Abstract
This study aims to evaluate water availability under changing climate scenarios in the Woybo catchment, Ethiopia. The bias-corrected outputs of multiple climate models’ ensemble mean were employed for the 2050 and 2080s against the reference period (1976–2005) under representative concentration pathways (RCPs) for both RCP4.5 and RCP8.5 scenarios. A semi-distributed physically based Hydrologic Engineering Center of Hydrologic Modeling System (HEC-HMS) was used to perform rainfall–runoff simulation. The projected rainfall and temperatures of the watershed will increase in the far future. The predictions from ensemble means of multiple climate models indicated that rainfall of the watershed will likely increase by 25% in the 2050s and 19% in the 2080s under RCP4.5 and RCP8.5, respectively. The discharge projection for the ensemble mean of all climate models shows an increment up to 20 and 19% under RCP 4.5 and RCP8.5, respectively, in the 2050s, whereas it will decline up to 15 and 28% in 2080s, under RCP4.5 and RCP8.5, respectively. This research plays a great role to reduce the impacts of changing climate for sustainable water resources management.
HIGHLIGHT
The output of this research plays a great role to reduce the impacts of changing climate for sustainable water resources management.
The ensemble mean of five climate models was employed in this study to analyze the impact on water availability.
A multiple climate model’s ensemble reduces uncertainty regarding the relative strengths and weaknesses of a single climate model.
Graphical Abstract
INTRODUCTION
Climate change is worsening on a regular basis across the globe, partially as a result of increased human activity (Boru et al. 2019). Internal and external forcing can affect climatic variables in a variety of ways, throughout a wide range of time periods and geographical areas (Abdollahbeigi 2020). Human-caused deforestation and industrial growth are regarded to be driving causes of climate change because they increase the greenhouse gas effect, disrupting the normal atmospheric composition (Nica et al. 2019).
According to recent studies, when greenhouses gas levels grow at all levels on the earth's surface, average temperatures in the atmosphere will continue to rise. For example, the world average temperature is predicted to rise by 1.5 °C in the 2050s (Pielke et al. 2022). The warming is likely to be larger than the global annual mean warming in Africa and this would result due to the likely increase in minimum temperature more rapidly than maximum temperatures.
Variations in precipitation on a hemispheric scale have yet to be realized. Some locations may see rises, while others will see losses, and still, others will see no major changes at all (Ukumo et al. 2022a, 2022b). Changes in precipitation occurrence, duration, and prediction are likely to be one of the most pervasive and deleterious effects of climate change in Ethiopia. The life and properties in the catchment are affected by the loss or rise in water volume. Water availability will be impacted by changes in regional precipitation, leading in decreased crop yields and the risk of widespread starvation. The influence of climate change requires great care to secure the environment for life.
Developing countries like Ethiopia will be more vulnerable as a result of climate change. Drought and desertification have affected a large area of the country in recent years. As a result, the country is concerned about climate change and its repercussions. As a result of climate change, more frequent and dramatic periodic changes in sea surface temperature and overlying atmospheric air pressure occurrences are projected, resulting in widespread hunger in the region (Keller 2009). Climate change effects must be studied, and adaption techniques developed as part of a larger strategy. Long-term development and water resource planning and management are harmed by climate change. The influence of climate change on streamflow varies by region due to differences in soil type, terrain, anthropogenic activity, and meteorological condition. Human civilization increased streamflow and caused major floods, while streamflow variability decreased (Ahmed et al. 2022). Runoff changes meaningfully and is mainly affected due to climate change. Because of climatic unpredictability, quantifying the effects of climate change on runoff in the Woybo watershed is critical for investigating flow regime behavior.
