A new extended-range weather forecast model- 

MLR/PC-CAR model 

(Yang Qiuming , Weather and Climate Laboratory of Jiangsu Meteorological Science Institute , January 18, 2014) 

Low-frequency component of rainfall and low-frequency principal component of circulation are used to construct a hybrid forecast method of the multivariate lagged regressive model and principal component-complex number autoregressive model (MLR/PC-CAR) for extended-range daily change forecast on the 20–30-day low-frequency component of the LYRV in the summer of 2011, so as to significantly prolong the lead time (Yang, 2014). 

The forecast experiments performed in June-August, 2011 show that the lead time will achieve about 50 days for the forecast of low-frequency rainfall in the LYRV on the time scale of 20-30 days. It is shown that the model with the principal component of 850 hPa low-frequency meridional wind at the Southern Hemisphere middle latitudes as a predictor has a significantly higher forecast precision than that with the low-frequency meridional wind in East Asia. It is also indicated that at the time scale of 20-30 days, the rainfall in the LYRV has a strongly lagged correlation with the principal component related to the Southern Hemisphere extratropical SCGT. The further forecast experiments of this MLR / PC-CAR hybrid model constructed on the basis of multi-year data with stronger 20–30-day oscillation showed that the SCGT is a significant signal for forecasting the changes in summer low-frequency rainfall in the LYRV in the future 50 days. 

It is shown that the MLR/PC-CAR hybrid forecast method may enable the forecast model to more accurately reflect the principal low-frequency lagged correlation of the rainfall to circulation and the changes in a longer period, and may also more accurately describe the evolution of the quasi-linear low-frequency dynamic process. This complex time series statistical method can not only describe the propagation characteristics of the low-frequency wave on a complex plane to obtain more information than fluctuations in real space, but can also reveal the temporal evolution of fluctuations more stably, which will aid in further extending the lead time. 

References 

Yang Qiuming, 2014: A study on the method of the extended-range forecast for the low frequency rainfall over the lower reaches of Yangtze river valley in summer based on the 20－30-day oscillation. Acta Meteor. Sinic, doi：10.11676/qxxb2014.028 (in press) (in Chinese). 

http://www.cmsjournal.net/qxxb_cn/ch/reader/view_abstract.aspx?flag=1&file_no=20200274&journal_id=qxxn_cn

[Figure (a)Correlation between the rainfall over the lower reaches of the Yangtze River valley (LYRV) and global meridional wind anomaly of 850 hPa on the time scale of the 20―30-day from May 1 into August 31 in 2011 ;(b)The third mode of the meridional wind anomaly of 850 hPa on the time scale of the 20―30-day in the region:0°－360°, 10°－65°S; Prediction (dashed line) and observation (solid line) of the 20―30-day rainfall over LYRV for the period from 1 into 50 days in the summer of 2011, the bar represents the time series of the daily precipitation over the LYRV (unit: mm), initial date (d) 31 May,(e)10 June,(f)20 June, which are forecasted by means of the MLR/PC-CAR model]