wang, N. (2020). Assessing temporal trends of soil erosion and sediment redistribution in the Hexi Corridor region using the integrated RUSLE-TLSD model. Catena, 195, 104756. Liu, J., Zhang, C., Kou, L., & Zhou, Q. (2017). Effects of climate and land use changes on water resources in the Taoer river. Advances in Meteorology, 2017.

161

Lorber, A., Wangen, L. E., & Kowalski, B. R. (1987). A theoretical foundation for the PLS algorithm. Journal of Chemometrics, 1(1), 19–31. Luo, M., Liu, T., Meng, F., Duan, Y., Frankl, A., Bao, A., & De Maeyer, P. (2018). Comparing Bias Correction Methods Used in Downscaling Precipitation and Temperature from Regional Climate Models: A Case Study from the Kaidu River Basin in Western China. Water, 10(8), 1046. https://doi.org/10.3390/w10081046 Ma, Z., Kang, S., Zhang, L., Tong, L., & Su, X. (2008). Analysis of impacts of climate variability and human activity on streamflow for a river basin in arid region of northwest China. Journal of Hydrology, 352(3–4), 239–249. MacDicken, K., Jonsson, Ö., Piña, L., Maulo, S., Contessa, V., Adikari, Y., Garzuglia, M., Lindquist, E., Reams, G., & D’Annunzio, R. (2016). Global Forest Resources Assessment 2015: How are the world’s forests changing? FAO. https://agris.fao.org/agris- search/search.do?recordID=XF2017001127 Maeda, E. E., Pellikka, P. K., Siljander, M., & Clark, B. J. (2010). Potential impacts of agricultural expansion and climate change on soil erosion in the Eastern Arc Mountains of Kenya. Geomorphology, 123(3–4), 279–289. Maitima, J. M., Mugatha, S. M., Reid, R. S., Gachimbi, L. N., Majule, A., Lyaruu, H., Pomery, D., Mathai, S., & Mugisha, S. (2009). The linkages between land use change, land degradation and biodiversity across East Africa. African Journal of Environmental Science and Technology, 3(10). Malthus, T. R. (1959). Population: The first essay (Vol. 31). University of Michigan Press. Mango, L. M., Melesse, A. M., McClain, M. E., Gann, D., & Setegn, S. G. (2011). Land use and climate change impacts on the hydrology of the upper Mara River Basin, Kenya: Results of a modeling study to support better resource management. Hydrology and Earth System Sciences, 15(7), 2245– 2258. Maraun, D. (2012). Nonstationarities of regional climate model biases in European seasonal mean temperature and precipitation sums. Geophysical Research Letters, 39(6). Marhaento, H., Booij, M. J., & Hoekstra, A. Y. (2018). Hydrological response to future land-use change and climate change in a tropical catchment. Hydrological Sciences Journal, 63(9), 1368–1385. Marko, K., Zulkarnain, F., & Kusratmoko, E. (2016). Coupling of Markov chains and cellular automata spatial models to predict land cover changes (case study: Upper Ci Leungsi catchment area). IOP Conference Series: Earth and Environmental Science, 47(1), 012032. McCarthy, J. J., Canziani, O. F., Leary, N. A., Dokken, D. J., & White, K. S. (2001). Climate change 2001: Impacts, adaptation, and vulnerability: contribution of Working Group II to the third assessment report of the Intergovernmental Panel on Climate Change (Vol. 2). Cambridge University Press. McDowell, G., Ford, J., & Jones, J. (2016). Community-level climate change vulnerability research: Trends, progress, and future directions. Environmental Research Letters, 11(3), 033001. McSweeney, C., New, M., & Lizcano, G. (2008). UNDP climate change country profiles: Ethiopia. Profiles for 52 Countries. Meinshausen, M., Smith, S. J., Calvin, K., Daniel, J. S., Kainuma, M. L. T., Lamarque, J.-F., Matsumoto, K., Montzka, S. A., Raper, S. C. B., & Riahi, K. (2011). The RCP greenhouse gas concentrations and their extensions from 1765 to 2300. Climatic Change, 109(1–2), 213. Mekonnen, D. F., Duan, Z., Rientjes, T., & Disse, M. (2018). Analysis of combined and isolated effects of land-use and land-cover changes and climate change on the upper Blue Nile River basin’s streamflow. Hydrology and Earth System Sciences, 22(12), 6187–6207. Mekonnen, Z., Berie, H. T., Woldeamanuel, T., Asfaw, Z., & Kassa, H. (2018). Land use and land cover changes and the link to land degradation in Arsi Negele district, Central Rift Valley, Ethiopia. Remote Sensing Applications: Society and Environment, 12, 1–9. Melaku, N. D., Renschler, C. S., Holzmann, H., Strohmeier, S., Bayu, W., Zucca, C., Ziadat, F., & Klik, A. (2018). Prediction of soil and water conservation structure impacts on runoff and erosion processes using SWAT model in the northern Ethiopian highlands. Journal of Soils and Sediments, 18(4), 1743–1755.

162

Melesse, A. F., Awulachew, S. B., & McCornick, P. G. (2011). Introduction: Hydrology of the Niles in the face of climate and land-use dynamics. Melesse, A. M., Loukas, A. G., Senay, G., & Yitayew, M. (2009). Climate change, land-cover dynamics and ecohydrology of the Nile River Basin. Hydrological Processes: An International Journal, 23(26), 3651–3652. Mengistu, D., Bewket, W., Dosio, A., & Panitz, H.-J. (2021). Climate change impacts on water resources in the Upper Blue Nile (Abay) River Basin, Ethiopia. Journal of Hydrology, 592, 125614. Mengistu, D., Bewket, W., & Lal, R. (2014). Recent spatiotemporal temperature and rainfall variability and trends over the Upper Blue Nile River Basin, Ethiopia. International Journal of Climatology, 34(7), 2278–2292. Mengistu, D., Bewket, W., & Lal, R. (2015). Soil erosion hazard under the current and potential climate change induced loss of soil organic matter in the Upper Blue Nile (Abay) River Basin, Ethiopia. In Sustainable intensification to advance food security and enhance climate resilience in Africa (pp. 137–163). Springer. Mengistu, D., Waktola, D. K., & Woldetsadik, M. (2012). Detection and analysis of land-use and land- cover changes in the Midwest escarpment of the Ethiopian Rift Valley. Journal of Land Use Science, 7(3), 239–260. Meshesha, D. T., Tsunekawa, A., & Haregeweyn, N. (2016). Determination of soil erodibility using fluid energy method and measurement of the eroded mass. Geoderma, 284, 13–21. Meshesha, D. T., Tsunekawa, A., Tsubo, M., & Haregeweyn, N. (2012). Dynamics and hotspots of soil erosion and management scenarios of the Central Rift Valley of Ethiopia. International Journal of Sediment Research, 27(1), 84–99. Mevik, B.-H., & Wehrens, R. (2015). Introduction to the pls Package. Help Section of The “Pls” Package of R Studio Software; R Foundation for Statistical Computing: Vienna, Austria, 1–23. Midekisa, A., Holl, F., Savory, D. J., Andrade-Pacheco, R., Gething, P. W., Bennett, A., & Sturrock, H. J. (2017). Mapping land cover change over continental Africa using Landsat and Google Earth Engine cloud computing. PloS One, 12(9), e0184926. Migiro, S. O., & Magangi, B. A. (2011). Mixed methods: A review of literature and the future of the new research paradigm. African Journal of Business Management, 5(10), 3757–3764. Miheretu, B. A., & Yimer, A. A. (2018). Land use/land cover changes and their environmental implications in the Gelana sub-watershed of Northern highlands of Ethiopia. Environmental Systems Research, 6(1), 7. Mishra, V., Cherkauer, K. A., Niyogi, D., Lei, M., Pijanowski, B. C., Ray, D. K., Bowling, L. C., & Yang, G. (2010). A regional scale assessment of land use/land cover and climatic changes on water and energy cycle in the upper Midwest United States. International Journal of Climatology, 30(13), 2025–2044. MoFED. (2010). Growth and transformation plan of Ethiopia (GTP). Moges, D. M., & Bhat, H. G. (2017). Integration of geospatial technologies with RUSLE for analysis of land use/cover change impact on soil erosion: Case study in Rib watershed, north-western highland Ethiopia. Environmental Earth Sciences, 76(22), 765. Moges, D. M., & Bhat, H. G. (2018). An insight into land use and land cover changes and their impacts in Rib watershed, north-western highland Ethiopia. Land Degradation & Development, 29(10), 3317– 3330. Moges, D. M., Kmoch, A., Bhat, H. G., & Uuemaa, E. (2020). Future soil loss in highland Ethiopia under changing climate and land use. Regional Environmental Change, 20(1), 1–14. Moghadam, H. S., & Helbich, M. (2013). Spatiotemporal urbanization processes in the megacity of Mumbai, India: A Markov chains-cellular automata urban growth model. Applied Geography, 40, 140–149. Mohamed, A., & Worku, H. (2020). Simulating urban land use and cover dynamics using cellular automata and Markov chain approach in Addis Ababa and the surrounding. Urban Climate, 31, 100545.

