Hydrological response to climate change in the Deme watershed, Omo-Gibe Basin, Ethiopia

Nowadays, scientific evidence is demonstrating the mean temperature of the earth's surface is rising as a result of greenhouse gas emissions, and other natural and human activities (Shiferaw et al. 2018). Earth's temperature is determined by the balance between the inward solar radiation and the outward terrestrial radiation. The energy coming in the form of the sun can pass through the atmosphere and heats the surface of the earth. Then, the radiation emitted from the surface of the earth is partly absorbed by some gases in the atmosphere and some of it is re-emitted downwards (Legesse et al. 2015). This increase in surface air temperature has a high probability to affect the hydrologic cycle. The hydrological cycle is the driving factor for the physical and ecological processes on the earth's surface and has a huge impact on the survival of living organisms, particularly human beings (Rakhimova et al. 2020). Changes in the hydrological cycle led to the increasing occurrence of extreme events, related to each aspect of human activity; influencing agricultural activities, ecosystems, groundwater, water supply, energy production, land use, and the quantity and quality of regional water resources (Biniyam & Abdella 2017). Climate change affects the function and operation of existing water infrastructures including hydropower, structural, drainage, and irrigation systems as well as water management practices (Chaemiso et al. 2016). The atmosphere and ocean have warmed, the quantities of snow and ice have reduced, and the sea level has risen. Many aspects of climate alteration and its associated effects will continue for centuries, even though anthropogenic emissions of greenhouse gases are stopped (IPCC 2014; Raneesh 2014). The warmer temperatures will upsurge the water-holding capacity of air resulting in higher moisture contents, thus creating forceful rainfall and snow events (Gunathilake et al. 2020). Therefore, understanding the hydrological process is essential for planning and managing water resources. Greenhouse gas enhancement is unequivocal at global scales resulting from increasing driving forces for future expansion of emissions of substances. Climate change and warming of temperature is a global issue, and its effect is not restricted to the unique region.

Watersheds are very sensitivity to climate change and uncertainty due to the hydrological impacts of climate change in the watershed areas. The comparison between monthly and annual streamflows between the baseline and future periods shows a high variation in the four watersheds despite their proximity in the upper catchments of the Nzoia river basin in western Kenya (Musau et al. 2015). The watersheds located at the same region can definitely show different degrees of hydrological responses to the same external climate changes. The differences in spatially distributed physical catchment properties play a main role in the sensitivity of catchments to varying climate conditions. Expected rise in precipitation and temperature resulted in upsurged total available streamflow, with a lower spring and summer streamflow, but a considerably higher winter streamflow. Also, significant changes in flow durations with lower chances of both high and low flows can be projected in boreal Sweden in the future (Teutschbein et al. 2015). It is vital to evaluate variations in low flow across the watershed and basin, because it is suffering from water shortage and salt water intrusion in the dry season. The variation of river discharge is likely to aggravate water stress. Particularly the reduction of low flow is a problematic. The results of the study in the Pearl River basin in the south of China, using five different climate models under the RCP4.5 and 8.5 scenarios, indicate a reduction in average low flows under all climate models. The decline varies across the basin and in between 6 and 48% under the RCP4.5 scenario. The river discharge in the dry season is expected to reduce throughout the basin. However, in the wet season, river discharge tends to increase in the middle and lower reaches and decrease in the upper reach of the Pearl River basin (Yan et al. 2015).

Developing countries, particularly sub-Saharan Africa, are likely to be especially vulnerable to climate change as recurrent floods and droughts continue to bring misery to millions in Africa, due to lack of institutional capacity and economic development (IPCC 2013; Wagesho et al. 2013). Therefore, linking watershed management with climate change adaptation and mitigation is mandatory in African countries (Joosten & Grey 2017).

Ethiopia is one of the most vulnerable countries to climate change and the least ready to improve resilience. The severe drought of 2015–2016 was worsened by the strongest El Nino, causing successive harvest failures and widespread livestock deaths in some regions. It has experienced even more major floods in different parts of the country (Disasa et al. 2019). Generally, the climate change impact in Ethiopia is signified by frequent and severe droughts, floods, increases in temperature, erratic and uneven rainfall, shifts in the onset and cessation of seasonal rainfalls, water stress, and scarcity. There have also been increased heat waves and windy days, increased health risks (malaria, diarrhea, and malnutrition), hailstorm and frost in some areas, increased landslides, and soil erosion problems in many parts of the country. Accordingly, crop failure is the most common problem in Ethiopia (Hagos 2019). Deforestation is the main cause of climate variability, and it is the major problem in Ethiopia (Tesfaye 2017).

