Automatic Cell Detection and Tracking

Abstract

Algorithms for the automatic detection and tracking of precipitation cells have been developed for the processing of digital weather radar data. The basic premise of the processing procedure is to represent the essential information in the weather radar data with a significantly smaller data set that describes the characteristics of the radar echo cells and their surrounding echo regions. Radar echo cells are defined by contours 3 dB below enclosed reflectivity peaks which do not enclose other cells. Attributes such as reflectivity, area, and location are tallied for each detected cell. Similar attributes are also tallied for the larger echo regions enclosed by preselected reflectivity value contours that encompass the cells. The attributes are used to extablish and maintain cell tracks within a tilt scan sequence and from one scan sequence to the next. Detected cells persist and produce tracks which last an average of 15 min. The detected cells are relatively small, typically 3 km in diameter, and often occur in clusters. The automatic detection and tracking programs provide a new tool for the analysis of large volumes of weather radar data.