Modeling climate change projection and its impact on the streamflow in the Yadot watershed, Genale Dawa basin, Ethiopia

There is scientific consensus that the global average temperature increased during the 20th century and will continue for the next several decades. According to the sixth assessment report of the Intergovernmental Panel on Climatic Change, global surface temperature has increased by 0.99 °C from 1850 to 1900 to the first two decades of the 21st century (2001–2020) and by 1.09 °C from 1850–1900 to 2011–2020 (IPCC 2021). These changes in global temperature have been accompanied by changes in climate patterns in various ways (Feng et al. 2014). Water resources are currently under severe pressure because of the impacts of climate change and human activities, which include land use change, increasing population growth, and economic development (IPCC 2013). Human-induced climate change has driven detectable changes in the global water cycle since the mid-20th century, and it is projected to cause substantial further changes at both global and regional scales (IPCC 2021). Changes in climate may cause changes in hydrological processes such as evapotranspiration, surface runoff, and flood events (Neupane & Kumar 2015). Temperature and precipitation patterns are two significant characteristics of climate change that have a direct impact on nearly all other hydrological responses (Azari et al. 2016). The changes in temperature and precipitation have a direct consequence for the quantity of the evapotranspiration component, and on both the quality and quantity of the streamflow component. Therefore, understanding and quantifying their respective influence is of great importance to water resource management and socioeconomic activities as well as policy and planning for sustainable development (Yang et al. 2017).

East Africa stands as one of the hot spot locations for the impacts of climate change (IPCC 2022). The future projected change in annual precipitation and temperature shows an increase in most parts of East Africa such as Ethiopia, Kenya, Uganda, and Tanzania (IPCC 2014; Gebrechorkos et al. 2023). Future temperature projections indicate a potential increase of 1.8–4.3 °C by 2100 in the region (Trisos et al. 2022). Similarly, rainfall will increase in the 21st century relative to the 20th century (Ongoma et al. 2018). Generally, despite the inherent uncertainties that exist in GCMs, most studies agree that the rainfall over the larger part of the Great Horn of Africa is likely to increase in the 21st century (Shongwe et al. 2011; Christensen et al. 2013; Tierney et al. 2015). Accordingly, the projected change in annual streamflow is projected to increase in the major river basins of Ethiopia, Kenya, and Uganda in the 2020s, 2050s, and 2080s (Gebrechorkos et al. 2023).

Recent modeling studies on the impact of climate change on hydrological and meteorological parameters reported a future increase in temperature and streamflow (Li & Fang 2021; Lotfirad et al. 2021, 2023; Shokouhifar et al. 2022). For example, Li & Fang (2021) studied climate change impacts on the streamflow for the Mun River in the Mekong Basin, Southeast Asia and reported a consistent increase in T_min and T_max in three future time slices (2030s, 2060s, and 2080s) under three RCPs (2.6, 4.5, and 8.5). Similarly, a study of climate change effects on flood frequency by Shokouhifar et al. (2022) in the Hablehroud basin, north-central Iran reported an increase in temperature and rainfall and a subsequent increase in runoff and flooding. Despite there being a consistent warming trend, the sign and magnitude of future streamflow changes vary between climate simulations and river basins, mainly due to the uncertainties in the hydrologic projections (Hirpa et al. 2019).

Ethiopia has become warmer over the past century and future climate change will bring further warming over the next century (Gashaw & Mahari 2014). In Ethiopia, over the past few decades, changes in rainfall and temperature have affected the various components of the hydrological cycle in major river basins (Gebremichael et al. 2013). In Ethiopia, the main climate change manifestations such as frequent and severe floods and hydrological droughts have become a common problem. Future predictions also indicate that changes in climate will lead to recurrent droughts and heavy rainfall in different parts of Ethiopia, reducing the amount of land that can be used for agriculture and decreasing crop productivity (Admassie & Abebaw 2021). Almost all food crops and most industrial crops in Ethiopia are produced by rain-fed agriculture, which is very sensitive to climate variations. Water scarcity, wildfires, flooding, and droughts have all been linked to the possible consequences of climate change on the country's hydrological cycles (Cheng et al. 2017).

