Performance Evaluation of NEX-GDDP-CMIP6 Models and Selection of a Multi-Model Ensemble for Climate Simulation in Northern Togo

Local-scale evaluation of climate models is essential to ensure the reliability of climate projections. This study evaluates the performance of 32 NEX-GDDP-CMIP6 models and constructs a multi-model ensemble for simulating monthly precipitation, maximum temperature (Tmax), and minimum temperature (Tmin) at four stations in northern Togo (Dapaong, Mango, Kara, Niamtougou) over the 1983–2014 period. Model performance was quantified using four metrics (bias, RMSE, Pearson correlation coefficient R, and Willmott’s d index), integrated into a Comprehensive Rating Metric (RM) for the overall ranking. Results showed that most models reproduced temperatures more accurately than precipitation. For Tmax, RM scores ranged from 0.03 (MPI-ESM1-2-HR, NESM3) to 0.97 (INM-CM4-8, the best score in the study), with generally high performance. For Tmin, greater spread among models was observed, with RM scores ranging from 0.00 (CMCC-CM2-SR5) to 0.91 (KACE-1-0-G and UKESM1-0-LL). For precipitation, RM scores ranged from 0.23 (CanESM5) to 0.86 (HadGEM3-GC31-LL), with systematic underestimation across nearly all models. Based on the overall ranking across the three variables, 18 models achieved an RM score above the 0.5 threshold. These models demonstrated reliable performance in reproducing the climate of northern Togo and were retained to form the multi-model ensemble for climate projections. These results provide a robust and reproducible evaluation framework for climate modeling in the Sudano-Sahelian region of West Africa.