Study On The Potential Prediction Of Thunderstorm And Hail In Northern Xinjiang

Severe convective weather, such as thunderstorm and hail, are main meteorological disasters. By the influence of the particular topography and weather systems, the southern and northern sides of Tianshan mountain regions are easily effected by thunderstorms and hail, but it’s difficult to forcast it. For now, the study on the potential forecasting of thunderstorm and hail in Northern Xinjiang only summarized the weather types or discussed the roles of the convection parameters, lack of forecast combined with the automatic identification of weather patterns or physic diagnosis. Therefore, it is necessary to study the potential forecast of thunderstorm and hail in Northern Xinjiang.Based on the data of 25meteorological stations in Northern Xinjiang from 1981 to 2013, the climatic characters of spatial and temporal distribution of thunderstorm were analyzed in detail by using methods of Climatic Statistical Analysis, Concentration Degree and Concentration Period Analysis. Combined with the NCEP/NCAR reanalyzed data during 2002-2013, the circulation background of thunderstorm or hail in Northern Xinjiang were studied, and four main synoptic patterns of weather situation were summarized. Also, the criterion of the four patterns were determined and they can be used to identify weather situation automatically. By diagnosing the physical quantities and convective parameters, the physical quantity and parameters which had good correlation with thunderstorm and their thresholds of different weather patterns were determined. Then, potential forecasting models of thunderstorm and hail in Northern Xinjiang were built with the stepwise regression method and artificial neural networks method, and the forecasting effects were verified. The main results are as follows:(1) Analysis of temporal and spatial distribution characteristics of thunderstorm in Northern Xinjiang showed that:the western region of Northern Xinjiang and the western Tianshan Mountains are regions which were easy for the occurrences of thunderstorm, that closely associated with the terrain and topography. Bayinbuluk, which located in the Tianshan mountains, had the largest number of annual thunderstorm days(36.2d), the annual mean thunderstorm days of the frequency occurring regions in northern and west mostly between 16d and 27d, while Dabancheng had the least number(0.7d).The annual thunderstorm days in Northern Xinjiang presented a slowly descending tendency between 1981～2000 then increased between 2000-2009, and there was a 6～7 years period of oscillation. The thunderstorm mostly occured in summer (from June to August), which accounted for 78.2%, the peak was in July and mainly occurred from 14:00 to 20:00 in a day, accounted for 51.8%. Thunderstorm had a high concentration in a year, the mainly occurrence period was from late June to early July, but has been decreasing in recent years.(2) Statistics of the circulation background of thunderstorm in Northern Xinjiang showed that:the sunoptic patterns can be classified into four kinds:vortex of mid Siberia, jet pattern, trough of Northern Xinjiang and northwest air current. The vortex of mid Siberia is the most common pattern among them, which accounted for 30.8%, mostly occurred in June and Jury; the jet pattern often occurred in June and August, accountined for 22.4%; trough of northern Xinjiang and northwest air current mainly happened in July, accountined for 20.1% and 23.0%. According to the characteristics of circulation situation and the high and low levels air circulation configuration, the criterions of different weather patterns were determined. After the automatic identification of weather patterns, the number of samples of thunderstorm decreased form 2354 to 1515, and the voiding rate was 35.6% and CSI was 19.4%.(3)Diagnosis of the physical quantities and convective parameters showed that quantities such as BCAPE, DCAPE, BLI, CIN, WINDEX, K, TT, CT, Tdd, Qdiv,(?)se,SHR3,etc were good predictors for thunderstorm and hail, while different weather patterns had different kinds of quantities. The dew-point deficit in 850hPa had a good correlation with the thunderstorm and hail in all four weather patterns. Over all, trough of Northern Xinjiang pattern needed ample water vapor condition while vortex of mid Siberia and northwest air current needed the instability of atmosphere. Using the scope of threshold value of each quantities or parameters, there were 482 empty samples eliminated from 1515 samples, and the voiding rate was 31.8% and the CSI of the forecast model based on the automatic identification of weather patterns and diagnosis of physical quantities was 27.8%. The forecasting result of 2013 was ideal and the CSI was 26.4%.(4)Potential forecasting models of thunderstorm in Northern Xinjiang have been built with stepwise regression method and artificial neural networks method to forecast thunderstorm or hail from April 2013 to October 2013.The results showed that either of the method could accurate the results to station, while because of the advantage to solve problems between linear and nonlinear, artificial neural networks method was better than stepwise regression method. The forecasting effectiveness based on weather patterns was better, relatively speaking, vortex of mid Siberia had batter effect. Based on the results of the regional potential forecast, the specific station was forecasted by using the neural network method, and CIS was between 13.2% and 18.5%.In conclusion, the regional potential forecasting model of Northern Xinjiang was established, based on the automatic identification of weather patterns and diagnosis of physical quantities; then the single station forecasting model with the stepwise regression method and the neural network method was built. These work added anew technical support for the prediction of thunderstorm storm and hail in this region.