天气雷达图像中对流云区识别算法

Convective cloud identification algorithm based on weather radar image

对流云和层状云是形成暴雨的重要因素,准确地识别两者,对降水精度估测有积极的作用。为此,提出了一种小波分析区域识别算法(WLS)。该算法借鉴了小波分析的突变点检测原理,对天气雷达原始反射率数据和顶高数据进行小波变换,进而对检测出的模极大值点进行奇异性分析,滤除噪声点的干扰,最后用数学形态学方法检测边缘并填充对流云区域。实验中对采自呼和浩特雷达站的真实数据进行了算法识别分析,WLS方法较准确地识别出特征云体的相应区域,并将实验结果与采用BL和SHY95方法的识别结果进行了对比,表明WLS方法不仅对特征云体进行较好的识别,同时还有效地处理了杂波和边缘问题。

Both the convective and stratiform regions take an important role in the precipitation. Correct recognition of them facilitates precise prediction of the rainfall amount and duration. An automatic algorithm for the partitioning of radar reflectivity into convective and stratiform rain classifications named WLS was developed. Theory of abrupt-change detection based on wavelet analysis was adopted in this algorithm. First, wavelet transform was carried out on preprocessed raw reflectivity data and echo top data. Second, the singularity detection of modulus maximum value was done and noise points were filtered too. Finally, the edge was detected and convective region was filled by using mathematical morphology. Experiment uses the representative squall line on 2150 UTC 25 August 2008 and mixed precipitation 2222 UTC 8 August 2008 at Hohhot. Compared with WLS algorithm, BL algorithm and SHY95 algorithm, the experimental results show that WLS algorithm is more effective, and it can exactly partition the regions of convective clouds and restrain noise points.