孝感汛期降水预测模型研究

Research on Precipitation Prediction Model in Xiaogan Flood Season

本研究利用国家气候中心下发的130项气候监测数据和孝感国家基准气候站1981—2019年汛期(5—9月)降水数据,分析130项气候监测数据与孝感汛期降水的相关性,通过逐步回归的方法选取相关系数大于0.3的特征指数,建立孝感汛期降水预测模型。结果表明:具有统计学意义的降水预测模型与实际降水的相关系数R达到0.825,明显大于单个因子的相关性。通过预测模型计算汛期降水,38年中能正确预测31年降水偏多偏少的趋势。在降水预报趋势错误的7年中,仅1年误差超过30%,其余都不超过20%。说明该模型对孝感汛期降水具有较好的预测能力。

This study used the 130 items of climate monitoring data issued by the National Climate Center and the precipitation data of the Xiaogan National Reference Climate Station from 1981 to 2019 in the flood season(May-September) to analyze the correlation between the climate monitoring data and the precipitation in the flood season in Xiaogan. The stepwise regression method was used to select the characteristic index with the correlation coefficient > 0.3 to establish the precipitation prediction model in Xiaogan flood season.The results showed that the correlation coefficient R between the statistically significant precipitation prediction model and the actual precipitation reaches 0.825, which was significantly larger than the correlation of a single factor. The precipitation in the flood season was calculated by the prediction model, and the trend of more and less precipitation in 31 years could be correctly predicted in 38 years. Among the seven years in which the precipitation forecast trend was wrong, only one year had an error of more than 30%, and the rest did not exceed 20%. The model had good prediction ability for Xiaogan flood season precipitation.