Large parts of Ethiopia are arid and semi-arid regions that are highly vulnerable to climate change. Also, a large number of populations are poor with an agricultural-based economy, and the country faces extensive deforestation. Moreover, it has a low adaptive capacity to deal with the consequences of climate variability (World Bank 2014). Several studies are done regarding climate change impacts in Ethiopia. Among them, the hydrological response to climate change in the Bilate catchment, Ethiopia was assessed, and the outcomes showed that the temperature and evapotranspiration (ET) are expected to rise, while runoff, groundwater, and water yield are projected to decline in the watershed (Kuma et al. 2021). Similarly, the effect of climate change on streamflow in the Gelana watershed, Rift valley basin, Ethiopia was evaluated, and the results revealed that the temperatures and ET were predicted to increase. These changes translate to possible reductions in the mean annual rainfall and streamflow, with a consequent higher decline of surface runoff, groundwater, and water yield. Moreover, the rainfall and streamflow are expected to face a higher decline in wet seasons (Daniel & Abate 2022).

According to the study by Dero et al. (2021), the land use land cover change was increased for settlement, bare land, and croplands in the upper Deme watershed. But, water body, forest, and grassland were decreased in the upper watershed in the period of 1986–2019. Besides, during these periods, high percentages of other land use and landcover in the upper Deme watershed were changed into settlement areas. This demonstrates that the urbanization causes a change in social, economic, land use land cover, and watershed management activities in the watershed. Moreover, in the Deme watershed, the society lives in rural areas, and is highly dependent on agricultural activities, and water demand and irrigation are increasing due to some investors and organizations. These activities coupled with the future global warming may result in water stress in the watershed. Besides, the local stakeholders and farmers have not had a good understanding of the climate variability impacts at a relevant scale until now. Therefore, this study tries to address the hydrological response to climate change in Deme Watershed, Omo-Gibe Basin, Ethiopia.

The spatial data (DEM) of Deme watershed of (12.5 m × 12.5 m resolution) were downloaded from Alaska satellite facility (https://asf.alaska.edu/) and used to delineate the watershed, and calculate the sub-basins parameters. Then, the soil map and land use/land cover map data obtained from Ethiopia Ministry of Water, Irrigation, and Energy (MoWIE) were added into SWAT and we classified the slope to create the Hydrological Response Unit (HRU) based on their homogeneity.

Next, the collected input data from institutions, such as climate data (maximum and minimum temperature, rainfall, relative humidity, wind speed, solar radiation) from 1987 to 2019 were collected from the Ethiopian National Meteorological Agency. The Average and Normal ratio methods were used to fill the missing meteorological data records, because of their computational simplicity and significant precisions. Double Mass Curve analysis was used for checking the consistency of recorded data. Homogeneity of the selected stations rainfall records was checked by the non-dimensional parameterization method. The streamflow data are necessary for calibration and validation of the SWAT model and are used to evaluate the streamflow response to climate change. The Deme streamflow was gauged at Oreta-Alem station and data from 1987 to 2010 was collected from the Ministry of Water and Energy. This station was used for calibration and validation of the hydrological model (SWAT).

In addition, the climate data were used by downscaling the output of the RACMO22T-RCM using CORDEX-Africa under RCP4.5 and RCP8.5 emission scenarios for the base period (1991–2010), and future periods (2031–2070). The climate variables were extracted from CORDEX-Africa Regional Climate Model (RCM) output by using Arc-GIS to the study area by considering the meteorological data gauging stations' latitude and longitude coordinates.

Further, the biases of RCM simulations were adjusted comparatively with historical observed data by using Distribution Mapping (DM) of Precipitation for precipitation, and Variance Scaling (VARI) of Temperature for temperature correction. Moreover, the performances of the selected RCM and bias correction method were evaluated by using the statistical indicators techniques. Then, the climate variables such as maximum and minimum temperature, and precipitation were projected for future periods (2031–2050 and 2051–2070) under RCP4.5 and RCP8.5 scenarios.

