Research Of Medium-small Scale Cyclone And Adverse Wind Area Recognition Algorithm

Mesocyclone is a kind of vortex weather condition with the scope of several km to dozens of km. According t o mass data statistics made out by w eather e xperts, in China, more t han 92% [1] of the st orms ar e a ccompanied b y t he o ccurrence an d development of mesocyclones, and in American is more than 90%[2]. Especially, a mesocyclone actually forms for a period of time before a severe storm hits. In modern society, t he r apid pa ce of pe ople's lif e and pr oduction a ctivities, t he continuous development of science a nd t echnology, make hi gher and hi gher demands of timeliness and accu racy o f t he weather f orecast. T hus i t has great im portance to identify the m esocyclone quickly and accurately, t hen t o s tudy t he d evelopment relation between the storm and shear.In t his pa per, w ith pr ogramming l anguage Visual C ++ a nd O penCV, ba sed on Doppler Radar Reflectivity a nd Velocity I mage, i ntegrated us ing of knowledge of Computer Vision, P attern R ecognition a nd Meteorology, t o d esign t he a utomatic recognition system of mesocyclone and draw the time-varying figure of the shear value. This paper has mainly completed the following two tasks:1. Analyze the basic characteristics of mesocyclone in Doppler radar reflectivity image a nd ve locity image at f irst, based on original algorithm, us ing image recognition technology, add several improvements in image pretreatment, configuring velocity pairs, simplify the short range velocity. Besides, researches also have been made i n 3D feature collecting, r elationship b etween s hear a nd s torm, small-scale cyclone recognition. Professional application system have been finished, too.2. Based o n t he image characteristics o f Adverse W ind R egion, using t he technology of binaryzation, region growing, image segmentation and edge detection, design and realize the automatic AWR detection system.The algorithm mentioned a bove ha s been realized by programming. Weather expert has tested it with lots of cases. Compared with PUP which is being widely used, it has higher recognition rate, lower false-alarm and false-reject rate.