The Application Of Characteristic Quantities Of Atmospheric Circulations For Climate Prediction In Flood Season In Xinjiang

It is demonstrated that global warming has severely impacted on natural ecosystem, especially in arid and semi-arid regions. Xinjiang is one of typical arid area. In recent 40 years, especially from 1987, the climate has evidently become warming and wetting in Xinjiang. On the whole, it is beneficial to the rehabilitation and melioration of environment. Nevertheless, it also results in latent danger on making a further violent melt and shrinkage of modern glaciers, and increasing the frequent occurrence of some meteorological disasters, such as rainstorms, hot days, droughts and floods, and some secondary and derivative geological disasters, such as debris flows and landslides. In order to reply the effects from climate change, we should make full use of meteorological high-tech, including remote-sensing information technique etc, intensify monitoring and assessment of climate and environment change, improve climate prediction accuracy. Climate prediction is still one of hotpots and difficulties in the research field of atmospheric science. At present, Statistical methods and climate-dynamic models are still the main methods of climate prediction. In recent years, climate models have developed rapidly, but they are difficult to satisfy with the requirement of climate prediction. Therefore, statically prediction method is still one of dominating methods.In this paper, the climate change in flood season in recent 46 years in Xinjiang has been analyzed. It shows that climate is experiencing warming and wetting in Xinjiang. Xinjiang is divided into 10 regions for climate prediction in flood season. 74 characteristic quantities of atmospheric circulations from 1960 to 2001, which are reanalyzed by National Climate Center of China, are applied and analyzed firstly in Xinjiang. The prediction models of seasonal precipitation, monthly precipitation, seasonal mean temperature or monthly mean temperature from June to August at 10 regions are separately developed by using characteristic quantities. Through correlation analysis, 4 to 6 characteristic quantities are selected for each model. The precipitation and mean temperature in flood season in 10 regions are closely related to the previous characteristic quantities of Northern Hemisphere. Among them, the dominant factors are different in different regions or period. All the prediction models in flood season pass the confidence test of 0.01. The average prediction accuracies of all the prediction factors are from 73.0% to 78.6%. And the average prediction accuracies of seasonal and monthly mean temperature in these regions are from 74% to 90%; the average prediction accuracies of precipitation in flood season, June and July in these regions are from 64% to 72%. They also have been proved being valuable for climate prediction of Xinjiang in flood season, and the prediction accuracies of temperature is higher than that of precipitation in flood season.