Climate change impact studies that evaluated the biases of climate models' simulations showed the presence of large systematic errors in their outputs. However, many studies continue to arbitrarily select bias correction methods for error reduction. This work evaluated the implications of bias correction methods on the projections of climate change impact on streamflow of the Gidabo sub-basin, Ethiopia. Climate outputs from four global climate model and regional climate model (GCM–RCM) combinations for the representative concentration pathway (RCP4.5) scenario were used. Five bias correction methods were used to reduce the systematic errors of the simulated rainfall data. The future changes in rainfall pattern, evapotranspiration, and streamflow were analyzed by using their relative percentage difference between the projected and the baseline period. The distribution mapping method provided better results in mean and extreme rainfall cases. This is also reflected in streamflow projections, as the daily interquartile range value indicates the lowest variability of the projected streamflow. The wet season streamflow will likely decrease in the future, whereas the short rainy season streamflow will increase. Our findings show that climate models and bias correction methods considerably limit the magnitude of future projections of streamflow. However, similar research should be conducted in other catchments to extend the conclusions of this study.

  • COordinated Regional Climate Downscaling Experiment (CORDEX)-Africa domain climate model data were used for impact analysis.

  • Projected climate change impact on streamflow is affected both by climate models and bias correction methods.

  • The distribution mapping method provided better results in correcting the errors of mean and extreme rainfall projection.

  • Wet season streamflow of the study area will likely decrease in the future.

Graphical Abstract

Graphical Abstract
Graphical Abstract

The world's water resources are highly impacted by rapid population growth, a dramatic expansion of urbanization, excessive water demand for agriculture, and pollution of water sources (Shrestha 2014; Versini et al. 2016; Taye et al. 2021). Climate change is already causing additional pressure on the water resources of basins across the world. Over the period 1888–2012, the emissions of greenhouse gases to the atmosphere increased due to a combination of human-induced emissions and natural phenomena, which led to a warming of the globe by 0.85 °C (IPCC 2013). Such climate change influences the hydrological processes at global and regional scales. Therefore, understanding the hydrological impacts of climate change is crucial for developing a management strategy and also adaptation mechanisms to tackle the adverse effects of such change. So far, various studies have investigated the potential impact of climate change on water availability in Ethiopia (e.g., Ayele et al. 2016; Haile et al. 2017; Taye et al. 2018; Worqlul et al. 2018). However, the application of multiple bias correction methods in climate change impact studies is limited in Ethiopia.

One of the challenges involved in climate change studies is to ensure a reliable estimation of climate change impact on water resources due to various sources of uncertainty (Andersson et al. 2011; Versini et al. 2016). According to the IPCC (2007), the major sources of uncertainty are global climate models (GCMs), hydrological models, downscaling techniques or bias correction methods, and emission scenario development. To reduce the levels of uncertainty using outputs of multiple GCMs and evaluating multiple downscaling and bias correction approaches is a recommended approach (Fowler et al. 2007; Dessu 2013; Maraun 2016).

Bias correction is an important component in developing useful climate projections since it reduces erroneous results which might lead to unfair decision-making (Maraun 2016). Many investigators have used quantile mapping and other nonlinear regression methods to correct the systematic errors of GCMs–RCMs data collected for conducting impact studies in various regions (Teutschbein & Seibert 2013; Yira et al. 2017; Charles et al. 2019; Enayati et al. 2021). There is no single bias correction method that performs best, or is most suitable, for all regions (Teutschbein & Seibert 2013; Fang et al. 2015; Luo et al. 2018). This is mainly due to spatiotemporal differences in rainfall properties and varying assumptions of the methods. Hence, the selection of methods to remove the systematic errors of climate model outputs needs to be based on a comparison of multiple bias correction methods.

The main assumption in bias correction is time-invariance of biases. It is assumed that the relationship between the observed and the simulated rainfall properties remains the same in historical and future periods. Velazquez et al. (2015) challenged the time-invariance assumption and evaluated its implication in hydrological impact studies. However, they suggested their experiments be repeated for other basins to extend the conclusions. This study's limitation would be the use of a single bias correction method (Delta method), which leads to misleading results in climate change impact studies (Chen et al. 2013). Teutschbein & Seibert (2012) found that all bias correction methods significantly improved or reduced rainfall biases; however, the implications on streamflow simulation were not considered. Chen et al. (2015) showed that bias correction could successfully remove biases over the calibration period. However, large biases (reaching up to 9.5% for mean rainfall amount) can remain when the same bias correction is applied outside the calibration period. Chen et al. (2020) applied a pseudo-reality approach, i.e., one climate model as a reference dataset to correct the data of another climate model. They reported differences in the performance of the bias correction method over historical and future periods. They found that these were due to the differences between the climate sensitivity of the model, reference data, and internal variability that manifested itself through low-frequency oscillation in a long time series. Laux et al. (2021) evaluated four bias correction methods for precipitation and temperature. Based on their findings, they suggested an ensemble of bias correction methods in the generation of climate projections for climate change impact study. Tan et al. (2020) found that the hybrid bias correction methods performed better than the single method for precipitation and streamflow simulation.

Wörner et al. (2019) evaluated six bias correction methods for the future projection of climate change impact on high flows and found that none of these methods were explicitly preferred over the others. However, they acknowledged that distribution-based bias correction methods better reduced the systematic errors than mean-based linear bias correction methods. Teng et al. (2015) showed that four bias correction methods did not affect the climate change signal in precipitation. However, these methods generated additional uncertainties on the change signals of high precipitation, which influences simulated discharge. However, the authors failed to show the propagation of errors by bias correction methods in streamflow simulations. Hence, this reflects the need to evaluate the implication of bias correction methods on projections of water resource availability.

Projections for future water availability in different parts of Ethiopia are not consistent. For instance, studies conducted in the Upper Blue Nile River Basin show contradictory outputs on streamflow due to climate change. Some studies reported a likely increase in future water availability (Setegn et al. 2009; Dile et al. 2013; Gebre et al. 2015), whereas others reported a decrease in various flow (low, medium, and high flow) conditions, including the wet season (Beyene et al. 2011; Cherinet 2013; Abebe & Kebede 2017). According to Koch & Cherie (2013) and Mostafa et al. (2015), water resource availability may show a reduction of −5.96 to 69.3% in the 2050s and 2090s depending on downscaling techniques and climate scenario conditions. A likely decline of future water availability is reported for the western parts, Baro-Akobo, Southern Rift Valley, and the Omo-Gibe Basin (Legesse et al. 2010; Demissie et al. 2013; Tedla et al. 2015; Tekle 2015; Chaemiso et al. 2016; Biniyam & Kemal 2017). Similar trends are reported for the Tekeze River Basin (Goitom et al. 2012; Ashenafi 2014) in northern Ethiopia. We would like to note the difficulty in being conclusive about future changes in rainfall pattern. Nearly all of the above studies prioritized evaluation of the projected impacts of climate change on water resource availability without quantifying the sources of errors, including bias correction.

