Interannual Variability in Subseasonal Prediction Skill for East Asian Summer Monsoon Precipitation

Accurate subseasonal prediction of East Asian summer monsoon (EASM) precipitation, especially over south of the Yangtze River, remains a grand challenge due to the complex interaction between tropical and extratropical intraseasonal variability (ISV). This study systematically examines the interannual variation in subseasonal prediction skill for EASM precipitation using historical hindcasts from multiple subseasonal-to-seasonal (S2S) models, with a focus on the ECMWF model. Predictions of 7- and 14-day lead (Week 2 and Week 3) show markedly different interannual variability, with Week 3 exhibiting roughly twice the variability of Week 2. By identifying years with notably good and poor prediction skill, we show that windows of opportunity for enhanced skill are closely tied to favorable background states and the presence of precursors. In the years with good 7-day lead prediction skill, EASM ISVs are characterized by strong tropical convection over the western North Pacific and robust midlatitude Rossby wave trains propagating from Eurasia. By contrast, the years with poor skill show no coherent precursor signals from the midlatitudes, although tropical signals remain robust. In ECMWF, the variability of 7-day lead prediction skill for EASM precipitation is more closely related to midlatitude ISV forcing than to tropical forcing. In good years, enhanced midlatitude upper-level ISV forcing over northern China is linked to increased seasonal-mean precipitation north of the Yangtze River, following pre-winter warming in the northern Indian Ocean and tropical Atlantic. For 14-day lead prediction, enhanced Madden–Julian Oscillation (MJO) activity contributes to higher prediction skill in the years characterized by a combined La Niña and negative Pacific Meridional Mode. These years are typically preceded by a pre-winter warm–cold–warm tripolar SST pattern in the North Atlantic. The results highlight the crucial role of mean-state–modulated tropical–extratropical coupling in shaping year-to-year fluctuations of EASM subseasonal prediction skill, thereby advancing our understanding of EASM predictability.