Impact of Annual Cycle Bias of ECMWF S2S Model on the Subseasonal Rainfall Forecast Skill Over East Asia in 2024

Abstract The subseasonal forecast of rainfall beyond 10 days over the East Asian monsoon region is largely affected by cross‐timescale modulations, particularly the synoptic time scale (<10 day) and annual cycle (>90 day) components, which are related to initial atmospheric condition and external forcing, play important roles in determining the subseasonal forecast skill of subseasonal‐to‐seasonal (S2S) models. Based on time scale decomposition, we investigate the multi‐time scale errors of the ensemble rainfall forecast generated by the European Centre for Medium‐Range Weather Forecasts (ECMWF) S2S model in 2024. Our results show that the model exhibits a faster annual cycle beyond 1 week forecast lead. This rapid drift restricts the model forecast skill for rainfall by one day via the impact of atmospheric internal processes. The systematic bias in the model is mainly observed in the phase and amplitude of the first harmonic rainfall component, which stems from the model's quick and strong response to the forcing of La Niña–like conditions during the Meiyu season in China.