Skillful, timely, and reliable seasonal forecasts are crucial in mitigating the adverse impacts of climate-induced risks in Ethiopia and the rest of Greater Horn of Africa region. However, access to skillful and usable forecasts is currently challenging. This study evaluated the skill of raw and bias-corrected deterministic and probabilistic summer (JJA) rainfall forecasts from the Climate Forecast System Version 2 (CFSv2) for Ethiopia, spanning lead times from 0.5 to 4.5 months. The investigation also considered the influence of increased ensemble size on forecast skill by comparing performance of CFSv2 with the North American Multi-Model Ensembles (NMME). The findings indicated that CFSv2 exhibited limited skill for operational use in seasonal rainfall forecasting over Ethiopia. In contrast, NMME displayed promise, suggesting that with some value addition efforts such as bias correction, statistical downscaling, and identification of smaller subset of best performing models, could position it as a valuable component in Ethiopia’s seasonal climate forecast services.

Despite their potential usefulness, coarse resolution global models like CFSv2 fail to meet users’ need for forecasts at local to regional scales. To address this limitation, the Weather Research and Forecasting model (WRF) was explored for its potential to enhance CFSv2 forecasts through downscaling. A sensitivity study using WRF identified optimal parameterization schemes, utilizing Climate Forecast System Reanalysis (CFSR) for initial and boundary conditions. Subsequently, the WRF model, configured with the identified optimum model configuration, was employed to downscale operational CFSv2 summer season rainfall forecasts. Despite downscaling a small subset of ensemble members, the WRF model demonstrated value in refining raw CFSv2 forecasts. However, additional research is essential to further fine-tune WRF configurations, potentially minimizing biases and enhancing forecast skill.

The findings of this study are expected to contribute towards improving access to skillful and usable seasonal predictions that could help decision-makers in mitigating adverse impacts of climate induced risks.