Comparative Analysis of Multiple Hydrological Models in Assessing Climate Change Impacts on the Mille Watershed, Awash Basin, Ethiopia

Abstract

As industrial and economic growth intensifies, greenhouse gases are released into the atmosphere, leading to a shift in global warming and climate change patterns. The Mille watershed faces significant challenges such as flooding, drought, irrigation, and water supply scarcity, as well as health issues stemming from climate change within the community. Thus, this study aims to assess the impact of climate change on hydrology in the Mille River, Awash River Basin, Ethiopia, utilizing multiple hydrological and climate models. The study examines three global circulation models (MIROC-6, CMCC, and MRI) operating under two shared socioeconomic pathways emission scenarios (SSP2-4.5 and SSP5-8.5) for both mid-term (near future) (2041–2070) and long-term (far future) (2071–2100) periods. Precipitation and temperature scenarios data were obtained using the CMhyd Tool and then bias-corrected using various methods based on the base time period (1985–2014). The projected annual rainfall is expected to increase by 8.91-18.68% and 8.09-18.65%, while the average temperature is projected to increase by 1.08–3.04°C and 1.59–4.05°C in the 2050s (2041–2070) and 2080s (2071–2100), respectively. The SWAT model shows daily responses with NSE (Nash–Sutcliffe efficiency) values of 0.77 for calibration and 0.79 for validation, R2 (coefficient of determination) values of 0.80 for calibration and 0.81 for validation, and PBIAS (percent bias) values of −10.6 for calibration and −8.6 for validation. Similarly, the HBV model shows NSE values of 0.683 for calibration and 0.706 for validation, R2 values of 0.705 for calibration and 0.71 for validation, and PBIAS values of −4.25 for calibration and −6.669 for validation. The results indicate a decrease in average annual streamflow ranging from −5.95% to −39.29% for SWAT and from −12.28% to −35.04% for HBV in the near future (2050s) and Far future (2080s) compared to the base period (1985–2014). The significance of this study lies in its contribution to understanding climate-hydrology interactions in a vulnerable region, providing actionable insights for adaptation planning, policy formulation, and sustainable resource management in the face of climate change. Extreme high and low changes in flow were used to quantify this impact. Therefore, based on the observed trends of decreased streamflow volume, recommendations for the study area include the development of water sources such as microdams, ponds, and water wells, implementation of water harvesting techniques, improvement of land use and land cover practices, proper utilization and management of available discharge, drought assessment, and environmental impact assessment. These measures are crucial for mitigating the adverse effects of climate change and ensuring the resilience of the region’s water resources.

This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1. Introduction

As the intensity of industrial transformation and economic growth increases, greenhouse gases, mainly carbon dioxide, are released into the atmosphere indefinitely causing global warming which leads to climate change [1]. Concentrations of carbon dioxide and other greenhouse gases in the atmosphere have steadily risen over the past century. Without implementing mitigating measures, a further multiplication of CO₂ concentration is anticipated by the middle of the next century [2]. The Earth experienced a temperature increase of 0.5°C over the course of the 20th century, a trend that was attributed to natural climatic fluctuations. In simpler terms, climate can be defined as the measured description of the average and varying amounts of temperature, precipitation, and wind over several decades [3]. As per the Intergovernmental Panel on Climate Change (IPCC), Africa is anticipated to experience a reduction in rainfall, coupled with a substantial increase in temperature, attributed to climatic variability and climate change [4]. This climate change and variability is affecting the river flow which leads to an increasing floods and drought [5]. Ethiopia boasts abundant water resources that could be efficiently harnessed to promote the socioeconomic progress of its people. Unfortunately, the underutilization of these resources has rendered Ethiopians susceptible to significant challenges, including the impacts of floods and droughts, a scarcity of clean water, and inadequate energy supply [1, 6].

The Awash River Basin stands out as one of Ethiopia’s most crucial and extensively utilized basins, while also being environmentally sensitive. This river consistently supplies irrigation water, drinking water, and hydropower consumption in a sizable and profitable to the people [7, 8]. The water resources in the region face a serious threat due to the expansion of irrigation projects in the upper reaches of the basin and the occurrence of droughts. These factors have significant impacts on the socioeconomic activities of the local population and the diversity of the ecosystem, especially in the lower reaches of the basin [9, 10]. The major tributaries of the Awash River have experienced significant environmental stress, similar to other rivers. This decline can be attributed to climate change, prolonged and frequent droughts, and the resulting insecurity related to flooding.

The Mille watershed in the Awash River Basin serves as a notable example where there is a lack of information concerning climate change, its influence on hydrological processes, and its effects on sediment loss. This is particularly noteworthy as the stream plays a significant role as a tributary to the Awash River. In general, the occurrence of summer flooding in the Mille watershed could lead to instability, despite the fact that the water resource is a crucial element in the social and economic development of the communities. The study leverages the CMIP6 Model, which represents the latest advancements in climate modeling. By utilizing this state-of-the-art model, the study aims to provide more accurate and reliable projections of future climate scenarios, enhancing the understanding of how climate change will affect hydrological processes in the Mille Watershed.

The Mille Watershed in the Lower Awash Basin of Ethiopia is a vulnerable region facing multiple hydrological challenges, including flooding, drought, and water scarcity. By focusing on this specific area, the study contributes valuable insights into the localized impacts of climate change on hydrology, which can inform adaptation strategies and resource management practices tailored to the region’s needs.

The study takes a holistic approach by considering multiple hydrological and climate models, including the CMIP6 Model, to comprehensively evaluate the effects of climate change on the hydrology of the Mille Watershed. This approach allows for a more robust assessment of the potential changes in precipitation, temperature, and streamflow, providing a more complete picture of the hydrological impacts of climate change in the region. By evaluating the effects of climate change on hydrology, the study aims to provide actionable insights for adaptation planning, policy formulation, and sustainable resource management in the Mille Watershed. These insights can help stakeholders make informed decisions to mitigate the adverse effects of climate change and enhance the resilience of the region’s water resources.