Previous investigations in the Woybo catchment were described, including their findings and limitations. The authors paid less attention to long-term climate change and climatic data bias correction. The studies mostly focused on the 2020 and 2050s, as well as single climate model output, making the results uncertain. No single climate model can capture the entire range of possibilities for all factors, regions, or seasons. Working with a restricted selection of models, as (Turco et al. 2013) demonstrated, can result in inconsistency in climate change signals. Global circulation models’ (GCMs’) intrinsic flaws and uncertainties are explored as a result of the simplification of highly complicated atmospheric physics. Multiple climate model ensembles were a better fit for the circumstance than the individual GCM, owing to the compensation of individual faults (Turco et al. 2013). Demissie et al. (2016), assessed the impact of climate change on flood frequency in the catchment and summarized that the catchment will experience a significant climate change for the next 30 years. However, the authors did not clarify the type and accuracy of the models used. Moreover, these studies focused on a short-term period ignoring the mid-term (2041–2070) and long-term (2071–2100) climate change effects. They came to the conclusion that rainfall and evapotranspiration are the two most important meteorological factors to consider when conducting rain-fed agriculture (Mohammed 2013; Mengistu et al. 2020). They did not employ a climate model to forecast the catchment's rainfall variability. Tekle (2015a) uses the GCM and regional circulation model (RCM) with A2a and B2a as inputs. According to their findings, there is an increase in water demand due to an increase in human water consumption for various purposes. According to them, the Statistical Downscaling Model (SDSM) can recreate historical maximum and lowest temperatures but not precipitation. However, they did not utilize bias correction, and hence the systematic error is expected in the study's final output. The finding is suspect because of their usage of a single GCM. (Negash 2014) on the other hand observed that the surface runoff component in the study catchment increases progressively since the 1970s. Edamo et al. (2022) also analysed the water resources potential of the catchment with respect to precipitation variability. They concluded that precipitation depends on altitude in the catchment. The variables of climate data are limited, which makes the results questionable. The GCM and RCM were downloaded from the Coordinated Regional Climate Downscaling Experiment (CORDEX) for the investigations. CORDEX provides RCP4.5 projections, which assumes that atmospheric greenhouse gas concentrations stabilize at values corresponding to radiative forcing of 4.5 W m−2 in 2100 (Tenfie et al. 2022) and RCP8.5, which assumes continued increases in greenhouse gas emissions throughout the 21st century. The main goal of this research is to evaluate water availability in the Woybo catchment using the ensemble mean of multiple climate models to reduce individual models’ uncertainty. As a result, the findings of this study are critical for long-term river basin management.
MATERIALS AND METHODS
Area under study
Location map of the study area: (a) Ethiopian basins, (b) Omo Gibe basin, and (c) Woybo catchment.
Location map of the study area: (a) Ethiopian basins, (b) Omo Gibe basin, and (c) Woybo catchment.
Data
The Ethiopian National Meteorological Agency (NMA) provided secondary data such as rainfall, temperature, relative humidity, and sunlight hours. The Ministry of Water, Irrigation, and Energy provided streamflow statistics (MoWIE) (Table 1).
Observed meteorological data from MoWIE for stations in the Woybo catchment
Stations . | Angacha . | Areka . | Bodity School . | Wolaita Sodo . |
---|---|---|---|---|
Longitude (°C) | 37.86 | 37.71 | 37.96 | 37.73 |
Latitude (°C) | 7.34 | 7.06 | 6.95 | 6.81 |
Altitude(m) | 2317 | 1,752 | 2,043 | 1,854 |
Data type with year of record | RF (1987–2017) | RF (1987–2017) | RF (1986–2017) | RF (1986–2017) |
Tmax (1987–2017) | Tmax (1987–2017) | Tmax (1986–2017) | Tmax (1986–2017) | |
Tmin (1987–2017) | Tmin (1987–2017) | Tmin (1986–2017) | Tmin (1986–2017) | |
No | No | No | SSH (1986–2017) | |
No | No | No | RH (1986–2017) | |
No | No | No | WS (1986–2017) | |
No | No | No | SF (2000–2012) |
Stations . | Angacha . | Areka . | Bodity School . | Wolaita Sodo . |
---|---|---|---|---|
Longitude (°C) | 37.86 | 37.71 | 37.96 | 37.73 |
Latitude (°C) | 7.34 | 7.06 | 6.95 | 6.81 |
Altitude(m) | 2317 | 1,752 | 2,043 | 1,854 |
Data type with year of record | RF (1987–2017) | RF (1987–2017) | RF (1986–2017) | RF (1986–2017) |
Tmax (1987–2017) | Tmax (1987–2017) | Tmax (1986–2017) | Tmax (1986–2017) | |
Tmin (1987–2017) | Tmin (1987–2017) | Tmin (1986–2017) | Tmin (1986–2017) | |
No | No | No | SSH (1986–2017) | |
No | No | No | RH (1986–2017) | |
No | No | No | WS (1986–2017) | |
No | No | No | SF (2000–2012) |
Note: RF, rainfall; Tmax, maximum temperature; Tmin, minimum temperature; SSH, sunshine hour; RH, relative humidity; WS, wind speed; SF, streamflow.