Several studies have examined the impacts of climate change on the streamflow of catchments and basins in Ethiopia (Abraham et al. 2018; Boru et al. 2019; Negewo & Sarma 2021; Tarekegn et al. 2021; Abdulahia et al. 2022; Daniel & Abate 2022; Edamo et al. 2022; Gurara et al. 2023). Most of these studies have assessed the projections of rainfall and temperature and their impact on streamflow. Heterogeneous precipitation changes and consistent warming trends caused variations in streamflow projection in different watersheds and basins of Ethiopia (Boru et al. 2019; Hirpa et al. 2019; Chakilu et al. 2020; Bekele et al. 2021; Edamo et al. 2022; Gurara et al. 2023). For instance, the mean annual and wet season (Kiremt) streamflow projections decreased in the Blue Nile, Upper Awash, OmoGhibe and Baro River basins (Hirpa et al. 2019; Daba & You 2020; Orkodjo et al. 2022). However, projected annual and wet season (Kiremt) streamflow increased in the Lake Tana, Tekeze and Genale river sub-basins (Gizaw et al. 2017; Fentaw et al. 2018; Chakilu et al. 2020; Abdulahia et al. 2022). The dry season streamflow is also expected to decrease in most of the sub-basins, such as in the Arjo-Didessa, Upper Awash, and Lake Tana sub-basins (Bekele et al. 2021; Abdulahia et al. 2022; Chakilu et al. 2022). Thus, such varied streamflow response to climate between river basins and seasons highlights the importance of further research on different basins and watersheds in different seasons to help plan adaptation options at watershed scale. Understanding climate change and its impact on hydrological variability is critical for water management.

Surface water resource sustainability is still a current area of focus due to the significant impact of climatic variability, which can be better mitigated by understanding the past and predicting its future consequences (Boko et al. 2020). Regional hydrology is extremely vulnerable to changing climatic conditions; consequently, it is crucial to evaluate climate change projections in order to determine if the availability of water will likely be sufficient to meet demand in the future (Tessema et al. 2021). According to Worqlul et al. (2018) and Sesana et al. (2019), the future availability of streamflow could have an impact on agricultural production, socioeconomic systems, and environmental sustainability. Therefore, it is necessary to understand how climate change may affect future streamflow fluctuations in order to adapt various mitigation techniques for planning water resource management.

Possible future changes in the precipitation and temperature extremes can be predicted by global circulation models (GCMs). However, GCMs face large uncertainties and biases due to their mathematical formulation, spatial resolution, initial condition and forcings, and model calibration processes, which limits their application in regional and local scale impact assessment studies (Khan et al. 2018; Salman et al. 2018; Kamruzzaman et al. 2021). Thus, climate change impact studies usually require climate projection information at finer spatial and temporal scales than the typical GCM grid resolutions and thus, dynamic downscaling with regional climate models (RCMs) is capable of addressing this scale gap (Givati et al. 2019; Sanjay Shekar & Vinay 2021).

The Soil and Water Assessment Tool (SWAT) is a physical-based, semi-distributed, basin-scale hydrological model and is frequently employed in research to simulate the hydrologic processes of watersheds with an emphasis on the impact of climate change (Zhang et al. 2016). Despite this, machine learning methods are promising for simulating the complex hydrological processes under climate change conditions (Anaraki et al. 2021, 2023; Nguyen et al. 2023). The SWAT hydrologic model continues to be a widely applicable model. For example, various recent studies have examined the impacts of climate change on the streamflow using the SWAT model in Ethiopia (Negewo & Sarma 2021; Tarekegn et al. 2021; Abdulahia et al. 2022; Daniel & Abate 2022; Edamo et al. 2022; Gurara et al. 2023) and elsewhere (Li & Fang 2021; Ismail et al. 2022; Reshma and Arunkumar, 2023). Moreover, based on the recent comparative study of SWAT and HEC-HMS for streamflow simulation in the Lower Godavari Basin and sub-humid tropical Hemavathi catchment, India, SWAT has shown exceptional simulating ability in calibration and validation (Sanjay Shekar & Vinay 2021; Vogeti et al. 2023).

