基于速度对结构检测的中气旋自动识别方法

Mesocyclone Automatic Recognition Method Based on Detection of Velocity Couplets

针对传统中气旋自动识别算法误报率较高的问题,提出了一种基于速度对结构检测的中气旋自动识别方法.设计树形数据结构检测不同尺度的嵌套式速度极值区域,通过空间匹配得到符合兰金条件且不受环境风影响的二维速度对结构和三维涡旋结构,通过跟踪算法得到三维涡旋的持续时间属性,借助训练数据集获得中气旋的诊断参数.实验结果表明:与美国灾害天气中心的中气旋自动检测算法相比,所提出的算法将误报率降低了0.24,临界成功指数提高了0.16,能够更好地预报对流性灾害大风.

To deal with the high false alarm rates in conventional mesocyclone automatic identification algorithms,a mesocyclone automatic recognition method based on detection of velocity couplets is proposed.A tree data structure is designed to detect nested velocity extreme regions at various scales. Two-dimensional（2D）velocity couplets and threedimensional（3D）vortices that meet Rankine conditions and are not affected by environmental wind are obtained via spatial matching.The durations of 3D vortices are obtained through tracking,and the diagnostic parameters for mesocyclones are obtained by training on a dataset.As compared with the National Severe Storms Laboratory mesocyclone detection algorithm,the experimental results show that the proposed method decreases the average false alarm rate by 0.24,increases the average critical success index by 0.16,and therefore can better predict the convective wind damage.