摘要：

The Madden-Julian Oscillation (MJO)/Boreal SummerIntraseasonalOscillation (BSISO) has been considered an important climate mode of variability onsubseasonaltimescales for East Asian summer. However, it is unclear how well the MJO/BSISO index would serve as guidance forsubseasonalforecasts. Using a probabilistic forecast model determined through multiple linear regression (MLR) with MJO, ENSO, and long-term trend as predictors, we examine lagged impacts of each predictor on East Asia extended summer (May-October) climate from 1982 to 2015. The forecast skills of surface air temperature (T2m) contributed by each predictor is evaluated for lead times out to five weeks. We also provide a systematic evaluation of three commonly used, real-time MJO/BSISO indices in the context of lagged temperature impacts over East Asia. It is found that the influence of the trend provides substantial summertime skill over broad regions of East Asia onsubseasonaltimescales. In contrast, the MJO influence shows regional as well as phase dependence outside the tropical band of the main action centers of the MJO convective anomalies. All three MJO/BSISO indices generate forecasts that yield high skill scores for week 1 forecasts. For some initial phases of the MJO/BSISO, skill reemerges over some regions for lead times of 3-5 weeks. This emergence indicates the existence of windows of opportunity for skillfulsubseasonalforecasts over East Asia in summer. We also explore the dynamics that contribute to the elevated skills at long lead times over Tibet and Taiwan-Philippine regions following the initial state of phases 7 and 5, respectively.