【作者】 刘丹；

【导师】 韩雷；

【作者基本信息】 中国海洋大学 ， 信号与信息处理， 2014， 硕士

【摘要】 强对流天气是我国的主要灾害性天气之一，多普勒天气雷达是强对流性天气的主要监测工具之一。本文提出了一种以高分辨率的三维多普勒雷达资料为基础的对流初生（Convective Initiation，简称CI）监测和预警的算法，这里CI定义为天气雷达首次检测到反射率因子大于或者等于35dBZ。本文研究内容主要包括分为以下三个部分：第一部分是雷达原始观测资料的预处理，其中重点研究了雷达极坐标体扫资料插值成笛卡尔坐标系下网格数据的处理过程。本文采用NN插值和NVI插值算法将空间分辨率不均匀的原始雷达数据（存储方式为仰角、方位角以及斜距）插值成空间分辨率均匀的网格数据（存储方式为经度、纬度以及高度）。结合雷达资料的实际存储方式，利用排序、折半查找等算法来对插值算法实现的细节进行优化，提高插值算法的处理速度。第二部分是三维对流风暴的自适应阈值识别算法，即给定一组阈值然后按阈值从小到大顺序依次判断该阈值是否为对流体的最终阈值，这样做的原因主要是为了同时识别出独立、反射率因子较小的对流单体，和在体积较大的对流体中存在的反射率因子阈值较大的对流体。本文的对流风暴自适应阈值识别算法是在单阈值三维风暴识别算法基础上使用递归思想完成的。第三部分是对流初生预警算法的设计与实现。主要利用描述对流风暴三维系统的指标对流强度比（Convective Intensity Ratio，简称CIR）以及高层回波分布信息等对CI进行约0-20分钟的预警。对流强度比的定义是对流系统中高层强回波面积与自身面积的比值。该预警算法通过追踪找到当前时刻对流体对应的上一时刻的对流体，然后通过计算当前对流体高层回波分布信息将对流体状态进行分类，利用相邻时刻匹配对流体的CIR值变化率及对流体其它特征对CI进行预警。更多还原

【Abstract】 Convection weather is one of the severe weather in china and often poses seriousthreats on the product and life of human beings. Doppler weather radar is a majorshort-term severe weather monitoring tool. This paper presents a convection initiationalgorithm based on Doppler weather radar data, CI defined in this study is the first todetect the radar reflectivity of clouds produce greater than or equal to35dBZ. Themain contents of this paper are divided into the following sections:The first part is the raw radar observation data converts into Cartesian coordinatesystem grid data. In this paper, the NVI interpolation algorithm converts raw radardata into three-dimensional Cartesian grid. Combined with the actual radar datastorage, we use sort, binary search, and other basic algorithms to optimize theimplementation details of the interpolation algorithm.The second part is a convective storm adaptive threshold recognition algorithm.Adaptive threshold recognition convective storm algorithm is based on a given set ofthresholds to automatically select the appropriate threshold and threshold valuescorresponding storm body. The main reason for using this method is to simultaneouslyidentify the independent convective cells with smaller reflectivity threshold andconvective cells with larger reflectivity threshold in large convective cells. Convectivestorm adaptive threshold recognition algorithm presents in this paper use the recursivethought based on a single threshold storm identification algorithm.The third part is the design and implementation of convective storm initiationalgorithm. Using the three-dimensional convective storm indicators CIR anddistribution of high-level echo information forecast the CI in0-20minutes, CIR isdefined in this study is the convection intensity ratio. The first step is the need offinding the current time convective cells corresponding to the cells of last volumescan. By calculating current information on the distribution of high-level echoes, weclassify the convective cells into five states. Then we forecast convective initiation by calculatingof the change rate of the convective cells’CIR between two volume scan.更多还原