163

Mohammed, S., Alsafadi, K., Talukdar, S., Kiwan, S., Hennawi, S., Alshiehabi, O., Sharaf, M., & Harsanyie, E. (2020). Estimation of soil erosion risk in southern part of Syria by using RUSLE integrating geo informatics approach. Remote Sensing Applications: Society and Environment, 100375. Mölders, N. (2011). Land-use and land-cover changes: Impact on climate and air quality (Vol. 44). Springer Science & Business Media. Molla, T., & Sisheber, B. (2017). Estimating soil erosion risk and evaluating erosion control measures for soil conservation planning at Koga watershed in the highlands of Ethiopia. Solid Earth, 8(1), 13. Mondal, B., Das, D. N., & Bhatta, B. (2017). Integrating cellular automata and Markov techniques to generate urban development potential surface: A study on Kolkata agglomeration. Geocarto International, 32(4), 401–419. Mondal, Md. S., Sharma, N., Garg, P. K., & Kappas, M. (2016). Statistical independence test and validation of CA Markov land use land cover (LULC) prediction results. The Egyptian Journal of Remote Sensing and Space Science, 19(2), 259–272. https://doi.org/10.1016/j.ejrs.2016.08.001 Montanarella, L., Badraoui, M., Chude, V., Costa, I., Mamo, T., Yemefack, M., AULANG, M., Yagi, K., Hong, S. Y., & Vijarnsorn, P. (2015). Status of the world’s soil resources: Main report. Embrapa Solos-Livro Científico (ALICE). Montanarella, L., Pennock, D. J., McKenzie, N., Badraoui, M., Chude, V., Baptista, I., Mamo, T., Yemefack, M., Aulakh, M. S., & Yagi, K. (2016). World’s soils are under threat. Embrapa Solos- Artigo Em Periódico Indexado (ALICE). Moore, I. D., & Burch, G. J. (1986a). Physical basis of the length-slope factor in the universal soil loss equation 1. Soil Science Society of America Journal, 50(5), 1294–1298. Moore, Ian. D., & Burch, Gordon. J. (1986b). Modelling erosion and deposition: Topographic effects. Transactions of the ASAE, 29(6), 1624–1630. Moriasi, D. N., Arnold, J. G., Van Liew, M. W., Bingner, R. L., Harmel, R. D., & Veith, T. L. (2007). Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the ASABE, 50(3), 885–900. Mukherjee, A., & Kamarulzaman, N. H. (2016). Mixed method research: A concept. In Mixed Methods Research for Improved Scientific Study (pp. 39–64). IGI Global. Mukhopadhyay, A., Mondal, P., Barik, J., Chowdhury, S. M., Ghosh, T., & Hazra, S. (2015). Changes in mangrove species assemblages and future prediction of the Bangladesh Sundarbans using Markov chain model and cellular automata. Environmental Science: Processes & Impacts, 17(6), 1111– 1117. Mullan, D., Favis-Mortlock, D., & Fealy, R. (2012). Addressing key limitations associated with modelling soil erosion under the impacts of future climate change. Agricultural and Forest Meteorology, 156, 18–30. Munthali, M. G., Mustak, S., Adeola, A., Botai, J., Singh, S. K., & Davis, N. (2020). Modelling land use and land cover dynamics of Dedza district of Malawi using hybrid Cellular Automata and Markov model. Remote Sensing Applications: Society and Environment, 17, 100276. Musau, J., Sang, J., Gathenya, J., & Luedeling, E. (2015). Hydrological responses to climate change in Mt. Elgon watersheds. Journal of Hydrology: Regional Studies, 3, 233–246. https://doi.org/10.1016/j.ejrh.2014.12.001 Mustard, J. F., Defries, R. S., Fisher, T., & Moran, E. (2012). Land-use and land-cover change pathways and impacts. In Land change science (pp. 411–429). Springer. Mwangi, H. M., Julich, S., Patil, S. D., McDonald, M. A., & Feger, K.-H. (2016). Relative contribution of land use change and climate variability on discharge of upper Mara River, Kenya. Journal of Hydrology: Regional Studies, 5, 244–260. Myint, S. W., & Wang, L. (2006). Multicriteria decision approach for land use land cover change using Markov chain analysis and a cellular automata approach. Canadian Journal of Remote Sensing, 32(6), 390–404. https://doi.org/10.5589/m06-032