This study focuses on the Gidabo sub-basin, which is found in the Main Ethiopian Rift Valley Lakes Basin. Similar to other locations in Ethiopia, it experiences an increasing population growth, urbanization, industrialization, and irrigation water demand that impose immense challenges on future water availability (Debisso 2009; Mechal 2015; Shanka 2017). This requires a better estimation of water resource potential and designing relevant integrated water resource management projects for dealing with an uncertain future. Hence, this study was conducted to estimate the impact of climate change on surface water resources. We also evaluated the implication of using multiple bias correction methods and related errors or uncertainties to simulate streamflow. The findings of this study will be a great addition to the scientific literature and surface water management strategies for the Gidabo sub-basin due to climate change impact.

Descriptions of the study area

The Gidabo sub-basin is located in the Main Ethiopian Rift Valley system. The sub-basin covers a drainage area of 3,299 km2 and travels a length of 120 km before joining Lake Abaya (Figure 1). The river originates from the high mountain area of Gelala and covers the wide area of Sidama, Gedeo, and west Guji zones in the southern part of Ethiopia. It flows through the agroforestry landmarks and agricultural lands to feed Lake Abaya. The sub-basin has three sections, which are the lower rift floor near Lake Abaya, middle or escarpment, and highland plateau areas (Mechal 2015). The elevation of Gidabo ranges from 1,175 m.a.s.l. at the lower rift valley floor to 3,213 m.a.s.l. at Gelala mountain. The major land-use and land-cover classes in the sub-basin are forest, agriculture, coffee, agroforestry, and pastureland.

Figure 1

Location map of the study area in the Ethiopian Rift Valley Lakes Basin.

Figure 1

Location map of the study area in the Ethiopian Rift Valley Lakes Basin.

Close modal

The climate of the Gidabo sub-basin ranges from semi-arid at the lower rift floor to humid and temperate at mount Gelala. The area experiences a bimodal rainfall pattern with an annual rainfall of less than 800 mm/year at the lower rift floor and exceeds 1,600 mm/year at the highland and escarpments of the basin. Its mean annual rainfall is 1,100 mm/year. There are two peak periods of rainfall magnitude in the sub-basin. It has a short rainy season from March to May (MAM) and a long rainy season from August to October, whereas the dry season is between November and February. The mean annual minimum and maximum temperature of the basin ranges between 21 and 25 °C in the lower rift floor to 11.5 and 13.5 °C at the highland section of the basin (Debisso 2009; Mechal 2015).

Methods

Data set

Observed meteorological data, i.e., mean daily precipitation and minimum and maximum temperatures, were collected from the Ethiopian National Meteorology Agency (NMA). The Digital Elevation Model (DEM) and streamflow data were collected from the Ministry of Water, Irrigation and Energy (MoWIE) of Ethiopia. The meteorological data cover the period from 1971 to 2005, whereas the streamflow data cover from 1998 to 2005. The DEM had a spatial resolution of 30 m×30 m and was generated by Shuttle Radar Topography Mission (SRTM). Moreover, a land-use and land-cover map of 2013 was obtained from the MoWIE.

Climate outputs of multiple climate models were obtained from the COordinated Regional Climate Downscaling Experiment (CORDEX)-Africa domain, which had a 50 km×50 km spatial resolution. In order to select the climate models, the performance of 10 climate models was compared in terms of capturing aspects of the observed rainfall over the baseline period (1971–2000). Four outstanding models that better captured the study area rainfall conditions were selected for this study (Worako et al. 2002).

Climate models and bias correction methods

Among 10 GCM–RCM combinations, the top four best-performing climate models were selected for the impact study. The selection was done by using the entropy method and Preference Ranking Organization Method of Enrichment Evaluation (PROMETHEE-2) ranking (Worako et al. 2002).

The entropy method is a logarithmic function that helps to weight the performance indicators without the intervention of decision-makers. It enhances objectivity in the selection and prioritization of bias correction methods. Procedurally, it can be done by normalizing each indicator value by dividing the actual value to its sum value. Then the normalized value is transformed by the natural logarithmic function to change the value into entropy, and this process is called the entropy method. The detailed procedure is described in Raju & Kumar (2018).

A description of the selected GCMs–RCMs is presented in Table 1. The RCP4.5 scenario was selected since it considers the stabilization of radiative forcing at 4.5 W/m2 for the year 2100 without ever exceeding the value. Water resource management and planning use the 30-year climate change conditions for future periods. The RCP4.5 scenario for the mid-term (2041–2070) was used for this study.

Table 1

CORDEX-Africa domain climate models used in this study

GCMRCMInstitute/origin
CNRM-CM5 CLMcom-CCLM4-8-17 Centre National de Recherches Meteoroliques, France 
CSIRO-MK3-6-0 SMHI-RCA4 Commonwealth scientific and industrial research organization in collaboration with the Queensland Climate Change Center of Excellence, the Australia 
MIROC5 SMHI-RCA4 Atmosphere and Ocean Research Institute (The University of Tokyo), the National Institute for Environmental Studies, and the Japan Agency for Marine-Earth Science and Technology, Japan 
GFDL-ESM2M SMHI-RCA4 Geophysical Fluid Dynamics Laboratory, USA 
GCMRCMInstitute/origin
CNRM-CM5 CLMcom-CCLM4-8-17 Centre National de Recherches Meteoroliques, France 
CSIRO-MK3-6-0 SMHI-RCA4 Commonwealth scientific and industrial research organization in collaboration with the Queensland Climate Change Center of Excellence, the Australia 
MIROC5 SMHI-RCA4 Atmosphere and Ocean Research Institute (The University of Tokyo), the National Institute for Environmental Studies, and the Japan Agency for Marine-Earth Science and Technology, Japan 
GFDL-ESM2M SMHI-RCA4 Geophysical Fluid Dynamics Laboratory, USA 

All climate models have a 50 km×50 km spatial resolution.

After GCM–RCM selection, five rainfall bias correction methods were identified for this study. These methods are delta change (DT), linear scaling (LS), local intensity scaling (LOCI), power transformation (PT), and distribution mapping (DM), and one temperature bias correction method (DM) was applied to reduce the systematic biases of climate model outputs. Table 2 gives a description of the five bias correction methods.

Table 2

Description of the five bias correction schemes, which are considered for comparison in this study area

MethodsDescription
Delta change The ratio of the simulated and observed monthly mean rainfall is used to correct projected rainfall amounts. 
Linear scaling Adjusts the projected rainfall based on the average difference between monthly simulated and observed data. 
Local intensity scaling Corrects the mean rainfall amount and the wet day frequencies and wet day intensity of rainfall. 
Power transformation Uses a power equation to adjust the mean and variance of monthly rainfall. 
Distribution mapping Adjusts the mean, variance, standard deviation, wet-day frequencies, and intensities by assuming the same distribution of observed and simulated rainfall data. 
MethodsDescription
Delta change The ratio of the simulated and observed monthly mean rainfall is used to correct projected rainfall amounts. 
Linear scaling Adjusts the projected rainfall based on the average difference between monthly simulated and observed data. 
Local intensity scaling Corrects the mean rainfall amount and the wet day frequencies and wet day intensity of rainfall. 
Power transformation Uses a power equation to adjust the mean and variance of monthly rainfall. 
Distribution mapping Adjusts the mean, variance, standard deviation, wet-day frequencies, and intensities by assuming the same distribution of observed and simulated rainfall data. 

