西北太平洋迅速加强热带气旋的统计特征和识别预报试验

利用2000—2014年热带气旋(TC)最佳路径、最终分析资料和静止卫星红外云顶亮温(TBB)资料,对比分析了西北太平洋(WNP),以及南海(SCS)的迅速加强(RI),与非迅速加强(non-RI)TC样本的环境背景和TBB统计特征,其中non-RI样本细分为不同的强度变化率即:缓慢加强(SI),强度稳定、缓慢减弱和迅速减弱等。结果表明,相对于SI,WNP海域的RI样本处于海表温度较高、海洋上层热容量较大、最大可能强度较大、高层辐散较强、风垂直切变(VWS)较弱和高层纬向风(U200)偏东分量较大等环境背景条件下;SCS海域的RI样本较易发生在VWS较弱的环境背景条件下。此外,相对于non-RI,支持RI发展的有利条件还包括中低层相对湿度较大、高层环境温度较低等。RI样本通常具备的TBB特征为TC内核的对流云覆盖率较大、TBB平均值相对较小。采用K最近邻分类算法进行RI预报试验,交叉检验结果表明,该方法对RI样本有一定的识别预报能力,RI样本概括率达到74.2%,技巧评分达到0.717。

Characteristics and Identification of Rapidly Intensifying Tropical Cyclones in Western North Pacific Basin

The best tracks and satellites borne infrared cloud top temperatures (TBB) observations of the National Centers for Environmental Prediction Final Analysis are employed to examine the large scale and inner core convection and characteristics of tropical cyclones (TCs) undergoing different intensity changes in the western North Pacific (WNP) and South China Sea (SCS) ocean basins from 2000 to 2014. The mean conditions of TCs cases that undergo rapid intensification (RI) are compared to those of the non RI cases. In addition, the non RI cases are defined in four other intensity change bins: slow intensification (SI), intensification stable (IS), slow weakness (SW) and rapid weakness (RW). For the environmental variables, statistically significant differences are found between RI cases and non RI samples, especially in IS, SW and RW groups. In both basins, RI events tend to form in environments with weaker environmental vertical wind shear than SI cases. In the WNP, RI events occurred in high maximum potential intensity environments characterized by warmer sea surface temperature, greater upper oceanic heat content, and stronger upper level divergence. RI events tend to occur in favorable environments with higher lower level relative humidity, and cooler upper level temperatures than non RI cases. For TBB observations, RI cases have larger cold TBB covering areas in the inner core regions, and lower average symmetry TBB within the eyewalls of the storms. The K Nearest Neighbor (KNN) rule is employed to identify the RI cases using leave one out cross validation. The results show that the verification of forecasts is generally skillful.