The Dynamic Statistical PP Forecast Model of Rainstorms in Southeast Gansu
Dong Yan-xiong · 2006
Based on the re-analyzed grid data of NCEP,the observed precipitation of 31 stations in southeast Gansu,and the 15~40 cm ground temperature data of 4 representative stations from 1960 to 2001,86 primarily selected factors were designed with 11 kinds of physical quantities,the forecasted area was divided into 3 districts of east,south and west,weather circulation was classified as 4 patterns,3 primary and 12 two-stage discriminant functions of rainstorms in southeast Gansu were obtained through discriminant analysis approach,and the dynamic statistical PP forecast model is composed of these functions.The results of examining in return indicate that in the situation of 90%~93% rainstorm days included by positive examples,excluding rate of forecast mistake reaches 63%~68%;The two-stage dynamic discriminant in east and south districts is good,the forecast skill is basically above 60%,of which for the composite flow patterns in east district it can reach above 80%,and it is not good in west district on the whole,but relatively to actual forecast level of rainstorms,12 functions have certain reference value.The examination conducted by using observed data from 2004 to 2005 July shows that the model forecast skill can reach 39.3%;Through experimental running,applying 12 and 36 hours forecasting data of T213,the model explanation has good reflection to the rainstorm occurred in southeast Gansu from June 30 to July 2,2005,two aspects both show that the model has practical value.