1.1. Limitations of the Study

The availability and quality of historical hydrological and meteorological data can significantly impact the study’s accuracy. In many regions, including Ethiopia, there might be gaps or inconsistencies in the data due to various reasons such as lack of infrastructure or resources for systematic data collection. The choice of Global Climate Models (GCMs) and Regional Climate Models (RCMs) introduces uncertainty. Different models might produce varying results for the same region due to differences in their structure, assumptions, and parameterizations.

2. Methodology

2.1. Description of the Study Area

The Mille River originates in the Ethiopian highlands west of Sulula in Tehuledere woreda and stands as one of the major tributaries of the Awash River Basin. It drains portions of the North Wollo and Debub Wollo of the Amhara Region, entering the lower Awash River basin in the Afar Region. The catchment area spans 4862.3 km² and is geographically situated between coordinates 532762.196N and 674509.97N UTM, as well as 1227053.29E and 1271751.47E UTM. Between the headwaters and downstream, the elevation of the Mille watershed ranges from 411 to 3548 a.m.s.l has shown Figure 1 and a mean elevation of 1979.5 m. The higher elevation is found at the western part of the watershed, Merrsa, Haike, Srinka and Wuchale while the lower elevation is found at the eastern part of the watershed, Bati, and Mille.

[figure(s) omitted; refer to PDF]

Agricultural land, bare land, eucalyptus, dispersed acacia, acacia, forest, grassland, dispersed shrub, shrub land, rocky bare land, and settlement constitute the primary land use and land cover types (Figure 2(a)). The predominant land use type is rocky bare land, constituting 45.28%.

[figure(s) omitted; refer to PDF]

The case-study watershed is characterized by twelve soil groups, namely, chromic chambisols, chromic luvisols, calcaric fluvisols, dystric nitisols, eutric cambisols, eutric regosols, haplic xerosols, leptosols, orthic solonchaks, vertic andosols, vitric cambisols, and water bodies (Figure 2(b)). The dominant soil type of the case-study area is eutric regosols (23.65%) and followed by Calcaric fluvisols (19.54%)

There is a variation in slope within the case-study watershed (Figure 2(c)). The dominant slope classes include flat or almost flat (0–3), covering 32.3% of the total area, followed by gently sloping (3–8) at 21.82%, undulating slope (15–30) at 18.78%, very steep slope (mountainous) (>30) at 13.42%, and moderately steep slope (8–15) at about 13.68%, respectively.

2.2. Data Sources and Descriptions

2.2.1. Observed Hydrometrological and Spatial Data

The meteorological and spatial data obtained were utilized as input for the SWAT model to simulate the stream flow in the case-study watershed.

(1) Metrological Data: The hydrological model development and bias correction of the GCM data in the watershed were conducted using the meteorological data. The selection of representative meteorological stations is influenced by factors such as the length of the record period, the availability of climatic variables, the distance from the catchment, and the choice of the base period used for analyzing current and future climatic characteristics. Merrsa, Srinka, Haike, Woldia, Wuchale, Werebabu, Titta, Kombolcha, Bati, and Mille are the stations situated inside and around the Mille watershed. The National Meteorological Agency of Ethiopia (NMA) supplied all stations with meteorological data used from 1987 to 2014.

(2) Soil: Soil data were collected from Ministry of Water and Energy (1 : 250,000 scale). Therefore, the soil data were extracted from Awash basins soil data from Ethiopia Basins Authority.

(3) Land Use land cover: The land use and land cover data (1 : 250,000) were acquired from the Ministry of Irrigation, Energy, and Water Resource.

(4) DEM: The digital elevation model (DEM) (30 m∗30 m) downloaded at http://earthexplorer.usgs.gov/.

(5) Stream flow: The hydrological (discharge) data were collected from the Ministry of Irrigation, Energy, and Water Resources. This data was employed for the calibration and validation of the SWAT and HBV models.

ArcGIS 4.1, ArcSWAT 10.4, Google Earth Pro, HBV, SWAT-weather database-v01803, XLSTAT 2014, Minitab 18, and SWAT-CUP were the resources and materials utilized for the study.

2.2.2. Global Climate Model with CMIP6

To predict future stream flow, it is essential to acquire forthcoming data on temperature (both maximum and minimum) and precipitation for various time horizons. The future climate projection data used in this study was obtained from the Coupled Model Intercomparison Project Phase Six (CMIP6) scenarios dataset, accessible on the Earth System Grid Federation (ESGF) website: https://esgf-node.llnl.gov/projects/cmip6/. The globally downscaled climate models from the World Climate Research Program (WCRP) were utilized in this study. In the Coupled Model Intercomparison Project Phase Six (CMIP6), there were 116 global climate models; however, some GCMs were unavailable, some models lacked scenarios data (precipitation and temperature), and others had low spatial resolution (>100 km or more), making them unsuitable for our watershed. Consequently, our watershed had access to 19 global climate models, from which three models (GCMs) were chosen under SMHI through the evaluation of climate model performance (RMSE, MAE, and PBIAS). The selection considered factors such as their resolution, ability to represent past and present climate, and relevance to other studies related to the impact of climate change on the Awash River Basin. The selected Shared Socioeconomic Pathways (SSPs) for this study were SSP4.5 and SSP8.5. The downscaling of each global climate model (GCM) from the CMIP6 was carried out using the method developed by the Swedish Meteorological and Hydrological Institute (Table 1).

Table 1

Model evaluation using different statistical parameters.

Model type	RMSE	Correlation coefficient	PBIAS
Precipitations (mm/day)
MIROC-6	2.18	0.83	−0.35
CMCC	5.10	0.52	−0.89
MRI-ESM2-0	5.9	0.38	−0.06
canESM2	7.28	0.31	−0.016
E3SM-1-0	9.3	0.30	−0.06
CAS_FGOALS-g3	10.11	0.39	−0.14
BCC-CSM2-MR	9.67	0.49	−0.18
INM-CM5-0	10.19	0.19	−0.09
NESM3	13.73	0.24	−0.09
ACCESS-ESM1-5	16.10	0.34	−0.002
NorESM2-LM	10.31	0.26	−0.39

The author draws Figure 3 of all climate models compared to baseline period as shown below three global circulation models (MIROC-6, CMCC, and MRI ESM2-0) has closet approach as compared to baseline period.

