Multi-model Ensemble Research Articles

Assessing climate sensitivity of the Upper Indus Basin using fully distributed, physically-based hydrologic modeling and multi-model climate ensemble approach

The Upper Indus Basin (UIB) of Pakistan is home to three largest mountain ranges: Karakoram, Hindukush, and the Himalayas. There’s a hovering danger of reduced water resources due to climate-induced warming, since the Indus Basin relies mostly on snow and glacier melt runoff from these mountains. In this study, the hydrology of these areas is studied to estimate variability in water resources under changing climate. A coupled hydrologic/hydraulic model, MIKE SHE/MIKE HYDRO RIVER, was used to calibrate and validate the model using streamflow data at the Shatial gauge station. The bias-corrected multi-model ensemble dataset accessed from the coordinated regional climate downscaling experiment (CORDEX) for two scenarios RCP 4.5 and RCP 8.5 were used up to the end of twenty-first century for the projection of stream flow. The calibrated model depicts an increase of 86% for RCP4.5 and 97% for RCP8.5, dominated by increased snow melt contribution due to consistent warming trends (average increase of 2.3 and 4 °C, annually for RCP 4.5 and 8.5, respectively). The most prominent increase was observed in winter months when predicted flows increased by 300% from historic. With predicted annual average available water of 4500 m3/s under RCP8.5, the study concludes that the available water would relieve some of the water stress in the country. However, the increased availability of water can also cause catastrophic floods. More flood defense and storage structures are needed to improve management practices in the downstream areas, particularly during wet and dry years.

Next-Generation Drought Intensity–Duration–Frequency Curves for Early Warning Systems in Ethiopia’s Pastoral Region

The pastoral areas of Ethiopia are facing a recurrent drought crisis that significantly affects the availability of water resources for communities dependent on livestock. Despite the urgent need for effective drought early warning systems, Ethiopia’s pastoral areas have limited capacities to monitor variations in the intensity–duration–frequency of droughts. This study intends to drive drought intensity–duration–frequency (IDF) curves that account for climate-model uncertainty and spatial variability, with the goal of enhancing water resources management in Borana, Ethiopia. To achieve this, the study employed quantile delta mapping to bias-correct outputs from five climate models. A novel multi-model ensemble approach, known as spatiotemporal reliability ensemble averaging, was utilized to combine climate-model outputs, exploiting the strengths of each model while discounting their weaknesses. The Standardized Precipitation Evaporation Index (SPEI) was used to quantify meteorological (3-month SPEI), agricultural (6-month SPEI), and hydrological (12-month SPEI) droughts. Overall, the analysis of historical (1990–2014) and projected (2025–2049, 2050–2074, and 2075–2099) periods revealed that climate change significantly exacerbates drought conditions across all three systems, with changes in drought being more pronounced than changes in mean precipitation. A prevailing rise in droughts’ IDF features is linked to an anticipated decline in precipitation and an increase in temperature. From the derived drought IDF curves, projections for 2025–2049 and 2050–2074 indicate a significant rise in hydrological drought occurrences, while the historical and 2075–2099 periods demonstrate greater vulnerability in meteorological and agricultural systems. While the frequency of hydrological droughts is projected to decrease between 2075 and 2099 as their duration increases, the periods from 2025 to 2049 and from 2050 to 2074 are expected to experience more intense hydrological droughts. Generally, the findings underscore the critical need for timely interventions to mitigate the vulnerabilities associated with drought, particularly in areas like Borana that depend heavily on water resources availability.

Advanced sleep disorder detection using multi-layered ensemble learning and advanced data balancing techniques

Sleep disorder detection has greatly improved with the integration of machine learning, offering enhanced accuracy and effectiveness. However, the labor-intensive nature of diagnosis still presents challenges. To address these, we propose a novel coordination model aimed at improving detection accuracy and reliability through a multi-model ensemble approach. The proposed method employs a multi-layered ensemble model, starting with the careful selection of N models to capture essential features. Techniques such as thresholding, predictive scoring, and the conversion of Softmax labels into multidimensional feature vectors improve interpretability. Ensemble methods like voting and stacking are used to ensure collaborative decision-making across models. Both the original dataset and one modified using the Synthetic Minority Oversampling Technique (SMOTE) were evaluated to address data imbalance issues. The ensemble model demonstrated superior performance, achieving 96.88% accuracy on the SMOTE-implemented dataset and 95.75% accuracy on the original dataset. Moreover, an eight-fold cross-validation yielded an impressive 99.5% accuracy, indicating the reliability of the model in handling unbalanced data and ensuring precise detection of sleep disorders. Compared to individual models, the proposed ensemble method significantly outperformed traditional models. The combination of models not only enhanced accuracy but also improved the system's ability to handle unbalanced data, a common limitation in traditional methods. This study marks a significant advancement in sleep disorder detection through the integration of innovative ensemble techniques. The proposed approach, combining multiple models and advanced interpretability methods, promises improved patient outcomes and greater diagnostic accuracy, paving the way for future applications in medical diagnostics.

