The purpose of this study is to predict future changes in precipitation in the Central Ethiopian Main Rift, which is vulnerable to climate change. The Long Ashton Research Station Weather Generator (LARS-WG) model was applied to project precipitation based on five global climate models (GCMs) (EC-EARTH, MIROC-ESM, HadGEM2-ES, INM-CM4, and CCSM4) from Coupled Model Intercomparison Project phase 5 (CMIP5) under two representative concentration pathways (RCP4.5 and RCP8.5) in the periods of 2041–2060 compared to the baseline period of 1976–2005. The model's calibration and validation results showed that it could predict future precipitation. According to the analysis, the mean rainfall is expected to increase in January (up to 14.2%) and December (up to 27.8%) under the RCP 4.5 and RCP 8.5 scenarios, respectively. However, a drop is anticipated in June (up to 8.2%) and May (up to 7%) under the RCP 4.5 and RCP 8.5 scenarios, respectively. In both scenarios, summer precipitation (usually the rainy season) is predicted to fall, while winter precipitation (usually the dry season) is expected to climb. Furthermore, annual and spring precipitation forecasts are anticipated to decrease in most locations. The findings of this research will be utilized to guide future water resource management in the study region.

  • The multi-model ensemble (MME) mean of CMIP5 global climate model (GCM) outputs was used to address uncertainty associated with a single GCM.

  • Precipitation is anticipated to decrease during the regularly wet season while increasing during the normally dry season in the years 2041–2060.

  • The projected mean annual, seasonal, and monthly rainfall could be used to influence future water resource management in the Central Ethiopian Main Rift.

Graphical Abstract

Graphical Abstract
Graphical Abstract

The increase in the atmospheric concentrations of greenhouse gases, most notably carbon dioxide, has led to long-term climate change, which has been observed at global, regional, and ocean basin scales. Among them are changes in precipitation amounts and timings, arctic temperatures, and various types of extreme weather, such as heavy rain, droughts, and heatwaves (IPCC 2007). Africa, particularly East Africa, is extremely vulnerable to climate change, with more extreme events such as frequent droughts, floods, and heavy rains expected in the future (Gebrechorkos et al. 2019). A number of research studies on the future climate change scenario in the Central Ethiopian Main Rift (CEMR) have been conducted. For example, Zeray et al. (2009) used the statistical down scaling model (SDSM) to downscale precipitation in the Ziway Lake watershed of CEMR, from 2001 to 2099, based on HadCM3 (Hadley Centre Coupled Model, version 3) GCM (general circulation model) outputs under A2 and B2 SRES (Special Report on Emission Scenarios). They discovered that, based on the seasonal pattern, the spring season is more likely to have a lower total precipitation share than the other seasons. The winter season, on the other hand, may see a rise in overall share. Abraham et al. (2018) employed multi-model results from three Coupled Model Intercomparison Project phase 5 (CMIP5) GCMs (HadGEM2-ES, CSIROMK-3-6-0, and CCSM4 models) under representative concentration pathway (RCP) scenarios (RCP4.5 and RCP8.5) over the Lake Ziway Catchment in the CEMR. The results of the GCMs were retrieved from the KNMI climate explorer web page. Their findings indicated that precipitation will decrease in the future periods (the 2020s, 2050s, and the 2080s). Kassie et al. (2014) used four GCMs (HadCM3, CSIRO2, CGCM2, and PCM) and two emission scenarios, SRES (A2 and B1) to investigate future rainfall projections in the CEMR for the 2080s relative to the 1971–90 baseline periods. The climate change scenario data for these GCM-SRES combinations were extracted from the Tyndall Centre for Climate Change's TYN CY 3.0 data set. According to their findings, rainfall will increase outside of the growing season (November–December), but decrease during the growing season.

The outputs of GCMs are coarse; they must be downscaled to very fine resolution, or even to a local or station scale. In general, there are two types of downscaling methods widely used to transform coarse-scale information to a finer scale: dynamic and statistical downscaling (Ekström et al. 2015). As a result, Gebrechorkos et al. (2019) recommended adopting statistical downscaling models rather than dynamic downscaling methods for East Africa. This could be due to the fact that statistical models are quick, easy, and effective, requiring fewer computational resources and costs. Semenov & Barrow (1997) and Semenov (2008) used the Long Ashton Research Station Weather Generator (LARS-WG), which is one of the most well-known statistical downscaling tools for assessing the impact of climate change. It is a computationally inexpensive downscaling tool for generating local-scale climate scenarios from global or regional climate models. It simulates time series of daily weather at a single site.