Thus, the specific objectives of this study are to assess the projected climate for the near- and mid-term and to examine the separate and combined impact of climate and LULC changes during the near-term (2021–2050) and mid-term (2051–2080) periods under medium (RCP4.5) and high (RCP8.5) emission scenarios using SWAT hydrological model and downscaled climatic data from CORDEX-RCMs output.

Two different wet seasons occur in Ethiopia's southern regions as the ITCZ passes through on its way to its southern position. With two wet seasons, the entire Genale River sub-basin is subjected to a ‘bi-modal’ rainfall regime. Rainfall is most abundant from March to May, with a peak flow in May, and then again from September to November, with high rainfall in October. The watershed's mean annual rainfall ranges from 975 mm in Rira to 1,059 mm in Delo Mena, with a mean annual areal rainfall of 1,029 mm. It is commonly acknowledged that temperature distribution in the region is substantially influenced by elevation. At Delo Mena and Rira meteorological stations, the mean annual minimum and maximum temperatures range from 21.6 to 28.72 °C and 9.78 to 15.62 °C, respectively. During the study period, the watersheds' mean annual temperature ranged from 12.7 to 25.16 °C.

The daily precipitation, maximum and minimum temperature data used for this study from Delo Mena and Rira meteorological stations located within and around the Yadot watershed were obtained from Ethiopia's National Meteorological Agency. The climate data cover 31 years, from January 1985 to December 2015.

The future daily precipitation, maximum and minimum temperatures were obtained from the Coupled Model Intercomparison Project (CMIP5), which dynamically downscaled RCM data from the CORDEX-Africa domain at a spatial grid resolution of 0.44° (50 km). The methodology involves running a multi-model ensemble of three RCMs namely (Regional Climate Limited-area Modeling (CCLM4.48), Rossby Center regional atmospheric model (RCA4), and KNMI Regional Atmospheric Climate Model, version 22 (RACMO22T) and downloaded from a public website, (https://climate4impact.eu/impactportal/data/esgfsearch.jsp) were used in this study (Table 1). The selected RCMs were widely applied in Africa (Lennard et al. 2018) and East Africa (Endris et al. 2013). The use of an ensemble mean of different RCMs helps to reduce the greatest and lowest projections and allows for the minimization of uncertainties. The downloaded climatic data from CORDEX-RCM is saved in NetCDF format with a rotated pole of longitude and latitude, which is then retrieved using ArcGIS10.4.1. The quantities of precipitation and temperature were kg/m2/s and Kelvin, respectively. Thus, before utilizing the data in any impact study, the rotated pole and units must be corrected. As a result, 86,400 multiplied the above data (24*60*60) to produce precipitation in mm/day and to obtain temperature in Celsius by subtracting 273.15 from downscaled data using the above conversion factors. These RCMs were downscaled using HadGEM2-ES, MPI-ESM-LR and EC-EARTH GCMs.

The hydrological SWAT model was developed to investigate how land use, land cover, and management affect water and sediment at the watershed level over the course of a day, monthly, and annual time increments.

The SWAT is a semi-distributed open-source model with a wide and rising number of model applications spanning from watershed to continental dimensions (Neitsch et al. 2011; Arnold et al. 2012). It looks at how LULC and climate change affect water resources in a basin with different soil, land use, and management techniques through time (Arnold et al. 2012). Thus, the SWAT hydrological model has been utilized effectively in investigations of the effects of land use/cover, and climate change on water resources in basins with variable soil, land use, and management techniques across time (Arnold et al. 2012). The SWAT model has clearly produced great accuracy for short- and long-term simulations of annual and monthly mean streamflow (Zuo et al. 2016; Anand et al. 2018).

Finally, the saturated hydraulic conductivity was determined from the soil data by SWAT.

For the extraction of flow direction, flow accumulation, stream network construction, and delineation of the watershed and sub-basins, a digital elevation model (DEM) with a resolution of 12.5 by 12.5 meters was employed. For this study, the DEM of 12.5 × 12.5 m was obtained from https://vertex.daac.asf.alaska.edu/website.