Advantages of the Multimodel Ensemble Approach for Subseasonal Precipitation Prediction in China and the Driving Factor of the MJO

Advantages of the Multimodel Ensemble Approach for Subseasonal Precipitation Prediction in China and the Driving Factor of the MJO

Projecting future snow changes at kilometer scale for adaptation using machine learning and a CMIP6 multi-model ensemble.

Assessing future snow cover changes is challenging because the high spatial resolution required is typically unavailable from climate models. This study, therefore, proposes an alternative approach to estimating snow changes by developing a super-spatial-resolution downscaling model of snow depth (SD) for Japan using a convolutional neural network (CNN)-based method, and by downscaling an ensemble of models from the Coupled Model Intercomparison Project Phase 6 (CMIP6) dataset. After assessing the coherence of the observed reference SD dataset with independent observations, we leveraged it to train the CNN downscaling model; following its evaluation, we applied the trained model to CMIP6 climate simulations. The downscaled mean ensemble reproduced the spatial distribution and seasonality of the reference observations. We found an average decrease in the snow-covered area by about 20% in winter and 25% in early spring, an altitude-dependent of the SD changes, and a delayed snow cover appearance by the middle of the 21st Century under a high emission scenario. Overall, the downscaling model captures physically plausible relationships, enables high-resolution assessments of future SD based on a multi-model ensemble, produces results consistent with regional climate models, and provides valuable insights into how future snow changes will affect winter tourism and water resources, highlighting its potential benefits for a wide range of adaptation studies.

Multifaceted changes in water availability with a warmer climate

Climate warming alters spatial and seasonal patterns of surface water availability (P-E), affecting runoff and terrestrial water storage. However, a comprehensive assessment of these changes across various hydroclimates remains lacking. We develop a multi-model ensemble approach to classify global terrestrial hydroclimate into four distinct regimes based on the mean and seasonality of P-E. P-E is projected to become increasingly variable across space and time. Wet regions with low and high seasonality are likely to experience more concentrated increases in wet-season runoff by up to 20%, highlighting potential increases in flood-related vulnerability. Low-seasonality regions exhibit faster wet-season increases and more rapid dry-season decreases in soil moisture, heightening the likelihood of water scarcity and drought. Conversely, dry regions with high seasonality are less sensitive to climate change. These findings underscore the multifaceted impacts of climate change on global water resources, necessitating the need for tailored adaptation strategies for different hydroclimate regimes.

Development of stacking algorithm for bias-correcting the precipitation projections using a multi-model ensemble of CMIP6 GCMs in a semi-arid basin, India

Development of stacking algorithm for bias-correcting the precipitation projections using a multi-model ensemble of CMIP6 GCMs in a semi-arid basin, India

Selecting CMIP6 models to assess the impact of climate change on energy need for heating and cooling in Europe

The CMIP project has reached the end of its sixth phase, with a large set of simulations that represent a very large volume of data. The exploitation of all the simulations in impact studies would be extremely costly in computing time and storage. Moreover, these models suffer from several sources of uncertainties (coarse spatial resolution, physical parameterizations, ...) which produce large systematic biases.This study details an approach developed for the selection of a subset of simulations from the CMIP6 multi-model ensemble to assess the impact of climate change on the energy demand in Europe. At a climatic time scale, besides socio-economic factors, energy demand variability and evolution depend mainly on air temperature. The developed process of selection combines two main criteria applied to the daily mean, maximum, and minimum air temperature from all CMIP6 simulations where these variables are available. The first one consists of choosing a subset of models that represents the whole range of possible temperature changes in the future compared to the historical climate. The second criterion considers the skills of the CMIP6 historical simulations over Europe with respect to ERA5 climatology in order to discard the climate simulations with the poorest representation of the present climate. The methodology of selection allows the maximization of the diversity of climate projections among the best-performing models.

