Analysis And Experimental Forecast Of Monthly Mean Temperature Variability In Heilongjiang Province

By using NCEP/NCAR reanalysis data of monthly mean geopotential height of 500hPa, monthly MSLP, monthly mean surface temperature and monthly mean temperature in Heilongjiang Province, temperature variability of different time scale, simultaneous & previous general circulation is studied with method of diagnostic analyzing of extreme cold or warm in Heilongjiang Province, and predictive relationship has been developed by using method of screening regression and SSA-MEM as well. The results show that temperature of Heilongjiang Province is increasing trend in century-scale. In this background, increasing process has apparent decadal-scale variety: there is a distinctness period of temperature increasing before 1950s', from 1950s' to 1970s' early it is temperature decreasing period, then outstanding temperature increasing is followed after 1970s'. Moreover, seasonal cycle also has apparent decadal-scale variety from 1909 to 2002: monthly mean temperature in winter time from 1983 to 2002 is obviously warmer than that from 1909 to 1931 (3.7℃ warmer in February) , while colder in summer (0.2℃ colder in July and August) and annual range of temperature is lower 2.5℃ . With composition analyzing and significance test of simultaneous & previous general circulation and surface temperature of extreme warm or cold January & July in Heilongjiang Province, distinct difference can be concluded in terms of statistics between previous general circulation or surface temperature and the tempreture of Heilongjiang Province, which gives some evidence for predicting extreme warm or cold temperature of Heilongjiang Province's winter and summer. By use of correlation analyzing of temperature in Heilongjiang Province and previous general circulation or surface temperature, provision predictors are chosen to develop a predictive relationship of temperature anomaly in January or July of Heilongjiang Province with the theory of screening regression. Also, a predictive model is established with method of SSA-MEM, which can be used to predict monthly mean temperature of the first half year in Heilongjiang Province. RMS error of these two methods are both less than climatological forecast's, especially, independent sample is adopted in SSA-MEM method.