164

Nainggolan, D., Termansen, M., Fleskens, L., Hubacek, K., Reed, M. S., de Vente, J., & Boix-Fayos, C. (2012). What does the future hold for semi-arid Mediterranean agro-ecosystems?–Exploring cellular automata and agent-based trajectories of future land-use change. Applied Geography, 35(1– 2), 474–490. Nash, J. E., & Sutcliffe, J. V. (1970). River flow forecasting through conceptual models part I—A discussion of principles. Journal of Hydrology, 10(3), 282–290. Nazarenko, L., Schmidt, G. A., Miller, R. L., Tausnev, N., Kelley, M., Ruedy, R., Russell, G. L., Aleinov, I., Bauer, M., & Bauer, S. (2015). Future climate change under RCP emission scenarios with GISS M odelE2. Journal of Advances in Modeling Earth Systems, 7(1), 244–267. Negese, A. (2021). Impacts of Land Use and Land Cover Change on Soil Erosion and Hydrological Responses in Ethiopia. Applied and Environmental Soil Science, 2021. Nie, W., Yuan, Y., Kepner, W., Nash, M. S., Jackson, M., & Erickson, C. (2011). Assessing impacts of Landuse and Landcover changes on hydrology for the upper San Pedro watershed. Journal of Hydrology, 407(1–4), 105–114. Nigussie, T. A., Fanta, A., Melesse, A. M., & Quraishi, S. (2014). Modeling rainfall erosivity from daily rainfall events, upper Blue Nile basin, Ethiopia. In Nile River Basin (pp. 307–335). Springer. Nkonya, E., Mirzabaev, A., & Von Braun, J. (2016). Economics of land degradation and improvement–a global assessment for sustainable development. Springer Nature. NMA. (2007). Climate change national adaptation programme of action (Napa) of Ethiopia. Nyesheja, E. M., Chen, X., El-Tantawi, A. M., Karamage, F., Mupenzi, C., & Nsengiyumva, J. B. (2019). Soil erosion assessment using RUSLE model in the Congo Nile Ridge region of Rwanda. Physical Geography, 40(4), 339–360. Nyssen, J., Poesen, J., Haile, M., Moeyersons, J., & Deckers, J. (2000). Tillage erosion on slopes with soil conservation structures in the Ethiopian highlands. Soil and Tillage Research, 57(3), 115–127. https://doi.org/10.1016/S0167-1987(00)00138-0 Nyssen, J., Poesen, J., Haile, M., Moeyersons, J., Deckers, J., & Hurni, H. (2009a). Effects of land use and land cover on sheet and rill erosion rates in the Tigray highlands, Ethiopia. Zeitschrift Für Geomorphologie, 53(2), 171-197. https://doi.org/10.1127/0372-8854/2009/0053-0171 Nyssen, J., Poesen, J., Haile, M., Moeyersons, J., Deckers, J., & Hurni, H. (2009b). Effects of land use and land cover on sheet and rill erosion rates in the Tigray highlands, Ethiopia. Zeitschrift Für Geomorphologie, 53(2), 171–197. Oliver, T. H., & Morecroft, M. D. (2014). Interactions between climate change and land use change on biodiversity: Attribution problems, risks, and opportunities. WIREs Climate Change, 5(3), 317– 335. https://doi.org/10.1002/wcc.271 Olson, J. M., Alagarswamy, G., Andresen, J. A., Campbell, D. J., Davis, A. Y., Ge, J., Huebner, M., Lofgren, B. M., Lusch, D. P., & Moore, N. J. (2008). Integrating diverse methods to understand climate–land interactions in East Africa. Geoforum, 39(2), 898–911. Oluwatomiwa, A. (2014). Assessment of climate variability impact on water resources within the Alabama River basin. Omrani, H., Tayyebi, A., & Pijanowski, B. (2017). Integrating the multi-label land-use concept and cellular automata with the artificial neural network-based Land Transformation Model: An integrated ML- CA-LTM modeling framework. GIScience & Remote Sensing, 54(3), 283–304. Overmars, K. P., & Verburg, P. H. (2005). Analysis of land use drivers at the watershed and household level: Linking two paradigms at the Philippine forest fringe. International Journal of Geographical Information Science, 19(2), 125–152. Pachauri, R. K., Allen, M. R., Barros, V. R., Broome, J., Cramer, W., Christ, R., Church, J. A., Clarke, L., Dahe, Q., & Dasgupta, P. (2014). Climate change 2014: Synthesis report. Contribution of Working Groups I, II and III to the fifth assessment report of the Intergovernmental Panel on Climate Change. Ipcc.

165

Pan, S., Liu, D., Wang, Z., Zhao, Q., Zou, H., Hou, Y., Liu, P., & Xiong, L. (2017). Runoff Responses to Climate and Land Use/Cover Changes under Future Scenarios. Water, 9(7), 475. https://doi.org/10.3390/w9070475 Parker, D. C., Manson, S. M., Janssen, M. A., Hoffmann, M. J., & Deadman, P. (2003). Multi-agent systems for the simulation of land-use and land-cover change: A review. Annals of the Association of American Geographers, 93(2), 314–337. Parry, M., Parry, M. L., Canziani, O., Palutikof, J., Van der Linden, P., & Hanson, C. (2007). Climate change 2007-impacts, adaptation and vulnerability: Working group II contribution to the fourth assessment report of the IPCC (Vol. 4). Cambridge University Press. Perović, V., Kadović, R., Djurdjević, V., Braunović, S., Čakmak, D., Mitrović, M., & Pavlović, P. (2019). Effects of changes in climate and land use on soil erosion: A case study of the Vranjska Valley, Serbia. Regional Environmental Change, 19(4), 1035–1046. Perry, M., & Hollis, D. (2005). The development of a new set of long-term climate averages for the UK. International Journal of Climatology, 25(8), 1023–1039. https://doi.org/10.1002/joc.1160 Phinzi, K., & Ngetar, N. S. (2019). Land use/land cover dynamics and soil erosion in the Umzintlava catchment (T32E), Eastern Cape, South Africa. Transactions of the Royal Society of South Africa, 74(3), 223–237. Plangoen, P., Babel, M. S., Clemente, R. S., Shrestha, S., & Tripathi, N. K. (2013). Simulating the impact of future land use and climate change on soil erosion and deposition in the Mae Nam Nan sub- catchment, Thailand. Sustainability, 5(8), 3244–3274. Pontius Jr, R. G. (2002). Statistical Methods to Partition Effects of Quantity and Location During Comparison of Categorical Maps at Multiple Resolutions. Photogrammetric Engineering and Remote Sensing, 68(10), 1041–1050. Rahman, M. T. U., Tabassum, F., Rasheduzzaman, M., Saba, H., Sarkar, L., Ferdous, J., Uddin, S. Z., & Islam, A. Z. (2017). Temporal dynamics of land use/land cover change and its prediction using CA-ANN model for southwestern coastal Bangladesh. Environmental Monitoring and Assessment, 189(11), 565. Ralha, C. G., Abreu, C. G., Coelho, C. G., Zaghetto, A., Macchiavello, B., & Machado, R. B. (2013). A multi-agent model system for land-use change simulation. Environmental Modelling & Software, 42, 30–46. Ramankutty, N., Graumlich, L., Achard, F., Alves, D., Chhabra, A., DeFries, R. S., Foley, J. A., Geist, H., Houghton, R. A., & Goldewijk, K. K. (2006). Global land-cover change: Recent progress, remaining challenges. Land-Use and Land-Cover Change, 9–39. Rellini, I., Scopesi, C., Olivari, S., Firpo, M., & Maerker, M. (2019). Assessment of soil erosion risk in a typical Mediterranean environment using a high resolution RUSLE approach (Portofino promontory, NW-Italy). Journal of Maps, 15(2), 356–362. Renard, K. G. (1997). Predicting soil erosion by water: A guide to conservation planning with the Revised Universal Soil Loss Equation (RUSLE). United States Government Printing. Renard, K. G., & Freimund, J. R. (1994). Using monthly precipitation data to estimate the R-factor in the revised USLE. Journal of Hydrology, 157(1), 287–306. https://doi.org/10.1016/0022- 1694(94)90110-4 Reshmidevi, T. V., Eldho, T. I., & Jana, R. (2009). A GIS-integrated fuzzy rule-based inference system for land suitability evaluation in agricultural watersheds. Agricultural Systems, 101(1), 101–109. https://doi.org/10.1016/j.agsy.2009.04.001 Ridwansyah, I., Yulianti, M., Apip, Onodera, S., Shimizu, Y., Wibowo, H., & Fakhrudin, M. (2020). The impact of land use and climate change on surface runoff and groundwater in Cimanuk watershed, Indonesia. Limnology, 21(3), 487–498. https://doi.org/10.1007/s10201-020-00629-9 Rogger, M., Agnoletti, M., Alaoui, A., Bathurst, J. C., Bodner, G., Borga, M., Chaplot, V., Gallart, F., Glatzel, G., Hall, J., Holden, J., Holko, L., Horn, R., Kiss, A., Kohnová, S., Leitinger, G., Lennartz, B., Parajka, J., Perdigão, R., … Blöschl, G. (2017). Land use change impacts on floods at the