In addition to the entropy method, we used six rainfall indices to select the best-fit bias correction method for the study region. These indices are indicated in Expert Team on Climate Change Detection and Indices (ETCCDI) (https://www.wcrp-climate.org/etccdi). The indices are consecutive dry days (CDD, where a dry day has rainfall amounts lower than 1 mm), consecutive wet days (CWD, where a wet day has rainfall amounts lower than or equal to 1 mm), annual total wet-day rainfall (PRCPTOT), number of heavy rain days (R10 mm, days with rainfall amount≥10 mm), number of heavy rain days (R20 mm, days with rainfall amount≥20 mm), and average daily wet-day rainfall intensity (SDII) (Table 3). These indices are selected as they are related to normal and extreme events (Obada et al. 2021; Tang et al. 2021).

Table 3

Climate extreme indices used to show the performance of bias correction methods in the study area

 
 

Note: The shaded pixels show the method that performs best for each index.

General description of the Hydrologiska Byråns Vattenbalansavdelning model

Hydrologiska Byråns Vattenbalansavdelning (HBV) is a semi-distributed conceptually based hydrological model. It was developed to primarily simulate runoff of large river basins (Bergström 1976, 1992). Several studies demonstrated that the HBV model sufficiently simulates the streamflow of various watersheds (Enyew et al. 2014; Versini et al. 2016; Adera & Alfredsen 2019). HBV received a wide range of applications for the evaluation of climate change impact (Abdo et al. 2009; Teutschbein & Seibert 2012; Al-Safi & Sarukkalige 2017; Soriano et al. 2019). This is because it is more robust, requires less data, and is simple to operate for streamflow simulation for a climate change impact study.

The water balance equation that has been used in the HBV-96 model to simulate hydrological processes or components is in the form of the following equation:
formula
(1)
where P is precipitation, E is evapotranspiration, Q is runoff, SP is snowpack, SM is soil moisture, UZ is the lower groundwater zone, LZ is the lower groundwater zone, and Lakes is the lake volume.

HBV divides each sub-basin into four zones representing the land surface, soil moisture store, upper sub-surface reservoir, and lower sub-surface reservoir. Thus, there are four routines in the HBV-96 model, which are precipitation routine, soil moisture routine, river routing, and a response routine. Precipitation routine partitions the precipitation into rainfall and snow and generates the infiltration input to the soil zone. The soil moisture routine simulates temporal changes in the soil moisture content of the catchment based on the amount of water infiltrated to the soil. The soil moisture routine is determined by three parameters, namely, field capacity (FC), limit of potential evaporation (lp), and beta or soil shape calibration parameter (β). The response routine or function transforms excess water from soil moisture to runoff. The equations for these routines are described in several studies (Rientjes et al. 2011; Al-Safi & Sarukkalige 2017; Haile et al. 2017).

Model sensitivity, calibration, validation, and performance evaluation

To set up the HBV-96 hydrological model, the input data were processed as per the model requirement. These inputs are DEM, vegetation zones or land use land cover, daily mean temperature, rainfall, and potential evaporation.

The model's sensitivity to its parameters is determined on the basis of their effects on streamflow simulation. Most previous studies considered three parameters of the model to affect the simulated hydrograph volume. These parameters are beta (β) (controls the indirect relation between soil moisture and indirect runoff as related by a power function), FC, and a limit above which potential evaporation occurs (lp) (Rientjes et al. 2011; Haile et al. 2017). The simulated peak flows are mainly controlled by Alfa (α) (a measure for the nonlinearity of the flow in the quick runoff reservoir) and recession coefficient for the upper reservoir (khq). The hydrograph recession is determined by the recession coefficient for the lower reservoir (k4) and the rate of percolation (perc) (Table 4). In general, the aforementioned sensitivity parameters are used in various studies to calibrate the HBV-96 model and are used in this study as well.

Table 4

Default, optimum, and current model parameters after calibration of the Gidabo sub-basin for the period 1999–2005

Model parameterParameter value rangeDefault valueCurrent optimum value
Alfa 0–1.5 0.6 0.575 
Beta 1–4 2.5 1.444 
FC 100–1,500 200 800 
LP <=1 0.9 0.8 
K4 0.001–0.1 0.01 0.0115 
Khq 0.005–0.5 0.1 0.121 
Perc 0.01–6 0.5 0.3723 
Cflux 0–2 0.5 
Model parameterParameter value rangeDefault valueCurrent optimum value
Alfa 0–1.5 0.6 0.575 
Beta 1–4 2.5 1.444 
FC 100–1,500 200 800 
LP <=1 0.9 0.8 
K4 0.001–0.1 0.01 0.0115 
Khq 0.005–0.5 0.1 0.121 
Perc 0.01–6 0.5 0.3723 
Cflux 0–2 0.5 

This study used 1 year for model warm-up, 5 years (1999–2003) for model calibration, and 2 years (2004–2005) for model validation. The HBV-96 model is manually calibrated by progressively changing the value of one parameter at a time keeping values of other parameters fixed. Based on the data quality assessment, we found that the measured flow data in October 2000 and in 4 months of 2001 (January, August, September, and October) show suspicious (outlier values) records. Hence, we omitted these data from serving as a reference dataset to evaluate the model calibration.

Model performance is evaluated through a visual inspection of the hydrographs (peaks and low flows) and using objective functions. The goodness of fit between measured and simulated variables is tested by using two objective functions. These are relative volume error (RVE) and Nash–Sutcliffe Efficiency (NSE). These two objective functions are widely used in studies of climate change impact (Worqlul et al. 2018; Bizuneh et al. 2021; Mohammed et al. 2021). RVE is a measure of the systematic difference (bias) between the observed and the simulated streamflow volume, whereas the NSE measures the level of match between the patterns of the observed and the simulated streamflow hydrographs.

These objective functions are defined as follows:
formula
(2)
formula
(3)
where Qiobs is the ith value of the observed streamflow, Qisim is the ith value of the simulated streamflow, Qmean is the mean of the observed streamflow data, and n is the total number of observations.

An NSE value of 1 indicates a perfect match between the patterns of the observed and the simulated streamflow. NSE values in between 0.8 and 0.9, 0.7 and 0.8, and 0.5 and 0.7 indicate very good, good, and satisfactory performance, respectively. An RVE value of 0 indicates a perfect match between the average volume of the observed and the simulated streamflow. An RVE value in between +5 and −5% indicates good performance of the model, whereas a value between +5 and +10% or between −5 and −10% indicates fair performance (Rientjes et al. 2011; Haile et al. 2017; Bizuneh et al. 2021).

