摘要
利用2011年2—8月逐日降水量序列及东亚地区850 h Pa经向风场资料建立多变量时滞回归(multivariable lagged regression,MLR)模型,对5—7月江西降水10~30 d和50~70 d低频分量分别进行延伸期逐日预报实验。结果表明:2011年江西降水存在显著的10~30 d和50~70 d的振荡周期。降水50~70 d低频分量延伸期预报技巧明显优于10~30 d低频分量延伸期预报技巧,平均预报技巧高达0.86。降水50~70 d低频分量延伸期预报可准确预报降水低频位相的正负转换,能为江西延伸期强降水过程发生的时段预测提供预报信号。
Based on the dai ly rainfal l series and 850 hPa meridional wind o f the NCEP reanalysis data from February to August in 2011, a multivariable lagged regressive (M LR ) model was constructed.The model was applied to do the dai ly extended forecast o f low frequency rainfal l over Jiangxi Area from May to July.The results show that there are two distinct oscillations in Jiangxi which are about 10-30 days and 50-70 days respectively.The predictive s k i l l of the 50-70 days low-frequency rainfalls is d is tinc tly better then that of the 10-30 days.The forecast score of 50-70 days low frequency rainfal l is up to 86% , which is able to accurately predict the positive and negative phase conversion of low frequency rainfall.This can provide extended range forecast signal for the heavy rainfal l process in Jiangxi.
出处
《沙漠与绿洲气象》
2018年第1期71-75,共5页
Desert and Oasis Meteorology
基金
2016年江西省气象局重点项目(动力气候模式产品在月季温度降水预测中的降尺度应用研究)
2016年江西省重点项目(延伸期(16-30天)逐日预报技术研究和业务建立)
江西省科学技术厅指导性计划项目(2013ZBBG70022)共同资助
关键词
延伸期预报
低频降水
MLR模型
extended range forecast
low frequency ra in fa l l
multivariable lagged regressive (MLR ) model