The LULC maps of 2015 were obtained from Landsat images of Landsat 8 OLI using ERDAS Imagine 2015 packages. The images were obtained from the USGS (https://earthexplorer.usgs.gov/) website.

Daily precipitation, maximum and minimum temperature, relative humidity, wind speed, and solar radiation data were gathered for 31 years, from 1985 to 2015, from Ethiopia's National Meteorological Agency (Table 2). The observed daily streamflow data of Yadot River at the Delo Mena gauging station were obtained from the Ministry of Water, Irrigation and Electricity from 1985 to 2008 for calibration and validation (Table 2).

Sensitivity analysis has been carried out in this study utilizing the SUFI-2 (Sequential Uncertainty Fitting-2) technique, which is the interface of SWAT–CUP (Soil and Water Assessment Tool–Calibration Uncertainty Program), (Malik et al. 2022). Using a global sensitivity analysis, a monthly time-step sensitivity analysis was done to determine the most sensitive parameters.

The calibration and validation for this investigation were done at the Delo Mena gauge station during a 24-year period (1 January 1985 to 31 December 2008). The first three years of data from 1 January 1985–31 December 1987 were used to warm up the model, whereas the calibration and validation periods were from 1 January 1988–31 December 2002 and 1 January 2003–31 December 2008, respectively. The model parameters were changed manually until the modeled discharge was within the statistically acceptable model performance. The SWAT model's performance was further evaluated using graphical and statistical methodologies until acceptable values for surface runoff were independently found. Validation at the watershed's outflow enables direct comparison of model outputs to an independent data set without any adjustments.

The impact of climate change on streamflow was determined by using near-term (2021–2050) and mid-term (2051–2080) period projected climate under medium (RCP4.5) and high (RCP8.5) emission scenarios and 2015 LULC data as input. The baseline climate data used for the simulation was from 1985 to 2015. Thus, the relative change was quantified by comparing the projected streamflow for the two 30-year time slices in the near-term (2021–2050) and mid-term (2051–2080) periods against the baseline period (1985 to 2015). For all simulations, we computed the projected change (as the percent change compared to the baseline) in the 30-year mean and seasonal streamflow relative to the baseline period. The first streamflow simulation was carried out using the baseline climate data (1985–2015). After the simulation of baseline streamflow, the near-term streamflow simulation was carried out using near-term (2021–2050) climate data under medium (RCP4.5) and high (RCP8.5) emission scenarios. Finally, the impact of climate change on streamflow under the near-term period under both scenarios was determined by subtracting the simulated streamflow during the near-term period under both medium (RCP4.5) and high (RCP8.5) emission scenarios from the baseline simulated streamflow. In a similar procedure, the impact of climate change on streamflow under the mid-term period under both scenarios was determined by subtracting the simulated streamflow during the mid-term period under both medium (RCP4.5) and high (RCP8.5) emission scenarios from the baseline simulated streamflow value.

The characteristics that regulate and govern streamflow in the watershed, as well as those linked to surface runoff, groundwater, and evaporation, were selected as the most sensitive parameters of the SWAT model (Table 3). The most important sensitive factors for streamflow predictions were determined to be maximum canopy storage (CANMX), SCS runoff curve number (CN2), saturation hydraulic conductivity (SOL K), and soil evaporation compensation factor (ESCO). Similarly, Negewo & Sarma (2021) recognized (CN2) and (SOL K) as the most sensitive parameters for runoff estimates in the Genale watershed. Similarly, Beyene et al. (2018) identified (CN2) and (CANMX) as highly sensitive parameters in the upper Awash Basin.

The findings of this study are in line with other previous studies in the region. For instance, Gragn et al. (2019) reported increasing mean annual precipitation during the near-term (2020) and mid-term (2050) period under both the RCP4.5 and RCP8.5 emission scenarios in the Awata River watershed of the Genale Dawa basin. A similar study by Gurara et al. (2023) reported a rising projected annual precipitation trend toward the end of the twenty-first century under RCP4.5 and RCP8.5 emission scenarios in the Upper Wabe Bridge Watershed in the Wabe Shebelle River Basin. According to Seyoum et al. (2017), the mean annual precipitation is projected to increase by about up to 9% in the 2080s in the Awash, Baro, Genale, and Tekeze rivers basins of Ethiopia. Precipitation in the upper Blue Nile River Basin in the Lake Tana Basin might increase by up to 25% (Getachew et al. 2021).

