基于Logistic方法构建的中国降雪判定方法研究

A distinguishing method for snowfall eventsin China using Logistic regression approach

基于中国区域逐日降雪、降水、气温、相对湿度、气压和风速等观测数据,构建了中国区域的Logistic降雪判定方法,并对该方法和当前广泛应用的其他降雪判定方法在中国区域的适用性展开对比研究。结果表明,单温度阈值法和S曲线法对[-3,4]℃气温区间内的降雪模拟不确定性相对较大。比较而言,Logistic拟合的系列方法成功率更高,对中国不同区域降雪识别也更为稳健,尤其是对青藏高原地区降雪事件的识别效果明显优于其他方法。在Logistic方法中,温度和相对湿度对降雪判定起决定作用,而气压和风速的影响相对较小。Logistic湿球温度方案(Log T_(w))和气温+相对湿度方案(LogT_(a)H_(R))均能很好地再现降雪量的空间分布和年际变化特征,且相应偏差均小于其他方法;总体上,这两种方案对降雪量识别效果差别不大。因此,可使用Log T_(w)方案或LogT_(a)H_(R)方案对中国区域降雪事件进行判别,尤其是对模式中降雪事件的识别。

Based on the observations of daily snowfall,precipitation,temperature,relative humidity,air pressure and wind speed in China,a distinguishing method for the snowfall events based on the Logistic regression approach is constructed,and the applicability of this method and other widely used snowfall distinguishing methods is compared.Results show that the single temperature threshold and S-curve methods are relatively uncertain for snowfall simulation within the temperature range from-3℃to 4℃.By contrast,the series of Logistic fitting methods have higher success rates in determining snowfall events,and are more robust for snowfall recognition in different regions of China,especially in the Tibetan Plateau.In the Logistic methods,temperature and relative humidity play a decisive role in determining snowfall,while the influences of air pressure and wind speed are relatively small.The Logistic wet-bulb temperature scheme(Log T_(w))and the air temperature+relative humidity scheme(LogT_(a)H_(R))can well reproduce the spatial distribution and interannual variation characteristics of snowfall,and the corresponding deviationsare smaller than other methods.On the whole,there is little difference between the two schemes for snowfall recognition.Therefore,the Log T_(w) or LogT_(a)H_(R) scheme can be used to identify snowfall events in China,especially the snowfall events in climate models.