For both RCP4.5 and RCP8.5 emission scenarios, downscaled rainfall and temperature data for the period 1976–2100 were downloaded in the form of NetCDF (Network Common Data Form) from the spatial grid resolutions of all CORDEX Africa programs (https://esgf-node.llnl.gov/projects/esgf-llnl/). The Pacific Northwest National Laboratory in the United States (US) developed the Intermediate Emissions scenario (RCP4.5). Here, radiative forcing is stabilized shortly after the year 2100, consistent with a future with relatively ambitious emissions reductions. RCP8.5 (high Emissions scenario), is consistent with a future with no policy changes to reduce emissions. The International Institute developed this scenario for Applied System Analysis in Austria. It is characterized by increasing greenhouse gas emissions that lead to high greenhouse gas concentrations over time. Five RCM models such as Climate Limited-Area Modeling Community Version 4 (CCLM4), Rossby Center Atmospheric Version 4 (RCA4), Regional Atmospheric Climate Model Version22 (RACMO22T), High-Resolution Hamburg Climate Model Version 5 (HIRAM5), and REgional MOdel (REMO2009) were used.
CORDEX provides a unique and novel set of dynamically downscaled climate forecasts for several domains around the world, including Africa, utilizing various RCMs. Although all RCMs simulate basic climatic variables (daily precipitation, maximum (Tmax), and minimum (Tmin) surface air temperatures), systematic biases exist in each model, necessitating bias adjustment prior to using the climate data for hydrological research. (Demissie & Sime 2021) examined five RCMs, including REMO2009, HIRAM5, CCLM4-8, and RCA4, and found that the climate models failed to capture observed rainfall and temperature.
A subset of models can be chosen using a variety of criteria, such as their ability to reproduce previous climate (Pierce et al. 2009) and the range of projected climate changes. Others have used automated methods to determine a representative subset of climate models based on the clustering of climate extreme indices (Farjad et al. 2019).
However, no single climate model can capture the entire range of possibilities for all factors, regions, or seasons. Working with a restricted selection of models, as (Turco et al. 2013) demonstrated, can result in inconsistency in climate change signals.
GCMs’ intrinsic flaws and uncertainties are explored as a result of the simplification of highly complicated atmospheric physics. They discovered that many model ensembles were a better fit for the circumstance than individual GCMs, owing to the compensation of individual faults.
In this study, five climate models were selected based on previous studies of the catchment which perform better than other climate models. As they studied, the previous knowledge of selecting climate models from multiple GCM–RCMs is better to limit the number of climate models.
Different research institutes and universities from various nations created the selected GCM–RCMs. HIRAM5 (van Meijgaard et al. 2008) is a completely new RCM, based on a subset of the High-Resolution Limited Area Model (HIRLAM) and European Centre Hamburg Model (ECHAM) models, combining the dynamics of the former model with the physical parameterization schemes of the latter. The physical parameterization has also greatly been modified (Christensen et al. 2007).
The next version of the hydrostatic KNMI regional climate model RACMO has been developed and implemented in a number of projects over the previous few years, starting in 2005. This version, referred to as RACMO2.2 in this document, is an update to the RACMO2 cycle. The hydrostatic RCA4 model has data produced with several horizontal resolutions but for this analysis 0.44° resolution data is used.