After that, the sensitivity analysis, calibration, validation, and uncertainty analysis of the SWAT model were done by SUFI-2 embedded in the SWAT-CUP, and its performance was evaluated by using the statistical indicators.

Finally, the streamflow was simulated for future periods under RCP4.5 and RCP8.5 scenarios by using the SWAT model. Moreover, the climate change effect on surface runoff, groundwater, total water yield, potential evapotranspiration (PET) and actual ET were evaluated. Similarly, the climate change impact in monthly, seasonal, and annual streamflow for future periods relatively with baseline period were evaluated.

According to the land use/land cover classification, seven major land uses and land cover types were identified in the Deme watershed. The classes are coded by four letters according to the SWAT database as Agricultural land (AGRL), Shrubland (RNGB), Bareland (BARR), Forest-mixed (FRST), Settlement (URBN), Grass land (PAST), and Water bodies (WATR). The SWAT model requires a basic physical–chemical property of the soil types (Dibaba et al. 2020). Its simulation depends on the soil data to determine the variety of hydrological characteristics found in each sub-basin within the watershed. The soil map of the study area was extracted using the Deme watershed shape file. The soil types of the study area were classified into three major groups. These are Eutric Cambisols, Eutric Nitosols, and Ochric Andosols.

There are six meteorological stations in and around the Deme watershed as shown in Table 1.

The data were successfully analyzed by using the different materials and software. Arc-GIS 10.4.2 was used to obtain the spatial information of the watershed, to generate the climate data from the RACMO22T climate model to the Deme watershed. The Arc-SWAT 2012 model was used to delineate the Deme watershed, assess the water balance components, and simulate the current and future periods streamflow. SWAT-CUP 2012 was used to analyze the sensitivity, calibration, validation, and uncertainty of the Arc-SWAT 2012 model. Google Earth was used to obtain and check the coordinate and elevation data, and provides a view of the watershed. Climate model data for hydrologic modeling (CMhyd) software were used to extract and correct the bias of climate data (precipitation, temperatures) obtained from RACMO22T RACMO22T. The homogeneity of the hydrological data (Deme stream flow) was checked by RAINBOW software. PCPSTAT and DEW02 were used to prepare the weather data generator algorithm for the SWAT model.

The modeling approach and the resolution of the model vary from one model type to another. The climate models driven by Global Climate Model (GCM) projections can limit the exact simulations of regional climatology owing to the inability to accurately simulate features of local or regional climate including topography, orthography, cloudiness, and land use due to the inherent coarse spatial resolution ranging between 100 and 250 km. Whereas, the resolution of RCMs is tens of kilometers, in the range of 12–50 km, in the proximity of the watershed scale. The RCM is also called the limited-area model (Dibaba et al. 2019; Gunathilake et al. 2020).

The downscaled RCM data were obtained from https://esgf-data.dkrz.de/search/cordex-dkrz. The CORDEX-Africa model data at longitude 0.44° and latitude 0.44° horizontal resolution and a multi-model ensemble of regional climate models with their driving GCMs provided the boundary conditions (Kuma et al. 2021). Africa was nominated as the target area of CORDEX for three main reasons. These are: the high vulnerability to climate changeability, relatively low adaptive capacity of its economies, and substantial changes in rainfall and temperature patterns (Giorgi et al. 2009).

The major driving forces for future expansion of greenhouse gas emissions of substances are technological innovation, energy choice, socioeconomic and demographic growth, and their integrations (Daniel & Abate 2022). According to a classification by Van Vuuren et al. (2011), the Representative concentration pathways, RCP4.5 and RCP8.5 scenarios were used for the study. Moreover, among the several RCMs, based on the vintage, resolution, validity, and representativeness, the RACMO22T model was selected, which was endorsed by Dibaba et al. (2019); and Daniel & Abate (2022). The RACMO22T model data of precipitation, maximum and minimum temperatures were extracted by Arc-GIS 10.4.1 software through a multidimensional tool and NetCDF table view. The grid points data were extracted, which are the nearest for observed meteorological stations by using the latitude and longitude of observed stations.

CMhyd software was obtained from https://swat.tamu.edu/software/. It was used to correct the bias of precipitation, minimum and maximum temperature simulation for historical (1991–2010) and future periods (2031–2070) of analysis under both RCP4.5 and RCP8.5 scenarios.