Seizure Prediction on EEG Signals using Feature Augmentation based Multi Model Ensemble

Background: Epilepsy is a neurological disorder that leads to seizures. This occurs due to excessive electrical discharge by the brain cells. An effective seizure prediction model can aid in improving the lifestyle of epilepsy patients. After analyzing various patents related to seizure prediction, it is observed that monitoring electroencephalography (EEG) signals of epileptic patients is an important task for the early diagnosis of seizures. Objective: The main objective of this paper is to assist epileptic patients to enhance their way of living by predicting the seizure in advance. Methods: This paper builds a feature augmentation-based multi-model ensemble-based architecture for seizure prediction. The proposed technique is divided into 2 broad categories; feature augmentation and ensemble modeling. The feature augmentation process builds temporal features while the multi-model ensemble has been designed to handle the high complexity levels of the EEG data. The first phase of the multi-model ensemble has been designed with heterogeneous classifier models. The second phase is based on the prediction results obtained from the first phase. Experiments were performed using the seizure prediction dataset from the University Hospital of Bonn. Results: Comparison indicates 98.7% accuracy, with improvement of 5% from the existing model. High prediction levels indicate that the model is highly capable of providing accurate seizure predictions, hence ensuring its applicability in real time. Conclusion: The result of this paper has been compared with existing methods of predicting seizures and it indicated that the proposed model has better enhancement in the accuracy levels.

Evaluating CMIP6 Global Climate Models Performances Over Nigeria: An Integrated Approach

ABSTRACTThe choice of global climate models (GCMs) for climate or hydrological studies remains a challenge due to their temporal and spatial variations and different performances in different parts of the globe. This study assesses the performances of 33 GCMs of the Coupled Model Intercomparison Project Phase 6 (CMIP6) for precipitation, maximum temperature and minimum temperature over Nigeria in order to select the best performing GCMs for aggregation into a multi model ensemble (MME). The study uses three statistical metrics (SM) and Random Forest (RF) machine learning method for the evaluation of the GCMs. In addition, the GCM performances were also estimated using spatial assessment, boxplots, scatter plots and mean monthly comparison at each grid point over the period 1985–2014. Finally, the average was used to generate variations of MMEs by increasing the number of models in the MME considering the inclusion of the better ranking ones first in order to determine the optimum MME for the variables. The highest ranking GCMs based on the average of the scores of the SM and RF were NESM3, CMCC‐ES, IPSL‐L‐R‐INCA and IPSL‐L‐R, MPI‐HAM and SAMO‐UNICON for precipitation; BCC‐C‐MR, MRI‐ESM, BCC‐ESM1, ACC‐ESM1‐5 and GISS‐E2‐CC for maximum temperature; and GFDL‐ESM, AWI‐C‐MR, IPSL‐L‐R, CAS‐ESM2 and AWI‐C‐LR for minimum temperature. The highest‐ranking model for all variables is ACC‐ESM1‐5, which ranked highest with a score of 0.6920 followed by BCC‐C‐MR with 0.6898, CAS‐ESM2 with 0.6597 and BCC‐ESM1 with 0.6545 score. The results of the spatial assessment, boxplots, scatter plots and the mean monthly comparison aligns with this. In the aggregation of MME for the three variables, the optimum number of models was obtained after averaging of the first four best ranking GCMs. This study presents a localised study, which is expected to reduce uncertainty in the projection of climate over Nigeria.

Precipitation Forecasting and Drought Monitoring in South America Using a Machine Learning Approach

Climate forecasting is essential for energy production, agricultural activities, transportation, and civil defense sectors, serving as a foundation for decision-making and risk management. This study addresses the challenge of accurately predicting extreme droughts in South America, a region highly vulnerable to climate variability. By employing a supervised neural network (NN) within a machine learning framework, we developed a methodology to forecast precipitation and subsequently calculate the Standardized Precipitation Index (SPI) for predicting drought conditions across the continent. The proposed model was trained with precipitation data from the Global Precipitation Climatology Project (GPCP) for the period 1983–2023. It provided monthly drought forecasts, which were validated against observational data and compared with predictions from the North American Multi-Model Ensemble (NMME). Key findings indicate the neural network’s ability to capture complex precipitation patterns and predict drought conditions. The model’s architecture effectively integrates precipitation data, demonstrating superior performance metrics compared to traditional approaches like the NMME. This study reinforces the relevance of using machine learning algorithms as a robust tool for drought prediction, providing critical information that can assist in decision-making for sustainable water resource management.

