基于三维雷达图像数据的风暴体追踪算法研究

Storm tracking algorithm development based on the three-dimensional radar image data

TITAN算法是目前临近天气预报的主要方法之一,主要用于对强对流天气的识别、跟踪和预报。在原算法中速度约束是固定的、预报方法只适用于结构较稳定的风暴,它们会对算法的总体性能有较大的影响。针对这些不足,通过使用动态速度约束和光流法对算法进行了改进。通过实验证明,这些改进对风暴结构变化较大的情况有较好的效果,可以使算法的总体性能平均提高6%左右。

TITAN algorithm, one of the main ways for short-term weather forecast, uses controid method to identify, track and forecast strong convective storms. But in the original algorithm, the speed constraint is fixed and the prediction method is only applicable to the storms whose structure is stable. They exert negative influence on the algorithmic pefformance. To surmount these disadvantages, dynamic speed constraint and optical flow method were introduced. The experimental results show that these improvements can result in better forecasting and higher success rate of tracking the more complicated strong convective storm.