新疆昌吉州一次暴雪过程的多模式数值预报检验

Multi-Mode Numerical Forecast Test of a Snowstorm Process in Changji, Xinjiang

摘要: 利用常规地面观测资料、EC_thin、T639_thin、ED_thin、DOGRAFS资料对2015年12月10日~12日昌吉州暴雪过程多模式数值预报效果进行检验分析。通过分析EC_thin、T639_thin、ED_thin、DOGRAFS四家数值模式的预报，分别对降水时间段、落区、量级和Ts评分进行检验，结果表明：降雪时段的预报以开始时间偏早，结束时间偏晚为主，24 h时效的预报效果较48 h和72 h好，且EC_thin和T639_thin预报效果较好；3个时效的降雪落区准确，但大值区(暴量)范围小，主要集中在中西部，东部预报偏小一个量级；晴(雨)、一般性降水(暴雪)、漏报率，ED_thin具有很好的参考价值；各模式预报的天气系统强度略偏弱，是造成预报与实况有偏差的主要原因，而模式预报的差异程度造成了预报降雪量级存在差异。

Abstract: Using conventional ground observation data, EC-thin, T639-thin, ED-thin and DOGRAFS data, the multi-model numerical forecast effect of the heavy snow process in Changji from 10 to 12 Dec, 2015 was tested and analyzed. By analyzing the forecast of four numerical models of the precipitation time interval, falling area, magnitude and Ts score were tested respectively. The results showed that: the forecast of snowfall time interval was mainly early at the beginning and late at the end. The forecast effect of 24-hour time interval was better than that of 48 and 72 hours, and EC-thin and T639-thin were better than others. The three time-dependent snowfall areas were accurate, but the range of large value areas (storm) was small, mainly concentrated in the central and Western regions, while the eastern prediction was one order of magnitude smaller; ED-thin had good reference value in sunshine (rain), general precipitation (snowstorm), and missing rate; the intensity of weather system forecasted by each model was slightly weak. The main reason for the deviation be-tween the forecast and the actual situation was that the difference degree of the model forecast results in the difference of the forecast snowfall magnitude.