[figure(s) omitted; refer to PDF]

2.3. Data Processing and Analysis

2.3.1. Data Quality Analysis

In order to provide the model with accurate and precise data and achieve a satisfactory simulated outcome, it is essential to conduct a data quality test. Evaluations of data quality, including the identification of outliers, handling missing data, and performing homogeneity and consistency checks, are typically conducted prior to the initiation of a hydrological study (Figure 4). The best approach for filling in missing temperature and rainfall data was determined by examining and contrasting their minimum standard errors. For data consistency checks, the double mass curve method was modified. Pettitt assessed the homogeneity of the temperature and rainfall time series at a 95% significance level using XLSTAT software. In the absence of a trend test, the Mann–Kendall trend test was conducted using XLSTAT in conjunction with Excel.

[figure(s) omitted; refer to PDF]

The average of the maximum and minimum temperatures underwent testing using XLSTAT software at a significance level of 95%, while the homogeneity of the annual rainfall time series in the watershed was evaluated using nondimensional plot analysis (Figure 5).

[figure(s) omitted; refer to PDF]

2.3.2. GCMs Data Bias Correction under CMIP6

In hydrological studies addressing the impact of climate change, bias correction is achieved through various techniques. Some commonly employed methods include quantile mapping, power transformation, variance scaling, and linear scaling. Because of its ease of use and suitability for bias correction at the daily bias of temperature and precipitation data, the linear-scaling approach [11] was employed for precipitation data and variance-scaling approach was used in the study [12]. It is the easiest bias correction method that has been used in a number of studies. The multiplier and additive terms, respectively, correct the temperature and precipitation model data.$$\begin{array}{c}\text{(1)}& P_{GCM,daily}^{\mathrm{\ast }}={\mathrm{P}}_{\text{GCM},\text{daily}}\mathrm{X}\frac{\mu P_{\text{observed},\text{daily}}}{\mu P_{\text{GCM},\text{daily}}},T_{\text{GCM},\text{daily}}^{\mathrm{\ast }}={\mathrm{T}}_{\text{GCM},\text{daily}}+\mu T_{\text{observed},\text{daily}}-\mu T_{\text{GCM},\text{daily}},\end{array}$$where $P_{\text{GCM},\text{daily}}^{\mathrm{\ast }}$ future daily bias-corrected precipitation $P_{\text{GCM},\text{daily}}$ daily GCM simulated precipitation $T_{\text{GCM},\text{daily}}^{\mathrm{\ast }}$ future daily bias-corrected temperature $T_{\text{GCM},\text{daily}}$ daily GCM simulated temperature data $\mu$: mean value.

For each GCM model on Table 2, daily bias corrections were applied between the simulated and observed variables during the control period (1987–2014). Each GCM’s rainfall, maximum and minimum temperature for the extracted nearby grids were bias-corrected using power transformation and variance scaling techniques, respectively. The corrected meteorological data were then compared to the observations in the baseline (1986–2014). Consequently, each GCM output was graphically represented for all stations, independently downscaled, as depicted in Figure 6. Finally, the bias-corrected data were used as input.

Table 2

Details of global climate models considered in this study.

S/NO	Institution ID	Source ID	GCM	Resolution (KM)
1	MIROC-6	MIROC-6	MIROC-6	75
2	MRI-M	MRI-ESM2	MRI-M	50
3	CMCC	CMCC-ESM1	CMCC	50

[figure(s) omitted; refer to PDF]

After bias correction, the outputs of the Global Climate Model data were evaluated to identify the models best suited for this study. The performance of 19 GCM models was assessed using statistical metrics, including correlation coefficients, percent bias, and root mean square error (RMSE). These statistics were utilized in the evaluation of climate datasets in this study.$$\begin{array}{c}\text{(2)}& \text{Root\hspace{0.17em}mean\hspace{0.17em}square\hspace{0.17em}error\hspace{0.17em}}\text{RMSE}=\sqrt{\frac{1}{\mathrm{n}}\sum {\text{Simulated}-\text{Observed}}^2,}\text{Percent\hspace{0.17em}of\hspace{0.17em}Bais\hspace{0.17em}}\text{PBIA}=\frac{1}{\mathrm{n}}\sum {\text{Simulated}-\text{Observed}}^2\mathrm{X}\times 100,\\ \text{Correlation\hspace{0.17em}coefficient}=\frac{\text{covariance}\text{simulated},\text{observed}}{\text{St}\text{simulated}\mathrm{\ast }\text{St}\text{observed}}.\end{array}$$

After bias correction should be done global climate model (CMIP6) must be evaluated by using different methods to select the best models among them for our study area. In general, there are two methods of selecting the best fit climate model in the particular those are statistical inductor method and graphical method. Therefore, in our study, the performances of 11 GCM models were evaluated by both statistical indicators techniques (RMSE, $R^2$, and PRBIS) and graphical method compared with baseline period described below, from those three of climate models (Three Global Circulation Models (MIROC-6, CMCC, and MRI ESM2-0) operating under two shared socioeconomic pathways emission scenarios (SSP2-4.5 and SSP5-8.5) were selected.

2.3.3. Trend Analysis of Future Temperatures and Rainfall Change

Various statistical techniques have been employed in research to detect trends and other changes in hydrological and climatic variables. This group includes both parametric and nonparametric approaches. Parametric methods assume a normal distribution of variables, in contrast to nonparametric approaches. The Mann–Kendall test is robust against outliers, does not necessitate the removal of outliers before trend detection, and makes no assumptions about the distribution of the data [13].

To determine the significance trends of rainfall from (1985–2014) at the ten meteorological stations, the Mann–Kendall statistic was utilized. The results indicated that the computed P value was greater than the significance level (α = 0.05), and as a result, the null hypothesis was not rejected. Therefore, it was concluded that if there was a trend in rainfall in the study area, it was not significant. The results of maximum and minimum temperature at the five stations in the watershed indicated a significant increasing trend. After bias-correction of the projected climate data, 58% of the average annual rainfall for the projected future period in the Mille watershed showed a nonsignificant decreasing trend, whereas 42% exhibited a nonsignificant increasing trend based on the selected GCMs in both scenarios.