In recent years, many studies have been conducted to assess the future precipitation changes using the LARS-WG model. For instance, Khoi et al. (2021) used the LARS-WG tool to predict future climate in the Be River Basin in Southern Vietnam using an ensemble of five CMIP5 GCMs (Can-ESM2, CNRM-CM5, HadGEM2-AO, IPSL-CM5A-LR, and MPI-ESM-MR) under two RCP scenarios (RCP4.5 and RCP8.5). They discovered that annual precipitation will decrease by 4.0% in the 2030s and rise by 1.6 and 6.4% in the 2050s and 2070s for the RCP4.5. According to RCP8.5, annual precipitation will decrease by 4.5 and 0.6% in the 2030s and 2050s, respectively, then increase by 5.1% in the 2070s. Chisanga et al. (2017) tested the LARS-WG model to see if it could generate synthetic weather data for 2020 and 2055 based on HadCM3 and BCCR-BCM2 GCM predictions under SRB1 and SRA1B scenarios at Mount Makulu, Zambia. They found that precipitation amounts would increase in 2020 and 2055, based on the ensemble mean of HadCM3 and BCM2 GCMs.

Prior research in Ethiopia employed the previous version 5.5 LARS-WG model to estimate future climate change, which was based on Special Emissions Scenarios SRES (A1B and B1) from the Coupled Model Intercomparison Projects phase 3 (CMIP 3) GCMs and the IPCC's Fourth Assessment Report (AR4). For instance, Fenta & Disse (2018) used the LARS-WG 5.5 model to see if it could capture extreme events and daily precipitation distributions across the whole data range in Ethiopia's upper Blue Nile Basin. Furthermore, Disasa et al. (2019) applied the LARS-WG 5.5 model at the Hawassa site in Ethiopia's Rift Valley Lakes Basin and discovered that for all three time periods (2020s, 2055s, and 2090s), rainfall decreases in summer and increases in winter.

As a result, the current research used the LARS-WG model's latest version 6.0, which includes climate forecasts from the IPCC Fifth Assessment Report's (AR5) CMIP5 ensembles. According to Taylor et al. (2012), CMIP5 comprises more extensive models and enhanced experiments than CMIP3. Furthermore, higher spatial resolution models with a larger number of output fields are used in CMIP5.

Changes in precipitation patterns raise concerns about water resource management because extreme events such as droughts and floods are likely to become more frequent and severe in the future (Brekke et al. 2009). Hence, the LARS-WG model performed well in reflecting the statistical properties of observed climatic variables, including extreme events (Qian et al. 2008; Semenov 2008; Fenta & Disse 2018), which encouraged its adoption in the present study.

The prediction of changes in critical climate variables like precipitation is extremely important in order to make informed decisions about the management of the available water resources (Jayanta et al. 2015). The analysis of precipitation changes for the future is of great importance since the CEMR research region is mainly characterized by rain-fed agricultural activities. The purpose of this study is to project future precipitation change in the CMER by using the LARS-WG 6.0 model based on the ensemble mean of five CMIP5 GCMs (EC-EARTH, MIROC-ESM, HadGEM2-ES, INM-CM4, and CCSM4) under RCP4.5 and RCP8.5 scenarios for the period of 2041–2060. The study's findings will be used to guide future water resource management and planning in the CEMR region.

Study area descriptions

This study focused on the CEMR, which consists of five lakes (Ziway, Abijata, Langano, Shalla, and Hawassa) and their watersheds with the Feeder Rivers, as shown in Figure 1. The CEMR is located at 38.22 and 39.31° E longitudes, and 6.75 and 8.47° N latitudes, with a total geographical area of about 15,880 km2 (MoWR 2008). The topographic features of the CEMR were analyzed using the Shuttle Radar Topography Mission Digital Elevation Model (SRTM DEM) with a 30 m resolution (obtained from https://www.usgs.gov/). The maximum and minimum heights, based on DEM data, are 4,188 and 1,546 m (above mean sea level), respectively. The highlands are the sources of rivers and their tributaries that pour into the lakes, as shown in this diagram.

