2022年5月10—14日广东短时强降水可预报性分析

Analysis of the Predictability of the May 10-14,2022 Short⁃range Intense Rain in Guangdong Province

利用自动气象站观测数据、EC ERA-Interim 0.25×0.25再分析数据、探空观测数据等多源数据,对2022年5月10—14日在广东发生的强降水过程进行天气尺度分析;在此基础上通过提取500 hPa相对涡度、925 hPa散度和200 hPa散度3个物理量,考察其与短时强降水的相关关系,从而引入物理量的关系组合和机器学习中的KNN分类算法,探讨短时强降水落区和强度预测的可行性。结果表明:≥-5×10^(-5)s^(-1)的500 hPa涡度、≤5×10^(-5)s^(-1)的925 hPa散度和(-10~20)×10^(-5)s^(-1)的200 hPa散度在一定程度上都可以指示短时强降水的发生区域;条件组合(925 hPa散度≤0 s^(-1)且200 hPa散度≥0 s^(-1)且500 hPa涡度≥0 s^(-1))对于短时强降水落区的指示性好;基于距离权重的KNN算法对该次过程短时强降水量级预测具有良好的效果。

ed to investigate their correlation with the short⁃range intense rain.Then,combinations of related physical quantities and the KNN classification algo⁃rithm(a kind of machine learning)were introduced to discuss the feasibility of predicting the area and in⁃tensity of the short⁃range intense rain.The results are shown as follows.With 500 hPa relative vorticity≥-5×10^(-5) s^(-1),925 hPa divergence≤5×10^(-5) s^(-1) and 200 hPa divergence in the range of(-10~20)×10^(-5) s^(-1),the area where the short⁃range intense rain took place can be indicated to some degree,and when the combined condition is met for which 925 hPa divergence≤0 s^(-1),200 hPa divergence≥0 s^(-1) and 500 hPa relative vorticity≥0 s^(-1),the area receiving the short⁃range intense rain can be well indicated.The KNN algorithm,which is based on distance weight,did a good job in predicting the rainfall amount of the process of interest.