Climate Change Impact on Hydro-climatic Fluxes in Kantamal Catchmentof the Middle Mahanadi River Basin, India

Climate change is now considered as a newly added threat for natural resource management, which has significant impact on agriculture and allied sectors. Climate change studies play a pivotal role in developing sustainable natural resource management strategies. The present study assesses the effect of climate change on hydro-climatic fluxes in Kantamal catchment of the Mahanadi River basin, India. Utilizing the modified Mann-Kendall test, analysis of long-term climatic variables revealed a decreasing trend in rainfall and increasing trend in temperature. Employing a bias-corrected, multi-model ensemble of three regional climate models (RegCM4-4, RCA4, REMO2009) under the Representative Concentration Pathways (RCPs) of RCP4.5 and RCP8.5 scenarios, a rise in the average maximum temperature of 0.86°C under RCP4.5 and 1.16°C under RCP8.5, as well as an increase in the average minimum temperature of 2.35°C under RCP4.5 and 2.89°C under RCP8.5, by 2099 were projected. Rainfall is projected to decrease by 29.53% (RCP4.5) and 24.34% (RCP8.5), with surface runoff decreasing by 13.91% (RCP4.5) and 9.94% (RCP8.5), actual evapotranspiration declining by 7.81% (RCP4.5) and 7.77% (RCP8.5), soil moisture reducing by 11.17% (RCP4.5) and 9.69% (RCP8.5), and water yield is projected to decline by 39.45% (RCP4.5) and 33.05% (RCP8.5) as compared to the baseline period. Water stress situation is anticipated in the catchment emphasizing the need for planning and management of water resources, and sustainable agriculture.

Assessing predictive attribution in NMME forecasts of summer precipitation in eastern china using deep learning

Due to systematic errors in models and the special geographic location of eastern China, most global climate models exhibit significant biases in predicting summer precipitation in this region. This study evaluates the North American Multi-Model Ensemble (NMME) forecasts for eastern China, with a lead time of six months.While NMME simulates precipitation climatology well, it poorly predicts anomalies. Using the Res34-Unet deep learning post-processing method, which has been proven to enhance NMME’s forecasts, we explore that Western Pacific Subtropical High (WPSH) and sea surface temperature (SST) are critical in enhancing forecast accuracy. Among the four models evaluated, only GEM-NEMO (correlation of 0.538 with the WPSH) and CanSIPS-IC3 (which partly captured the impact of SST anomalies on precipitation) partially reflected the key factors identified by deep learning. Simulating these factors more accurately could greatly enhance NMME’s predictive skill for summer precipitation.

Selection of optimum GCMs through Bayesian networks for developing improved machine learning based multi-model ensembles of precipitation and temperature

Selection of optimum GCMs through Bayesian networks for developing improved machine learning based multi-model ensembles of precipitation and temperature

Comparison and Integrated Application for Runoff Simulation Models in Small and Medium-Sized River Basins of Southeast China Coastal Area

Runoff simulation is of fundamental importance for hydrological research. This study evaluated the applicability of multiple hydrological models and their ensembles for simulating runoff in small and medium-sized river basins of southeastern coastal China, focusing on the Xixi tributary of Jinjiang River and the Songxi and Chongyang tributaries of Minjiang River in Fujian Province. Four lumped hydrological models were selected for analysis: GR4J, IHACRES, TVGM, and MISDc-2L. The Bayesian model averaging method was utilized to compare the performance of each individual model and the multi-model ensemble in runoff simulation. Results: (1) For the calibration and validation periods of four hydrological stations, the mean values of KGE, NS, and R2 for the models GR4J, IHACRES, TVGM, and MISDc-2L were all above 0.7, and the mean values of |RE| were below 8.3%, without significant simulation accuracy variations when basin size changes, demonstrating strong regional applicability for runoff simulation; (2) The multi-model ensemble simulations using Bayesian model averaging of GR4J, TVGM, and MISDc-2L exhibited higher accuracy than individual models; (3) The MISDc-2L model demonstrated strong applicability in daily runoff simulations for both small and medium-sized river basins in Fujian Province and the large-sized Dongting Lake basin, showing that it is worthy of further application in other river basins across China. The findings of this study provide a reference for the selection and application of hydrological models for runoff simulation in small and medium-sized river basins of southeastern coastal China.