REMO2009 is a three-dimensional hydrostatic regional climate model created at the Max Planck Institute of Meteorology. Similarly, CCLM4 is a fifth-order upwind non-hydrostatic regional climate model (Zentek & Heinemann 2020). The magnitude of climatic factors is varied in each RCM model. The climate models used in this study revealed variations in capturing the observed rainfall and temperature in previous studies. All models have a 0.44° × 0.44° grid resolution. The details of the RCMs used in this research were also described in Table 2.
Details of CORDEX–RCMs and the driving GCM–RCMs
RCM . | Institute . | Driving GCM . | Projection . | Reference . |
---|---|---|---|---|
CCLM4 | Climate Limited-area Modelling Community (CLMcom) | MPI-M-MPI-ESM-LR | Rotated pole | Rockel et al. (2008) |
RCA4 | Sveriges Meteorologiska och Hydrologiska Institute (SMHI) | CNRM-CERFACS-CNRM-CM5 | Rotated pole | Samuelsson et al. (2015) |
RACMO22T | Koninklijk Nederlands Meteorologisch Instituut, (KNMI), Netherlands | ICHEC-EC-EARTH_ | Rotated pole | van Meijgaard et al. (2008) |
HIRHAM5 | Danmarks Meterologiske Institut (DMI) | ICHEC-EC-EARTH | Rotated pole | Christensen et al. (2007) |
REMO2009 | Helmholtz-Zentrum Geesthacht, Climate Service Center, Max Planck Institute for Meteorology | ICHEC-EC-EARTH | Rotated pole |
RCM . | Institute . | Driving GCM . | Projection . | Reference . |
---|---|---|---|---|
CCLM4 | Climate Limited-area Modelling Community (CLMcom) | MPI-M-MPI-ESM-LR | Rotated pole | Rockel et al. (2008) |
RCA4 | Sveriges Meteorologiska och Hydrologiska Institute (SMHI) | CNRM-CERFACS-CNRM-CM5 | Rotated pole | Samuelsson et al. (2015) |
RACMO22T | Koninklijk Nederlands Meteorologisch Instituut, (KNMI), Netherlands | ICHEC-EC-EARTH_ | Rotated pole | van Meijgaard et al. (2008) |
HIRHAM5 | Danmarks Meterologiske Institut (DMI) | ICHEC-EC-EARTH | Rotated pole | Christensen et al. (2007) |
REMO2009 | Helmholtz-Zentrum Geesthacht, Climate Service Center, Max Planck Institute for Meteorology | ICHEC-EC-EARTH | Rotated pole |
Until recently, computer power and operational time limits on resolutions where the hydrostatic approximation is almost perfectly true limited the resolution of operational numerical weather forecast models. As a result, operational forecasts have primarily depended on hydrostatic models, which performed admirably in numerical simulations of global circulation in both the atmosphere and the ocean at the spatial and temporal resolutions available at the time (Gibbon & Holm 2011).
Parallel to this, for more than four decades, researchers have been working on non-hydrostatic atmospheric models that maintain the vertical acceleration term and so capture significant vertical convection, particularly for mesoscale analyses of abrupt storms. Over the last decade, the spatial resolution of both numerical weather prediction and climate simulation models has increased as computer systems have become faster and memory has become more affordable. This advancement has made it easier to move away from highly developed hydrostatic models and toward non-hydrostatic models. Over the last ten years, atmospheric research organizations have begun to replace operational hydrostatic models with non-hydrostatic equivalents (Davies et al. 2005).
Rainfall and temperature
METHOD
Bias correction of climate data
Without bias correction, data from climate models cannot be used directly for impact assessment. To develop a better climate projection, the original climate data were modified. To compensate for the overestimation or underestimate of the mean of downscaled climatic data, bias correction was used. The bias correction method of Quantile Mapping (QM, also known as probability mapping) was employed to improve the persistence of the raw climate model for expected changes in temperature and precipitation (Ringard et al. 2017). Depending on the climate variable, this includes establishing a statistical link between observed and model-simulated outputs by replacing observed values with simulated values at the same cumulative density function of the chosen distribution.


