Abstract:The basic characteristics of precipitation probability forecasts (PPS) are analysed. Theprobabilistic quantitative precipitation forecast equation is given by the use of REEP methodbased on the BMNFM. Moreover,the objective results are verified by new weather data andforecaster's knowledge,and the objective PPF and subjective PPF are estimated by B score.

Abstract:Based on the data analysis of the wind speed,wind direction and temperature observedon Tianjin tower during 8 months in the winter season of 1990-1992,the characteristics ofwind shear in the boundary layer below 250m in Tianjin are given,such as the temporal andspatial variations of the index pin the power law of the wind profile,the frequency distribution of the wind shear in different layers and different time intervals,the relation between thewind shear and stability. The vertical distribution of wind in the tower layer over the city ispresented. The variations of the wind shear and temperature during the passage of cold frontare also analysed.

Abstract:The band echo of the gust front of a severe hail storm system is found out during radardetection of the hail storm. The properties )structure and causes of the band echo of the gustfront are discussed. The reactions of the band echo of the gust front on the hail storm systemare analyzed. The band echo of the gust front is different from that of the gust in a classicalmodel. The self-trigger mechanism for continuation and conveyance of the hail storm systemis given. Application of the band echo of the gust front to nowcasting is considered tentatively.

Abstract:Through lots of data training and calculations,the selected BP network is improved to beadapted to the typhoon track prediction. After 2159 times of training,the neural network output of the typhoon track——westward,northward and northwestward,is compared with thehistorical data. Its fitted ratio is up to 97%,The probability of the typhoon moving trend isgiven by means of human-computer dialogus.

Abstract:Using the method of maths morphology,the area-spliting of typhoon clouds is conducted,and the possibility density of grain features of typhoon clouds is statistically analysed.Combined with the method of fuzzily discriminant and multi-level neural network,the typhoon clouds' picture recognition of GMS satellite image is studied. The result suggests thatthe method in successful.

Abstract:Based on the analysis of operational process of typhoon prediction,an objective comprehensive model of probability ellipse is designed to determine the typhoon track forecast. It isa new taste for synthetical application of each track forecasting methods and automatical objective decision. The difficult technique of automatic distinguish and the selection of objectivedecision reference point in software design are solved. The result shows it is better than thegeneral regression decision level.

Abstract:Applying the grid point field data of ECMWF,the diagnostic analyses to nine typhoonsfrom 1980 to 1992 which took a west turn while they moved north were made. The resultshows that when typhoons took west turns,there are warm advection,vapour flux convergence,stratification instability,aeroligical divergence and entire level cyclonic circlllation addition,and there is an entire level height descent center in the coverage of those additionarea,which result in typhoons to take a west turn while they move north.

Abstract:The medium-range forecasting method of cold air activity is presented,which is a compositive decision based on NWP product,experience,synoptics,statistics,energetics,artificalintelligence and extended-NWP etc.

Abstract:The numerial prediction model of the local torrential rain enhancement is an operationalsystem of regional numerical prediction and the established system is to be realized on localstation's microcomputers.The model is introduced from Lanzhou Plateau Atmosphere Institude to this observatory.The contrast experiment of the numerical model has been done inthe mean time.It was processed in batches during the plum rain season from 15th June to15th July in 1994.The result shows that the model runs with no problems,the calculation isstable.It possesses certain ability to precipitation forecasts.

Abstract:The location,development and structure of mesoscale cloud burst cluster are analysed,using water vapour images from NOAA,as well as GMS cloud charts and Doppler radar data. The forecasting method is given.

Abstract:The sectional torrential rain forecast in Anhui province was discussed during flood period with the application of the step wise regression and triple discriminatory with T42 predicting data of 48 hours instead of the real-time diagnosis. This test was conducted during 1993-1994. Its results show that (1) the CSI of the torrential rain prediction was from 0.30 to 0.71,both methods could be used in the torrential rain forecast in Anhui; (2) the predictingeffect of the regression was better than discriminatory; (3)the predicting effect in surplusMei-yu rainfall was better than that in deficient Mei-yu years.

Abstract:12-36h rainfall forecast has been studied in 5 feature regions of Hubei and Wuhan station during mei-yu period. It was divided into 3 grades (light rain.moderate rain and heavyrain). Based on the grey forecast model,the precipitation predictand's consensus of vary numerical model (Beijing LAFS、Beijing B、Japan etc.)was made for the station;Using methodof regression analysis,rainfall predictions was given with LAFS data. Semi-operational prediction was conducted in 1993 and 1994. Better forecasting results were achieved.