Potential evapotranspiration

In order to determine the potential evapotranspiration (PET), we used the Hargreaves method (Hargreaves & Samani 1985). This method requires readily available data, which are in the form of temperature and radiation data (Equation (4)). It considers the terrestrial radiation effect in evapotranspiration, and the equation reads:
formula
(4)
where PET is the potential evapotranspiration (mm day−1), Tmean is the mean average daily temperature (°C), Tmax and Tmin are the daily maximum and minimum temperatures (°C), respectively, and Ra is terrestrial radiation (MJ m−2 day−1).

Analysis of the projected streamflow

The effect of the bias correction methods was evaluated by estimating the relative percentage difference of the simulated rainfall, PET, and streamflow for baseline and future periods. Temperature is computed by the difference between the projected and the baseline period. The equations read:
formula
(5)
formula
(6)
formula
(7)
formula
(8)
where RVP, RVQ, RVET, and refer to the relative percentage difference of rainfall, streamflow, PET, and temperature of the future and baseline period simulated discharge. P refers to rainfall, while other terms are as defined previously.

Boxplots are used to show the influence of the bias correction methods on the daily rainfall and streamflow simulation. The plots were prepared for the five bias correction methods and four climate models. These plots provide the minimum, median, maximum, and interquartile range (IQR) values of rainfall and streamflow simulation. Hence, this information can be used to evaluate the effects of the bias correction methods on low, medium, and high streamflow and the corresponding variability.

In Worako et al., PT is selected on the basis of evaluating which bias correction method performed better (submitted article). We evaluated five bias correction methods and found that PT outfits the study area, and that is why we selected PT for streamflow simulation.

Hydrographs of streamflow

Figure 2 shows a good match between the patterns of the simulated and the observed streamflow for the mean monthly time step. During the rising limb and recession limb, the model well captured the pattern of the observed flow. However, there is some overestimation of the observed flow magnitude in some months.

Figure 2

Mean monthly observed and simulated streamflow for the calibration period (1998–2003) and the validation period (2004–2005) in the Gidabo sub-basin.

Figure 2

Mean monthly observed and simulated streamflow for the calibration period (1998–2003) and the validation period (2004–2005) in the Gidabo sub-basin.

Close modal

The objective function indicators of the HBV-96 hydrological model such as RVE and NSE indicated that the model performance can be rated good considering that the study area is located in a data-scarce region. In the calibration period, RVE=11% and NSE=0.58 were obtained and similar performance was achieved in the validation (RVE=9.1 and NSE=0.70) period (Figure 2). Hence, this suggests that the model can simulate the streamflow outside of the calibration period and can be applied for a climate change impact study. Table 4 contains the calibrated values of the HBV parameters for the study area.

Bias correction methods’ evaluation using rainfall indices

In Table 3, the values of the six selected rainfall indices are given for raw climate data, observed rainfall, and bias corrected data for the various bias correction methods. For the CNRM climate model, the DM bias correction method best captures the observed values of CDD, whereas DM and PT resulted in R20 mm values that are closest to the observed value. Although the DT method perfectly matches the observation data in terms of the remaining indices, its major constraint lies in its assumption that it is the best bias correction method (Mendez et al. 2020).

For the CSIRO climate model, LS and PT are best at capturing the observed CDD and PRCPTOT. The LOCI method performs best at representing four of the rainfall indices. Hence, it is difficult to be conclusive about which bias correction method performs best at representing all climate indices. Each bias correction method performs best at capturing certain aspects (indices) of the observed rainfall. This applies for all the climate models considered in this study, as shown in Table 3. However, the entropy method and PROMETHEE-2 ranking that combine multiple objectives indicate the PT to be the best method for bias correction in the study area.

Implications of bias correction methods on rainfall and streamflow simulation

Figure 3 shows boxplots of the future projected daily rainfall for RCP4.5 scenarios for the period of 2041–2070 for five bias correction methods and four climate models. The IQR (size of the boxes) varies in magnitude for the different climate models and bias correction methods. For instance, the DT bias correction method resulted in the smallest IQR (2.27 mm day−1 in GFDL to 3.64 mm day−1 in CNRM) of rainfall for all climate models. This suggests low variability of the projected daily rainfall amounts. Bias correction using PT resulted in the largest rainfall variability, i.e., 7.48 mm day−1 in GFDL to 13.27 mm day−1 in the CSIRO model. Each bias correction method indicated varying conditions in the future median rainfall in the sub-basin. The lowest median rainfall values are projected using the DT method, which suggests that the area will receive light rainfall amounts contrary to higher projections using the other bias correction methods. Both bias correction methods and climate models affect the magnitude of the projected median rainfall of the study area. The projected magnitude of the extreme rainfall (whisker) is the largest when the PT bias correction method is applied to the data from all climate models. However, DT resulted in the lowest extreme rainfall projection for the future. The difference in the projected extreme rainfall can be more than double because of using different bias correction methods.

Figure 3

Boxplots of the future projected daily rainfall for the RCP4.5 scenario for the period 2041–2070 for five bias correction methods and four climate models. DT refers to the delta change method, DM is the distribution mapping method, LS is linear scaling, LOCI is local intensity scaling, and PT is the power transformation method.

Figure 3

Boxplots of the future projected daily rainfall for the RCP4.5 scenario for the period 2041–2070 for five bias correction methods and four climate models. DT refers to the delta change method, DM is the distribution mapping method, LS is linear scaling, LOCI is local intensity scaling, and PT is the power transformation method.

Close modal

Figure 4 shows boxplots of future projected daily streamflow for RCP4.5 scenarios for the period of 2041–2070 for five climate models and four bias correction methods. The IQR varies in magnitude similar to what was observed in rainfall projection. For instance, DM has shown the smallest IQR (0.9 m3 s−1 in CSIRO to 2.3 m3 s−1 in MIROC) of streamflow for all climate models. This suggests low variability of the projected magnitude of streamflow. LOCI and LS bias correction methods resulted in the largest variability of the simulated future streamflow in all climate models. The magnitude (IQR) varies from model to model, i.e., 2.2 m3 s−1 in GFDL to 5.3 m3 s−1 in the MIROC model. Except for the LOCI and LS methods, each bias correction method indicated varying median streamflow projection. DM has shown lowest median values of projected streamflow as compared with other bias correction methods. The DT method indicated the highest median value of simulated streamflow in two climate models, i.e., GFDL and CSIRO models. With regard to extreme values, both the LOCI and LS methods indicated the largest magnitude in projected streamflow. However, DM has shown the lowest extreme value in future projected streamflow magnitude. Thus, both climate models and bias correction methods affect the magnitude of future projected streamflow conditions in the sub-basin.