166

catchment scale: Challenges and opportunities for future research. Water Resources Research, 53(7), 5209–5219. https://doi.org/10.1002/2017WR020723 Roth, V., Lemann, T., Zeleke, G., Subhatu, A. T., Nigussie, T. K., & Hurni, H. (2018). Effects of climate change on water resources in the upper Blue Nile Basin of Ethiopia. Heliyon, 4(9), e00771. Routschek, A., Schmidt, J., & Kreienkamp, F. (2014). Impact of climate change on soil erosion—A high- resolution projection on catchment scale until 2100 in Saxony/Germany. Catena, 121, 99–109. Safriel, U. N. (2007). The assessment of global trends in land degradation. In Climate and land degradation (pp. 1–38). Springer. Sahli, Y., Mokhtari, E., Merzouk, B., Laignel, B., Vial, C., & Madani, K. (2019). Mapping surface water erosion potential in the Soummam watershed in Northeast Algeria with RUSLE model. Journal of Mountain Science, 16(7), 1606–1615. Samndong, R. A., Bush, G., Vatn, A., & Chapman, M. (2018). Institutional analysis of causes of deforestation in REDD+ pilot sites in the Equateur province: Implication for REDD+ in the Democratic Republic of Congo. Land Use Policy, 76, 664–674. Sanderson, E. W., Redford, K. H., Vedder, A., Coppolillo, P. B., & Ward, S. E. (2002). A conceptual model for conservation planning based on landscape species requirements. Landscape and Urban Planning, 58(1), 41–56. Sang, L., Zhang, C., Yang, J., Zhu, D., & Yun, W. (2011). Simulation of land use spatial pattern of towns and villages based on CA–Markov model. Mathematical and Computer Modelling, 54(3–4), 938– 943. Schmidt, E., & Zemadim, B. (2013). Hydrological modelling of sustainable land management interventions in the Mizewa watershed of the Blue Nile basin. Rainwater Management for Resilient Livelihoods in Ethiopia, 48. Schneiderman, E. M., Steenhuis, T. S., Thongs, D. J., Easton, Z. M., Zion, M. S., Neal, A. L., Mendoza, G. F., & Todd Walter, M. (2007). Incorporating variable source area hydrology into a curve-number- based watershed model. Hydrological Processes: An International Journal, 21(25), 3420–3430. Schulze, R. E. (2000). Modelling hydrological responses to land use and climate change: A southern African perspective. Ambio, 12–22. SCS. (1972). National engineering handbook: Section 4: Hydrology. USDA Washington. Serpa, D., Nunes, J. P., Santos, J., Sampaio, E., Jacinto, R., Veiga, S., Lima, J. C., Moreira, M., Corte-Real, J., & Keizer, J. J. (2015). Impacts of climate and land use changes on the hydrological and erosion processes of two contrasting Mediterranean catchments. Science of the Total Environment, 538, 64–77. Setegn, S. (2008). Hydrological and sediment yield modelling in Lake Tana basin, Blue Nile Ethiopia [PhD Thesis]. KTH. Setegn, S., Dargahi, B., Srinivasan, R., & Melesse, A. M. (2010). Modeling of Sediment Yield From Anjeni-Gauged Watershed, Ethiopia Using SWAT Model 1. JAWRA Journal of the American Water Resources Association, 46(3), 514–526. Setegn, S., Melesse, A. M., Rayner, D., & Dargahi, B. (2014). Climate change impact on water resources and adaptation strategies in the Blue Nile River Basin. In Nile River Basin (pp. 389–404). Springer. Setegn, S., Rayner, D., Melesse, A. M., Dargahi, B., & Srinivasan, R. (2011). Impact of climate change on the hydroclimatology of Lake Tana Basin, Ethiopia. Water Resources Research, 47(4). Setegn, S., Rayner, D., Melesse, A. M., Dargahi, B., Srinivasan, R., & Wörman, A. (2011). Climate change impact on agricultural water resources variability in the Northern Highlands of Ethiopia. In Nile River Basin (pp. 241–265). Springer. Seutloali, K. E., & Beckedahl, H. R. (2015). A review of road-related soil erosion: An assessment of causes, evaluation techniques and available control measures. Earth Sciences Research Journal, 19(1), 73– 80. Sewnet, A. (2015). Land Use/Cover Change at Infraz Watershed, Northwestren Ethiopia. Journal of Landscape Ecology, 8(1), 69–83.

167

Shamir, E., Megdal, S. B., Carrillo, C., Castro, C. L., Chang, H.-I., Chief, K., Corkhill, F. E., Eden, S., Georgakakos, K. P., & Nelson, K. M. (2015). Climate change and water resources management in the Upper Santa Cruz River, Arizona. Journal of Hydrology, 521, 18–33. Sharma, A., Tiwari, K. N., & Bhadoria, P. B. S. (2011). Effect of land use land cover change on soil erosion potential in an agricultural watershed. Environmental Monitoring and Assessment, 173(1–4), 789– 801. Sharpley, A. N., & Williams, J. R. (1990). EPIC-Erosion/Productivity Impact Calculator. I: Model documentation. II: User manual. Technical Bulletin-United States Department of Agriculture, 1768. Shaw, I. G. R., Dixon, D. P., & Jones III, J. P. (2010). Theorizing our world. GOMEZ, B.; JONES, JP Research Methods in Geography. Oxford: Blackwell Publishing, 9–25. Shawul, A. A., Chakma, S., & Melesse, A. M. (2019). The response of water balance components to land cover change based on hydrologic modeling and partial least squares regression ( PLSR ) analysis in the Upper Awash Basin. Journal of Hydrology: Regional Studies, 26(August), 100640. https://doi.org/10.1016/j.ejrh.2019.100640 Shi, P., Ma, X., Hou, Y., Li, Q., Zhang, Z., Qu, S., Chen, C., Cai, T., & Fang, X. (2013). Effects of land- use and climate change on hydrological processes in the upstream of Huai River, China. Water Resources Management, 27(5), 1263–1278. Shi, Z., Ai, L., Li, X., Huang, X., Wu, G., & Liao, W. (2013). Partial least-squares regression for linking land-cover patterns to soil erosion and sediment yield in watersheds. Journal of Hydrology, 498, 165–176. Shiferaw, B., & Holden, S. T. (1998). Resource degradation and adoption of land conservation technologies in the Ethiopian highlands: A case study in Andit Tid, North Shewa. Agricultural Economics, 18(3), 233–247. Shrestha, B., Cochrane, T. A., Caruso, B. S., & Arias, M. E. (2018). Land use change uncertainty impacts on streamflow and sediment projections in areas undergoing rapid development: A case study in the Mekong Basin. Land Degradation & Development, 29(3), 835–848. https://doi.org/10.1002/ldr.2831 Shrestha, M. (2015). Data Analysis Relied on Linear Scaling Bias Correction (V. 1.0) Microsoft Excel File. Shrestha, Manish, Acharya, S. C., & Shrestha, P. K. (2017). Bias correction of climate models for hydrological modelling – are simple methods still useful? Meteorological Applications, 24(3), 531– 539. https://doi.org/10.1002/met.1655 Simane, B., Zaitchik, B. F., & Ozdogan, M. (2013). Agroecosystem analysis of the Choke Mountain watersheds, Ethiopia. Sustainability, 5(2), 592–616. Simonneaux, V., Cheggour, A., Deschamps, C., Mouillot, F., Cerdan, O., & Le Bissonnais, Y. (2015). Land use and climate change effects on soil erosion in a semi-arid mountainous watershed (High Atlas, Morocco). Journal of Arid Environments, 122, 64–75. Singh, S. K., Mustak, S., Srivastava, P. K., Szabó, S., & Islam, T. (2015). Predicting spatial and decadal LULC changes through cellular automata Markov chain models using earth observation datasets and geo-information. Environmental Processes, 2(1), 61–78. Sisay, K., Thurnher, C., & Hasenauer, H. (2017). Daily climate data for the Amhara region in Northwestern Ethiopia. International Journal of Climatology, 37(6), 2797–2808. Solomon, S., Manning, M., Marquis, M., & Qin, D. (2007). Climate change 2007-the physical science basis: Working group I contribution to the fourth assessment report of the IPCC (Vol. 4). Souverijns, N., Thiery, W., Demuzere, M., & Van Lipzig, N. P. (2016). Drivers of future changes in East African precipitation. Environmental Research Letters, 11(11), 114011. Srivastava, P. K., Han, D., Rico-Ramirez, M. A., & Islam, T. (2014). Sensitivity and uncertainty analysis of mesoscale model downscaled hydro-meteorological variables for discharge prediction. Hydrological Processes, 28(15), 4419–4432.