Figure 1

Location map showing the CEMR with rivers, five lakes, and five meteorological stations online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/wcc.2022.440.

Figure 1

Location map showing the CEMR with rivers, five lakes, and five meteorological stations online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/wcc.2022.440.

Close modal

Meteorological data

The National Meteorological Agency (NMA) of Ethiopia provided daily time series of meteorological data (rainfall) for the period 1976–2005 (30 years) for this study. These data were utilized as a baseline for future climate data simulation (precipitation). Due to the limited data availability, five stations were used in this study, which are located in the northwest (e.g. Butajira), in the northeast (e.g. Assela), in the central (e.g. Ziway and Bulbula), and in the southern part of the basin (e.g. Hawassa). Figure 1 depicts the distribution of meteorological stations.

The climate of the study region is divided into three seasons: (i) the main rainy season in the summer (July–September), locally known as Kiremt rains, which account for 50–70% of annual rainfall totals; (ii) the short rainy season in the spring (March–May), known locally as Belg, which accounts for 20–30% of the yearly amount; and (iii) the dry season in the winter (October–January/February), known locally as Bega, which may have sporadic rains, accounts for 10–20% of the annual total (Legesse et al. 2004). Table 1 shows that the study region receives the annual mean rainfall between 717.68 and 1,089.27 mm, the summer mean rainfall between 407.74 and 617.89 mm, the spring mean rainfall between 205.68 and 353.14 mm, and the winter mean rainfall between 71.14 and 166.82 mm, regarding the time-series data from 1976 to 2005 (the baseline period of this study).

Table 1

Mean rainfall for various time scales (annual, spring, summer, and winter) for 1976–2005 (the baseline period)

StationLatitudeLongitudeAltitudeMean rainfall
AnnualSummerSpringWinter
Butajira 8.15 38.37 2,000 1,087.93 560.55 353.14 115.16 
Ziway 7.933 38.7 1,640 753.12 435.17 209.53 71.14 
Assela 7.956 39.14 2,413 1,089.27 617.89 322.48 108.98 
Bulbula 7.72 38.65 1,606 717.68 407.4 205.68 78.06 
Hawassa 7.065 38.48 1,694 964.45 444.95 303.81 166.82 
StationLatitudeLongitudeAltitudeMean rainfall
AnnualSummerSpringWinter
Butajira 8.15 38.37 2,000 1,087.93 560.55 353.14 115.16 
Ziway 7.933 38.7 1,640 753.12 435.17 209.53 71.14 
Assela 7.956 39.14 2,413 1,089.27 617.89 322.48 108.98 
Bulbula 7.72 38.65 1,606 717.68 407.4 205.68 78.06 
Hawassa 7.065 38.48 1,694 964.45 444.95 303.81 166.82 

General circulation models

In the present study, as shown in Table 2, the output from the five CMIP5 GCMs (EC-EARTH, MIROC-ESM, HadGEM2-ES, INM-CM4, and CCSM4) were utilized to generate the future scenarios of rainfall for the period 2041–2060 (2050s) under two RCP scenarios (RCP 4.5 and RCP 8.5). RCP4.5 is an intermediate stabilization scenario with an equivalent concentration of CO2 varying from 580 to 720 ppm in 2100, while RCP8.5 is a high emission scenario representing an equivalent concentration of CO2 above 1,000 ppm in 2100 (IPCC 2013).