2.4. Hydrological Modeling

2.4.1. Soil and Water Assessment Tool (SWAT) Modeling

The semidistributed, physically based, time-continuous SWAT model was developed by the US Department of Agriculture’s Agricultural Research Service. This model requires geographic information, meteorological parameters, and hydrological information. The model divides a basin into subbasins using topographic data, and these subbasins are further segmented into minimal hydrologic response units (HRUs) based on soil type, slope, and land use. Key components of SWAT include watershed delineation, HRU creation, initiation of the SWAT editor for input preparation and execution of SWAT, and visualization of results. SWAT simulates stream flow based on the water balance equation [14].$$\begin{array}{}\text{(3)}& {\text{SW}}_t={\text{SW}}_o+{\sum }_{i=1}^t{R}_{\text{day}}-Q_{Surf}-E_{\mathrm{a}}-W_{\text{seep}}-Q_{\text{Qw}},\end{array}$$where ${\text{SW}}_{\mathrm{t}}$: final soil water content (mm); ${\text{SW}}_o$: initial soil water content on a day i (mm H₂O), t is the time (days); $R_{day}$: the amount of precipitation on a day i (mm H₂O); $Q_{\text{Surf}}$: the amount of surface runoff on a day i (mm); $E_{\mathrm{a}}$: the amount of evapotranspiration on a day i (mm H₂O); $W_{\text{seep}}$: the amount of water entering the vadose zone from the soil profile on a day i (mm H₂O); and $Q_{\text{Qw}}$: the amount of return flow on a day i (mm H₂O).

Consequently, the watershed was divided into 108 hydrologic response units (HRUs) and 27 subbasins. The delineation reports a total area of 4862.3 km² for the delineated watershed. The land use threshold of 20%, soil threshold of 10%, and slope threshold of 10% for each subwatershed area’s slope were set to their default values. In total, 108 HRUs were defined throughout the subwatershed and the entire watershed.

2.4.2. Hydrologiska Byråns Vattenbalansavdelning (HBV) Modeling

A watershed model’s ability to accurately estimate constituent yields and stream flow for a given application is evaluated using sensitivity analysis, model calibration, and model validation [15]. Sensitivity analysis is crucial during optimization. An insensitive parameter, one that doesn’t significantly change with varying values, may be kept at a fixed value. This strategy reduces the effective number of parameters to be optimized, promoting better convergence towards the optimum value of the objective function. All other parameters must be kept constant by setting the lower limit, upper limit, and starting number to the same value (Table 3). The model contains four routines for discharge simulation, among which (i) the snow routine was ignored in this study owing to the lack of snow in the study area, (ii) the soil moisture routine was ignored in this study due to the lack of soil moisture in the study area, (iii) a response function, and (iv) a routing routine.

Table 3

Parameters of HBV-Light model.

Parameter	Explanation	unit
Soil routine
FC	Maximum soil storage (storage in the soil)	mm
LP	Soil moisture value above which Etact reaches	—
BETA	Relative condition to runoff from rain, snow and	—
Response function
PERC	Maximum percolation rate from upper to lower zone	mm/day
UZL	Threshold for ko outflow	mm
$K_0$	Recession coefficient (upper storage)	1/day
$K_1$	Recession coefficient (upper storage)	1/day
$K_2$	Recession coefficient (lower storage)	1/day
Routing routine
MAXBAS	Length of triangular weight function

2.4.3. Model Evaluation: Sensitivity Analysis, Calibration, and Validation

The uncertainties of model prediction are analyses using Calibration and Uncertainty Procedures (SWAT-CUP for SWAT model and Monte Carlo method for HBV model), which is the program for integrated sensitivity analysis, calibration, and validations [16]. The Sequential Uncertainty Fitting (SUFI-2) in the SWAT-CUP is used for model sensitivity analysis, calibration, and validation. Sensitivity analysis helps to identify parameters that strongly influence the flow process [16]. The global sensitivity analysis procedure is used for the evaluation of stream flow and sediment parameters’ relative sensitivity using the Latin hypercube “one-at-a-time’’ regression systems. The coefficient of a parameter over its standard error (t-stat) was used for parameter sensitivity and ranking, while the significance of the sensitivity is determined by the p value.

The performance of the model simulation is determined by comparing the observed stream flow and sediment yield against their simulated data. Statistics like coefficients of determination ($R^2$), Nash–Sutcliffe efficiency (NSE), and percent bias (PBIAS) were used to test the goodness of fit between the simulated and observed values. By choosing model parameters, the best possible correspondence between simulated and observed runoff from a catchment must be found. The entire amount of observed flow data that was available from 1985 to 2014 at the gauged station was used, consisting of five years (2010–2014) for validation and 23 years (1987–2009) for calibration.

2.5. Impact of Climate Change on Stream Flow

The model was executed over two-time horizons, the baseline (1987–2014), and future climate change scenarios after calibration and validation using observed climate data (2050s and 2080s). The study concluded by investigating the impact of climate change on average monthly seasonal stream flow, annual stream flow, and maximum and minimum stream flow in comparison to the base period flow, based on the simulated water balance under baseline and climate change scenarios.

Climate change will significantly impact the regime of flow extremes. It is crucial to examine the effects of climate change on stream flow extremes (low or high) events to mitigate the associated risks. Utilizing average discharge data (daily, monthly, and annual), flow duration curves represent cumulative frequency distributions illustrating the percentage of time a specific discharge is equal to or greater than during a given period. The classification comprises three categories: (a) high flow segment (0–20%), indicating a catchment response to a significant rainfall event; (b) mid-flow segment (20–70%), indicating flows regulated by moderate precipitation events combined with medium-term base flow; and (c) low-flow segment (70–100%), indicating a catchment response dominated by long-term base flow during prolonged dry periods [17]. The general methodology is described in Figure 7.

