Non-uniform error correction for daily precipitation forecasting in Eastern China on the S2S model

Subseasonal-to-seasonal (S2S) precipitation forecasting bridges the gap between weather forecasts and monthly or seasonal outlooks. Due to the accumulation of model errors over time, error correction has become a critical step in improving forecast accuracy. This study proposes an error correction method tailored for subseasonal precipitation forecasting: the Non-uniformity Error Correction (NUEC) method. NUEC is a dynamic correction method that accounts for the temporal and spatial heterogeneity of model forecast errors. It allocates errors proportionally across each year and day, resulting in a more nuanced and precise error correction process. In contrast, the Uniform Error Correction (UEC) method, which is widely recognized, applies a simpler approach by distributing uniform errors based on multi-year average errors across individual days. Validations show that NUEC consistently outperforms UEC in precipitation forecasting correction from 1998 to 2017 in eastern China, particularly over the forecasting with longer leading time. NUEC achieved 10.2 % enhancement in mean anomaly correlation coefficient (ACC), surpassing UEC's 3.0 % improvement. NUEC also demonstrated a higher reduction in errors of the root mean square error (RMSE) across all time frames than those of UEC. Independent sample validation for the years 2018 to 2020 further confirms the effectiveness of NUEC, particularly in eastern China, where substantial error reduction demonstrates its potential for enhancing S2S precipitation forecasting. • The proposed method captures non-uniform errors more effectively than previous method. • The proposed method improves anomaly correlation coefficient by 10.2 %. • The proposed method reduces root mean square error more than previous method. • Independent sample testing confirms the proposed method's superior performance.