Table 2

Five CMIP5 GCMs used in this study

Institute (modeling center)Model nameAtmospheric resolution
European Community Earth-System, consortium, Europe EC-EARTH 1.125°×1.125° 
Met Office Hadley Centre, UK HadGEM2-ES 1.25°×1.88° 
National Institute for Environmental Studies, The University of Tokyo, Japan MIROC-ESM 2.8125°×2.8125° 
Institute for Numerical Mathematics(INM), Russia INM-CM4 1.5°×2° 
Community Climate System Model(CCSM), National Centre for Atmospheric Research(NCAR), USA CCSM4 0.9°×1.25° 
Institute (modeling center)Model nameAtmospheric resolution
European Community Earth-System, consortium, Europe EC-EARTH 1.125°×1.125° 
Met Office Hadley Centre, UK HadGEM2-ES 1.25°×1.88° 
National Institute for Environmental Studies, The University of Tokyo, Japan MIROC-ESM 2.8125°×2.8125° 
Institute for Numerical Mathematics(INM), Russia INM-CM4 1.5°×2° 
Community Climate System Model(CCSM), National Centre for Atmospheric Research(NCAR), USA CCSM4 0.9°×1.25° 

The uncertainties originate from the selection of the emission scenarios, GCM outputs, downscaling techniques, and hydrological models (Hoan et al. 2018; Vesely et al. 2019). In this study, for assessing the future precipitation simulation, the multi-model ensemble (MME) mean approach with different climate scenarios was utilized, which could help to reduce the uncertainty associated with using a single climate emission scenario and GCM in climate predictions (Fenta & Disse 2018; Bekele et al. 2021; Khoi et al. 2021).

Description of LARS-WG stochastic weather generator

This study employed the latest version, 6.0 LARS-WG, to simulate future climate change for the period 2041–2060 (2050s), which integrates climate forecasts from the CMIP5 used in the IPCC Fifth Assessment Report (AR5) (Table 2). Version 6.0 of the LARS-WG is available for download at https://sites.google.com/view/lars-wg/. LARS-WG is a stochastic weather generator, which can be used for the simulation of weather data in the form of daily time series at a single site under both current and future climate conditions. LARS-WG uses observed daily weather data for a given site to compute a set of parameters for probability distributions of weather variables as well as correlation between them. These parameters are then used to generate synthetic weather time series of arbitrary length by randomly selecting values from the appropriate distributions. LARS-WG uses a semi-empirical distribution (SED) that is known as the cumulative probability distribution function (CDF) approximate probability distribution of daily weather data (Semenov & Barrow 1997; Semenov & Stratonovitch 2010).

Hence, for each climatic variable v, a value of climatic variable vi corresponding to the probability pi is computed using the following equation:
formula
(1)
where

Because the probability of very low daily precipitation (<1 mm) is typically relatively high and such low precipitation has very little effect on the output of a process-based impact model, only two values are used, v1=0.5 mm and v2=1 mm to approximate precipitation within the interval [0, 1] with the corresponding probabilities calculated as: . To account for extremely high long dry and wet series, two values close to 1 are used in SEDs for wet and dry series, pn−1=0.99 and vn−2=0.98 (Semenov 2007; Semenov & Stratonovitch 2010).

Summary of the modeling process with LARS-WG

The modeling method for generating synthetic weather data in LARS-WG can be divided into three steps: model calibration, model validation, and generation of synthetic weather data. The following are brief descriptions of these steps: (i) model calibration is done to use the function ‘SITE ANALYSIS’ in LARS-WG, which analyses observed weather data (e.g. precipitation) to determine their statistical characteristics; (ii) model validation helps analyze and compare the statistical characteristics of the observed and synthetic weather data to assess the ability of LARS-WG to simulate the parameters in the chosen sites in order to determine whether or not it is suitable for use in the study; and (iii) the parameter files derived from the observed weather data during the model calibration process can also be used to generate synthetic weather data corresponding to a particular climate change scenario simulated by GCMs (Racsko et al. 1991; Semenov & Barrow 2002).

LARS-WG model performance evaluation

In order to test the performance of the LARS-WG model and ensure its ability to predict future precipitation in this study, the coefficient of determination (R2) and the root-mean-square error (RMSE) were employed to compare observed and simulated rainfall data (Hassan et al. 2014; Bayatvarkeshi et al. 2020; Khoi et al. 2021). R2 is a dimensionless measure whose optimal value is one, as shown in Equation (2). Moreover, in Equation (3), the RMSE denotes the model error rate, with the optimal value being zero.
formula
(2)
formula
(3)
where x and y are the observed and simulated values, respectively; and are the average of X and Y; and n is the total number of data.

