Abstract:The middle and lower reaches of the Yangtze River basin (MLRYB) are prone to flooding because their orientation is parallel to the East Asian summer monsoon rain belt. Since the East Asian summer monsoon presents pronounced intraseasonal variability, the subseasonal prediction of summer precipitation anomalies in the MLRYB region is an imperative demand nationwide. Based on rotated empirical orthogonal function analysis, 48 stations over the MLRYB with coherent intraseasonal (10-80-day) rainfall variability are identified. Power spectrum analysis of the MLRYB rainfall index, defined as the 48-station-averaged intraseasonal rainfall anomaly, presents two dominant modes with periods of 20-30 days and 40-60 days, respectively. Therefore, the intraseasonal (10-80-day) rainfall variability is divided into 10-30-day and 30-80-day components, and their predictability sources are detected separately. Spatial-temporal projection models (STPM) are then conducted using these predictability sources. The forecast skill during the period 2003-2010 indicates that the STPM is able to capture the 30-80-day rainfall anomalies 5-30 days in advance, but unable to reproduce the 10-30-day rainfall anomalies over MLRYB. The year-to-year fluctuation in forecast skill might be related to the tropical Pacific sea surface temperature anomalies. High forecasting skill tends to appear after a strong El Niño or strong La Niña when the summer seasonal mean rainfall over the MLRYB is enhanced, whereas low skill is apparent after neutral conditions or a weak La Niña when the MLRYB summer seasonal mean rainfall is weakened. Given the feasibility of STPM, the application of this technique is recommended in the real-time operational forecasting of MLRYB rainfall anomalies during the summer flooding season.