The DM method is applied to correct the distribution function of the RCM outputs and to align them with the observed distribution function. This can be done by creating a transfer function to shift the occurrence distributions of precipitation. It is based on the assumption that both the RCM-simulated and observed climatic variables obey a specific frequency distribution (Teutschbein & Seibert 2012). The DM method uses the transferring of function to adjust the cumulative distribution of estimated data to the cumulative distribution of rain gauges, and it reproduces precipitation very well (Dibaba et al. 2020; Daniel & Abate 2022).

The LOCI method aims to simultaneously correct the precipitation intensity and frequency. Initially, the rainfall intensity threshold for each month is confirmed. Accordingly, the number of wet days in RCM precipitation that exceed this threshold matches the number of days for which observed precipitation is determined. This approach can effectively eliminate the drizzle effect because too many drizzly days are often included in original RCM outputs (Luo et al. 2018).

VARI of the temperature method is developed to correct both the mean and variance of temperature (Luo et al. 2018).

There are some pre-requisites for the selection of climate model and bias correction methods. In addition, their performances were evaluated by using statistical indicators. The performance of the selected RCM and bias correction methods were evaluated using four approaches: Pearson correlation coefficient (R), Root Mean Square Error (RMSE), Nash–Sutcliffe Efficiency (NSE), and Percent bias (PBIAS).

The Soil and Water Assessment Tool (SWAT) is a model that simulates the hydrology of the watershed. It is a deterministic hydrologic model, and it is the semi-distributed model. It was developed in the early 1990s by the Agricultural Research Service of the United States Department of Agriculture (Neitsch et al. 2002). The study-dependent factors to select the SWAT model are as follows:

• It is physically based.

• It uses readily available inputs.

• The model is computationally efficient and simulates the major hydrological process in the catchments.

• It is continuous in time and capable of simulating for long periods.

• The model is verified by numerous studies for the assessment of climate change on the hydrological cycles in different parts of the world.

• It is an easily available model.

The SWAT model simulation of the routing phase of the hydrological cycle can be explained as the movement of water over the channel network of the watershed to the outlet. The SWAT input files were organized and the model was set to run, at the end, it simulates. Daily climate data such as rainfall, temperature (maximum and minimum temperature), solar radiation, wind speed, and relative humidity were required for the SWAT modeling. The 24 years including 2-year warm-up period of the six meteorological stations (Wolaita sodo, Gessuba, Morka, Dara malo, Areka, and Bele stations) from January 1, 1987 to December 31, 2010 were used for SWAT simulation depending on data availability.

The SWAT model uses the concept of infiltration excess runoff mechanism and it assumes the runoff comes about when the rainfall intensity is more than the infiltration rate. This phenomenon happens in areas with significant soil erosion that occurs during very high rainfall intensity storm events in the irrigated field and urban areas (Shiferaw et al. 2018). For this research, the Soil Conservation Services (SCS) curve number was used to estimate the surface runoff due to its capability to use daily input data.

The SCS runoff equation is an empirical model that came into common use in the 1950s. The model was developed to provide a consistent basis for estimating the amounts of runoff under varying land use and soil types.

SWAT calculates the peak runoff rate with a modified rational method. The rational method assumed that precipitation of intensity, I, begins at time t = 0 and continues indefinitely, while the rate of runoff will increase until the time of concentration, t = t_conc.

Sequential Uncertainty Fitting version 2 (SUFI-2) was embedded in the Arc-SWAT-CUP, which was used for the sensitivity analysis, calibration, and validation of the SWAT model. This algorithm was found to be quite effective (Abbaspour et al. 2015).

The SUFI-2 was given several iterations to reach acceptable results. Each iteration provides the suggested values for the new parameters to be used in the next iteration. Until the best fit curve of simulated versus observed streamflow data was obtained, the sensitive parameters were changed again and again in the allowable range recommended by SWAT-CUP. Finally, it achieved an acceptable result with the values of the Nash–Sutcliffe, Coefficient of determination, Percent of bias, and other uncertainty analysis statistical parameters. The sensitivity parameters were identified with their ranges which is necessary to reduce the computational time required for SWAT model calibration. This process is called sensitivity analysis. All SWAT model parameters involved in the hydrological process were reduced into the most sensitive parameters with different degrees of sensitivity based on a low P-value and a high absolute value of t-Stat (Dibaba et al. 2020; Daniel & Abate 2022).