The mean rainfall is compared graphically to the observed data to boost confidence in the analyses' performance. The model's patterns and variations can be identified using these graphical comparisons. As a result, the LARS-WG was calibrated and validated in this work utilizing rainfall from five rain gauges from 1976 to 2005 (Figure 1). The calibration and validation processes took place from 1976 to 1990 and 1991 to 2005, respectively.

The LARS-WG model was used to generate future rainfall scenarios based on five GCM outputs acquired from CMIP5 GCMs (Table 2) for the 2050s (2041–2060) under two emission scenarios: RCP4.5 and RCP8.5, after successful calibration and validation against historical data from 1976 to 2005. The LARS-WG model's future precipitation projections in the study area were compared to previous studies.

Analysis of the LARS-WG calibration and validation results

The daily time series of precipitation data from 1976 to 2005, selected as the baseline year, was used in this study for the calibration and validation of the LARS-WG model. The calibration and validation processes took place from 1976 to 1990 and 1991 to 2005, respectively.

The performance statistics results of the downscaled monthly rainfalls simulated by LARS-WG during calibration and validation periods are shown in Table 3. The R2 value between observed and simulated monthly precipitation is reasonable, ranging from 0.92 to 0.99 in the calibration period and 0.826 to 0.908 in the validation period. Furthermore, for the calibration and validation periods, the RMSE values ranged from 7.0 to 13.74 and 11.44 to 25.431 mm, respectively. Monthly mean values of observed and synthetic data for the calibrated and validated periods are also remarkably consistent across all rain gauges, as shown in Figure 2(a) and 2(b), with monthly simulated and observed precipitation differences ranging from 0.52 to 2.69 mm/month for the calibration and 0.99 to 8.86 mm/month for the validation.

Based on these findings, LARS-WG performance in reproducing historical climate is deemed satisfactory, and it can be used to forecast the future precipitation with reasonable accuracy.

Table 3

Performance statistics of LARS-WG results against historical observed rainfall data during calibration and validation periods

Meteorological stationCalibration period (1976–1990)
Validation period (1991–2005)
R2RMSER2RMSE
Butajira 0.921 13.741 0.826 25.431 
Ziway 0.971 7.358 0.941 11.44 
Assela 0.973 10.574 0.894 22.446 
Bulbula 0.987 7.002 0.908 12.114 
Hawassa 0.957 8.125 0.865 14.365 
Meteorological stationCalibration period (1976–1990)
Validation period (1991–2005)
R2RMSER2RMSE
Butajira 0.921 13.741 0.826 25.431 
Ziway 0.971 7.358 0.941 11.44 
Assela 0.973 10.574 0.894 22.446 
Bulbula 0.987 7.002 0.908 12.114 
Hawassa 0.957 8.125 0.865 14.365 
Figure 2

Monthly mean observed and simulated rainfall for calibration period (a) and validation period (b).

Figure 2

Monthly mean observed and simulated rainfall for calibration period (a) and validation period (b).

Close modal

Analysis of future precipitation changes

Analysis of the future monthly mean precipitations changes

The percentage changes in the monthly mean precipitation values for the future periods compared to the baseline period are shown in Table 4 and in Figure 3(a) and 3(b). According to Table 4, we can see that in most stations, the forecast of the monthly mean precipitation changes by CCSM4 and HadGEM2-ES is expected to increase. This result agrees with Jury (2014). Also in Table 4, it can be seen that CCSM4 predicts the biggest rise (up to 18.83%) in the monthly mean precipitation for Hawassa, while INM-CM4 predicts the highest drop (up to 4.75%) for Assela. The monthly mean rainfall is expected to decrease and increase in most of the stations, according to the GCM outputs of EC-EARTH, MICRO-ESM, and INM-CM4.

