时空投影法在上海地区梅汛期降水延伸期预报中的应用

本文基于2001—2010年上海市11个基本气象站的逐日降水量和澳大利亚气象局的逐日大气低频振荡(MaddenJulian Oscillation,MJO)指数(包括RMM1和RMM2)资料,选取MJO指数作为预报因子,上海地区梅汛期降水量作为预报对象,建立了基于时空投影法(spatial-temporal projection model,STPM)的上海地区梅汛期降水延伸期预报模型。利用该模型对近6年(2011—2016年)的梅汛期降水进行回报试验,其预报技巧评估结果表明:该模型对未来10～25 d的降水具有较好预报效果,可较准确地预报出梅汛期3/4左右的降水量级和降水发生时段。其中,预报时效为10～20 d的预报技巧较高,而提前21～25 d的预报技巧略有下降。总体而言,基于MJO活动的STPM预报模型在上海地区梅汛期延伸期降水预报中具有较好的参考价值。

Application of Spatial Temporal Projection Model for Extended Range Forecast During Meiyu Season in Shanghai

Recently, an extended range (10-30 days) forecast method, namely the spatial temporal projection model (STPM) has been applied in predicting the low frequency precipitation in tropics and shows comparable skills. Based on the daily precipitation data from 11 weather stations in Shanghai, and the daily MJO index (RMM1 and RMM2) provided by Australian Meteorological Bureau during 2011-2010, the STPM model for the extended range forecast of precipitation during Meiyu season in Shanghai was constructed. Depending on the relationship between the quasi period MJO activities and the precipitation, the coupled patterns between temporal varying real time multivariate MJO (RMM) index and regional low frequency precipitation during Meiyu season is derived based on the singular value decomposition (SVD). Therefore, using the MJO index as predictor and projecting it onto the spatial temporal coupled patterns, the forecast of regional low frequency precipitation is obtained. The STPM model is then used in the precipitation prediction during 2011-2016. The evaluation demonstrates that the prediction of the STPM model achieves a useful skill at a lead time of 25 days. Three quarters of precipitation occurrences and intensity during Meiyu season can usually be predicted. Among them, the STPM model attains a higher skill in the prediction of 10-20 d lead time. Generally, the STPM model provides a useful method in the extended range forecast during Meiyu season in Shanghai.