Abstract:By analysis of the climate characteristic of extreme temperature，the representative meteorological stations in Qingyang were
selected，and the correlation between the representative stations and other stations was analyzed．At first，the highest and lowest tem-
perature prediction model of the representative stations was established by using the method of the support vector machines(CMSVM)
and ECMWF product，and based on that the linear regression statistics relationship of the representative stations and other stations was
established．The forecast test indicated that the accuracy rate of highest and lowest temperature forecast was improved using this method
compared with that of the numerical forecasting release at every station in 2009，and forecast of highest temperature was good in effect
and the mean accuracy rate has enhanced 5%．So the model could be applied in the day－to－day business directly．

The predictionmodel for the highestand lowest temperature is established by the CMSVM regressionmethod based on ECMWF data and the extreme temperature observations from 8 automaticweather stations inQingyang ofGansu Province from 2003 to2007.The effect test in 2008 operation forecast indicated that the rate ofaverage forecastaccuracy is64% and 71% for the highestand lowest temperature in 5 days, which has the good instruction for the real-timeweather service. The forecast accuracy rate of the highest and lowest temperature decreased with the forecast time increase, and the accuracy rate for the lowest temperature is higher than thatof the highest temperature. The forecast effect for the lowest temperature is better in spring and summer than that in autume and winter, but it is opposite for the highest temperature. The absolute and the average forecasterrors for the highestand lowest temperature are added with the forecast time increase, and the absolute forecasterror for the highest temperature is bigger than thatof the lowestbut the average error is close and positive. The forecast effect test in Septemper of 2008 indicated that the forecast values presented the similar trend to real time observations for the highest and lowest temperature.

The augural characteristic ofDoppler radarVWP productwas revealed by analyzing two typical cases of the hail and the strong rainfall occurred in Xifeng in 2006. The result indicated that the strong convection generally occurred after arid region(no cloud region,“ND”in the figure ofVWP)vanishing, and the wind shearwas obvious in lower level during two strong convective weather processes. The hail appeared in the cold advection environmentwith anti-clockwisewind direction shear, but the strong precipitation appeared in thewarm advection environmentwith clockwise wind direction shear, and this vertical variation ofwind field can keep a long time. The northwest air current expanded from high-level to downward 2. 7 km in hailweather, and the southwest vapor transferred above 5. 5 km in the strong precipitationwith jet stream in the low leve.l This verticaldistribution characteristic ofwind field has good instruction to short-timeweather forecast.

The new generation weather radar echoes data of the strong rainstorm weather occurred on July 8, 2006 in Qingyang citywas analyzed. Results show that upper shearline resulted in thisweather event, and the difference in intensity, velocity echoes and physical quantity p roduction between the rainstorm and hailweather was not so obvious, but the damage was more severe. So for weather forcast, weather events like this should be focused, and the same way as hail supp ression shoud be adop ted to reduce p recip itation intensity and decrease damages.

Based on the Doppler radar data，the characteristics of the echo intensity，velocity，top height and vertically integrated liq-uid water(VIL)of the strong hail on May 30 and the rainstorm on July 1．’2 in 2005 in Qingyang region were analyzed，and compari-sOn between these two weather events Was carried out．Results show that difierence of the reflectivity between the hail and rainstorm Was great；the hail echo developed and moved quickly at the rate of 50 km／h，and the central intensity of hail echo exceeded 50 dBZ；there Was some speed fuzzy area on velocity map，and middle cyclone Was detected by the radar ；山e t0p of cloud Was more than 12 km；VIL increased suddenly before haifal1．and hail occurred when VIL in central was over 20 ks／m ．While the echo intensity of the rainstorm Was only 25—35 dBZ，and rainstorm echo moved slowly；the 0 speed line was regular and it can reflect the low level jet and develop—ment of air mass，and the top height of cloud Was only 5—8 km and VIL changed little during rainfall process．

The chosen factors in the temperature and precipitation equations established using HLAFS information in May to September are concluded in order to find out the factors used most frequently. Results show that factors chosen in the temperature forecasting equation for 24 hours are mainly before 24 hours，while in that for 48 hours，the factors are mainly after 24 hours，and the factors in real time are chosen in large probability，which is consistent with forecasterˊs；in precipitation forecasting equations，the first chosen factor is the divergence wind，and then the divergence and vorticity of Q vector. The heavy storm case in Qingyang on 4 July 1999 was used to analyze the real time and forecasting fields using most factors in forecasting equation，and the chosen factors have clear physical meaning，which provides reference basis for selecting precipitation factors in MOS forecast.

Based on the analysis of Xifeng new generation weather radar echo data of regional hail weather on 30 May 2005 in Qingyang city，the hail cloud characteristics in new generation weather radar intensity echo，velocity echo and physical quantity products were given out. Results show that the shearline at the bottom of the low votex resulted in this hail weather，and it was not the zonal hail cloud moving to Qingyang city but the supercell hail cloud forming over the city in the moving course of the squall line that caused this strong hail event.

The support vector machine(SVM) is a new statistical study theory method which developed in recent years,  by using this method in
hail in the method is this paper',  We analyzed  the classific;ation prediction of hail weather based on summarizing the circulation(harac;teristic;s of hail in east region of Cansu provinc;e} and(ompared with traditional forecast under different weather types.It is confirmed that SVM method is not need to classify weather types and the effect of forecast model being established is better on many samples. This has provided new method and idea for hasin stations nn weather forecast.

The drought index is defined by precipitation and evaporation in Longdong region during July and August in this paper，and it is used to denote drought.The drought climate feature is discussed by analyzing its precipitation and the drought index of 15 stations，and it is found 500 hPa mean situation is different in drought and rain yyear.The cause of drought is also discussed by the correlation between drought and Pacific temperature and 500hPa height feild，and finally the forecast model has been established