Table 4

Percentage change in monthly mean precipitation in the future period (2040–2061) based on each of the five CMIP5 GCMs under RCP 4.5 and RCP 8.5 scenarios compared to the baseline period (1976–2005)

StationScenarioEC-EARTHMIROC-ESMCCSM 4HadGEM2-ESINM-CM4
Butajira RCP4.5 3.67 0.33 3.92 0.50 3.42 
 RCP8.5 10.08 0.17 7.42 1.58 3.42 
Assela RCP4.5 0.83 2.75 8.25 2.00 4.75 
 RCP8.5 12.92 0.17 12.50 3.08 7.00 
Ziway RCP4.5 1.33 1.42 6.58 0.75 3.75 
 RCP8.5 10.50 0.08 10.67 0.50 7.83 
Bulbula RCP4.5 1.33 0.92 9.58 1.00 3.67 
 RCP8.5 9.33 0.50 12.42 0.42 8.00 
Hawassa RCP4.5 1.92 0.25 18.83 2.17 2.67 
 RCP8.5 5.67 1.42 18.25 3.50 9.17 
StationScenarioEC-EARTHMIROC-ESMCCSM 4HadGEM2-ESINM-CM4
Butajira RCP4.5 3.67 0.33 3.92 0.50 3.42 
 RCP8.5 10.08 0.17 7.42 1.58 3.42 
Assela RCP4.5 0.83 2.75 8.25 2.00 4.75 
 RCP8.5 12.92 0.17 12.50 3.08 7.00 
Ziway RCP4.5 1.33 1.42 6.58 0.75 3.75 
 RCP8.5 10.50 0.08 10.67 0.50 7.83 
Bulbula RCP4.5 1.33 0.92 9.58 1.00 3.67 
 RCP8.5 9.33 0.50 12.42 0.42 8.00 
Hawassa RCP4.5 1.92 0.25 18.83 2.17 2.67 
 RCP8.5 5.67 1.42 18.25 3.50 9.17 
Figure 3

Percentage changes in the average monthly rainfall from the ensemble mean of five GCMs under (a) the RCP 4.5 scenario and (b) the RCP 8.5 scenario.

Figure 3

Percentage changes in the average monthly rainfall from the ensemble mean of five GCMs under (a) the RCP 4.5 scenario and (b) the RCP 8.5 scenario.

Close modal

As a result, there are inconsistencies in the GCM outputs for each site/area. These findings suggest that utilizing a single GCM to forecast future precipitation is fraught with uncertainty; hence, an attempt was made in this work to calculate the mean value of the precipitation forecasts from five GCMs, which are mentioned in the subsequent sections (the annual and seasonal mean precipitation forecasts) in order to further illustrate the future change in the period from 2041–2060.

According to Figure 3(a) and 3(b), the percentage changes in the monthly average precipitation for all stations for both scenarios (RCP 4.5 and RCP 8.5) in the entire catchment area for the future periods followed almost a similar pattern. In those figures, we can also see that all sites would expect the highest increased monthly rainfall (floods) in January (up to 14.2%) and December (up to 27.8%) in the RCP 4.5 and RCP8.5 scenarios, respectively. Meanwhile, the highest decline (droughts) is predicted in June (up to 8.2%) and in May (up to 7%) under the RCP 4.5 and RCP 8.5 scenarios, respectively. As a result of these data, it can be concluded that rainfall is expected to increase continuously during the dry season (October–January), while rainfall is expected to decrease or stay close to the baseline monthly rainfall during the wet months (March–September) of the study area under both the scenarios.

Analysis of future changes in annual and seasonal mean precipitation

Figures 4(a) and 4(b) and 5(a) and 5(b) depict the ensemble averages of precipitation forecasts from the five CMIP5 GCMs (Table 2), calculated to further illustrate the future changes in the annual and seasonal timescales over the period (2041–2060) compared to the baseline period (1976–2005) under the RCP 4.5 and RCP 8.5 scenarios. Figure 4(a) and 4(b) shows that rainfall in the basin is predicted to drop by up to 3.8 and 5.5% in the annual and spring seasons, respectively, under both scenarios for all stations except Hawassa (under both scenarios) and Assela (close to the baseline rainfall under the RCP8.5 scenario). Figure 5(a) and 5(b) shows that summer and winter rainfall patterns are forecast to be opposite in all sites; hence, rainfall is expected to decrease in summer (extended rainy season), while rainfall is expected to increase in winter (dry season) under both the scenarios. This finding is in line with previous studies in the research area (Kassie et al. 2014; Abraham et al. 2018; Disasa et al. 2019). Figures 4(a) and 4(b) and 5(a), comparing all stations, Butajira sites (north western part of the basin), exhibit relatively the highest decrease in annual, spring, and summer rainfall under both the scenarios. Moreover, Figure 5(a) and 5(b) depicts under the RCP8.5 scenario, unseasonal rainfall will be expected during the dry season, but the RCP 4.5 scenario predicts a noticeable reduction in summer rainfall for all sites. According to those figures, we can infer that extreme precipitation events, such as droughts in the summer (normally the rainy season) and flooding in the winter (normally the dry season), are predicted under the RCP4.5 and RCP8.5 scenarios respectively.

