Quality Analysis Of Temperature Forecast In Shenyang Area And Study On The Correction Method Of Towns Forecast

The experiment analyze the municipal temperature guidance forecast in Shenyang area in 2013-2015,calculating the accuracy,the absolute error and the error grade.Secondly,using the guidance forecast,history data of national stations and the township site automatic meteorological station,taking Shenbei New Area as an example,the experiment forecast maximum and minimum temperature of township by using statistical downscaling method.Comparing the results and analyzing the relevant factors,then choose the best forecast method for each site.The method used in this paper are difference value method,inverse distance weighting interpolation method(IDW)and multiple regression method.Each method is divided into monthly,quarterly and annual calculation according to the length of training period.The influence of sunshine,precipitation and other factors on accuracy of prediction and forecast error are also considered when calculated maximum and minimum temperature.Test results and conclusions include:the forecast quality decreases with the increase of aging time,forecast of 24h performed better in accuracy and absolute error,also forecast of maximum temperature performed better than minimum temperature.The quality of forecast in each station is different,Xinmin station performed best and Shenbei station worst.The process of strong cooling and strong heating reduced the quality of temperature forecast.The prediction quality is the best at a median of 11.58 degrees,and the two sides are gradually reduced.With the increase of precipitation,the quality of the minimum temperature forecast increases gradually,while the maximum temperature quality decreases first and then increases.With the increase of sunshine hours,the quality of forecast is improved gradually.Overall,the quality of guidance forecast meet the requirements of the data source of the township.Both of the methods can effectively improve the accuracy of the temperature prediction of minimum and maximum temperature.Due to the theoretical reason for statistical downscaling,quality of each method are dependent on the quality of guidance forecast.The minimum temperature accuracy can be increased to 66.7%and the maximum to 80.5%when using the arithmetic value method.While using IDW,the minimum temperature accuracy can be increased to 66.7%,the maximum to 78.4%.Multiple regression method performed the best with annual training period,which increased the minimum temperature accuracy to 72.2%,8%higher than current business and reduced the average absolute error 0.38 degrees;while maximum temperature accuracy to 82.6%,20%higher than current business and reduced the average absolute error 0.37 degrees.Sunshine and precipitation factors did not join in this experiment for there was no Pearson relationship.This study has some positive reference when revising forecast at other sites with similar influence factors,for the theoretical basis and calculation method used in this paper are relatively simple and convenient.