168

Steenhuis, T. S., Easton, Z. M., Awulachew, S. B., Ahmed, A. A., Bashar, K. E., Adgo, E., Selassie, Y. G., & Tilahun, S. A. (2013). The Nile Basin sediment loss and degradation, with emphasis on the Blue Nile. In The Nile River Basin (pp. 134–154). Routledge. Stocker, T., Qin, D., Plattner, G.-K., Tignor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex, V., & Midgley, P. M. (2013). Climate change 2013: The physical science basis. Contribution of working group I to the fifth assessment report of the intergovernmental panel on climate change (Vol. 1535). Cambridge university press Cambridge, United Kingdom and New York, NY, USA. Stonestrom, D. A., Scanlon, B. R., & Zhang, L. (2009). Introduction to special section on impacts of land use change on water resources. Water Resources Research, 45(7). Streile, G., Shields, K. D., Stroh, J. L., Bagaasen, L. M., Whelan, G., McDonald, J. P., Droppo, J. G., & Buck, J. W. (1996). The Multimedia Environmental Pollutant Assessment System. Pacific Northwest National Lab., Richland, WA (United States). Sun, W., Shao, Q., & Liu, J. (2013). Soil erosion and its response to the changes of precipitation and vegetation cover on the Loess Plateau. Journal of Geographical Sciences, 23(6), 1091–1106. Sunde, M. G., He, H. S., Hubbart, J. A., & Urban, M. A. (2018). An integrated modeling approach for estimating hydrologic responses to future urbanization and climate changes in a mixed-use midwestern watershed. Journal of Environmental Management, 220, 149–162. Swain, S. S., Mishra, A., Sahoo, B., & Chatterjee, C. (2020). Water scarcity-risk assessment in data-scarce river basins under decadal climate change using a hydrological modelling approach. Journal of Hydrology, 590, 125260. Tabari, H., Taye, M. T., & Willems, P. (2015). Statistical assessment of precipitation trends in the upper Blue Nile River basin. Stochastic Environmental Research and Risk Assessment, 29(7), 1751–1761. Tadele, K., & Förch, G. (2007). Impact of land use/cover change on streamflow: The case of Hare River Watershed, Ethiopia. Catchment and Lake Research, Proceedings 2nd Lake Abaya Research Symposium (LARS), Arba Minch, Ethiopia. Tadesse, L., Suryabhagavan, K. V., Sridhar, G., & Legesse, G. (2017). Land use and land cover changes and Soil erosion in Yezat Watershed, North Western Ethiopia. International Soil and Water Conservation Research, 5(2), 85–94. Taffese, T., & Zemadim, B. (2013). Hydrological modelling of a catchment using the SWAT model in the upper Blue Nile basin of Ethiopia. ILRI. Talib, A., & Randhir, T. O. (2017). Climate change and land use impacts on hydrologic processes of watershed systems. Journal of Water and Climate Change, 8(3), 363–374. Tamene, L., & Le, Q. B. (2015). Estimating soil erosion in sub-Saharan Africa based on landscape similarity mapping and using the revised universal soil loss equation (RUSLE). Nutrient Cycling in Agroecosystems, 102(1), 17–31. Tamene, L., Le, Q. B., & Vlek, P. L. (2014). A landscape planning and management tool for land and water resources management: An example application in northern Ethiopia. Water Resources Management, 28(2), 407–424. Tamene, L., & Vlek, P. L. (2008). Soil erosion studies in northern Ethiopia. In Land use and soil resources (pp. 73–100). Springer. Tan, M. L., Ibrahim, A. L., Yusop, Z., Duan, Z., & Ling, L. (2015). Impacts of land-use and climate variability on hydrological components in the Johor River basin, Malaysia. Hydrological Sciences Journal, 60(5), 873–889. Tankpa, V., Wang, L., Awotwi, A., Singh, L., Thapa, S., Atanga, R. A., & Guo, X. (2020). Modeling the effects of historical and future land use/land cover change dynamics on the hydrological response of Ashi watershed, northeastern China. Environment, Development and Sustainability, 1–30. Tashakkori, A., & Creswell, J. W. (2007). The new era of mixed methods. Sage Publications. Tashakkori, A., Teddlie, C., & Teddlie, C. B. (1998). Mixed methodology: Combining qualitative and quantitative approaches (Vol. 46). Sage.