Assessment of Historical and Future Mean and Extreme Precipitation Over Sub‐Saharan Africa Using NEX‐GDDP‐CMIP6: Part I—Evaluation of Historical Simulation

ABSTRACTThis study assesses the performance of 28 NASA Earth Exchange Global Daily Downscaled Climate Projections (NEX‐GDDP‐CMIP6) models and their multi‐model ensemble (MME) in simulating mean and extreme precipitation across sub‐Saharan Africa from 1985 to 2014. The Multi‐Source Weighted‐Ensemble Precipitation (MSWEP) and Climate Hazards Group InfraRed Precipitation with Station Data (CHIRPS) are used as reference datasets. Various statistical metrics such as the mean bias (MB), spatial correlation coefficients (SCCs), Taylor skill scores (TSS) and comprehensive ranking index (CRI) are employed to evaluate the performance of NEX‐GDDP‐CMIP6 models at both annual and seasonal scales. Results show that the NEX‐GDDP‐CMIP6 can reproduce the observed annual precipitation cycle in all the subregions, with the model spread within observational uncertainties. The MME also successfully reproduces the spatial distribution of mean precipitation, achieving SCCs and TSSs greater than 0.8 across all subregions. The biases in mean precipitation are consistent across different reference datasets. However, most of the NEX‐GDDP‐CMIP6 models show trends of mean precipitation opposite to observations. While the MME can generally reproduce the spatial distribution of extreme precipitation, its performance varies with the reference dataset, particularly for the number of rainy days (RR1) and maximum consecutive dry days (CDD). TSS values for extreme precipitation indices differ significantly by region, reference data and index, with the lowest values over South Central Africa and the highest over West Southern Africa. The CRI indicates that no single model consistently outperforms others across all subregions, even within the same region, when compared to both MSWEP and CHIRPS. These results may be helpful when using NEX‐GDDP‐CMIP6 models for future projections and impact assessment studies in sub‐Saharan Africa.

Larval dispersal predictions are highly sensitive to hydrodynamic modelling choices

AbstractLarval dispersal is a critical ecological process in marine ecosystems, responsible for connecting and replenishing populations in patchy habitat. Because empirical measurements of larval dispersal are very challenging, coupled biological and oceanographic simulations (“biophysical models”) of larval dispersal are commonly used to answer ecological questions and support conservation management decisions. In the process of creating biophysical models, a series of choices must be made that do not have a single correct answer—sometimes because the oceanographic or ecological processes are uncertain; sometimes because trade-offs are required between different goals (e.g. computational time versus spatial resolution). In this paper, we demonstrate that larval dispersal estimates at management scales are strongly affected by these choices. Using three different hydrodynamic models of the Great Barrier Reef, we estimated the dispersal of crown-of-thorns starfish larvae in the spawning seasons between 2018 and 2021. Despite sharing similar physical forcings and using similar models of larval behaviour, we find that the different hydrodynamic models produce divergent predictions of larval dispersal between the reefs. If used to support crown-of-thorns starfish control decisions, these different predictions would recommend different priority reefs. Our results caution against the use of single models of larval dispersal, and suggest that multi-model ensembles may offer a valuable new perspective on dispersal patterns in marine environments.

Assessing the Effect of Bias Correction Methods on the Development of Intensity–Duration–Frequency Curves Based on Projections from the CORDEX Central America GCM-RCM Multimodel-Ensemble

