湖南冬季积雪日数影响因子及其预测模型

The Impact Factors and Prediction Model of Winter Snow Cover Days in Hunan Province

基于湖南省89个气象观测站积雪气象观测资料、NCEP/NCAR再分析资料、Hadley中心月平均海温资料及ERA5逐月再分析表层土壤湿度资料,采用相关分析、年际增量法和多元线性回归等方法,研究了影响湖南冬季积雪日数的主要因子。结果表明,前期夏季东亚夏季风指数的年际增量、79月太平洋海温指数、10月AAO的年际增量和9月土壤湿度指数是湖南冬季积雪日数的4个关键影响因子。选取上述4个预报因子,建立了湖南冬季积雪日数年际增量统计预测模型。模型模拟效果较好,积雪日数年际增量拟合值和实况的相关系数为0.83,积雪日数距平拟合值与实况相关系数达0.77,通过了0.001的显著性水平检验。利用20102018年独立样本对模型进行检验,结果显示,积雪日数距平预测值与实况同号率为7/9。基于4个关键因子建立的全省逐站冬季积雪日数预测模型,时间相关系数检验表明,湘中一带模拟效果最好。

Based on the snow cover data collected at 89 meteorological stations in Hunan Province,NCEP/NCAR reanalysis data,monthly mean sea surface temperature data from Hadley Centre and the ERA5 monthly surface soil moisture reanalysis data from ECMWF,we studied the main factors affecting winter snow days in Hunan Province by using correlation analysis,interannual increment method and multiple linear regression method.The results show that the interannual increment of East Asian summer monsoon index in early summer,Pacific SST index from July to September,AAO interannual increment in October and soil moisture index in September are the four key impact factors for winter snow days in Hunan Province.By selecting the above four impact factors,we established the interannual increment statistical prediction model of winter snow cover days in Hunan.The statistical prediction model of the interannual increment of snow days has a good simulation result.The correlation coefficient between the fitting value of the annual increment of days with snow cover and the actual situation is 0.83.The correlation coefficient between the anomaly fitting value of snow cover days and the observations reaches 0.77,which has passed the significance test at 0.001 level.The independent samples from 2010 to 2018 are used for the test,and the results show that the same sign rate between the predicted and observed anomaly values of snow cover days is 7/9.Based on the four key factors,the station-by-station prediction model of winter snow days in Hunan Province is established.The time correlation coefficient test shows that the simulation effect in central Hunan is the best.