Calibration is the determination to better parameterize a model to a given set of local conditions, through comparing the model prediction with the observed data, thus reducing the prediction uncertainty (Arnold et al. 2012). Calibration was done next to identifying the sensitive parameters by comparing model-simulated streamflow with observed streamflow data from the period of January 1, 1989 to December 31, 2000.

To utilize the calibrated model for estimating the effectiveness of future potential management practices, the model should be tested against an independent set of measured data without further adjustments; this testing of a model with an independent data set is commonly referred to as model validation (Dibaba et al. 2020). It is the final step of the hydrologic modeling, verifying the performance of the SWAT model for simulated flows in periods different than the calibration periods of January 2001 to December 2010. Generally, the objective function used in SWAT-CUP (SUFI-2) for calibration and validation of the SWAT model was NSE in this study. Because it is very commonly used, the best objective function for reflecting the overall fit of a hydrograph provides extensive information, and it is less sensitive to high extreme values due to the squared differences (Moriasi et al. 2007).

Statistics techniques such as coefficient of determination, NSE, and PBIAS were used to express the good fitness of the model simulation with the observed streamflow (Daniel & Abate 2022).

PET is determined primarily by net radiation and temperature, but also by the moisture-holding capacity of the air and other factors. Increasing temperature will lead to more evaporation, although the effect is complicated and alters the hydrologic regime in and around the catchment. Climate change manifests itself through increasing temperatures that lead to an increase in actual ET, which will accelerate the hydrological cycle (Jiménez et al. 2014).

The most sensitive parameters which affect the SWAT model output were evaluated by using the SWAT-CUP (SUFI-2). The small valued sensitivity parameters do not expressly affect the hydrological model output. Nevertheless, from medium to very high ranked sensitive parameters that significantly affect the SWAT model output were used to calibrate and validate the model. The result of SWAT sensitivity analysis, ranges, and their fitted values in the Deme watershed at Oreta-Alem gauging station are tabulated in Table 2. It indicates ALPHA_BF.gw (Baseflow alpha factor in days), CH_K2.rte (Effective hydraulic conductivity in main channel alluvium), RCHRG_DP.gw (Deep aquifer percolation fraction), ESCO.hru (Soil evaporation compensation factor), SLSUBBSN.hru (Average slope length), GWQMN.gw (Threshold depth of water in the shallow aquifer required for return flow to occur in mm were found to be the six topmost sensitive parameters at the Oreta-Alem gauging station in the Deme watershed.

The calibration results on mean monthly flow show that the SWAT model can capture the observed streamflow with R², NSE, and PBIAS of 0.77, 0.73, and 4.6, respectively. The average monthly flow validation indicates R², NSE, and PBIAS of 0.82, 0.75, and 14.2, respectively. Generally, the hydrological model showed a reasonably good agreement in the Oreta-Alem gauging station. Hence, according to descriptions by Moriasi et al. (2007), the SWAT model has an acceptable performance in the Deme watershed.

The bias of RACMO22T (RACMO22T) simulations was corrected by the DM for adjusting the precipitation, and VS of temperature were used for adjusting the maximum and minimum temperature.

Generally, the DM method for precipitation and VS method for maximum and minimum temperature bias correction showed very good performance as tabulated in Table 3, and were used to adjust the climate variables of future periods. The related conclusion was reported by Daniel & Abate (2022). Also, it is supported by Geleta & Gobosho (2018) and Dibaba et al. (2020), in the Finchaa watershed in Ethiopia. Besides, the results revealed that the RACMO22T climate model performed better in reproducing daily precipitation, maximum, and minimum temperatures in the Deme watershed. The corresponding study was also done by Dibaba et al. (2019) and Daniel & Abate (2022). Thus, the RACMO22T climate model was used in the study to model climate variables.

To evaluate the climate change effect, the precipitation and temperature were obtained from the RACMO22T model, then, the bias-corrected precipitation and temperature by DM and VS methods, respectively, for future periods (2031–2070), are compared to the base period datasets (1991–2010). Additionally, the hydrological responses of the watershed are investigated by computing and comparing the future water balance components such as surface runoff, groundwater flow, and total water yield. Besides, the PET and actual ET were analyzed for RCP4.5 and RCP8.5 scenarios.