Simulated rainfall and temperature from climate models
Statistical measurements are used to evaluate model performance. The bias, root mean squared error (RMSE), correlation coefficient (corr), and coefficient of variation (CV) are all expressed in percentage (Mengistu et al. 2021).
The average tendency of the simulated data is higher or lower than the observed values, which is measured by bias. Bias is measured in percentages; the lower the absolute magnitude of the bias, the better the model's performance will be. The unit of observed variable in the RMSE makes its interpretation very simple. A model's performance is improved when the RMSE is near zero. The linear relationship between observed and modeled rainfall levels is evaluated using the correlation coefficient (corr). A value of 1.0 indicates a perfect linear relationship between two variables.



Climate change impact analysis
Ensemble mean of multiple climate models
HEC-HMS model
HEC-HMS is designed to simulate the rainfall-runoff processes particularly for dendritic watershed (Kastali et al. 2022). The input data used for HEC-HMS were streamflow, sunshine hour, relative humidity, and rainfall.
Calibration considered the least sensitive model parameter first and subsequently followed the more sensitive parameters by systematic adjustment of the initial values to optimize the candidate parameters which provides the best fit between the observed streamflow and simulated flow (Kastali et al. 2022). Model calibration was carried out by lessening the difference between the models computed and observed streamflow data. Then model validation was conducted, using the model parameters fixed during the model calibration and other set of data. Therefore, of the total sample data, two-third of the observed data was used for calibration and one-third was used for validation.
Performance of HEC-HMS model
RESULTS AND DISCUSSION
Bias correction
Some research that focused on the impact of climate change using single RCM output should re-evaluate their findings based on the output of numerous climate models. According to (Dibaba et al. 2019), all RCMs are not equal when it comes to their performance in a localized study area. The performance varies from model to model and region-to-region. RCMs that achieve good results in some areas may fail in other places.
Annual cycle of corrected maximum and minimum temperature (1987–2005).
When compared to GCM, the forced RCMs to the local area are imprecise and biased. The bias correction strategy of distribution mapping of precipitation and temperature minimized model errors and fit the observed minimum temperature. Bias correction of generated data from climate models was endorsed by researchers (Hosseinzadehtalaei et al. 2021).
Rainfall and temperature variability in reference period (1976–2005)
Bias-corrected annual rainfall cycle of GCM–RCM projected climate data (1976–2005).
Bias-corrected annual rainfall cycle of GCM–RCM projected climate data (1976–2005).
Bias-corrected annual maximum and minimum temperature cycle of GCM–RCM projected climate data under the baseline period (1976–2005).
Bias-corrected annual maximum and minimum temperature cycle of GCM–RCM projected climate data under the baseline period (1976–2005).
In August, all climate models predicted a low minimum temperature for the Woybo watershed (7.2 °C). In March, however, the projections predicted a high lowest temperature (13 °C). In December, the CNRM-RCA4 model forecasted a temperature of 12.8 °C.
Projected changes in rainfall and temperature in mid-term period (2041–2070) or 2050s
Bias-corrected annual rainfall cycle of GCM–RCM projected climate data (2050s).
The predicting from ensemble means of multiple climate models indicated that rainfall of the watershed will likely increase by 25% in the 2050s and 19% in 2080s under RCP4.5 and RCP8.5, respectively. Thus, the use of ensemble means of multiple GCM–RCMs models reduced prediction uncertainties (Ukumo et al. 2022a, 2022b). Multiple climate model ensembles were a better fit for the circumstance than individual GCMs, owing to the compensation of single climate uncertainty (Ukumo et al. 2022a, 2022b). Precipitation will change by 32 to −33% in the 2050s under the B2a scenario. Nonetheless, the conclusions are suspect due to the use of a single climate model as discussed in literature (Turco et al. 2013).
Bias-corrected annual maximum and minimum temperature cycle of GCM–RCM projected climate data under RCP4.5 (2050s).
Bias-corrected annual maximum and minimum temperature cycle of GCM–RCM projected climate data under RCP4.5 (2050s).