The magnitude of the projected minimum temperature during the mid-term period is slightly higher than the near-term period under both scenarios for both dry and wet seasons and annually (Figure 9).

The change in projected annual maximum and minimum temperature showed a consistent increase in the near-term and mid-term periods under both scenarios with a higher rate of increase toward the end of the century. It also noted that the rate of change of maximum temperature is higher than the rate of change of minimum temperature. The pattern of change of projected annual maximum and minimum temperature almost follows the same pattern as that of seasonal changes.

The increased temperature result in this research is consistent with other previous studies in the region and elsewhere. Similarly, the studies in the Genale Dawa basin reported a more warming projected temperature (Muleta 2017; Gragn et al. 2019; Negewo & Sarma 2021). Similar studies in the Finchaa catchment and Upper Wabe Bridge Watershed in the Wabe Shebele River Basin by Dibaba et al. (2020) and Gurara et al. (2023) showed a consistent maximum and minimum temperature increase for all time horizons, with a higher rate of increase toward the end of the century. Overall regional studies reported a future projected warming in sub-Saharan Africa and increased precipitation in East Africa (Serdeczny et al. 2017).

The overall projected temperature change (maximum and minimum temperature) during both near-term and mid-term periods consistently showed an increase. However, the magnitude of change during the mid-term period is slightly higher than the near-term period under both scenarios for both dry and wet seasons and annually.

The seasonal variation of projected streamflow from the baseline period was computed for wet (March to May and August to November) and dry (January, February, June, July and December) seasons. Future changes in streamflows for wet and dry seasons are important to understand the hydrological impact of climate change. The wet season mean streamflow of the watershed is expected to increase by 8.48 and 6.23% from the baseline flow in the near-term under RCP4.5 and RCP8.5 scenarios, respectively. Similarly, mid-term mean wet season streamflow may increase by 9.36 and 7.21% from the baseline streamflow under RCP4.5 and RCP8.5 scenarios, respectively (Table 5). The dry season mean streamflow is projected to increase in the watershed by 5.46 and 4.57% from baseline flow in the near-term while, in the mid-term dry season, mean streamflow is expected to increase by 5.41 and 3.16% from the baseline streamflow under RCP4.5 and RCP8.5 scenarios, respectively.

The simulated mean annual streamflow may rise by 7.63 and 8.25%, near- and mid-term period, respectively, under the RCP4.5. Similarly, the mean annual streamflow increased by 5.76 and 6.07%, respectively, under RCP8.5 scenarios.

The projected increase in streamflows into the watershed is in agreement with other studies that reported how mean annual streamflow consistently increased in line with the predicted changes in rainfall and temperature patterns in the future period 2021–2080 under the RCP 4.5 and RCP 8.5 emission scenarios of Genale watershed (Negewo & Sarma 2021). The study of Gragn et al. (2019) indicated that the average annual, seasonal and monthly flow volume is observed to increase between the 2018 and 2077 period corresponding to an increase in mean annual, seasonal and monthly precipitation during the future scenario of the Awata river watershed, Genale Dawa Basin. Similarly, Anwar et al. (2016) from a study in the Upper Gilgel Abay Watershed, Blue Nile basin concluded that streamflow projections for future periods will likely increase.

Generally, the annual and seasonal streamflow changes are expected to increase in all future periods in Yadot watershed under the RCP4.5 and RCP8.5 scenarios.

The changes in precipitation and temperature influence streamflow patterns. The correlation between precipitation and the streamflow is positive, while the relationship between the temperature and hydrological process is positive with regard to potential evapotranspiration. Although there is a general tendency toward increasing maximum and minimum temperature, the overall increase in annual and seasonal rainfall contributed more to the seasonal and annual streamflow increase. This implies that the increase in streamflows due to increased precipitation is not counter balanced by the increased evapotranspiration from the warming temperature trend. This indicates that streamflow is more highly sensitive to precipitation change than to change in potential evapotranspiration resulting from temperature increase (Yira et al. 2017; Daniel & Abate 2022; Idrissou et al. 2022).