The annual declining rainfall in the two future periods under RCP4.5 and RCP8.5 scenarios is shown in Table 3. The rainfall will be declined by 30.14 and 27.6% under RCP4.5 and RCP8.5 scenarios, respectively, in the 2031–2050 period. Similarly, it will decline by 32.87 and 26.21% under RCP4.5 and RCP8.5 scenarios, respectively, in 2051–2070. The related projections were also confirmed by Dibaba et al. (2020) in the Finchaa sub-basin, Ethiopia, and by Daniel & Abate (2022) in the Gelana watershed, in Rift valley basin, Ethiopia.

The annual maximum temperature in the Deme watershed will increase on average by 0.97 and 1.10 °C under the RCP4.5 and RCP8.5 scenarios, respectively, for the near future period. Correspondingly, an increment by 1.70 and 2.00 °C was expected under both scenarios, respectively, for the far future period with respect to the base period. Similarly, the annual minimum temperature will be increased on average by 1.19 and 1.67 °C under the RCP4.5 scenario for the near and far future periods, respectively. Likewise, it will be increased by 1.62 and 2.29 °C for the near and far future, respectively, under the RCP8.5 scenario as shown in Table 4. Moreover, the variations are higher for the minimum temperature than the maximum temperature.

The overall results indicate that maximum and minimum temperatures increase under both RCPs throughout the study years, showing the warming trends in the watershed. The highest and lowest change in temperatures under the RCP8.5 scenario is more than the RCP4.5 scenario for all time horizons, especially in the far future period concerning the baseline period. This indicates that the RCP8.5 scenario will be warmer than the RCP4.5 scenario in the Deme watershed.

Therefore, the maximum and minimum temperature predictions for future time horizons (2031–2070) under both emission scenarios are within the range confirmed by IPCC (2014). Moreover, the maximum and minimum temperatures in the Bilate watershed from 2051 to 2080 are expected to become warmer than from 2021 to 2050 under both emission scenarios (Kuma et al. 2021). The related study was reported by Daniel & Abate (2022) in the Gelana watershed, in Rift valley basin, Ethiopia, which concluded that the highest temperature variation with respect to the base period was seen under the RCP8.5 scenario.

The PET and actual ET variation between the base period (1991–2010), near future (2031–2050), and far future (2051–2070) were evaluated as shown in Table 3. The mean annual change of PET is expected to increase by 6.78 and 7.32% under both scenarios, respectively, for the near future period. Likewise, its increment is expected by 9.15 and 10.89% under both RCP scenarios, respectively, for the far future period comparatively with the base period. Furthermore, the high variation in actual ET under both RCPs was analyzed for future years. According to evaluation the actual ET will increase on average by 5.14 and 7.17% under a medium emission scenario for the near and far future periods, respectively. Likewise, it will increase by 6.34 and 9.24% for the near and far future periods, respectively, under a high emission scenario in the Deme watershed.

Generally, the results show that the variation rate of PET and actual ET under the RCP8.5 scenario will be more than the RCP4.5 scenario in both future time horizons due to a large increment of the maximum and minimum temperature. Moreover, the magnitude of changes will be greater in the far future than in the recent future under both emission scenarios. Therefore, changes in temperature and ET are correlated positively in the Deme watershed. Accordingly, the rise in temperature resulted in an increase in the potential and actual ET in the watershed. The increase in ET suggests increased crop water requirements in future crop production. The conforming results were done by Kuma et al. (2021) in the Bilate watershed, and by Daniel & Abate (2022) in the Gelana watershed, in Rift valley basin, Ethiopia.

The variations in water balance components due to climate change were evaluated for future periods in Deme watershed, and the effects have resulted from the integration of rainfall, temperature, and ET with hydrological cycles. The hydrological responses of the watershed to climate changes were investigated for two future periods under the RCP4.5 and RCP8.5 scenarios. It was considered in terms of the process that contributes to the water balance components of annual surface runoff, groundwater, total water yield, PET and actual ET as shown in Table 5. The projected rainfall and temperatures for future periods from 2031 to 2070 under both RCP scenarios cause significant changes in hydrological components in the watershed.

