Monitoring And Forecasting Strong Convective Cloud Based On Satellite Image

Strong convective weather events are potentially damaging and dangerous to lives and property. There has been a significant interest in monitoring and forecasting for convective acitivities. Satellite as high-tech large scale surveillance devices, has higher temporal and spatial resolution as compared with conventional meteorological observation instruments. The paper studied strong convective clouds based on satellite imagery in two aspects: monitoring and forecasting.When atmospheric strong convective activities happens, radiant charactoristics apperances in water-vapor and long-wave inferred band. The paper researchs monitoring strong convective clouds by using light-temperature difference data of the above two channels, and combines radar iamges and satellite light-temperature difference images at the same time. Then compares charactoristics of strong convective clouds in two forms of remote sence images, and uses statistic methods to analyze the light temperature charactoristics of the satellite channel difference image. From this, we can obtain distribution of light temperature in the satellite channel difference images during the clouds developing periods, which can be used to monitor the convective activities.This paper leads target tracking theory and methods in the field of computer vision into the research on cloud images of meteorological satellite, by the analysis of long-wave inferred channels of satellite data to realize the recognizing, tracking and forecasting to cumulus. In the system flow, the most important aspect is the tracking technology of rain clouds. The classical algorithems on cumulus analysis by using satellite images often have some limitations and one-sidedness. This paper references Sophisticated algorithms based on radar data, puts it onto the satellite data to try to accomplish cumulus tracking and forecasting. In the actual process, two-dimensional clustering method is used to identify cumulus. Then area overlap and combinatorial optimization method are used to obtain the cumulus tracking. At last, a linear extra- polation method is used to give 30,60,90 and 120min forecasting results. As the authentication of the forecast result by the relevant data, the experimental system has fine reliability and accuracy.