多变量时滞回归模型在江南地区初夏降水低频分量延伸期预报中的应用

利用江南地区77个台站的日降水资料及NCEP/NCAR再分析资料,基于不同时间尺度的江南地区降水低频分量和东亚地区850 h Pa低频经向风主成分,建立了多变量时滞回归(Multivariable Lagged Regression,MLR)模型,并对2011年5—7月江南降水低频分量进行延伸期逐日预报试验。结果表明,50~70 d时间尺度的江南低频降水的平均预报技巧高达0.92,可准确预报持续性强降水过程和降水低频位相的正负转换。对利用2001—2012年资料分别构建的MLR模型的历史回报预测试验表明,在50~70 d振荡较强和正常的年份,模型能提前30 d做出初夏江南低频降水分量预报。模型结果也表明,850 h Pa低频经向风的发展和演变是影响初夏江南低频降水未来30 d变化的显著信号,可作为延伸期强降水预报的关键因子。

Application of multivariable lagged regression model in extended range forecast of low frequency components of precipitation in Jiangnan area in early summer

Based on the daily precipitation data at 77 stations in Jiangnan area of China and the NCEP/NCAR reanalysis data,the different time scale low frequency precipitation in Jiangnan area and the principal component of 850 hPa low frequency meridional wind in East Asia are employed to construct a multivariable lagged regression(MLR) model,which is applied to complete the daily extended range forecast test for the low frequency precipitation in Jiangnan area during May-July of 2011.Results show that the average prediction skill of the 50-70 d low frequency precipitation in Jiangnan area is up to 0.76,which is able to predict accurately the period of persistent heavy rainfall and the conversion of positive and negative phase of low frequency precipitation.The hindcast experiments from 2001 to 2012 show that the MLR model can predict well the low frequency precipitation in Jiangnan area ahead 30 days in the years of stronger or normal 50-70 d oscillation.The model results also show that the development and evolution of 850 hPa low frequency meridional wind are the outstanding signals for the change of low frequency precipitation in the next 30 days and can be considered a key factor of extended range heavy precipitation prediction in Jiangnan area in early summer.