Figure 4

Boxplots of the future projected daily discharge for the RCP4.5 scenario for the period 2041–2070 for five bias correction methods and four climate models. DT refers to the delta method, DM is the distribution mapping method, LS is linear scaling, LOCI is local intensity scaling, and PT is the power transformation method.

Figure 4

Boxplots of the future projected daily discharge for the RCP4.5 scenario for the period 2041–2070 for five bias correction methods and four climate models. DT refers to the delta method, DM is the distribution mapping method, LS is linear scaling, LOCI is local intensity scaling, and PT is the power transformation method.

Close modal

Future rainfall projection

Figure 5 shows the monthly rainfall change in percentage between the baseline period (1971–2000) and the future scenario period (2041–2070) for the four climate models. The figure illustrates high variability in the monthly difference of mean rainfall both in direction and in magnitude depending on the climate models. For instance, the highest decline in rainfall is projected in December and the maximum rainfall increase is in April for GFDL, whereas the highest decline is in October and the maximum increase is in January for MIROC climate models (Figure 5). CSIRO and MIROC showed a positive change (40–100%) in future rainfall amounts in January and February. However, the CNRM projections indicate a slight decline in the rainfall of these months, while GFDL projections indicate no change (except a slight decline in January). In November and December, all climate models showed a negative change (29–85%) in future projected rainfall, except CNRM, which indicated a positive change. During the short rainy season (MAM), particularly in March, most climate models showed a positive change (33–158%) in future rainfall magnitude, except CNRM, which showed a decline in rainfall. In April and May, all climate models showed likely increases in projected rainfall, except the CSIRO model. In most of the main rainy months (June, July, August, September, October (JJASO)), the future rainfall projections indicated likely decreases in rainfall amount. However, in MIROC, some months (July–September) showed a slightly positive change.

Figure 5

Mean monthly rainfall changes in percentage between the control period (1971–2000) and the RCP4.5 future scenario period (2041–2070) for the four climate models. The relative percentage rainfall difference using the power transformation bias correction method.

Figure 5

Mean monthly rainfall changes in percentage between the control period (1971–2000) and the RCP4.5 future scenario period (2041–2070) for the four climate models. The relative percentage rainfall difference using the power transformation bias correction method.

Close modal

Future projection of temperature

The basin shows increases in mean monthly minimum and maximum temperatures for the future period with reference to the baseline period (Figure 6). Mean monthly minimum temperature change increases from 0.8 to 1.5 °C, 0.8 to 3.6 °C, 0.6 to 2.1 °C, and 1.1 to 2.7 °C for the CNRM, CSIRO, GFDL, and MIROC models, respectively. Also, the mean monthly maximum temperature changes between the baseline and the future projected period show an increasing trend and varies from 0.3 to 1.2 °C, 0.2 to 2.5 °C, −0.4 to 2.3 °C, and 1.2 to 3.5 °C for the CNRM, CSIRO, GFDL, and MIROC models, respectively. The range of the difference in temperature varies with the climate models (Figure 6). The mean annual maximum temperature increases by 0.7, 1.6, 1.0, and 2.2 °C, whereas the mean annual minimum temperature increases by 1.3, 2.3, 1.4, and 1.8 °C for the CNRM, CSIRO, GFDL, and MIROC climate models, respectively (Table 5). In general, the future temperature is projected to increase in the study area.

Figure 6

Projected mean temperature change for the RCP4.5 scenario of the 2050 period using four climate model simulations compared with the control period (1971–2000): (a) mean minimum temperature change and (b) mean maximum temperature change.

Figure 6

Projected mean temperature change for the RCP4.5 scenario of the 2050 period using four climate model simulations compared with the control period (1971–2000): (a) mean minimum temperature change and (b) mean maximum temperature change.

Close modal

Future projection in PET

The projected relative difference between the baseline and the future scenario period indicated an increasing trend in PET in the sub-basin (Figure 7). Mean monthly PET shows an increasing trend that varies from 0.5% (May) to 17% (October) in the MIROC model, and other models also show increasing trends, except for a few months. The month that shows a decreasing trend of PET is April using the GFDL model that might be related to decreases in maximum temperature (Figure 7). The annual PET shows an increasing trend in all climate models, i.e., 51.5% in CNRM to 112% in the MIROC model.

Figure 7

Potential evapotranspiration relative percentage difference between the control period (1971–2000) and the future scenario period (2041–2070), RCP4.5, using four climate models.

Figure 7

Potential evapotranspiration relative percentage difference between the control period (1971–2000) and the future scenario period (2041–2070), RCP4.5, using four climate models.

Close modal

Future streamflow simulation

There is a high variability in flow conditions in the future period in the basin. However, the magnitude of flow variability depends on the climate models used to simulate streamflow. The relative percentage difference between the future and the baseline period streamflow shows minimum value during October in CSIRO and MIROC simulations and maximum value in March for CSIRO and in April for the MIROC model (Figure 8). The level of decline in mean monthly flow depends on the climate models. During half of the dry season (November and December), all climate models show a decline in future projected streamflow, except CNRM (will likely increases 93–97%). However, in January and February, most climate models show positive change in future projections of streamflow simulation. However, in GFDL, the whole dry season (November, December, January, February (NDJF)) indicates a decline in future projected streamflow simulation.

Figure 8

Mean monthly streamflow relative percentage difference between the control period (1971–2000) and the RCP4.5 future scenario period (2041–2070) for the four climate models. The relative percentage mean streamflow difference for the power transformation (PT) bias correction method.

Figure 8

Mean monthly streamflow relative percentage difference between the control period (1971–2000) and the RCP4.5 future scenario period (2041–2070) for the four climate models. The relative percentage mean streamflow difference for the power transformation (PT) bias correction method.

Close modal

During the short rainy season (MAM), all climate models show likely increases (16–294%) in the projected streamflow simulation. In the main rainy season (JJASO), CSIRO and GFDL show a decline in future projected streamflow simulation, whereas MIROC shows positive change in these months. In October, all climate models indicate likely decreases in future projected streamflow simulation, except CNRM (likely increases 39%). In general, the wet or Kiremt season streamflow shows a decreasing trend, whereas the short rainy or Belg season streamflow shows an increasing trend in the sub-basin.

Comparisons of climate models and bias correction methods in the direction of change for hydro-climatic variables

Most months show moderate agreement on the direction of rainfall change in the future period, i.e., three models show a decrease in future rainfall in 8 months. In June, all climate models strongly agree that there will be a decrease in rainfall amount in the future. However, in 3 months (January, February, and May) the climate models show weak agreement in the direction of the projected rainfall change (Table 6).

