Abstract Climate model ensembles and proper bias correction enhance climate projections; however, most previous predictions focused on a single model. This study aimed to project the climate of the Kessie watershed in the Ethiopian Abay Basin using best-fitting model ensembles, bias-corrected with relatively outperformed methods for the study area. The performance of the following methods was evaluated: power transformation, linear scaling, local intensity scaling, and distribution mapping for rainfall; and linear scaling, variance scaling, and distribution mapping for temperature. The method’s performance was assessed by its ability to correct errors in the model’s historical predictions, using observational time series from 1984 to 2014 as a reference. The capability of climate models to capture the case study area’s climate was evaluated using bias-corrected historical model predictions, from which various model ensembles were developed. The performance of individual and model ensembles was re-evaluated to select outperforming models for climate projections. The Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) was used to rank models and methods based on their performance, as measured by statistical metrics. The selected models’ future simulations were bias-corrected using the outperforming methods, and the resulting data were used to project the climate of the case-study area. It was done at the 2030s (2015–2044), the 2060s (2045–2074), and the 2090s (2075–2100), under four Shared Socioeconomic Pathway scenarios: SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP4-8.5. All methods reduced errors in model simulations, with the power transformation (variance scaling) comparatively excels for rainfall (temperature). ACCESS-CM2, MIROC6, MRI-ESM2-0, and BCC-CSM2-MR models relatively outperform for rainfall; ACCESS-ESM1-5, GFDL-ESM4, and NorESM2-LM for maximum temperature, and UKESM1-0-LL, ACCESS-CM2, and NorESM2-LM for minimum temperature. However, all model ensembles consistently outperform individual models for both rainfall and temperature. The long-term trend (1986 to 2100) showed significant increases ( p 0.05 ) except in the 2030s under SSP2-4.5. Both maximum and minimum temperatures are expected to increase significantly across all time periods, except that maximum temperature remains unchanged in the 2060s and 2090s under SSP1-2.6, and minimum temperature remains unchanged in the 2090s under SSP1-2.6 and increases, but not statistically significantly, in the 2090s under SSP4-8.5. The rainfall projections exhibit mixed and inconsistent trends, whereas temperature projections show a statistically significant upward trend across most scenarios and time periods. The study’s results suggest the need for climate change adaptation strategies to support sustainable development in the region.
Ayalew et al. (Thu,) studied this question.