[figure(s) omitted; refer to PDF]

3. Results and Discussion

3.1. Projected of Climate Change

3.1.1. Precipitation Changes

To analyse the variation in characteristics generated by GCM outputs, changes in rainfall between the baseline and future periods were considered. The projected mean monthly rainfall changes in the Mille watershed varied at different time horizons for each GCM (MIROC-6, MRI, and CMCC) under SSPs. In the near term or 2050s (2041–2070), the projected mean monthly rainfall changes in both scenarios ranged from −4.39 to 5.48%, −4.03 to 6.21%, and −3.97 to 6.80% under SSP4.5 and −4.31 to 6.94%, 3.88 to 7.37%, and −3.62 to 7.20% under SSP8.5, compared to the baseline period for MIROC-6, CMCC, and MRI, respectively. Similarly, changes in the late term or 2080s (2071–2100) ranged from −4.26 to 5.98%, −1.18 to 6.28%, and −4.90 to 4.98% under SSP4.5 and −4.44 to 6.10%, −0.74 to 7.36%, and 0.97 to 8.59% under SSP8.5 for MIROC-6, CMCC, and MRI, respectively, as shown in Figure 8. The annual maximum increment of rainfall was (15.52%, 18.67%) and (16.41%, 18.65%) in the near and late term under CMCC-SSP4.5 and CMCC-SSP8.5, respectively, compared to the baseline period, as illustrated in Figure 8.

[figure(s) omitted; refer to PDF]

3.1.2. Temperature Change

The maximum monthly percentage change in maximum temperature was observed in September (3.8% and 4.5%) and (5.31% and 5.67%) in the 2050s and 2080s under CMCC-SSP4.5 and SSP8.5, respectively. Similarly, the seasonal maximum minimum temperature was observed in autumn (3.17% and 3.78%) and (4.13% and 5.60%), respectively. The annual maximum temperature was observed in autumn (2.06% and 2.5%) and (2.84% and 4.31%), respectively, as shown in Figure 9.

[figure(s) omitted; refer to PDF]

The maximum monthly percentage change in minimum temperature was observed in January (3.67% and 3.9%) and in December (3.95% and 4.55%) in the 2050s and 2080s under CMCC-SSP4.5 and SSP8.5, respectively. Similarly, the seasonal maximum minimum temperature was observed in autumn (4.43% and 4.78%) and (4.41% and 5.87%), respectively, and the annual maximum temperature was observed in autumn (3.05% and 3.78%) and (3.44% and 3.34%), respectively, as shown in Figure 10.

[figure(s) omitted; refer to PDF]

3.2. Hydrological Modeling

3.2.1. SWAT Model Calibration and Validation

The sensitivity analysis was conducted to determine the order of sensitivity of stream flow to input parameters. Sensitivity analysis, utilizing daily flow data from 1985 to 2006, was performed using global sensitivity analysis. As a result, sixteen critical parameters were identified in the global sensitivity analysis due to their control over the hydrological processes of the examined area (Table 4). However, for the Mille watershed, the soil evaporation compensation factor (ESCO), soil depth, and SCS Curve Number II (CN2) were shown to be more important than other parameters.

Table 4

Parameter sensitivity analysis of SWAT model.

No	Parameters	Description	Fitted-value	Min_V	Max_V	$p$ value	t-stat value
1	R__CN2.mgt	Initial SCS runoff curve number for moisture condition I	−0.025	−0.25	1.25	0.001	−13
2	V__ESCO.hru	Soil evaporation compensation factor	0.45	0	1	0.001	−4.333
3	R__SOL_Z(…).sol	Soil depth (mm)	0.175	−0.25	0.25	0.002	2.512
4	V__LAT_TTIME.hru	Lateral flow travel time	114.012	0.25	120.3	0.046	2.24
5	R__CANMX.hru	Maximum canopy storage	2.17	0	10	0.12	1.736
6	V__GWQMN.gw	Shallow aquifer requires for return flow`	1.3	0	500	0.125	−1.71
7	V__RCHRG_DP.gw	Deep aquifer percolation fraction	0.38	0	1	0.15	1.71
8	V__ALPHA_BF.gw	Base flow alpha factor for bank storage	0.583	0	1	0.15	1.61
9	R__EPCO.hru	Plant uptake compensation factor	0.35	0	1	0.15	1.62
10	V__GW_DELAY.gw	Groundwater recap coefficient	175	0	500	0.2	−1.41
11	R__CH_N2.rte	Manning’s “n” value for the main channel	0.3833	0	1	0.28	−1.31
12	R__SOL_AWC(…).sol	Soil available water capacity (mm water/mm soil)	0.05833	−0.25	0.25	0.48	−1.21
13	R__REVAPMN.gw	Percolation to the deep aquifer	1.571	0	500	0.58	0.721
14	R__BIOMIX.mgt	Biological mixing efficiency	0.783	0	1	0.65	−0.6
15	V__ALPHA_BNK.rte	Base flow alpha factor for bank storage	0.293	0	1	0.66	−0.45
16	R__SOL_K(…).sol	Saturated hydraulic conductivity	0.125	−0.25	0.25	0.7	0.45

Note. (i) R-sensitive parameter indicates multiply by 1+ given fitted value, V-sensitive parameter replace the value by the given fitted value.

Flow forecasts were calibrated using daily flow data from 1985 to 2006 and validated using flow data from 2007 to 2014, employing the SUFI-2 Al algorithm. According to the model performance evaluation the daily coefficient of determination, Nash–Sutcliffe coefficient, $R^2$, percentage of bias and RSR were 0.80, 0.77, −10.6 and 0.47 for the calibration period, respectively, and 0.81, 0.79, −9.16, and 0.60 for the validation period (Figure 11).

[figure(s) omitted; refer to PDF]

3.2.2. HBV Model Calibration and Validation

The HBV model has more than 13 parameters, but seven model parameters are used to control the total volume and shape of the hydrograph. Out of those seven parameters, five were found to be the most sensitive: field capacity (Fc), soil drainage (BETA), limiting for evapotranspiration (LP), storage coefficient, and percolation. The model parameters of the soil moisture routine, response function, and routing routine were sensitive to predict discharge for the Mille watershed (Table 5).

