Evaluating CMIP6 Model Accuracy in Predicting and Clustering Precipitation and Temperature Trends in the Omo-Gibe River Basin, Ethiopia

Abstract

This study evaluated the performance of the latest generation of global climate models (CMIP6) in predicting and clustering climate variables in the Omo-Gibe River Basin (OGRB), Ethiopia—a region highly vul nerable to climate change. Historical data (1984–2014) were used to evaluate 12 CMIP6 models using statistical metrics (R², RMSE, MBE) and categorical metrics (POD, FAR, CSI), identifying the best-performing models for each cluster. The selected models underwent bias correction using distribution mapping to enhance projection accuracy under two socio-economic pathways: SSP2–4.5 (moderate emissions) and SSP5–8.5 (high emissions) for the near-term (2023–2053) and mid-term (2054–2084) periods. The study employed the K-means clustering method to classify spatial variations in precipitation and temperature, resulting in three clusters of precipitation and two clusters for temperature. Sen’s slope estimator and the Modified Mann-Kendall (MMK) test were utilized to analyze trends in precipitation and temperature time series. Key findings include model performance: For pre cipitation, INM-CM5-0, FGOALS-g3, and IPSL-CM6A-LR demonstrated strong performance in Clusters 1 and 3, while MPI-ESM1-2-LR and NorESM2-MM excelled in Cluster 2. For temperature, INM-CM5-0, BCC-CSM2-MR, and MPI-ESM1-2-LR exhibited the highest accuracy in Clusters 1 and 2. Significant precipitation increases were noted, particularly in Cluster 3, which shows a 74.7% rise by mid-century under SSP5–8.5. Temperature projec tions indicate maximum increases of 11.5% (absolute change: 3.1°C) in Cluster 1 and 8.4% (absolute change: 2.2°C) in Cluster 2, while minimum temperature rises reach as high as 23.5% (absolute change: 3.5°C). Spatial clustering revealed distinct climate patterns, with regions experiencing monomodal and bimodal rainfall cycles, providing a nuanced understanding of precipitation and temperature dynamics across the basin. These results reveal pronounced shifts in climate patterns, underscoring the importance of targeted adaptation strat egies. The findings provide critical insights into regional climate dynamics and support the development of climate-resilient water resource management and infrastructure planning. Collaborative efforts between policy makers, researchers, and communities are essential to address the anticipated challenges effectively. RÉSUMÉ [Traduit par la rédaction] La présente étude a évalué la performance de la dernière génération de modèles climatiques globaux (CMIP6) dans la prévision et le regroupement des variables climatiques dans le bassin de la rivière Omo-Gibe (OGRB), en Éthiopie — une région très vulnérable aux changements climatiques. Des données historiques (1984-2014) ont été utilisées pour évaluer 12 modèles CMIP6 au moyen de mesures