169

Taye, G., Vanmaercke, M., Poesen, J., Van Wesemael, B., Tesfaye, S., Teka, D., Nyssen, J., Deckers, J., & Haregeweyn, N. (2018). Determining RUSLE P-and C-factors for stone bunds and trenches in rangeland and cropland, North Ethiopia. Land Degradation & Development, 29(3), 812–824. Taye, M. T., Willems, P., & Block, P. (2015). Implications of climate change on hydrological extremes in the Blue Nile basin: A review. Journal of Hydrology: Regional Studies, 4, 280–293. Taylor, K. E., Stouffer, R. J., & Meehl, G. A. (2012). An overview of CMIP5 and the experiment design. Bulletin of the American Meteorological Society, 93(4), 485–498. Teferi, E., Bewket, W., Uhlenbrook, S., & Wenninger, J. (2013). Understanding recent land use and land cover dynamics in the source region of the Upper Blue Nile, Ethiopia: Spatially explicit statistical modeling of systematic transitions. Agriculture, Ecosystems & Environment, 165, 98–117. Teferi, E., Uhlenbrook, S., Bewket, W., Wenninger, J., & Simane, B. (2010). The use of remote sensing to quantify wetland loss in the Choke Mountain range, Upper Blue Nile basin, Ethiopia. Hydrology and Earth System Sciences, 14(12), 2415–2428. Tegene, B. (2002). Land-cover/land-use changes in the derekolli catchment of the South Welo Zone of Amhara Region, Ethiopia. Eastern Africa Social Science Research Review, 18(1), 1–20. Teklay, A., Dile, Y. T., Asfaw, D. H., Bayabil, H. K., & Sisay, K. (2021). Impacts of Climate and Land Use Change on Hydrological Response in Gumara Watershed, Ethiopia. Ecohydrology & Hydrobiology, 21(2), 315–332. https://doi.org/10.1016/j.ecohyd.2020.12.001 Teklay, A., Dile, Y. T., Setegn, S. G., Demissie, S. S., & Asfaw, D. H. (2019). Evaluation of static and dynamic land use data for watershed hydrologic process simulation: A case study in Gummara watershed, Ethiopia. CATENA, 172, 65–75. https://doi.org/10.1016/j.catena.2018.08.013 Tekle, K., & Hedlund, L. (2000). Land cover changes between 1958 and 1986 in Kalu District, southern Wello, Ethiopia. Mountain Research and Development, 20(1), 42–51. Tekleab, S., Mohamed, Y., & Uhlenbrook, S. (2013). Hydro-climatic trends in the Abay/upper Blue Nile basin, Ethiopia. Physics and Chemistry of the Earth, Parts A/B/C, 61, 32–42. Tekleab, S., Mohamed, Y., Uhlenbrook, S., & Wenninger, J. (2014). Hydrologic responses to land cover change: The case of Jedeb mesoscale catchment, Abay/Upper Blue Nile basin, Ethiopia. Hydrological Processes, 28(20), 5149–5161. Teklesadik, A. D., Alemayehu, T., Van Griensven, A., Kumar, R., Liersch, S., Eisner, S., Tecklenburg, J., Ewunte, S., & Wang, X. (2017). Inter-model comparison of hydrological impacts of climate change on the Upper Blue Nile basin using ensemble of hydrological models and global climate models. Climatic Change, 141(3), 517–532. Tenenhaus, M., Pagès, J., Ambroisine, L., & Guinot, C. (2005). PLS methodology to study relationships between hedonic judgements and product characteristics. Food Quality and Preference, 16(4), 315–325. https://doi.org/10.1016/j.foodqual.2004.05.013 Tesfaye, S., Guyassa, E., Joseph Raj, A., Birhane, E., & Wondim, G. T. (2014). Land Use and Land Cover Change, and Woody Vegetation Diversity in Human Driven Landscape of Gilgel Tekeze Catchment, Northern Ethiopia. International Journal of Forestry Research, 2014, 614249. https://doi.org/10.1155/2014/614249 Teutschbein, C., & Seibert, J. (2010). Regional climate models for hydrological impact studies at the catchment scale: A review of recent modeling strategies. Geography Compass, 4(7), 834–860. Teutschbein, C., & Seibert, J. (2012). Bias correction of regional climate model simulations for hydrological climate-change impact studies: Review and evaluation of different methods. Journal of Hydrology, 456, 12–29. Thomson, A. M., Calvin, K. V., Smith, S. J., Kyle, G. P., Volke, A., Patel, P., Delgado-Arias, S., Bond- Lamberty, B., Wise, M. A., & Clarke, L. E. (2011). RCP4. 5: A pathway for stabilization of radiative forcing by 2100. Climatic Change, 109(1–2), 77. Tibebe, D., & Bewket, W. (2011). Surface runoff and soil erosion estimation using the SWAT model in the Keleta watershed, Ethiopia. Land Degradation & Development, 22(6), 551–564. Tierney, J. E., Ummenhofer, C. C., & deMenocal, P. B. (2015). Past and future rainfall in the Horn of Africa. Science Advances, 1(9), e1500682.

170

Tolessa, T., Senbeta, F., & Kidane, M. (2017). The impact of land use/land cover change on ecosystem services in the central highlands of Ethiopia. Ecosystem Services, 23, 47–54. https://doi.org/10.1016/j.ecoser.2016.11.010 Touma, D., Ashfaq, M., Nayak, M. A., Kao, S.-C., & Diffenbaugh, N. S. (2015). A multi-model and multi- index evaluation of drought characteristics in the 21st century. Journal of Hydrology, 526, 196– 207. Trap, J., Hättenschwiler, S., Gattin, I., & Aubert, M. (2013). Forest ageing: An unexpected driver of beech leaf litter quality variability in European forests with strong consequences on soil processes. Forest Ecology and Management, 302, 338–345. Tshimanga, R. M., & Hughes, D. A. (2012). Climate change and impacts on the hydrology of the Congo Basin: The case of the northern sub-basins of the Oubangui and Sangha Rivers. Physics and Chemistry of the Earth, Parts A/B/C, 50–52, 72–83. https://doi.org/10.1016/j.pce.2012.08.002 Turner, B., Meyer, W. B., & Skole, D. L. (1994). Global land-use/land-cover change: Towards an integrated study. Ambio. Stockholm, 23(1), 91–95. UNEP. (2016). GEO-6 Regional Assessment for Africa. United Nations Environment Programme, Nairobi, Kenya. UNEP - UN Environment Programme. http://www.unenvironment.org/resources/assessment/geo-6-regional-assessment-africa Urama, K. C., & Ozor, N. (2010). Impacts of climate change on water resources in Africa: The role of adaptation. African Technology Policy Studies Network, 29, 1–29. Vaezi, A. R., Ahmadi, M., & Cerdà, A. (2017). Contribution of raindrop impact to the change of soil physical properties and water erosion under semi-arid rainfalls. Science of the Total Environment, 583, 382–392. van Dam, J. C. (2003). Impacts of climate change and climate variability on hydrological regimes. Cambridge University Press. Van Rompaey, A., Krasa, J., & Dostal, T. (2007). Modelling the impact of land cover changes in the Czech Republic on sediment delivery. Land Use Policy, 24(3), 576–583. van Vuuren, D. P., Edmonds, J., Kainuma, M., Riahi, K., Thomson, A., Hibbard, K., Hurtt, G. C., Kram, T., Krey, V., Lamarque, J.-F., Masui, T., Meinshausen, M., Nakicenovic, N., Smith, S. J., & Rose, S. K. (2011). The representative concentration pathways: An overview. Climatic Change, 109(1), 5. https://doi.org/10.1007/s10584-011-0148-z Vaze, J., Post, D. A., Chiew, F. H. S., Perraud, J.-M., Viney, N. R., & Teng, J. (2010). Climate non- stationarity–validity of calibrated rainfall–runoff models for use in climate change studies. Journal of Hydrology, 394(3–4), 447–457. Veldkamp, A., & Lambin, E. F. (2001). Predicting land-use change. Agriculture, Ecosystems & Environment, 85(1), 1–6. https://doi.org/10.1016/S0167-8809(01)00199-2 Verburg, P. H., Schot, P. P., Dijst, M. J., & Veldkamp, A. (2004). Land use change modelling: Current practice and research priorities. GeoJournal, 61(4), 309–324. Verheye, W. (2009). Land Cover, Land Use and the Global Change. Encyclopedia of Land Use, Land Cover and Soil Sciences-Land Cover, Land Use and the Global Change, 45–80. Vick, E. P., & Bacani, V. M. (2019). Prediction of the future scenario of the expansion of areas of forestry into the Brazilian Cerrado using a cellular automata Markov chain model. Mercator (Fortaleza), 18. Wagena, M. B., Sommerlot, A., Abiy, A. Z., Collick, A. S., Langan, S., Fuka, D. R., & Easton, Z. M. (2016). Climate change in the Blue Nile Basin Ethiopia: Implications for water resources and sediment transport. Climatic Change, 139(2), 229–243. Waithaka, M., Nelson, G. C., Thomas, T. S., & Kyotalimye, M. (2013). East African agriculture and climate change: A comprehensive analysis. Intl Food Policy Res Inst. Walliman, N. (2017). Research Basics. In Research Methods (pp. 7–15). Routledge. Wang, W., Lu, H., Yang, D., Sothea, K., Jiao, Y., Gao, B., Peng, X., & Pang, Z. (2016). Modelling hydrologic processes in the Mekong River Basin using a distributed model driven by satellite precipitation and rain gauge observations. PloS One, 11(3), e0152229.