Figure 4

Percentage changes in the mean precipitation for five sites (a) in annual rainfall and (b) in spring rainfall.

Figure 4

Percentage changes in the mean precipitation for five sites (a) in annual rainfall and (b) in spring rainfall.

Close modal
Figure 5

Percentage changes in the mean precipitation for five sites in (a) summer rainfall and (b) winter rainfall.

Figure 5

Percentage changes in the mean precipitation for five sites in (a) summer rainfall and (b) winter rainfall.

Close modal

In the present study, we used the LARS-WG stochastic weather generator to analyze future precipitation changes across the CEMR in the period of 2041–2060. The observed rainfall data for the period of 1976 to 2005 were selected as the baseline year for comparison with future scenarios. The calibration and validation processes took place from 1976 to 1990 and 1991 to 2005, respectively.

The performance evaluation of the LARS-WG model revealed that the model illustrates a satisfactory agreement between the observed and simulated rainfall. The results revealed the increased mean rainfall in January (up to 14.2%) and December (up to 27.8%) under the RCP 4.5 and RCP8.5 scenarios, respectively. However, a drop is expected in June (up to 8.2%) and May (up to 7%) under the RCP 4.5 and RCP 8.5 scenarios, respectively. Summer precipitation is expected to decrease, while winter precipitation is expected to increase at all locations in both scenarios. A noticeable reduction in summer rainfall is predicted under the RCP 4.5 scenario for all sites. The RCP 8.5 scenario predicts unseasonal rainfall during the dry season. Predicting rainfall below normal can improve the use of water abstraction techniques to alleviate the problem of water scarcity in the study region characterized by rain-fed agricultural activities. Predicting above average and unusual rainfall throughout the winter season is critical to reducing the risk of flooding and crop damage during harvest.

In the present work, in order to address the uncertainty associated with a single emission scenario and GCM, we used the ensemble mean of five CMIP5 GCM outputs (EC-EARTH, MIROC-ESM, HadGEM2-ES, INM-CM4, and CCSM4) under the RCP4.5 and RCP8.5 scenarios. Additional downscaling techniques (apart from LARS-WG) and hydrological modeling tools, on the other hand, are not considered in our analysis. More GCMs, emission scenarios, downscaling techniques, and hydrological modeling tools for climate change impacts on water resources are all needed to improve these results. Further research will be required to take into account the limitations of this study in addressing all uncertainties. The findings of this study can be used to establish future sustainable water resource management and planning for a variety of fields, including water resources and agriculture.

The authors would like to thank the National Meteorological Agency of Ethiopia for providing meteorological data for this study. We are also grateful to the editors and two anonymous reviewers for their insightful comments and suggestions, which have improved the quality of this paper.

The authors declare no conflict of interest.