Table 5

Parameter sensitivity of HBV.

Parameter	Lower limit	Upper limit	Optimized value	Rank
FC	100	1500	1053	1
BETA	0	5	2.52	2
LP	0.1	1	0.65	3
K2	5 × 10−5	1	0.037	4
UZL	0	70	48.76	5
MAXBAS	1	2.5	2.18	6

Parameter optimization was carried out for the Monte Carlo runs of the model by providing lower and upper ranges to obtain an optimum value for the three routines (Soil Moisture, Response, and Routing), except for the snow routine, which was excluded due to the absence of snow in the watershed. The mean difference between the observed and simulated values was 61 mm/year, and the model performance was evaluated using the Nash–Sutcliffe efficiency criteria, resulting in 0.685 for daily data, as shown in Figure 12.

[figure(s) omitted; refer to PDF]

3.3. Future Climate Change Impact on Stream Flow Using Hydrological Model

3.3.1. Stream Flow for SWAT

The maximum increment in monthly stream flow was observed in the range of 42.12–53.78%, 40.11–59.19%, and 47.68-82.51%, 61.65-67.84% under SSP4.5 and SSP8.5 for all GCMs in the 2050s- and 2080s-time horizons, respectively. Similarly, the maximum decrement in monthly stream flow was observed in the range of 34.42–41.11, 34.44-43.71, and 58.3–61.5, 50.92–61.22 under SSP4.5 and SSP8.5 for all GCMs in the 2050s- and 2080s-time horizons, respectively. The average annual flow showed the maximum decrement in CMCC (39.29% and 35.04%) under SSP4.5 and SSP8.5 in the near and late-term horizons, respectively (Figure 13). The annual stream flow also exhibited a decreasing trend in all GCMs time horizons under all SSPs scenarios.

[figure(s) omitted; refer to PDF]

3.3.2. Stream Flow for HBV

The maximum increment in monthly stream flow was observed in March (88.72 and 88.06%) under CMCC-SSP4.5 in the 2050s and 2080s. Similarly, the maximum increment in stream flow was observed in April (87.12 and 78.3%) under SSP8.5 in 2050s and 2080s. The maximum decrement in stream flow was observed in August (35.90) and in July (53.78) under CMCC and MRI-SSP4.5 in 2050s and 2080s, respectively. Similarly, the maximum decrement is 28.81 and 67.42% occurred in July under CMCC and MRI-SSP8.5 in the 2050s and 2080s. Generally, the annual stream flow shows a decreasing trend in all GCMs time horizons under all SSPs scenarios.

In all GCM scenarios for the Mille River watershed, the change in maximum annual flow for all future time periods (2050s and 2080s) exhibited a decreasing trend in both hydrological models for both SSP4.5 and SSP8.5 compared to the base period maximum flow (1985–2014) (Table 6).

Table 6

Percentage change of projected extreme stream flow as compared to the base period.

Model	SWAT-model				HBV-model
Time	Near-term		Late-term		Near-term		Late-term
Scenarios	SSP2-4.5	SSP5-8.5	SSP2-4.5	SSP5-8.5	SSP2-4.5	SSP5-8.5	SSP2-4.5	SSP5-8.5
MIROC-6	−38.6	−42.98	−43.05	−36.47	−31.22	−44.81	−38.52	−19.08
CMCC	−23.25	−47.37	−4.05	−40.86	−5.69	−0.89	−12.87	−1.9
MRI	−14.47	−7.89	−1.43	−29.9	−42.16	−2.78	−12.87	−1.9

Based on the flow duration curve, the impacts of climate change on extreme occurrences in the Mille watershed area have also been assessed. The results clearly indicate a reduction in stream flow size for all GCMs under both emission scenarios, and the curve becomes substantially flattened between Q30 and Q70 (mid-flow). This suggests that the groundwater contribution reaching the stream is significant and supports sustainability throughout the year. Additionally, FDCs show significant decreases in high flows (5% exceedance), mid-flows (30%–70% exceedance), and low flows (95% exceedance) during 2041–2100 compared to the base period (1987–2014) (Table 7).

Table 7

Percentage change of high flow and low flow for both SWAT and HBV models with respect to the baseline under both scenarios (SSP4.5 and SSP8.5) in the Mille watershed.

Model		SWAT model				HBV model
Scenarios	GCM	Near-term		Late-term		Near-term		Late-term
Probability		Q5	Q95	Q5	Q95	Q5	Q95	Q5	Q95
SSP2-4.5	MIROC-6	−34.18	21.7	−21.46	−1.36	−48.79	−13.51	−43.43	2.69
	CMCC	−23.37	−8.11	−6.26	−33.27	−41.88	−5.1	−41.89	−16.23
	MRI	−20.31	19.06	−23.65	−41.26	−40.19	−2.72	−41.89	−13.32
SSP5-8.5	MIROC-6	−19.28	−11.15	−15.66	−11.15	−46.13	−2.71	−33.81	−21.49
	CMCC	−9.17	−22.62	−18.32	−22.62	−29.47	−10.77	−37.3	−7.96
	MRI	−15.01	−5.16	−17.12	−5.16	−25.39	−13.28	−37.19	5.18

3.4. Comparisons of SWAT and HBV Hydrological Mode

Comparisons of SWAT and HBV Hydrological Mode. (see Tables 8 and 9).

Table 8

Performance of SWAT and HBV Hydrological Mode under different climate change scenario.

Scenarios	GCM	SWAT		HBV
		Near-term	Late-term	Near-term	Late-term
SSP4.5 (annual flow (%))	MIROC-6	−22.64	−29.28	−21.48	−19.41
	CMCC	−39.29	−19.43	−8.89	−17.37
	MRI	−17.27	−5.95	−9.28	−17.37
SSP8.5 (Annual flow (%))	MIROC-6	−25.47	−21.67	−21.76	−7.1193
	CMCC	−31.04	−35	6.115	−4.611
	MRI	−5.95	−22.99	6.028	−4.322

Table 9

Calibration and validation value of SWAT and HBV hydrological model.

