Study On Forecasting Method Of Summer Temperature And Precipitation In China Based On Dynamic - Statistics

Short-term climate prediction has been a worldwide problem. The forecast results is quite uncertainty.In recent years,under the background of global warming there have been widespread drought and flood, drought and flood sharp turn,and a wide range of high and low temperature extreme events in China in summer, to mention just a few. This is a new challenges to the short-term climate prediction. Study on the short-term climate prediction in China is no longer simply pay attention to the physical statistical forecast method or numerical model predictive method of application, but also attach importance to the combination of both.In view of this, this article is based on power-the basic principle of statistics, with the National Climate Center coupled ocean-atmosphere model (CGCM) as the foundation, according to the characteristics of the summer temperature and precipitation in China, the combination of dynamic statistical model of error correction method and the empirical mode decomposition coupling method to forecast the temperature and precipitation test are brought up. At the same time,the impact of global warming on prediction is analysisedUsing climate trend coefficient analysis the variation of observed and model predicted China summer temperature in the past30years. The dates come from NCEP/NCAR daily reanalysis date and National Climatic Center monthly models temperature date. Through empirical mode decomposition method fitting the raise trend of observed temperature and remove the trend in the context of global warming. Combined with systematic error correction and seasonal prediction method-Statistical method of combining power to analysis the trend on the impact of summer temperatures forecast. The results show that:China has a significant increase trend in summer temperature in last30years, In most areas the climate trend coefficient get through0.01significance level. But mode Temperature almost have no change in trend in last30years, there is obviously inadequate for global warming. Using the empirical mode decomposition method can effectively fitting and remove the trend of observed temperature. When forecasting the temperature we remove the trend first and add the trend in the last. Trough tests find that the method can significantly improve the result compare with direct forecast result in most area. And solve the problem of forecast result is lower than observed result. All the results show that it is necessary to remove the observed temperature increase trend in numerical prediction model results in post-processing.Taking Singular value decomposition to predictions of the National Climate Center business model CGCM, then get the improved forecast results. SVD decomposition by field coupling between the years1983-2011summer rainfall patterns forecast and actual rainfall values, the selection mode Forecast best3-7modal and live precipitation, these revised results of the different number of modal number of comparison, selected number of modal best of the China region Corrections effect, then get the revised results from the model predictions.21-26years of2004-2009cross hindcast experiment anomaly correlation coefficient (ACC) and the root mean square error (RMSE) as evaluation criterion to test the return result. The best come to take the first five modal2004-2009most of the year as the year revised the number of modal revised, Already in2004-2009mode and actual precipitation data and inspections are indeed in the best and most of the Year the five modal Corrections effect a result of systematic errors is of Dhamma compared. The results show that the SVD set Dhamma in most years the Dhamma revised effect than the systematic errors. And take2010as the forecast year results of the cross-examination of return to1983-2009vears28vears and forecast results to be analvzed to determine the number of modal. The method has the potential business value.