Evaluation of the ECMWF SEAS5 Climate Model in Indonesia Based on Dominant Climate Characteristics

Indonesia’s agriculture critically depends on seasonal climate prediction skill, particularly for anticipating rainfall onset and dry‐season duration. The ECMWF SEAS5 model offers high‐resolution coupled forecasts and representation of large‐scale drivers such as ENSO, IOD, and MJO. This study evaluates SEAS5 performance over Indonesia for the 1991–2020 period by benchmarking simulated precipitation and temperature against MSWEP and ERA5 using deterministic, probabilistic, and composite‐based approaches. Results show that SEAS5 successfully reproduces the main spatial and seasonal rainfall cycle, with annual‐cycle correlations above 0.8 in most regions. Monthly mean temperature correlations range from 0.7 to 0.9. Probabilistic evaluation reveals a positive median continuous ranked probability skill score (CRPSS) of ∼0.25, indicating that the model generally outperforms climatology. The model more accurately captures the driest and coldest months (61.3% and 62.1% match, respectively) than the wettest or hottest months (50.4% and 47.1%). Furthermore, composite analysis highlights that predictive skill for precipitation is notably higher and more spatially consistent during El Niño phases compared to La Niña phases. Combined correlation–hit rate classification identifies four regencies (Rote Ndao, Kupang, Kepulauan Tanimbar, and Maros) as top‐performing areas with very strong (>0.8) agreement. These areas are recommended as pilot regions for SEAS5‐based agricultural climate services, where forecast outputs can be further refined through bias correction and operational post‐processing procedures.