Bias-corrected annual maximum and minimum temperature cycle of GCM–RCM projected climate data under RCP8.5 (2050s).
Bias-corrected annual maximum and minimum temperature cycle of GCM–RCM projected climate data under RCP8.5 (2050s).
Under the RCP4.5 scenario in the 2050s, the monthly minimum temperature of the Woybo catchment will change by 1.1 °C from January to December for MPI-CCLM4. However, under RCP8.5, it will change by 1.0 °C over the same time period. The ICHEC-HIRAM5 model estimated a 1.1 °C increase in lowest temperature in the 2050s under RCP4.5. Under RCP8.5, it will also shift by 1.8 °C in the 2050s. The ICHEC-RACMO22T model anticipated a 0.95 °C increase in lowest temperature in the 2050s under RCP4.5. However, at RCP8.5, the model forecast 1.4 °C. The CNRM-RCA4 model anticipated a shift of 0.1–1.1 °C in the 2050s under RCP4.5. Under RCP8.5, the model expected a change of 1.92 °C over the same time period. The ICHEC-REMO2009 model estimated a 1.32 °C increase in rainfall in the 2050s under RCP4.5. Under RCP8.5, it likewise revealed a 1.1 °C shift in the 2050s (Figure 10).
Predicted changes of seasonal and annual rainfall and temperature in 2080s
Bias-corrected annual rainfall cycle of GCM–RCM projected climate data (2080s).
Bias-corrected annual rainfall cycle of GCM–RCM projected climate data (2080s).
Bias-corrected annual maximum and minimum temperature cycle of GCM–RCM projected climate data under RCP4.5 (2080s).
Bias-corrected annual maximum and minimum temperature cycle of GCM–RCM projected climate data under RCP4.5 (2080s).
Bias-corrected maximum and minimum temperature cycle of GCM–RCM projected climate data under RCP8.5 (2080s).
Bias-corrected maximum and minimum temperature cycle of GCM–RCM projected climate data under RCP8.5 (2080s).
Performance of simulated rainfall and temperature
Table 3 indicated that all models underestimate the observed rainfall. The accuracy of models was not the same in representing the rainfall of the catchment. In terms of bias, ensemble mean performs better (Bias = −4.01%) whereas ICHEC-REMO2009 accomplishes worst (Bias = −17.3%). This large bias (Bias = −17.3%) indicates that the RCM rainfall amount largely deviates from the observed rainfall. In terms of CV, ensemble mean performs best (CV = 2.3%) whereas ICHEC-REMO2009 performs worst (CV = 8.3%). ICHEC-REMO2009 performs worst (RMSE = 24.36 mm/year), while ensemble mean achieves better (RMSE = 1.09 mm/year). Furthermore, the ensemble mean performs better (R2 = 0.89) which implies that it shows linear relationship between observed and simulated rainfall. ICHEC-RACMO22T performs worst in terms of correlation coefficient (R2 = 0.25). The ensemble mean estimated the observed mean annual rainfall amount by 85.73% rainfall variability as compared to those five models. In order to reduce the differences between the simulated and observed rainfall, the biases are removed before the use of RCMs models’ simulations.