This work aims to examine the effect of bias correction (BC) methods on the development of Intensity–Duration–Frequency (IDF) curves under climate change at multiple temporal scales. Daily outputs from a 9-member CORDEX-CA GCM-RCM multi-model ensemble (MME) under RCP 8.5 were used to represent future precipitation. Two stationary BC methods, empirical quantile mapping (EQM) and gamma-pareto quantile mapping (GPM), along with three non-stationary BC methods, detrended quantile mapping (DQM), quantile delta mapping (QDM), and robust quantile mapping (RQM), were selected to adjust daily biases between MME members and observations from the SJO weather station located in Costa Rica. The equidistant quantile-matching (EDQM) temporal disaggregation method was applied to obtain future sub-daily annual maximum precipitation series (AMPs) based on daily projections from the bias-corrected ensemble members. Both historical and future IDF curves were developed based on 5 min temporal resolution AMP series using the Generalized Extreme Value (GEV) distribution. The results indicate that projected future precipitation intensities (2020–2100) vary significantly from historical IDF curves (1970–2020), depending on individual GCM-RCMs, BC methods, durations, and return periods. Regardless of stationarity, the ensemble spread increases steadily with the return period, as uncertainties are further amplified with increasing return periods. Stationary BC methods show a wide variety of trends depending on individual GCM-RCM models, many of which are unrealistic and physically improbable. In contrast, non-stationary BC methods generally show a tendency towards higher precipitation intensities as the return period increases for individual GCM-RCMs, despite differences in the magnitude of changes. Precipitation intensities based on ensemble means are found to increase with the change factor (CF), ranging between 2 and 25% depending on the temporal scale, return period, and non-stationary BC method, with moderately smaller increases for short-durations and long-durations, and slightly higher for mid-durations. In summary, it can be concluded that stationary BC methods underperform compared to non-stationary BC methods. DQM and RQM are the most suitable BC methods for generating future IDF curves, recommending the use of ensemble means over ensemble medians or individual GCM-RCM outcomes.

On the Use of Hindcast Skill for Merging NMME Seasonal Forecasts across the Western United States

Abstract Multimodel ensemble forecasts have gained widespread use over the past decade. A yet unresolved issue is whether forecast skill benefits from the use of prior skill from each model in providing a weighted combination. Here, we use the available seasonal ensemble forecasts of six models from the North American Multi-Model Ensemble (NMME) to study various aspects of prior skill–based weighting schemes and explore ways to merge multimodel forecasts. First, we postprocess each NMME model through quantile mapping and a simple spread error adjustment. Then, using an equal weighted combination as the baseline forecast, we test merging the models together through skill-based weights by varying the prior skill metric and varying how the metrics are aggregated across the different subbasins and time of year. Results confirm prior work that the combined forecasts do outperform individual models. When evaluating prior skill, equal weighting generally performed as well as or slightly better than all weighting schemes tried. The skill of the weighting scheme was not found to be strongly dependent on prior metric but did improve when aggregating all forecasted months and subbasins together to provide one overall weight to each model. Also, we found that including an offset to the prior metric that nudged the weights closer to equal weighting improves skill especially at longer leads where individual model skill is low. Results also show that the weighting schemes performed better than regression-based techniques including multiple linear regression and random forest. Significance Statement Here, we test how effective the past performance of seasonal climate models can be used for generating weights to merge multiple models together using the North American Multi-Model Ensemble (NMME) for forecasting temperature and precipitation across the western United States. Our results showed there was little benefit in using prior performance compared to weighting all six NMME models equally. The performance of the weighted forecasts was not strongly dependent on the choice of performance metric or over how many months or basins performance was pooled. An important finding is that when only one or two models were used for merging, performance was reduced relative to equal weighting; however, if three or more models were used, performance was nearly the same.

An evaluation of the Africa-CORDEX Regional Climate Model's performance in simulating air temperatures and precipitation in the Melka-Wakena Catchment, Southeast Ethiopia

Understanding climate science is essential for effective policy development, adaptation, mitigation, and risk management. Given the inherent limitations in climate models, this study evaluates the performance of CORDEX Africa regional climate models to simulate precipitation and temperatures over the Melka-Wakena catchment. To accomplish this, the performance evaluation utilizes techniques such as multi-metric weighted ranking to select top-1 (best individual model), specific multi-model ensembles (top-N ensemble), multi-model ensemble, and average hybrid (top-N ensemble with MME) approaches at various temporal scales. Also, Taylor diagrams and empirical cumulative distribution functions are other useful elements for model comparison visualization. This study finds that, top-3 ensemble (for temporal scale tasmin, monthly pr), top-7 ensemble (monthly and seasonal tasmax), average hybrid (top-1 with MME (daily pr and tasmax), HadGEM2ES_RACMO22T for seasonal and yearly pr), and HadGEM2ES_RCA4 for yearly tasmax) are preferable, with weighted scores of 0.29, 0.44, 0.22, 0.20, 0.35, 0.40, 0.39, 0.33, 0.45, 0.25, and 0.33, respectively. In short, RCA4 and RACMO22T efficiently mimic air temperatures and precipitation in the Melka-Wakena catchment. While ensemble approaches are generally more resilient and possible, single-best-performing model approaches can be quite effective for long-term periods. Finally, these studies have an important implication for improving climate change assessment, adaptation, and mitigation measures.