Table 5

Changes in mean temperature (°C) with the RCP4.5 scenario for the 2050s period using selected climate models

AnnualCNRMCSIROGFDLMIROC5
Tmax (2050 s) 0.7 1.6 1.0 2.2 
Tmin (2050 s) 1.3 2.3 1.4 1.8 
AnnualCNRMCSIROGFDLMIROC5
Tmax (2050 s) 0.7 1.6 1.0 2.2 
Tmin (2050 s) 1.3 2.3 1.4 1.8 
Table 6

Comparison of the agreements between the relative changes in climate model projection in the sub-basin

 
 

Note: Green refers to four models that show the same direction of change (strongly agree), yellow refers to three climate models that show the same direction (moderately agree), and red refers to two models that show the same direction (weakly agree).

Three of the climate models agreed on the direction of future streamflow change in 10 months. This suggests moderate agreement between the projections of the climate models. In April, all climate models strongly agree that the streamflow amount will decline in the future. However, in June, there is weak agreement in the projections of the climate models, i.e., two models show an increase, and two models show a decrease in the future streamflow. Most models show strong agreement in the direction of change in PET, except for April (which shows a decrement in projected PET by one model).

All bias correction methods (except DM) resulted in weak agreement between the projected changes in annual rainfall and streamflow (Table 7). When the DM method is applied on the projections of the four climate models, it resulted in a decreasing trend of the projected rainfall in the sub-basin. The direction of the relative streamflow change indicates moderate agreement in PT and strong agreement in the DM method. However, a weak agreement was obtained between the direction of projected annual streamflow change when the projection of the four climate models was bias-corrected using all methods, except PT and DM.

Table 7

Comparisons of the agreement between climate models in five bias correction methods

 
 

Note: Green (strongly agree) indicates four climate models that show the same direction of change; yellow (moderately agree) indicates three climate models that show the same direction of bias correction methods, and red (weakly agree) indicates two climate models that shows the same direction of change from five bias correction methods.

Projected changes in water availability are subject to uncertainties from various sources. In this study, the effect of bias correction of rainfall simulation by climate models is investigated for both historic and future periods. As indicated by boxplots, the DT method resulted in the lowest rainfall variability, whereas PT indicated the largest variability in future projected rainfall. Overall, the rainfall projection is subject to bias correction methods and climate models in the sub-basin. Different studies confirmed our findings in the various parts of the world (Teutschbein & Seibert 2012; Soriano et al. 2019). Similar to rainfall projections, the influence of bias correction is observed in streamflow projections, which indicated variation in streamflow magnitude. For instance, the DM method has shown the lowest variability in projected streamflow, whereas LOCI and LS indicated the largest variability. The magnitude of variation depends on the climate models as well. In general, the results indicate that comparing several bias correction methods is better than randomly selecting the single bias correction method for climate change impact analysis. Using a randomly selected single bias correction method may lead to wrong decision-making.

An evaluation of multiple bias correction methods creates an opportunity to select the best performing one that fits the study area and climate models (Teutschbein & Seibert 2012). Hence, this study used five rainfall bias correction methods to determine climate change impact on water resource availability in the region. The performance of bias correction methods shows large variability between climate models for streamflow projection in the basin. Based on the findings of Worako et al. (2002), the PT method performs better than the DM method in removing systematic errors of rainfall in the historical period for the study area. Other study findings also confirmed the superiority of PT over other bias correction methods (Fang et al. 2015; Luo et al. 2018). However, the DM method shows better performance in the projection of streamflow simulation in the sub-basin, which indicates that the bias correction method that works best in rainfall error reduction may not work best in streamflow projection. This may be due to the hydrological model uncertainty and also the nonlinearity of the rainfall–runoff relationship.

The mean monthly rainfall shows a decreasing tendency in most months in different climate models. However, the magnitude of the decrease varies with climate models and bias correction methods. Three climate models moderately agree on the decrement of projected rainfall in 6 months of a year. Regarding the bias correction method, the two climate models (CNRM and MIROC) indicate a strong agreement in terms of an increase in the projected future rainfall and vice versa in the rest of the models. The mean annual rainfall shows decreases in two models (GFDL and CSIRO) and an increase in the other two models (CNRM and MIROC). Moreover, the rainfall amount of the short rainy season rainfall will likely increase, and the mean rainfall amount in the main rainy season decreases in all climate models. This result confirms the finding of Gadissa et al. (2019) in the Central Rift Valley Lakes Basin and Taye et al. (2018) in the Awash River Basin.

The mean monthly, seasonal, and annual minimum and maximum temperatures will likely increase in the future period. The temperature increase is within the threshold of the global climate change agreement, i.e., 1.5 and 2.0 °C (UNFCCC 2015; Liu et al. 2017). All climate models strongly agree with regard to the likely increase of the projected future PET in the basin.

The mean monthly streamflow will likely decrease between June and February (GFDL), May and December (CSIRO), and August and December (MIROC) in the basin. Three climate models moderately agree with regard to the future decrement of streamflow in 6 months, whereas the projected streamflow will likely decrease in 5 months. The streamflow of the short rainy season (MAM) most likely increases in the future period, which is suitable for agricultural activities. On the contrary, the streamflow will decrease in the long rainy (JJASO) and dry (NDJF) seasons. The findings of our study agree with those of other studies in different basins in Ethiopia (Tedla et al. 2015; Tekle 2015; Chaemiso et al. 2016). The projected decrement in streamflow will lead to exacerbated water scarcity in the sub-basin due to climate change. Future studies can explore suitable adaptation measures to reduce the climate change impact in the region.

This study was designed to analyze the implications of bias correction methods on the projection of climate change impact on surface water resources in the Gidabo sub-basin. We used the CORDEX-African domain climate model for the future scenario period simulation of streamflow. The HBV hydrological model was applied to simulate the streamflow in the sub-basin in order to assess the impacts of climate change on water availability. The following conclusions were drawn on the basis of the findings of the work:

The implications of bias correction methods on water availability were evaluated by using boxplots to show how low, medium, and maximum daily streamflows are affected. The IQR varies in magnitude when simulated streamflow was based on rainfall simulations of different climate models and bias correction methods. For example, the DM method shows low variability, i.e., 0.9 m3 s−1 (CSIRO) to 2.3 m3 s−1 (MIROC), whereas the LOCI and LS methods indicate the largest variability, i.e., 2.2 m3 s−1 (GFDL) to 5.3 m3 s−1 (MIROC), in future streamflow projection.

The mean monthly rainfall shows a decreasing tendency in most months in different climate models. However, the magnitude of decrease varies with climate models and bias correction methods. Mean annual rainfall indicates a mixed pattern depending on the climate model. Moreover, the mean rainfall amount in the small rainy season will likely increase, and the rainfall amount in the main rainy season decreases in the future.

The mean monthly, seasonal, and annual minimum and maximum temperatures will likely increase in the future period. However, it is within the desired global warming level for the period of the 2050s. The warming increases the PET in the sub-basin, which, in turn, raises water consumption. The short rainy season, Belg, streamflow is projected to increase, whereas the long rainy season, Kiremt, streamflow is projected to decrease. Climate change will affect the surface water availability of the sub-basin, which calls for designing a better water management strategy under uncertainty.