Model	Calibration			Validation
	R2	NSE	PBIAS	R2	NSE	PBIAS
SWAT	0.8	0.77	−10.6	0.81	0.79	−8.16
HBV	0.705	0.683	−4.25	0.713	0.706	−6.669

3.5. Discussion

The findings of this study provide valuable insights into the potential effects of climate change on the hydrology of the Mille watershed in the Lower Awash Basin, Ethiopia. Through the utilization of the CMIP6 Model, several key observations have been made, which warrant further discussion and comparison with previous related literature. The projected increases in annual rainfall and average temperatures have significant implications for the hydrological dynamics of the Mille Watershed. The observed increases in precipitation align with the findings of several previous studies that have documented a general trend of rising rainfall in the East African region as a result of climate change [18, 19]. These increases in rainfall are expected to influence the frequency and intensity of flooding events in the watershed, exacerbating existing challenges related to water management and infrastructure. Similarly, the projected rise in average temperatures is consistent with global climate change trends and has been well-documented in previous literature [20, 21]. Higher temperatures can accelerate evaporation rates, leading to increased water stress and evapotranspiration in the watershed. This can further exacerbate drought conditions and reduce water availability for agricultural and domestic use, posing significant challenges for local communities reliant on water resources from the Mille River. The hydrological models employed in this study provide valuable insights into the potential impacts of climate change on streamflow dynamics in the Mille Watershed. The observed decreases in average annual streamflow, particularly in the near and far future scenarios, underscore the vulnerability of the watershed to changing climatic conditions. These findings are consistent with previous studies that have highlighted the sensitivity of hydrological systems to shifts in precipitation patterns and temperature regimes [22].

4. Conclusion

This study undertook an analysis of the effects of climate change on the streamflow and their corresponding ensemble mean, through the application of a physically based distributed hydrologic model known as Soil and Water Assessment Tool (SWAT) and HBV model in the Mille River, Awash River Basin, Ethiopia. The future climate was downscaled from the ESGF using three climate models (MIROC-6, CMCC, and MRI) in CMIP6 under two scenarios (SSP4.5 and SSP8.5). The projected annual rainfall trend indicates a 58% negative and 42% positive trend. The maximum annual projected rainfall increased by 18.67% and 18.65% under SSP4.5 and SSP8.5 in the 2050s and 2080s, respectively. The maximum average temperature also increased by 3.04°C and 4.05°C under CMCC-SSP8.5 in the 2050s and 2080s compared to the base period (1987–2014). Consequently, for the SWAT model, the maximum streamflow decreased by −39.29% and −35.04% under CMCC-SSP4.5 and SSP8.5 in the 2050s and 2080s. For HBV, it decreased by −21.76% and −19.41% under MIROC-SSP8.5 and SSP4.5 in the 2050s and 2080s, respectively. The maximum high flow extremes (Q5) decreased by −34.18% and −19.28% for SWAT and −48.79% and −46.13% for HBV under MIROC-SSP4.5 in the 2050s- and 2080s-time horizons.

Consequently, the catchment is highly susceptible to the impacts of climate change, such as warming and crop failures resulting from prolonged dry seasons. The observed decrease in runoff within the catchment has contributed to enhanced water security. The study’s analysis indicates that the model simulations provide adequate estimates of the impacts of climate change in the Mille River. To enhance our understanding of the present and future climate dynamics, urgent attention is required to improve the availability and quality of hydroclimatic data in the region. Therefore, these findings can inform watershed management and planning, environmental impact and drought assessment, artificial cloud rain, water harvesting, and the creation of water sources should be implemented in the Mille watershed.

Ethical Approval

The authors approved that there are no ethical issues related with this title.

Consent

The authors have consented to participate.

Disclosure

The authors acknowledge that the manuscript has been previously published as a preprint and updated the manuscript to reflect this, including the statement: A preprint of this manuscript has been published [23].

Authors’ Contributions

A.B.N., M.A., D.W.A., S.K., T.D.M., and A.E. contributed to the study conception and design. A.E.A., A.B.N., and D.W.A. proposed the methodology and contributed to data curation. A.B.N., G.F., and A.E.A. wrote the original draft. T.D.M., S.K., A.B.N., D.W.A., G.F., and M.A. reviewed and edited the manuscript. A.E.A., S.K., A.E., and M.A. supervised the study. All authors have read and agreed to the published version of the manuscript.

Acknowledgments

The authors would like to extend their gratitude to the Amhara National Metrological Agency (NMA) and Ministry of Water and Energy of Ethiopia for the provision of data and to Kombolcha Institute of Technology, Wollo University.

References

[1] K. Hailemariam, "Impact of climate change on the water resources of Awash River Basin, Ethiopia," Climate Research, vol. 12, pp. 91-96, DOI: 10.3354/cr012091, 1999.

[2] D. Singh, M. Vardhan, R. Sahu, D. Chatterjee, P. Chauhan, S. Liu, "Machine-learning- and deep-learning-based streamflow prediction in a hilly catchment for future scenarios using CMIP6 GCM data," Hydrology and Earth System Sciences, vol. 27 no. 5, pp. 1047-1075, DOI: 10.5194/hess-27-1047-2023, 2023.

[3] IPCC, "An introduction to simple climate model used in the IPCC Second Assessment report: contribution of working group I to the second assessment report of the intergovernmental panel on climate change," 1997. https://www.ipcc.ch/site/assets/uploads/2018/02/ipcc_sar_wg_I_full_report.pdf

[4] A. W. Worqlul, J. Jeong, Y. T. Dile, J. Osorio, P. Schmitter, T. Gerik, R. Srinivasan, N. Clark, "Assessing potential land suitable for surface irrigation using groundwater in Ethiopia," Applied Geography, vol. 85,DOI: 10.1016/j.apgeog.2017.05.010, 2017.

