冰雹指数的误差分析及其在冰雹云识别中的应用

Error analysis of hail index and its application to hail cloud identification

利用模拟测试方法分析几个重要因子对冰雹指数误差的影响,并根据遵义新一代多普勒天气雷达资料,以探测概率、临界成功指数和误报率为衡量指标,对2008年3月至2012年7月12次冰雹天气过程中25个降雹风暴和72个非降雹风暴的冰雹指数进行分析评估。结果表明:H0、反射率因子垂直廓线和目标距雷达的距离等因子对POSH误差的影响较大,Hm20对其影响较小。POH识别冰雹云的误报率较高,临界成功指数较小;POSH识别效果好于POH,通过增大阈值可以提高临界成功指数和减小误报率;但当阈值太大时,探测概率会较低,为了获得较大的探测概率和临界成功指数,减小误报率,以POSH=70为阈值时识别效果最好;在POSH识别冰雹云时加上VIL>20 kg·m-2的条件对探测概率没有影响,但是当阈值小于90时,可减少误报率和提高临界成功指数。

The effects of several important factors on the error of hail index were analyzed by a simulation test method. Using the measurement indexes of probability of detection （ POD ）, critical success index （ lcs ） and false a- larm ratio （RFA） ,the 25 hail storms and 72 non-hail storms among 12 hail processes during March of 2008 to July of 2012 were assessed based on the new generation Doppler weather radar data in Zunyi City. The results indicate that H0, the vertical profile of reflectivity factor and distance between the target and radar have great influence on POSH error, while Hr^0 has insignificant influence. There are high RFA and low lcs when POH are used to recog- nize hail cloud. Identification effect of POSH is better than that of POH. Increasing threshold can enhance Ics and decrease RFA. The PoD is low when the threshold is larger. POSH = 70 is the best threshold, with which high PoD and Ics and low RFA Can be obtained. PoD has no change when POSH and VIL 〉20 kg.m^-2 are used together to i- dentify hail cloud,but lower RFA and higher Ics can be obtained when the threshold of POSH is less than 90.