Research Of Hailstone Detection Based On Radar Echo Reflectivity Image

The work of this paper belongs to the project,"The research of automatic hailstone detection system based on Radar", which is cooperated by TianJin weather bureau and our lab.Hailstone is a main disaster in the Meteorological area. The hailstorm arises suddenly, moves fast, and withers away very quickly. Those characteristics make the hailstone forecasting very difficult. So the research on hailstone automatic detection system using Doppler radar becomes very important in the area of strong storm tracking. Actually now the research of hailstone forecasting based on radar image is during its developing stage. Therefore, this project enjoys a highly research value.This dissertation recognizes the hailstone from statistical characteristic (the echo reflectivity) and shape characteristic (hook echo). For the statistical characteristic, first according to the size of the echo reflectivity and the color theory of the color image, filter the radar reflectivity image. Secondly using morphological operator to erode the result image that has been got by the first step. Finally detect the edge in the eroded image to find the hail cloud. When choosing the morphological operator, I choose the elliptic operator but not the rectangular one on the base of the experience of forecasters that hail cloud are always elliptic.According to the experience of hailstone forecast, when the radial grads of the cloud is very big, there is always a hailstone descending. The difference between the hailstone and the rainstorm is that the hail cloud gathers more compact than that of rainstorm cloud. Therefore in this dissertation we give an innovative concept of hail cloud density to differentiate the hail cloud and the rainstorm cloud in the image. The result indicates this algorithm can recognize the hail cloud in some condition.At last this paper uses Hough Transform to detect line on the image of hail cloud edge to recognize the hook echo. The examination result shows that the algorithm can detect some clear hook echo but not the distinct one. In conclusion, based on the recognized hail cloud, this dissertation has will contribute to the construction of automatic hailstone recognizing system and expert knowledge database.