Data cannot be made publicly available; readers should contact the corresponding author for details.

Abdo
K. S.
,
Fiseha
B. M.
,
Rientjes
T. H. M.
,
Gieske
A. S. M.
&
Haile
A. T.
2009
Assessment of climate change impacts on the hydrology of Gilgel Abay catchment in Lake Tana basin, Ethiopia
.
Hydrol. Processes
23
,
3661
3669
.
Adera
A. G.
&
Alfredsen
K. T.
2019
Climate change and hydrological analysis of Tekeze river basin Ethiopia
.
J. Water Clim. Change
.
Andersson
L.
,
Samuelsson
P.
&
Kjellström
E.
2011
Assessment of climate change impact on water resources in the Pungwe river basin
.
Tellus A: Dyn. Meteorol. Oceanogr.
63
(
1
),
138
157
.
Ashenafi
A. A.
2014
Modeling Hydrological Responses to Changes in Land Cover and Climate in Geba River Basin, Northern Ethiopia
.
PhD Dissertation
,
Freie Universität Berlin
,
Germany
.
Ayele
H. S.
,
Li
M.-H.
,
Tung
C.-P.
&
Liu
T.-M.
2016
Impact of climate change on runoff in the Gilgel Abbay watershed, the Upper Blue Nile Basin, Ethiopia
.
Water
8
,
380
.
doi:10.3390/w8090380
.
Bergström
S.
1976
Development and Application of a Conceptual Runoff Model for Scandinavian Catchments. Report RHO 7
.
SMHI
,
Norrköping
,
Sweden
.
Bergström
S.
1992
The HBV Model – Its Structure and Applications. Hydrologi Report RH 4
.
SMHI
,
Norrköping
,
Sweden
.
Biniyam
Y.
&
Kemal
A.
2017
The impacts of climate change on rainfall and flood frequency: the case of Hare Watershed, Southern Rift Valley of Ethiopia
.
J. Earth Sci. Clim. Change
8
,
383
.
doi:10.4172/2157-7617.1000383
.
Bizuneh
B. B.
,
Moges
A. M.
,
Sinshaw
G. B.
&
Kerebih
M. S.
2021
SWAT and HBV models’ response to streamflow estimation in the upper Blue Nile Basin, Ethiopia
.
Water-Energy Nexus
4
,
41
53
.
Charles
S. P.
,
Chiew
F. H. S.
,
Potter
N. J.
,
Zheng
H.
,
Fu
G.
&
Zhang
L.
2019
Impact of downscaled rainfall biases on projected runoff changes
.
https://doi.org/10.5194/hess-2019-375
.
Chen
J.
,
Brissette
F. P.
,
Chaumont
D.
&
Braun
M.
2013
Finding appropriate bias correction methods in downscaling precipitation for hydrologic impact studies over North America
.
Water Resour. Res.
49
,
4187
4205
.
doi:10.1002/wrcr.20331
.
Chen
J.
,
Brissette
F. P.
&
Lucas-Picher
P.
2015
Assessing the limits of bias-correcting climate model outputs for climate change impact studies
.
J. Geophys. Res. Atmos.
120
,
1123
1136
.
doi:10.1002/2014JD022635
.
Chen
J.
,
Brissette
F. P.
&
Caya
D.
2020
Remaining error sources in bias-corrected climate model outputs
.
Clim. Change
.
https://doi.org/10.1007/s10584-020-02744-z
.
Cherinet
Z. N.
2013
Downscaling and Modeling the Effects of Climate Change on Hydrology and Water Resources in the Upper Blue Nile River Basin, Ethiopia
.
PhD Dissertation
,
Kassel University
,
Germany
.
Debisso
B.
2009
Groundwater Assessment in the Gidabo River Catchment, Southern Main Ethiopian Rift Province/Yirgalem-Wonago/
.
MSc Thesis
,
Addis Ababa University
,
Ethiopia
.
Demissie
T. A.
,
Saathoff
F.
,
Sileshi
Y.
&
Gebissa
A.
2013
Climate change impacts on the streamflow and simulated sediment flux to Gilgel Gibe 1 hydropower reservoir – Ethiopia
.
Eur. Int. J. Sci. Technol.
2
(
2
),
63
77
.
Dessu
S. B.
2013
Water Demand and Allocation in the Mara River Basin, Kenya/Tanzania in the Face of Land use Dynamics and Climate Variability
.
FIU Electronic Thesis and Dissertations
.
Paper 861
.
Enayati
M.
,
Bozorg-Haddad
O.
,
Bazrafshab
J.
,
Hejabi
S.
&
Chu
X.
2021
Bias correction capabilities of quantile mapping methods for rainfall and temperature variables
.
J. Water Clim. Change
12 (2), 401–419.
Enyew
B. D.
,
Van Lanen
H. A. J.
&
Van Loon
A. F.
2014
Assessment of the impact of climate change on hydrological drought in Lake Tana catchment, Blue Nile Basin, Ethiopia
.
J. Geol. Geosci.
3
,
174
.
doi:10.4172/2329-6755.1000174
.
Gadissa
T.
,
Nyadawa
M.
,
Mutua
B. M.
&
Behulu
F.
2019
Comparative assessment of the effect of climate change and human activities on streamflow regimes in Central Rift Valley Basin, Ethiopia
.
Am. J. Water Resour.
7
(
1
),
23
29
.
doi:10.12691/ajwr-7-1-4
.
Goitom
H.
,
De Smedt
F.
,
Yohannes
T. G.
,
Walraevens
K.
,
Gebrehiwot
K.
,
Bauer
H.
&
Deckers
J.
2012
Modeling climate change impact in the Geba Basin, Ethiopia
. In:
International Conference on Environmental, Biomedical and Biotechnology
.
IACSIT Press
,
Singapore
.
Haile
A. T.
,
Akawka
A. L.
,
Berhanu
B.
&
Rientjes
T.
2017
Changes in water availability in the Upper Blue Nile basin under the representative concentration pathways scenario
.
Hydrol. Sci. J.
doi:10.1080/02626667.2017.1365149
.
Hargreaves
G. H.
&
Samani
Z. A.
1985
Reference crop evapotranspiration from temperature
.
Appl. Eng. Agric.
1
,
96
99
.
IPCC
2007
Climate change: impacts, adaptation and vulnerability
. In:
Contribution of Working Group II to the Fourth Assessment, Report of the Intergovernmental Panel on Climate Change
(
Parry
M. L.
,
Canziani
O. F.
,
Palutikof
J. P.
,
van der Linden
P. J.
&
Hanson
C. E.
, eds).
Cambridge University Press
,
Cambridge
,
UK
.
IPCC
2013
Summary for policymakers//climate change: the physical science basis
. In:
Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
. (T. F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex & P. M. Midgley, eds.)