171

Wassie, A. (2017). Forest Resources in Amhara: Brief Description, Distribution and Status. In Social and Ecological System Dynamics (pp. 231–243). Springer. Weber, E. U. (2006). Experience-based and description-based perceptions of long-term risk: Why global warming does not scare us (yet). Climatic Change, 77(1–2), 103–120. Welde, K., & Gebremariam, B. (2017). Effect of land use land cover dynamics on hydrological response of watershed: Case study of Tekeze Dam watershed, northern Ethiopia. International Soil and Water Conservation Research, 5(1), 1–16. https://doi.org/10.1016/j.iswcr.2017.03.002 Whitehead, P. G., & Robinson, M. (1993). Experimental basin studies—An international and historical perspective of forest impacts. Journal of Hydrology, 145(3–4), 217–230. Wijitkosum, S. (2012). Impacts of land use changes on soil erosion in Pa Deng sub-district, adjacent area of Kaeng Krachan National Park, Thailand. Soil and Water Research, 7(1), 10–17. Williams, A. P., Funk, C., Michaelsen, J., Rauscher, S. A., Robertson, I., Wils, T. H., Koprowski, M., Eshetu, Z., & Loader, N. J. (2012). Recent summer precipitation trends in the Greater Horn of Africa and the emerging role of Indian Ocean sea surface temperature. Climate Dynamics, 39(9– 10), 2307–2328. Wischmeier, W. H., Johnson, C. B., & Cross, B. V. (1971). Soil erodibility nomograph for farmland and construction sites. https://trid.trb.org/view/125184 Wischmeier, W. H., & Smith, D. D. (1978). Predicting rainfall erosion losses. Agricultural Handbook 537. Agricultural Research Service, United States Department of Agriculture. Wold, S. (1995). PLS for multivariate linear modeling. Chemometric Methods in Molecular Design, 195– 218. Woldesenbet, T. A., Elagib, N. A., Ribbe, L., & Heinrich, J. (2017). Hydrological responses to land use/cover changes in the source region of the Upper Blue Nile Basin, Ethiopia. Science of the Total Environment, 575, 724–741. Woldesenbet, T. A., Elagib, N. A., Ribbe, L., & Heinrich, J. (2018). Catchment response to climate and land use changes in the Upper Blue Nile sub-basins, Ethiopia. Science of the Total Environment, 644, 193–206. Worku, G., Teferi, E., Bantider, A., Dile, Y. T., & Taye, M. T. (2018). Evaluation of regional climate models performance in simulating rainfall climatology of Jemma sub-basin, Upper Blue Nile Basin, Ethiopia. Dynamics of Atmospheres and Oceans, 83, 53–63. World Bank. (2017). World development indicators. World Bank. World Bank. (2020). Climate Risk Profile: Ethiopia: The World Bank Group. https://climateknowledgeportal.worldbank.org/country/ethiopia Worqlul, A. W., Ayana, E. K., Yen, H., Jeong, J., MacAlister, C., Taylor, R., Gerik, T. J., & Steenhuis, T. S. (2018). Evaluating hydrologic responses to soil characteristics using SWAT model in a paired- watersheds in the Upper Blue Nile Basin. Catena, 163, 332–341. Woznicki, S. A., Nejadhashemi, A. P., Tang, Y., & Wang, L. (2016). Large-scale climate change vulnerability assessment of stream health. Ecological Indicators, 69, 578–594. Wu, X., Shen, Z., Liu, R., & Ding, X. (2008). Land use/cover dynamics in response to changes in environmental and socio-political forces in the upper reaches of Yangtze River, China. Sensors, 8(12), 8104–8122. Wubie, M. A., Assen, M., & Nicolau, M. D. (2016). Patterns, causes and consequences of land use/cover dynamics in the Gumara watershed of lake Tana basin, Northwestern Ethiopia. Environmental Systems Research, 5(1), 8. Xie, Y., Batty, M., & Zhao, K. (2007). Simulating emergent urban form using agent-based modeling: Desakota in the Suzhou-Wuxian region in China. Annals of the Association of American Geographers, 97(3), 477–495. Yaekob, T., Tamene, L., Gebrehiwot, S. G., Demissie, S. S., Adimassu, Z., Woldearegay, K., Mekonnen, K., Amede, T., Abera, W., & Recha, J. W. (2020). Assessing the impacts of different land uses and soil and water conservation interventions on runoff and sediment yield at different scales in the central highlands of Ethiopia. Renewable Agriculture and Food Systems, 1–15.

172

Yalew, S., Mul, M. L., Van Griensven, A., Teferi, E., Priess, J., Schweitzer, C., & van Der Zaag, P. (2016). Land-use change modelling in the Upper Blue Nile Basin. Environments, 3(3), 21. Yan, B., Fang, N., Zhang, P., & Shi, Z. (2013). Impacts of land use change on watershed streamflow and sediment yield: An assessment using hydrologic modelling and partial least squares regression. Journal of Hydrology, 484, 26–37. Yan, R., Huang, J., Wang, Y., Gao, J., & Qi, L. (2016). Modeling the combined impact of future climate and land use changes on streamflow of Xinjiang Basin, China. Hydrology Research, 47(2), 356– 372. https://doi.org/10.2166/nh.2015.206 Yang, X., Zheng, X.-Q., & Chen, R. (2014). A land use change model: Integrating landscape pattern indexes and Markov-CA. Ecological Modelling, 283, 1–7. Yang, X., Zheng, X.-Q., & Lv, L.-N. (2012). A spatiotemporal model of land use change based on ant colony optimization, Markov chain and cellular automata. Ecological Modelling, 233, 11–19. Yesuf, M., Di Falco, S., Deressa, T., Ringler, C., & Kohlin, G. (2008). The impact of climate change and adaptation on food production in low-income countries: Evidence from the Nile Basin, Ethiopia. Intl Food Policy Res Inst. Yesuph, A. Y., & Dagnaw, A. B. (2019). Soil erosion mapping and severity analysis based on RUSLE model and local perception in the Beshillo Catchment of the Blue Nile Basin, Ethiopia. Environmental Systems Research, 8(1), 17. Yin, J., He, F., Xiong, Y. J., & Qiu, G. Y. (2017). Effects of land use/land cover and climate changes on surface runoff in a semi-humid and semi-arid transition zone in northwest China. Hydrology and Earth System Sciences, 21(1), 183. Yohannes, H., Soromessa, T., Argaw, M., & Dewan, A. (2020). Changes in landscape composition and configuration in the Beressa watershed, Blue Nile basin of Ethiopian Highlands: Historical and future exploration. Heliyon, 6(9), e04859. Yu, W., Zang, S., Wu, C., Liu, W., & Na, X. (2011). Analyzing and modeling land use land cover change (LUCC) in the Daqing City, China. Applied Geography, 31(2), 600–608. Zabihi, H., Ahmad, A., Vogeler, I., Said, M. N., Golmohammadi, M., Golein, B., & Nilashi, M. (2015). Land suitability procedure for sustainable citrus planning using the application of the analytical network process approach and GIS. Computers and Electronics in Agriculture, 117, 114–126. https://doi.org/10.1016/j.compag.2015.07.014 Zeleke, G. (2010). A study on mountain externalities in Ethiopia. Final Report, FAO Sustainable Agriculture and Sustainable Development, Mountain Policy Project, Addis Ababa. Zeleke, G., & Hurni, H. (2001). Implications of land use and land cover dynamics for mountain resource degradation in the Northwestern Ethiopian highlands. Mountain Research and Development, 21(2), 184–191. Zeleke, G., Kassie, M., Pender, J., & Yesuf, M. (2006). Stakeholder analysis for sustainable land management (SLM) in Ethiopia: Assessment of opportunities, strategic constraints, information needs, and knowledge gaps. Environmental Economics Policy Forum for Ethiopia (EEPFE), Addis Ababa, 96. Zerihun, M., Mohammedyasin, M. S., Sewnet, D., Adem, A. A., & Lakew, M. (2018). Assessment of soil erosion using RUSLE, GIS and remote sensing in NW Ethiopia. Geoderma Regional, 12, 83–90. Zewdu, S., Suryabhagavan, K. V., & Balakrishnan, M. (2016). Land-use/land-cover dynamics in Sego Irrigation Farm, southern Ethiopia: A comparison of temporal soil salinization using geospatial tools. Journal of the Saudi Society of Agricultural Sciences, 15(1), 91–97. https://doi.org/10.1016/j.jssas.2014.03.003 Zhang, M., Wei, X., Sun, P., & Liu, S. (2012). The effect of forest harvesting and climatic variability on runoff in a large watershed: The case study in the Upper Minjiang River of Yangtze River basin. Journal of Hydrology, 464, 1–11. Zhang, S., Fan, W., Li, Y., & Yi, Y. (2017). The influence of changes in land use and landscape patterns on soil erosion in a watershed. Science of the Total Environment, 574, 34–45.

