Abstract
Weather detection algorithms often rely on a simple rule base that is based on several features. Fuzzy logic can be used in the rule base, and the membership functions of the fuzzy sets can be tuned using a search or optimization algorithm that is based on the principles of natural selection.
The bounded weak echo region (BWER) detection algorithm was developed using a genetic algorithm to tune fuzzy sets. The run-time algorithm uses the tuning information produced by the genetic algorithm to differentiate between BWERs and non-BWERs and to assign confidence estimates to its detections. The genetic algorithm that was used to tune the fuzzy rule base of the BWER algorithm is described.
The paradigm of using a genetic algorithm to tune a fuzzy rule is a very general and useful one. It can be used to improve the performance of other weather detection algorithms. The paradigm makes it easy to change the behavior of a run-time algorithm according to locale and/or end users. The paradigm when applied to the BWER algorithm made it possible to tune the algorithm for use by forecasters as well as by a neural network.
Corresponding author address: Valliappa Lakshmanan, Stormscale Research and Applications Division, National Severe Storms Laboratory, 1313 Halley Circle, Norman, OK 73069.