The overall results show that the mean annual surface water runoff, groundwater, and total water yield will be expected to decrease, whereas actual ET and PET will be expected to increase in all future periods in the watershed. The maximum decline of surface runoff by 15.10%, ground water by 14.78%, and total water yield by 26.10% was seen in the near future (2031–2050) under the RCP8.5 scenario relative to the baseline period (1991–2010). These outcomes from the decreases of rainfall and increases of mean temperature, actual ET and PET led to a decline in surface runoff, groundwater, and total water yield in the Deme watershed.

The maximum rise in mean annual temperature seen by 2.15 °C in the far future period under the RCP8.5 scenario corresponded to the base period. The related study was done by Dibaba et al. (2020), which showed the annual mean temperature in the Finchaa watershed is projected to rise by 1.1–3.1 °C in the 2060s.

Consequently, the temperature rise resulted in increased PET and actual evaporation, which could be a critical factor for the reduction of total water yield. As the forms of water are exposed to losses, owing to the changes in temperature, the evaporation is also a factor for the future period reduction of the surface runoff and groundwater in the watershed. These indications of water shortage will be more pronounced in the future periods due to the warming of the atmosphere through climate changes. These results are consistent with the study done in the Bilate watershed by Kuma et al. (2021), who reported that a 15.39% decline in the rainfall resulted in around 9.22, 11.11, and 10.25% reduction in water yield, groundwater, and surface runoff, respectively, under the RCP4.5 scenario from 2051 to 2080. It is likely that a 2.0 °C increase in the mean annual temperature will lead to a rise in PET and actual evaporation by 19 and 13.9%, respectively, under the high emission scenario in 2051–2080. The corresponding study was done in the Gelana watershed by Daniel & Abate (2022), which predicted a decline of surface runoff by 22.23%, groundwater by 42.54%, and total water yield by 35.89% in 2051–2070 under the RCP4.5 scenario corresponding to the base period.

Relative changes in streamflow exhibited a large uncertainty when the projected temperature and precipitation changes were used in the SWAT model (Musau et al. 2015). The mean seasonal changes in the streamflow for the future periods under both RCP scenarios were predicted by using the projected rainfall and temperature, and will cause a significant variation of the future streamflow of the Deme watershed. However, the other climate variables (solar radiation, relative humidity, and wind speed); and land use land cover, soil map, and slope classification that were observed in the baseline period were considered a constant for the future periods. This is because varying these parameters may not have a significant influence on modeling the climate change in local hydrology (Dibaba et al. 2020).

Additionally, the annual changes in the streamflow for the future periods under both RCP scenarios were predicted as shown in Table 6. It suggested decreasing streamflow in the two future periods under both scenarios. The annual streamflow is expected to decline by 21.2% under RCP4.5 and 26.2% under RCP8.5 in the near future period. In the far future, it is likely to decline by 23.0% under the RCP4.5 scenario and 24.7% under the RCP8.5 scenario in the Deme river relative to the baseline period.

Generally, due to climate change, the temperature and ET are expected to increase, and the rainfall is expected to decrease in the watershed, due to the fact that the streamflow will decrease in the future periods from 2031 to 2070 with respect to the observed streamflow in baseline period from 1991 to 2010. This shows the direct relationship between temperature and ET, and the inverse relationship with rainfall and streamflow in the Deme watershed. The findings are related to the study done by Daniel & Abate (2022) in the Gelana river, which is expected to decline by 21.09% under RCP4.5 and 32.30% under RCP8.5 in the 2031–2050. Similarly, it declines by 44.14% under RCP4.5 and 24.27% under the RCP8.5 scenario in 2051–2070. Similarly, the mean annual and monthly streamflow in the future periods shows a declining trend in the four watersheds in the upper catchments of the Nzoia river basin in western Kenya (Musau et al. 2015).

