多光谱卫星云图降水区域与强度估计的模糊推理模型

Fuzzy inference model of rainfall estimation based on multi-spectrum satellite cloud pictures

在卫星云图降水分析中,针对降水天气和降水强度具有明显的“模糊”特性,提出了云图降水的模糊推理模型。从不同云类降水的经验知识和降水天气的卫星云图样本中,提取出若干云类降水规则。在此基础上,引入模糊推理方法,建立了卫星云图降水分布和降水强度估计的模糊推理模型。实际应用中,对卫星云图进行云分类,根据云分类结果,去除不可能产生降水的云类区域。对可能产生降水的云类区域,通过提取多光谱云图的灰度等特征进行推理判别,进而得出该区域是否降水以及降水强度和降水分布的估算。试验结果表明,该方法在客观性、定量性和自动化程度方面优于传统的降水估计方法,且建模所需的样本少,可适用于多种类型的气象卫星云图。

Given the fuzzy characteristic of rainfall weather and intensity, a Fuzzy inference model of rainfall distribution and intensity of satellite cloud images was proposed. Based on experiential knowledge of different cloud sorts bringing rainfall, combining multi-spectrum cloud images sample corresponding rainfall weather, some rainfall rules corresponding to cloud sorts, with the fuzzy inference method, the fuzzy inference model of rainfall distribution and intensity of the satellite cloud image was built. In practice, wiping off the cloud sort area of impossible rainfall can be ruled out based on cloud sort. By analyzing the gray character of the multi-spectrum cloud images, whether the area would rain or not can be deduced and the rainfall distributing and intensity can be estimated. Experiments indicate that this method excells the traditional rainfall estimation in objectivity, quantification and automation, and that it needs fewer samples to build models. This model is applicable to the multi-type satellite cloud image.