On The Extraction And Predictability Of Stable Components In Extended-range Predictions

In recent years, extended-range forecasting for the coming 10-30 days has become an important direction of development. But its scientific basis and business applications are still in the exploratory stage. Using Lanczos filtering and Empirical Orthogonal Function (EOF), a method is given to extract stable components in extended-range forecasting for the coming 10-30 days, and then basing on it we get climatic stable components and abnormal stable components. And then we studied its temporal and spatial variation features. At last, we studied its predictability in extended-range forecasting, and how to improve extended-range forecasting for the coming 10-30 days by using it.First, a method of extracting stable components in extended-range forecasting for the coming 10-30 days is given. Stable components depend on the temporal and spatial scales. Basing on scaling analysis, we divided the NCEP-DOE Reanalysis 2 geopotential field into 3parts:60d lowpass filtering,10-60d bandpass filtering and 10d highpass filtering. Stable components are most included in 60d lowpass filtering and 10-60d bandpass filtering, and chaotic components are included in 10d highpass filtering and also parts in 10-60d bandpass filtering. By using EOF, we introdutied how to get climatic stable components and abnormal stable components.Then we disscused their temporal and spatial variation features.Second, we studied further about stable components, especially about the temporal and spatial variation features of abnormal stable components in two examples (Continuous rain weather around Yangtze River Basin in Early 2009 and Low Temperature&Snowstorm from 31 Oct to 16 Nov 2009 in East China). The features of stable components are summarized as follows:It has large temporal and spatial scales, keeping long time (10 days above) on the stability or performance of the low-frequency changes or long-scale wave activity. It reflects the changes of zonal circulation index and the movement, adjustment activities of ultra-long wave and Planetary Motions. It's also closely related to continuous weather process.Last, we disscused the predictability of stable components in extended-range forecasting by Noliear Local Lyapunov Exponent and ensemble predictions. Climatic stable components and abnormal stable components have better predictability than chaotic components. Then by using T63L16 Numerical Model we found abnormal stable components and 60d lowpass filtering are more difficult to be predicte accurately, and which may get better results by statistical forecasting.