Evaluation of MSWX temperature forecast performance across Ethiopian climatic regions

Abstract This study evaluates the forecasting performance of the Multi-Source Weather (MSWX) product for daily minimum and maximum temperatures across Ethiopia’s climatic regions during 2015–2019. The evaluation employs continuous statistical metrics, graphical analyses, spatial temporal bias mapping, and the Critical Rating Index (CRI) to establish an overall performance ranking of MSWX ensemble members. Results indicate that MSWX3 exhibits superior skill in reproducing daily minimum temperatures, with higher correlation coefficients and lower RMSE values across most regions, while MSWX1 performs best for maximum temperatures. For instance, MSWX3 achieves correlation values exceeding 0.80 for minimum temperature in several regions, whereas MSWX1 records comparatively lower RMSE values for maximum temperature. A systematic bias is evident, with minimum temperatures generally overestimated and maximum temperatures underestimated. Specifically, daily maximum temperatures are most underestimated in Region 8 (28.5%), whereas minimum temperatures are most overestimated in Region 7 (16.5%). Despite these biases, MSWX ensemble members demonstrate a strong capability in capturing daily temperature extremes across Ethiopia’s diverse climatic regions. These findings highlight the potential of MSWX products to enhance short-term temperature forecasting, thereby supporting informed decision-making, climate resilience, and sustainable development across agriculture, water resources, health, energy, and disaster risk management sectors.