Cambridge University Press
, Cambridge, UK.
Koch
M.
&
Cherie
N.
2013
SWAT-Modeling of the Impact of Future Climate Change on the Hydrology and the Water Resources in the Upper Blue Nile River Basin, Ethiopia
.
doi:10.5675/ICWRER_2013
.
Laux
P.
,
Rotter
R. P.
,
Webber
H.
,
Dieng
D.
,
Rahimi
J.
,
Wei
J.
,
Faye
B.
,
Srivastava
A. K.
,
Bliefernicht
J.
,
Adeyeri
O.
,
Arnault
J.
&
Kunstmann
H.
2021
To bias correct or not to bias correct? An agricultural impact modelers’ perspective on regional climate model data
.
Agric. For. Meteorol.
304–305
,
108406
.
https://doi.org/10.1016/j.agrformet.2021.108406
.
Legesse
D.
,
Abiye
T. A.
,
Vallet-Coulomb
C.
&
Abate
H.
2010
Streamflow sensitivity to climate and land cover changes in Meki River, Ethiopia
.
Hydrol. Earth Syst. Sci.
14
,
2277
2287
.
Mechal
A.
2015
Hydrogeology of Complex Rift Valley System: The Example of Gidabo River Basin, Ethiopian Rift
.
PhD Dissertation
,
University of Graz
,
Austria
.
Mendez
M.
,
Maathuis
B.
,
Hein-Griggs
D.
&
Alvarado-Gamboa
L.-F.
2020
Performance evaluation of bias correction methods for climate change monthly precipitation projections over Costa Rica
.
Water
12
,
482
.
doi:10.3390/w12020482
.
Mohammed
S. A.
,
Solomatine
D. P.
,
Hrachowitz
M.
&
Hamouda
M. A.
2021
Impact of dataset size on the signature-based calibration of a hydrological model
.
Water
13
,
970
.
https://doi.org/10.3390/w13070970
.
Mostafa
H.
,
Saleh
H.
,
El Sheikh
M.
&
Kheireldin
K.
2015
Assessing statistical downscaling to quantify the impact of climate changes of rainfall over the Blue Nile Basin
.
Life Sci. J.
12
(
12
),
86
95
.
Obada
E.
,
Alamou
E. A.
,
Biao
E. I.
&
Zandagba
E. B. J.
2021
Interannual variability and trends of extreme rainfall indices over Benin
.
Climate
9
,
160
.
https://doi.org/10.3390/cli9110160
.
Raju
K. S.
&
Kumar
N. D.
2018
Selection of Global Climate Models
.
Springer Nature
.
https://doi.org/10.1007/978-981-10-6110-3_2
.
Rientjes
T. H. M.
,
Perera
B. U. J.
,
Haile
A. T.
,
Reggiani
P.
&
Muthuwatta
L. P.
2011
Regionalisation for lake level simulation-the case of Lake Tana in the Upper Blue Nile, Ethiopia
.
Hydrol. Earth Syst. Sci.
15
,
1167
1183
.
Setegn
S. G.
,
Dargahi
B.
,
Srinivasan
R.
&
Melesse
A. M.
2009
Modelling of sediment yield from Anjeni Gauged watershed, Ethiopia using SWAT model
.
J. Am. Water Resour. Assoc. (JAWRA)
46 (3), 514–526.
Shanka
A. S.
2017
Evaluation of climate change impacts on run-off in the Gidabo River Basin: Southern Ethiopia
.
Environ. Pollut. Clim. Change
1
,
129
.
doi:10.4172/2573-458X.1000129
.
Shrestha
S.
2014
Climate Change Impacts and Adaptation in Water Resources and Water use Sectors
.
Springer Water
.
doi:10.1007/978-3-319-09746-6_2
.
Tang
B.
,
Hu
W.
&
Duan
A.
2021
Assessment of extreme precipitation indices over Indochina and South China in CMIP6 models
.
J. Clim.
34
,
7507
7524
.
doi:10.1175/JCLI-D-20-0948.1
.
Taye
M. T.
,
Dyer
E.
,
Hirpa
F. A.
&
Charles
K.
2018
Climate change impact on water resources in the Awash basin, Ethiopia
.
Water
10
,
1560
.
doi:10.3390/w10111560
.
Taye
M. T.
,
Haile
A. T.
,
Fekadu
A. G.
&
Nakawuka
P.
2021
Effect of irrigation water withdrawal on the hydrology of the Lake Tana sub-basin
.
J. Hydrol.:Reg. Stud.
38
,
100961
.
https://doi.org/10.1016/j.ejrh.2021.100961
.
Tedla
E.
,
Gebremichae
M.
,
Hailu
D.
&
Seyoum
S.
2015
River basin scale long-term hydropower operational planning under changing climate
.
Int. J. Appl. Innovation Eng. Manage. (IJAIEM)
4
(
2
),
43
58
.
Tekle
A.
2015
Assessment of climate change impact on water availability of Bilate Watershed, Ethiopian Rift Valley Basin
. In
IEEE
.
978-1-4799-7498-6/15
.
Teng
J.
,
Potter
N. J.
,
Chiew
F. H. S.
,
Zhang
L.
,
Wang
B.
,
Vaze
J.
&
Evans
J. P.
2015
How does bias correction of regional climate model precipitation affect modelled runoff?
Hydrol. Earth Syst. Sci.
19
,
711
728
.
UNFCCC
2015
FCCC/CP/2015/L.9/Rev.1: Adoption of the Paris Agreement
.
UNFCCC
,
Paris
, pp.
1
32
Velazquez
J. A.
,
Troin
M.
,
Caya
D.
&
Brissette
F.
2015
Evaluating the time-invariance hypothesis of climate model bias correction: implications for hydrological impact studies
.
J. Hydrometeorol.
16
,
2013
2026
.
doi:10.1175/JHM-D-14-0159.1
.
Versini
P.-A.
,
Pouget
L.
,
McEnnis
S.
,
Custodio
E.
&
Escaler
I.
2016
Climate change impact on water resources availability – case study of the Llobregat River basin (Spain)
.
Hydrol. Sci. J.
doi:10.1080/02626667.2016.1154556
.
Worako
A. W.
,
Haile
A. T.
,
Rientjes
T. H. M.
&
Gebrekristos
S. T.
2002
Integrated multi-objective performance evaluation of CORDEX-Africa climate models and bias correction schemes
.
Physics and Chemistry of the Earth Journal
,
Submitted version
.
Wörner
V.
,
Kreye
P.
&
Meon
G.
2019
Effects of bias-correcting climate model data on the projection of future changes in high flows
.
Hydrology
6
,
46
.
doi:10.3390/hydrology6020046
.
Worqlul
A. W.
,
Dile
Y. T.
,
Ayana
E. K.
,
Jeong
J.
,
Adem
A. A.
&
Gerik
T.
2018
Impact of climate change on streamflow hydrology in headwater catchments of the Blue Nile basin, Ethiopia
.
Water
10
,
120
.
doi:10.3390/w10020120
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).