川东北极端强降水预报方法对比研究

Comparative Study on Forecasting Methods of Extreme Heavy Precipitation in the Northeast of Sichuan Province

【目的】为了建立川东北最佳极端强降水概率预报方程。【方法】利用ERA时间间隔为1 h的0.25°×0.25°再分析资料,计算1990—2019年5—9月逐日4个时次的物理量,统计分析极端强降水个例物理量异常度,采用4种实验方法对比分析并建立预报方程。【结果】(1)异常度较明显因子分别为700 hPa比湿、850 hPa比湿、700 hPa水汽通量散度、850 hPa水汽通量散度、500 hPa垂直速度、700 hPa垂直速度、850 hPa垂直速度,大部分平均异常度绝对值在1.5以上,K指数和假相当位温相当,08时总体比其他时次异常度明显。(2)预报方程预报结果。样本内检验,该预报概率平均值为0.652,60%的百分位值为0.623,预报概率最大值为0.999。样本外检验,预报指数阈值为0.6时,平均准确率达到了90%,平均空报率为9.3%,平均漏报率为0.7%,其中有13站的漏报率为0%;以0.65为阈值检验,平均准确率达到了92.4%,平均空报率为6.9%,平均漏报率为0.7%,其中有13站的漏报率为0%,预报效果更佳。【结论】极端强降水概率预报方程具有较好的预报效果。

Based on the ERA 0.25°×0.25°reanalysis data with 1 h interval,the physical quantities of four times per day from May to September in 1990—2019 are calculated,and the anomaly degree of physical quantities of extreme heavy precipitation cases is statistically analyzed.Four testing methods are used for comparative analysis with the objective of establishing the optimal probability prediction equation of extreme heavy precipitation in northeast of Sichuan Province.The following conclusions are drawn from the test results:(1)The more obvious anomaly factors are specific humidity at 700 hPa and 850 hPa,water vapor flux divergence at 700 hPa and 850 hPa,and vertical velocity at 500 hPa,700 hPa and 850 hPa.The absolute values of most of the mean anomalies are above 1.5,the K index is equivalent to the pseudo-equivalent potential temperature,and the overall anomaly at 08∶00 is more obvious than that at other times.(2)The probability prediction equation of extreme heavy precipitation has a positive forecasting effect.In sample test,the mean forecast probability is 0.652,and the percentile value of 60%is 0.623,and the max value of forecast probability is 0.999.In out-of-sample test,when the prediction index threshold is 0.6,the average accuracy reaches 90%,the false prediction ratio is 9.3%,and the average missing report rate is 0.7%,among which 13 meteorological stations have the missing report rate of 0%.When threshold is 0.65,the average accuracy reaches 92.5%,the false prediction ratio is 6.9%,and the average missing report rate is 0.7%,including the 0%missing report rate of 13 meteorological stations,which shows an even better prediction effect.