Annual cycle performance of rainfall
Climate models . | Annual rainfall (mm) . | Bias (%) . | CV (%) . | RMSE (mm/year) . | Correlation coefficient (–) . |
---|---|---|---|---|---|
Observed | 1,377.7 | – | 10.6 | – | – |
MPI-CCLM4 | 1,078.68 | −5.7 | 3.8 | 22.3 | 0.3 |
ICHEC-HIRAM5 | 1,096.7 | −5.6 | 3.3 | 20.5 | 0.4 |
ICHEC-RACMO22T | 1,014.4 | −5.8 | 3.9 | 23.1 | 0.25 |
CNRM-RCA4 | 1,102.30 | −4.99 | 4.7 | 19.05 | 0.6 |
ICHEC-REMO2009 | 935.2 | −17.3 | 8.3 | 24.36 | 0.06 |
Ensemble mean | 1,108.3 | −4.01 | 2.9 | 1.09 | 0.89 |
Climate models . | Annual rainfall (mm) . | Bias (%) . | CV (%) . | RMSE (mm/year) . | Correlation coefficient (–) . |
---|---|---|---|---|---|
Observed | 1,377.7 | – | 10.6 | – | – |
MPI-CCLM4 | 1,078.68 | −5.7 | 3.8 | 22.3 | 0.3 |
ICHEC-HIRAM5 | 1,096.7 | −5.6 | 3.3 | 20.5 | 0.4 |
ICHEC-RACMO22T | 1,014.4 | −5.8 | 3.9 | 23.1 | 0.25 |
CNRM-RCA4 | 1,102.30 | −4.99 | 4.7 | 19.05 | 0.6 |
ICHEC-REMO2009 | 935.2 | −17.3 | 8.3 | 24.36 | 0.06 |
Ensemble mean | 1,108.3 | −4.01 | 2.9 | 1.09 | 0.89 |
CNRM-RCA4 model tried to capture observed average annual rainfall (1102.3 mm/year). In RMSE, a value close to zero indicates the best performance of the RCM model. A value greater than zero indicates the poor performance of the RCM model between simulated and observed rainfall amounts. Errors in climate models can be caused by a range of factors. Errors or biases are due to limited spatial resolution (large grid sizes), simplified thermodynamic processes and physics, or incomplete understanding of the global climate system. RCMs are commonly used to transfer large-scale GCM data to smaller scales and to provide more detailed regional information. Due to systematic and random model errors, however, RCM simulations often show considerable deviations from observations.
The climate models underestimated observed maximum temperature of the catchment (Figure 10). CNRM-RCA4, ICHEC-REMO2009 and MPI-CCLM4 slightly underestimated observed maximum temperature but MPI-CCLM4 captured in February. ICHEC-HIRAM5 and ICHEC-RACMO22T entirely underestimated observed maximum temperature. Many studies also indicated that climate models may overestimate or underestimate observed maximum temperature (Demissie & Sime 2021).
Annual cycle of uncorrected minimum and maximum temperature (1987–2005).
HEC-HMS model calibration and validation
Noble enactment was obtained in terms of reproducing the observed pattern of the streamflow hydrograph during calibration and validation (NSE = 0.69 and 0.66), respectively. The model was accepted when evaluated using objective functions. The REV for the calibration period was 4.50% which suggested that the model showed very good performance in estimating observed streamflow volume. Very good performance in estimating observed streamflow volume during validation period (−4.94%) was established.
Despite advancements in modeling many processes, hydrological models remain unreliable (Tyralis & Papacharalampous 2021). The input and calibration data, model structure, and parameters all contribute to model uncertainty (Moges et al. 2021). Natural process variability and observation errors can both cause uncertainty. On the other hand, even if a model is a perfect depiction of the hydrologic system, parameter uncertainty might occur due to mistakes in the calibration data because of observation errors. Due to the lack of a unified theory, limited knowledge, and numerical and process simplifications, an exact description of a hydrological system is difficult (Beven 2006). Model structure has a significant impact on model performance.
Projected changes of rainfall and temperature in the Woybo catchment
The climate models’ simulations show a varied rainfall between mid-term (2050s) and long-term (2080s) with respect to baseline period (1976–2005).
The RCP4.5 scenario exhibits mixed signals in the direction of seasonal and annual rainfall. The average rainfall of the four seasons and annual will probably change by 6, 1.2, 6, −12 and −8%, respectively, for all climate models. The RCP 8.5 scenario shows a discrepancy in annual and seasonal rainfall. The average rainfall of the four seasons and annual will likely change by −5.3, 0.1, 0.72, −1.3, and 0.75%, respectively. The rainfall for Autumn (small rain season) and Summer (rainy season) shows positive increment under both RCP scenarios and will likely decrease in annual under RCP8.5. This shows that the area is more sensitive for climate change. This change may depend on the characteristics of the precipitation, and temperatures of the catchment. The Woybo catchment is vulnerable to climatic change in terms of temperature, and rainfall changes. Comparably, the Summer and Autumn season's rainfall show increment in the area. As result it may cause frequent variation on surface water availability in the catchment.
