Automated classification of rainfall systems using statistical characterization.

Abstract

A general, completely automated procedure for classifying rainfall systems is developed. The technique is flexible and universally applicable, in that any rainfall system can be classified regardless of size, location, time of day or year, degree of organization, etc. The knowledge obtained from previous research was used to develop a relatively straightforward and unique classification system. To test the performance of the method, results were validated against a subjective classification based upon objective criteria. From an independent random sample, the automated classification system accurately placed events into stratiform, linear, and cellular classes 85% of the time.