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

Abraham
T.
,
Woldemicheala
A.
,
Muluneha
A.
&
Abateb
B.
2018
Hydrological responses of climate change on Lake Ziway Catchment, Central Rift Valley of Ethiopia
.
Journal of Earth Science and Climatic Change
9
,
474
.
Bayatvarkeshi
M.
,
Zhang
B.
,
Fasihi
R.
,
Adnan
R. M.
,
Kis
O.
&
Yuan
X.
2020
Investigation into the effects of climate change on reference evapotranspiration using the HadCM3 and LARS-WG
.
Water
12
(
3
),
666
.
Bekele
W. T.
,
Haile
A. T.
&
Rientj
T.
2021
Impact of climate change on the streamflow of the Arjo-Didessa catchment under RCP scenarios
.
Journal of Water and Climate Change
12
(
6
),
2325
2337
.
Brekke
L. D.
,
Kiang
J. E.
,
Olsen
J. R.
,
Pulwarty
R. S.
,
Raff
D. A.
,
Turnipseed
D. P.
,
Webb
R. S.
&
White
K. D.
2009
Climate change and water resources management – a federal perspective
.
U.S. Geological Survey Circular
1331
,
65
.
Disasa
K. N.
,
Tura
F. S.
&
Fereda
M. E.
2019
Climate change downscaling using stochastic weather generator model in Rift Valley Basins of Ethiopia
.
American Journal of Climate Change
8
,
561
590
.
Ekström
M.
,
Grose
M. R.
&
Whetton
P. H.
2015
An appraisal of downscaling methods used in climate change research
.
WIREs Climate Change
6
,
301
319
.
Gebrechorkos
S. H.
,
Hülsmann
S.
&
Bernhofer
C.
2019
Statistically downscaled climate dataset for East Africa
.
Science Data
6
,
31
.
Hassan
Z.
,
Shamsudin
S.
&
Harun
S.
2014
Application of SDSM and LARS-WG for simulating and downscaling of rainfall and temperature
.
Theoretical and Applied Climatology
116
,
243
257
.
IPCC
2007
Summary for Policymakers
. In:
Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
(S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. Averyt, M. Tignor & H. L. Miller, eds.).
Cambridge University Press
,
Cambridge
,
UK
and New York, NY, USA
.
IPCC
2013
Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
.
Cambridge University Press
,
Cambridge
,
UK
and New York, NY, USA
.
Kassie
B. T.
,
Rötter
R. P.
,
Hengsdijk
H.
,
Asseng
S.
,
Van Ittersum
M. K.
,
Kahiluoto
H.
&
Van keulen
H.
2014
Climate variability and change in the Central Rift Valley of Ethiopia: challenges for rainfed crop production
.
The Journal of Agricultural Science
152
(
1
),
58
74
.
Khoi
D. N.
,
Sam
T. T.
,
Loi
P. T.
,
Hung
B. V.
&
Nguyen
V.
2021
Impact of climate change on hydro-meteorological drought over the Be River Basin, Vietnam
.
Journal of Water and Climate Change
12
(
7
),
3159
3169
.
MoWR (The Federal Democratic Republic of Ethiopia Ministry of Water Resources)
2008
Rift Valley Lakes Basin Integrated Resources Development Master Plan Study Project
.
Halcrow Group Limited and Generation Integrated Rural Development (GIRD) Consultants
,
Addis Ababa, Ethiopia
.
Racsko
P.
,
Szeidl
L.
&
Semenov
M. A.
1991
A serial approach to local stochastic weather models
.
Ecological Modelling
57
,
27
41
.
Semenov
M. A.
2007
Development of high-resolution UKCIP02-based climate change scenarios in the UK
.
Agricultural and Forest Meteorology
144
,
127
138
.
Semenov
M. A.
2008
Simulation of extreme weather events by a stochastic weather generator
.
Climate Research
35
(
3
),
203
212
.
Semenov
M. A.
&
Barrow
E. M.
2002
LARS-WG. A Stochastic Weather Generator for Use in Climate Impact Studies. Version 3.0. User Manual. Available from: http://resources.rothamsted.ac.uk/sites/default/files/groups/mas-models/download/LARS-WG-Manual.pdf (accessed 4 Novemeber 2021)
.
Taylor
K. E.
,
Stouffer
R. J.
&
Meehl
G. A.
2012
An overview of CMIP5 and the experiment design
.
Bulletin of American Meteorological Society
93
(
4
),
485
498
.
Vesely
F. M.
,
Paleari
L.
,
Movedi
E.
,
Bellocchi
G.
&
Confalonieri
R.
2019
Quantifying uncertainty due to stochastic weather generators in climate change impact studies
.
Scientific Reports
9
,
9258
.
Zeray
L. A.
,
Roehrig
J.
&
Alamirew
D. C
, .
2009
Climate Change Impact on Lake Ziway Watershed Water Availability, Ethiopia. In: CICD Series Vol.1: LARS 2007 Proceedings. May 7–11, 2007, Arba Minch, Ethiopia. Available from: https://www.uni-siegen.de/zew/publikationen/cicd_series/cicd_vol.1_lars_2007-08.02.10.pdf (accessed 5 October 2020).
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/).