An Automatic Tracking and Recognition Algorithm for Thunderstorm Cloud-Cluster (TRACER)

The technique of thunderstorm cloud clusters identificationAtracking and early-warning based on Doppler weather radar data is one of the most important techniques of nowcasting. The thunderstorm cloud clusters'borderline-correlation tracking technique(BCTT) discussed in this paper is one of a thunderstorm cloud clusters'extrapolation method which employs the pattern recognition to recognize and topologize the borderlines of the cloud clusters, establish the lifecycles of clouds and their ancestry, so as to extrapolate the movement of thunderstorm cloud clusters. BCTT identifies thunderstorm cloud clusters by their borderlines, then acquires the ancestry of each thunderstorm cloud cluster by six distinguishing factors: quad tree factorAsuperposition factorAsize factorAoutside rectangle factorAprofile similarity factorApart similarity factorBAnd the moving directionAvelocityAsizeAstrong echo center and other information (strengthening or weakeningAswelling or shrinking ) could also be obtained. Then a linearly extrapolation of each thunderstorm cloud cluster can be acquired. This technique can recognize thunderstorms very well, and manage the thunderstorm cloud clusters as objectives. The system of recognizing and tracking thunderstorms automatically("TRACER") developed by the method mentioned above can appoint the cloud cluster based on GIS and capture locality, contrail , evolvement and prediction of the chosen cloud.