The impact of climate change on hydro-meteorological variables has produced a significant alteration in the hydrology of the watershed. This study aims to investigate the hydrological responses to climate change in the Deme watershed using the SWAT for two consecutive periods of 2031–2050 and 2051–2070. To evaluate the hydrological responses to climate change, the precipitation and temperature were obtained from the RACMO22T model, then the bias-corrected precipitation and temperature within DM and VS methods, respectively, for future periods (2031–2070) were used for further analysis and compared to the base period datasets (1991–2010). The SWAT model was calibrated (and validated) for the 1989–2000 (2001–2010) period at the Oreta-Alem gauging station in the Deme watershed; the hydrological model showed a reasonably good agreement. The result shows a declining rainfall pattern in all seasons for future periods comparatively with the baseline period, except in the winter season. The highest seasonal rise in temperatures is expected from the high emission scenario compared to the medium emission scenario. Likewise, the maximum rise is projected in the far future period (2051–2070) compared to the near future period (2031–2050). Generally, temperatures will be expected to increase in all seasons with a high degree of variation with respect to a base period in the watershed. The more reflective fluctuations in projected streamflow were seen in seasonal bases concerning the annual bases in the watershed. Even though in the winter season there was a slight increase in the streamflow, most of the seasons in the future periods under both RCP scenarios show a declining trend relatively with an observed streamflow in the watershed. Moreover, the maximum decrease of flow variation is expected in the summer season. Specifically, it will decline by a rate of 62.20% under the RCP4.5 scenario from 2051 to 2070. Altogether, similar to rainfall patterns, the streamflow shows a decreasing trend in wet seasons rather than dry seasons in the Deme watershed. The mean annual rainfall will decline by 30.14 and 27.6% under the RCP4.5 and RCP8.5 scenarios, respectively, in the 2031–2050 period. Similarly, it will decline by 32.87 and 26.21% under the RCP4.5 and RCP8.5 scenarios, respectively, in 2051–2070.

The annual maximum temperature will increase on average by 0.97 and 1.10 °C under the RCP4.5 and RCP8.5 scenarios, respectively, for the near future period. Correspondingly, its increment was expected by 1.70 and 2.00 °C under both scenarios, respectively, for the far future period with respect to the base period. Similarly, the annual minimum temperature will increase on average by 1.19 and 1.67 °C under the RCP4.5 scenario for the near and far future periods, respectively. Likewise, it will increase by 1.62 and 2.29 °C for the near and far future, respectively, under the RCP8.5 scenario. Furthermore, the variations are higher for the minimum temperature than the maximum temperature. The overall results indicate that maximum and minimum temperatures increase under both RCPs throughout the study years, showing warming trends in the watershed. The RCP8.5 scenario will be warmer than the RCP4.5 scenario in the Deme watershed. Besides, the mean annual change of PET is expected to increase by 6.78 and 7.32% under both scenarios, respectively, for the near future period. Likewise, its increment is expected by 9.15 and 10.89% under both RCP scenarios, respectively, for the far future period compared with the base period. According to evaluation the actual ET will increase on average by 5.14 and 7.17% under the medium emission scenario for the near and far future periods, respectively. Also, it will increase by 6.34 and 9.24% for the near and far future periods, respectively, under the high emission scenario in the watershed. The overall results show that the mean annual surface water runoff, groundwater, and total water yield will be expected to decrease, whereas actual ET and PET will be expected to increase in all future periods in the watershed. The maximum decline of surface runoff by 15.10%, groundwater by 14.78%, and total water yield by 26.10% is expected in near future (2031–2050) under the RCP8.5 scenario relative to the baseline period (1991–2010). These outcomes from the decreases of rainfall and increases of mean temperature, actual ET and PET led to a decline in surface runoff, groundwater, and total water yield in the watershed. Additionally, the annual streamflow is expected to decline by 21.2% under RCP4.5 and 26.2% under RCP8.5 in the near future period. In the far future, it is likely to decline by 23.0% under the RCP4.5 scenario and 24.7% under the RCP8.5 scenario in the Deme River. Generally, due to climate change, the temperature and ET are expected to increase, and the rainfall is expected to decrease in the watershed; due to that the streamflow will decrease in the future periods from 2031 to 2070 with respect to the observed streamflow in baseline period from 1991 to 2010. This shows the direct relationship between temperature and ET, and their inverse relationship with rainfall and streamflow in the Deme watershed. Generally, the land use/land cover scenarios, and weather variables (solar radiation, windspeed, and humidity) were assumed constant for future periods during the study. However, these parameters directly contribute to hydrological systems. In future, this study should be extended by varying these parameters. Moreover, future studies should consider multiple RCMs with different scenarios for hydrological responses to climate change analysis, because it diminishes the uncertainties in climate change evaluations.

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

The authors declare there is no conflict.