The main reasons for variation or change of projected rainfall in 2050 and 2080s were dependence on characteristics and patterns of precipitation, temperature, and evapotranspiration (Ukumo et al. 2022a, 2022b); the varying characteristics of different climate models and the forcing of GCM to RCM thus leading to systematic errors with their respective uncertainties. According to Dibaba et al. (2019), all RCMs are not equal when it comes to their performance in a localized study area. The performance varies from model to model and region-to-region. RCMs that achieve good results in some areas may fail in other places.
The annual mean maximum temperature over the catchment for the 2050s will rise in magnitude up to 1.8 and 0.6 °C under RCP4.5 and RCP8.5, respectively whereas it will increase up to 2.0 and 1.4 °C in 2080s under RCP4.5 and RCP8.5, respectively, for all climate models.
The annual minimum temperature over the catchment for the 2050s increased in magnitude up to 2 °C under RCP4.5 and 2.2 °C for RCP8.5 with respect to baseline period. The annual mean minimum temperature of the catchment will increase up to 3.3 and 3.81 °C in 2080s under RCP4.5 and RCP8.5, respectively, for all climate models.
Simulated changes of seasonal and annual streamflow in the Woybo catchment
Seasonal streamflow changes in the Woybo catchment in 2050 and 2080s.
Under RCP4.5 scenario, there is a varied signal in the way of annual and seasonal streamflow change. On average the streamflow for the four seasons and annual will likely change by 1.05, 0.33, −1.65, −4.82, and −1.65%, respectively. Under RCP8.5 scenario, there is mixed signals. On average, the streamflow of the four seasons and annual will possibly change by −4.4, 0.51, 1.11, −4.84, and 0.7%, respectively.
The discharge projection for the ensemble mean of all climate models shows an increment up to 20 and 19% under RCP 4.5 and RCP8.5, respectively, in the 2050s, whereas it will decline up to 15 and 28% in 2080s, under RCP4.5 and RCP8.5, respectively.
According to Tekle 2015b, the average total annual flow of the catchment might decrease up to 17.01 and 15.2% for the A2a scenario and 16.6 and 15.7% for the B2a scenario of 2011–2040 and 2041–2070 periods, respectively, but they are missing the long-term (2071–2100) periods. The falling trend of the average annual flow volume is mainly associated with a relatively higher reduction in summer season flow volume by between 21.1 and 22.5% for the A2a scenario and between 19.9 and 21.1% for the B2a scenario.
CONCLUSIONS
The findings of this research indicated that climate change can affect the availability of water in the Woybo catchment. This study used five different climate models. The HEC-HMS model was used to simulate rainfall–runoff and generate the future flow. The climate change analysis was performed for baseline (1976–2005), 2050 and 2080s. The probable changes of projected precipitation, temperature and streamflow for all climate models were concluded. In the 2050s, yearly rainfall in the Woybo watershed is expected to change by 25% under RCP4.5 and 8.5% under RCP8.5. In the 2080s, annual rainfall will differ by 19.08% under RCP4.5 and 36.83% under RCP8.5. In the 2050s, yearly streamflow will most likely change by 20% under RCP4.5 and 7.20% under RCP8.5. In the 2080s, streamflow is expected to change by 26.20% under RCP4.5 and 19% under RCP8.5. Ensemble means of multiple climate models were employed in this study to analyze the impact on water availability. This study takes into account the ensemble means of these numerous climate models. Because a multiple model ensemble reduces uncertainty regarding relative strengths and weaknesses of single model, the considerable differences between the various models are always hidden. As a result, the study addresses a thorough investigation of ensemble of numerous models for future water availability projections.
ACKNOWLEDGEMENTS
The authors would like to acknowledge all data providers, namely MoWIEE (Minstry of Water Irrigation and Energy of Ethiopia), NMAE (National Meteorology Agency of Ethiopia) for providing required data. The authors are thankful to Arba Minch University who has provided all logistic support in conducting the research work.
FUNDING
No funding has been received from any source for this research work.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.