173

Zhang, T., Zhang, X., Xia, D., & Liu, Y. (2014). An analysis of land use change dynamics and its impacts on hydrological processes in the Jialing River Basin. Water, 6(12), 3758–3782. Zhang, X. C., & Nearing, M. A. (2005). Impact of climate change on soil erosion, runoff, and wheat productivity in central Oklahoma. CATENA, 61(2), 185–195. https://doi.org/10.1016/j.catena.2005.03.009 Zhang, X.-C., Nearing, M. A., Garbrecht, J. D., & Steiner, J. L. (2004). Downscaling monthly forecasts to simulate impacts of climate change on soil erosion and wheat production. Soil Science Society of America Journal, 68(4), 1376–1385. Zhang, Y., Zhou, C., & Zhang, Y. (2007). A partial least-squares regression approach to land use studies in the Suzhou-Wuxi-Changzhou region. Journal of Geographical Sciences, 234–244. https://doi.org/DOI: 10.1007/s11442-007-0234-3 Zhou, P., Luukkanen, O., Tokola, T., & Nieminen, J. (2008). Effect of vegetation cover on soil erosion in a mountainous watershed. CATENA, 75(3), 319–325. https://doi.org/10.1016/j.catena.2008.07.010

174

APPENDIX Appendix 1: Questionnaire to be completed by Muga Watershed household Head Dear respondents this questionnaire is collected for the partial fulfillment of PhD dissertation and the principal objective of this questionnaire is to collect information about the drivers of LULC change in the Muga watershed. The study is conveyed for academic purposes. Hence, your genuine response will contribute a lot to the research output and your response is confidential. Answer questions either by choosing from the given alternatives by circling or putting “√” mark or write your answer on the space provided. Thank you in advance for your cooperation. Id No: _____ Part 1. Background Information of the Respondents 1.1. Name of Rural Kebele ______1.2. Village Name ______1.3. Sex: 1=Male 2=Female 1.4. Age ______1.5. Household size/Family size_____

Part 2. Respondents’ response about landholding and property ownership 2.1. Do you have your land? 1. Yes 2. No 2.2. If your answer is yes for question number 1, what is your current total landholding in timad (i.e. 0.25 hectare)? ______2.3. Do you feel that land is becoming scarce in your community? 1. Yes, it is becoming scarce. 2. No, it is still abundant. 3. No change 2.4. If your answer to question number 3 is yes, what is the reason? 1. Population growth. 2. The proportion of fertile land is diminishing. 3. The non-availability of off-farm activities. 4. The portion of land becomes a wasteland because of degradation. 5. Land re-distribution. 6. Other, Specify ______Part 3. Respondents’ response about Land use/ land cover change 3.1.Have you noticed any change in the land use/land cover in your area over the past 30 years? 1. Yes 2. No 3.2.If yes, what has been the change? Please indicate your response by putting this symbol (√) based on your perception level.

175

No Observed land use/land cover 1=increasing 2=decreasing 3=The same change 1 Cropland 2 Forest cover 3 Water surface 4 Grassland 5 Settlement and infrastructure 6 If others, specify Part 4: Households response about drivers/causes of the land use/land cover change 4.1. Expected drivers of land use/land cover change over the last 30 years in the Muga watershed are listed below please indicate your answer by putting this symbol (√) based on your perception level. No Drivers of land use/land cover change Answer

disagree 1=Stron y agree 2=Agree 3=Neutral 4=disagree 5=Strongly 5=Strongly 1 Increasing demand for farming. 2 Expansion of settlement and urbanization 3 Advancement of technology 4 Expansion of market-oriented crops such as onion and banana 5 Land tenure system. 6 Distribution of bushlands and communal lands to landless youths and returned solders 6 Cutting of woody species for fuelwood and charcoal making 7 Expansion of small trade and enterprise 8 Improved access to extension services such as fertilizer, improved varieties and irrigation. 9 Expansion of small irrigation schemes. 10 Soil erosion and/or soil fertility decline 11 Slope and elevation 12 The financial capital of rural farmers (poverty) 13 Improved access to basic infrastructure such as road, market, health centre, school etc. 14 Climate variability (drought, rainfall variability) 16 Absence of alternative sources of income (off-farm activities) such as trade. 19 Others, specify

4.2.Is there a deforestation problem in your locality? 1. Yes 2. No.

176

4.3.If your answer to question number 4.2 is yes, what is the major cause of deforestation in your locality? 1. Expansion of agricultural land. 4. Cutting of trees to generate income. 2. Increasing demand for firewood. 5. Other, specify ______3. Cutting of trees for construction. 4.4.What is the main sources of energy you have been used for cooking? 1. Charcoal/fuelwood. 4. Natural gas. 2. Cow dung 5. Solar energy. 3. Biogas. 6. Electricity. 4.5.Do you think that the local community can cut woody species for charcoal/fuelwood? 1. Yes 2. no 4.10. If your answer is yes, what is the reason? 1. Shortage of alternative sources of energy for cooking. 2. For charcoal making. 3. Lack of awareness about the problem of cutting a tree. 4. To use as a source of income/alternative livelihood strategies.

177

Appendices II. Guideline for interview and Focus Group Discussion 1. Is there any land use land cover change in your locality over the past 30 years? 2. Have you observed Spatio-temporal variation of land use/land cover changes in your locality over the last 30 years? 3. What do you think about the major drivers of land use land cover change in your locality? 4. What are the major forms of land modification in your locality? 5. What is/are the major environmental problems you have observed in your locality related to land use/land cover change? 6. Would you discuss your observation on climate variability (e.g., temperature and precipitation) in recent years in your locality? 7. What is/are the off-farm alternative sources of income in your locality? 8. What are the indicators of soil erosion in your locality? 9. Are there rules and regulations implemented by the local government and community to manage the illegal expansion of farmland, settlement, deforestation? 10. is/are there environmental problems related to climate variability/ climate change in your locality?

178

179