[5] J. Bogner, R. Pipatti, S. Hashimoto, C. Diaz, K. Mareckova, L. Diaz, P. Kjeldsen, S. Monni, A. Faaij, G. Qingxian, Z. Tianzhu, A. A. Mohammed, R. T. M. Sutamihardja, R. Gregory, "Mitigation of global greenhouse gas emissions from waste: conclusions and strategies from the intergovernmental panel on climate change (IPCC) fourth assessment report. Working group III (mitigation)," Waste Management and Research: The Journal for a Sustainable Circular Economy, vol. 26 no. 1, pp. 11-32, DOI: 10.1177/0734242X07088433, 2008.

[6] R. Sahu, R. D. Gupta, "Snow cover analysis in chandra Basin of western Himalaya from 2001 to 2016," Applications of Geomatics in Civil Engineering, pp. 557-566, 2020.

[7] N. Tessema, A. Kebede, D. Yadeta, "Modelling the effects of climate change on streamflow using climate and hydrological models: the case of the Kesem sub-basin of the Awash River basin, Ethiopia," International Journal of River Basin Management, vol. 19 no. 4, pp. 469-480, DOI: 10.1080/15715124.2020.1755301, 2021.

[8] D. Bekele, "Analysis of rainfall trend and variability for agricultural water management in Awash River Basin, Ethiopia," ,DOI: 10.2166/wcc.2016.044, .

[9] R. Mizuta, H. Yoshimura, H. Murakami, M. Matsueda, H. Endo, T. Ose, K. Kamiguchi, M. Hosaka, M. Sugi, S. Yukimoto, S. Kusunoki, A. Kitoh, "Climate simulations using MRI-AGCM3 . 2 with 20-km grid," Journal of the Meteorological Society of Japan. Ser. II, vol. 90, pp. 233-258, DOI: 10.2151/jmsj.2012-A12, 2012.

[10] B. G. Balcha, "Environmental degradation and conflict in Borkena area," International Conference on Contemporary Development Issues in Ethiopia, .

[11] C. Teutschbein, J. Seibert, "Bias correction of regional climate model simulations for hydrological climate-change impact studies: Review and evaluation of different methods," Journal of Hydrology, vol. 456-457, pp. 12-29, DOI: 10.1016/J.JHYDROL.2012.05.052, 2012.

[12] Z. M. Nigatu, Hydrological Impacts of Climate Change On Lake Tana’s Water Balance Hydrological Impacts of Climate Change On Lake Tana’s Water Balance, 2013.

[13] G. S. Takele, G. S. Gebrie, A. G. Gebremariam, A. N. Engida, "Future climate change and impacts on water resources in the Upper Blue Nile basin," Journal of Water and Climate Change, vol. 13 no. 2, pp. 908-925, DOI: 10.2166/wcc.2021.235, 2022.

[14] J. G. Arnold, R. Srinivasan, R. S. Muttiah, P. M. Allen, "Continental scale simulation of the hydrologic balance," Jawra Journal of the American Water Resources Association, vol. 35 no. 5, pp. 1037-1051, DOI: 10.1111/j.1752-1688.1999.tb04192.x, 1999.

[15] C. B. Field, V. Barros, T. F. Stocker, Q. Dahe, D. J. Dokken, K. L. Ebi, M. D. Mastrandrea, K. J. M. Pauline, G.-K. Plattner, S. K. Allen, M. Tignor, P. M. Midgley, "Special report of the intergovernmental panel on climate change," Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation Special, 2011.

[16] K. C. Abbaspour, E. Rouholahnejad, S. Vaghefi, R. Srinivasan, H. Yang, B. Kløve, "A continental-scale hydrology and water quality model for Europe: calibration and uncertainty of a high-resolution large-scale SWAT model," Journal of Hydrology, vol. 524, pp. 733-752, DOI: 10.1016/j.jhydrol.2015.03.027, 2015.

[17] D. B. Boorman, C. E. M. Sefton, "Recognising the uncertainty in the quantification of the effects of climate change on hydrological response," Climatic Change, vol. 35, pp. 415-434, DOI: 10.1023/A:1005372407881, 1997.

[18] T. Dinku, K. Hailemariam, R. Maidment, E. Tarnavsky, S. Connor, "Combined use of satellite estimates and rain gauge observations to generate high‐quality historical rainfall time series over Ethiopia," International Journal of Climatology, vol. 34 no. 7, pp. 2489-2504, DOI: 10.1002/joc.3855, 2014.

[19] C. Funk, P. Peterson, M. Landsfeld, D. Pedreros, J. Verdin, S. Shukla, G. Husak, J. Rowland, L. Harrison, A. Hoell, J. Michaelsen, "The climate hazards infrared precipitation with stations—a new environmental record for monitoring extremes," Scientific Data, vol. 2 no. 1, pp. 150066-150121, DOI: 10.1038/sdata.2015.66, 2015.

[20] Z. T. Segele, P. J. Lamb, "Characterization and variability of Kiremt rainy season over Ethiopia," Meteorology and Atmospheric Physics, vol. 89 no. 1-4, pp. 153-180, DOI: 10.1007/s00703-005-0127-x, 2005.

[21] J. G. Canadell, C. Le Quéré, M. R. Raupach, C. B. Field, E. T. Buitenhuis, P. Ciais, T. J. Conway, N. P. Gillett, R. A. Houghton, G. Marland, "Contributions to accelerating atmospheric CO2 growth from economic activity, carbon intensity, and efficiency of natural sinks," Proceedings of the National Academy of Sciences, vol. 104 no. 47, pp. 18866-18870, DOI: 10.1073/pnas.0702737104, 2007.

[22] T. Alemayehu, "Modeling the IMACT of land use/landcover dynamics on stream flow and sediment yield," 2017. https://core.ac.uk/download/pdf/199939052.pdf

[23] A. Amognehegn, A. Nigussie, M. Assega, A. Adamu, G. Adamu, N. Assefa, A. E. Amognehegn, A. B. Nigussie, M. Ayana, A. Y. Adamu, G. Feleke, N. Yassab Assefa, "Analysis of future drought characterization under multiple drought indices and climate change impact conditions for developing best strategic measures with CMIP-6 climate model in Mille watershed, lower Awash Basin, Ethiopia," , 2023. https://sciety.org/articles/activity/10.21203/rs.3.rs-3705593/v1