The study implied that the projected increase in streamflows due to the impact of climate change could lead to increased sediment yield, which could have negative impacts on water quality and aquatic ecosystems downstream. Additionally, the overall projected streamflow increase, in combination with projected increased precipitation within the watershed, may cause flooding in the region. Generally, in the future, floods and extreme precipitation are expected to intensify in response to global warming (Tabari 2020).

Thus, understanding how streamflow is affected by climate change is crucial to inform adaptive land and water management for policymakers and help to improve future decision-making on water resources. This information can help stakeholders develop and prioritize management strategies that are better suited to the changing hydrological conditions. Therefore, the projected streamflow during annual, wet and dry seasons affects not only the livelihood of people residing in the Bale highlands but also all inhabitants in the downstream areas whose life depends on the flow of the Yadot River. Thus, to mitigate the adverse effects of the projected increase in streamflows, appropriate physical and biological soil and water conservation measures are highly recommended to protect against flooding, soil erosion, and downstream sedimentation in the region.

The study assessed the influence of climate change on streamflow in the Yadot River watershed Genale Dawa basin, Ethiopia, for the near-term (2021–2050) and mid-term (2051–2080) future periods under the RCP 4.5 and. RCP 8.5 emission scenarios. The SUFI-2 algorithm approach in the SWAT-CUP tool was utilized to parameterize the SWAT hydrological model, which was used to simulate streamflow. For streamflow, the uncertainty analysis, calibration (1988–2002), and validation (2003–2008) processes were all successful. CANMX, SCS runoff curve number (CN2), saturation hydraulic conductivity (SOL K), and soil ESCO were discovered to be the most sensitive parameters for streamflow predictions using the sensitivity analysis. The study used the ensemble mean of three RCM outcomes, which were bias-corrected in the CORDEX-RCM.

According to this study, the projected monthly precipitation did not indicate a clear pattern of change in the Genale Dawa basin, whereas the projected wet and dry season precipitation is likely to increase during the near- and mid-term periods under medium and high emission scenarios in the study area. The change in precipitation was higher during the mid-term period (2050–2080) under the high emission scenario (RCP8.5) than other periods and emission scenarios.

The projected maximum temperature increased on both dry and wet seasons during the near- and mid-term periods under both medium and high emission scenarios. The change in projected maximum temperature was highest (almost reaches 2 °C) in the dry season during the mid-term period under a high emission scenario. The mid-term period under the high emission scenario, generally, showed the highest magnitude projected maximum temperature increase than near-term and medium emission scenarios during both seasons. Generally, the change in projected maximum temperature will be higher during the dry season than the wet season during both periods and scenarios, i.e. the dry season will be getting warmer than the wet season. Overall, during the near-term and mid-term periods the study region will be getting warmer. The projected seasonal minimum temperature showed a consistent increase during both seasons. The change in projected annual maximum and minimum temperature showed a consistent increase in the near- and mid-term period under both scenarios with a higher rate of increase toward the end of the century.

The projected annual and seasonal changes of streamflow indicated the likely increase during both near- and mid-term periods under both scenarios (RCP4.5 and RCP8.5) in the Yadot watershed. The expected increase in the projected rainfall and a consistent projected warming will further increase the streamflow in the study area. Thus, increased streamflow expected from heavy rainfall will become more common and more intense due to higher temperatures, with floods expected to become more frequent in the region. The prediction of future hydrological scenarios will be useful in the development and implementation of effective watershed management mitigation strategies in the near and far future.

The authors are grateful to the Ethiopian Ministry of Water, Irrigation, and Electricity, the Department of Hydrology, and the Ethiopian National Meteorological Agency for supplying streamflow and meteorological data from the Yadot River Watershed, Genale Dawa Basin. We are also grateful to the Ministry of Agriculture for supporting the research financially.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.