Study On Synoptic Discrimination And Forecast Of Regional Low Visibility In North China

In recent years, the incidents of regional low visibility in North China have been widely concerned, which not only have very serious impact to traffic safety and agricultural production in North China, but also cause very serious injury to the health of people in the local. The occurrences of low visibility events are relevant to special circulation and weather system. Therefore, to identify circulation and weather system quantitatively, explore more effective forecasting methods and improve the accuracy in regional low visibility forecast, have been very realistic problems for meteorologists to solve urgently.In this paper, by classifying different types of weather situation of regional of low visibility in North China, establishing automatic identification system to screen weather situation, checking days meeting the type requires, regional low visibility forecasts were then got. Neural network and stepwise regression method were then used to forecast the concrete appearance sites in the forecasted regional low visibility day. The main results of this study are:(1) Low visibility weather process was studied with synoptic analysis methods. Results showed that, the weather situation of regional low visibility in 500hPa can be divided into 3 types:2 troughs and 1 ridge type, low trough type, zonal flow type.(2) By using the trough and ridge automatic identification system, regional low visibility weather situation meeting the types requires between 2002 and 2011 are all identified quantitative, and several threshold of the recognition system were then determined, as quantitative criterion to whether situation meet the types requires. The automatic identification system of regional low visibility situation in North China was then established. And all day of the 10 years were discriminated whether meet the types requires.403 samples were discriminated, which include 104 low visibility samples. The total sample number is 2422,2019 empty samples are eliminated.(3)Depend on type discrimination, K index, depression of dew point, and depression of pseudo equivalent temperature of 500hPa and 850hPa are taken as parameters to diagnosis a further step. The three parameters’threshold range of regional low visibility can be obtained. Using the three kinds of parameters to discriminant the situation meeting the type requires again, eliminate empty furthermore, the regional low visibility weather forecasts can be get.223 empty samples are eliminated,13 low visibility samples were leaked, and the accuracy rate of forecast was 50.5%. Regional low visibility forecast of 2012 were then did,3 days of regional low visibility weather is forecasted completely,6 low visibility samples were leaked.(4)Finally, establish forecasting relational expression between 55 physical parameters and low visibility of each site by using neural network and stepwise regression method, and give a forecast of the correct sites in regional low visibility days forecasted before. The accuracy rates of forecast were 18.2% and 32.7%.To sum up, the test results show that, by using classification, automatic identification, physical parameter diagnosis and neural network and stepwise regression, low visibility events can be forecasted more accurately. Effective science and technology support is given in this paper to improve accuracy in low visibility forecast.