不同插值算法对GRAPES模式在乌鲁木齐地区的降水分级检验比较  

Comparison of Precipitation Forecast Results in Different Grades with Three Interpolation Algorithms in GRAPES Model in Urumqi Area

作　　者：黄成荣 贾丽红 汤浩 蒋军 

机构地区：[1]新疆气象台,新疆乌鲁木齐830002

出　　处：《沙漠与绿洲气象》2011年第2期23-27,共5页Desert and Oasis Meteorology

摘　　要：采用三种不同的平面插值算法对GRAPES模式降水预报场资料在乌鲁木齐地区的24 h和48 h降水分级预报能力进行比较和检验。总体上,24 h的预报能力优于48h的。通过不同插值算法计算后,降水预报能力或保持原水平或有不同程度的提高。对比分析发现,插值算法的不同对≥0.1 mm且≤10 mm累积降水的影响不大,对降水量偏少的极端干旱区站点影响也不大。然而当累积降水≥10 mm时,经过高斯插值算法换算以后的站点,降水能力较维持原水平的最优插值法提高了5～7倍,从TS得分情况看,具有较好的预报效果。而距离平方法的降水预报能力提高幅度更大,约为10倍,不同的是,预报效果不但没有改善,反而有所下降,尤其是在累积降水≥10 mm且≤25 mm时,这时空报率高于高斯插值法计算的结果,TS评分有所下降。This paper compared and verified the 24 h and 48h precipitation forecast abilities of different grades by three plane interpolation algorithms with the precipitation forecast field data of GRAPES model in Urumqi area.Generally,the 24-hour forecast was better than 48 hours＇.Through different interpolation algorithm,the precipitation forecast capabilities maintained at the same level or improved in some degrees.The difference of interpolation algorithm had little influence on the accumulated precipitation between 0.1mm and 10mm as well as the that in extreme arid region.However,when the cumulative precipitation was more than 10mm,compared with those stations remaining the original level with the optimum interpolation method,the precipitation forecasting capacities for these stations calculated with the Gaussian interpolation algorithm improved 5 to 7 times,and the TS scores of these stations also showed better forecast results.The precipitation forecasting ability of the least square distances advanced even larger,about 10 times,but the forecast effect declined,particularly when the precipitation was no less than 10 mm no more than 25mm.Under this situation,the TS score decreased a little,and the forecast results of fitting accuracy rate were higher than that of Gaussian interpolation algorithm.

关 键 词：插值算法 GRAPES 降水 检验 

分 类 号：P456[天文地球—大气科学及气象学]