Anomalous propagation echo detection using artificial neural network and doppler velocity features

Anomalous propagation echo belongs to one of representative non-precipitation echoes which have significant influences in weather forecast process. It occurs due to super-refraction or ducting phenomena of radar beam path by atmospheric profile such as temperature or humidity distribution. In order to classify anomalous propagation echoes in radar data accurately, an automated procedure based on an artificial neural network classification scheme has been suggested in this paper. After applying spatial clustering in corrected reflectivity data and extracting coordinate information for Doppler velocity data, the artificial neural network classification method is applied. The experimental result with actual appearance case of the anomalous propagation echo describes that the suggested system with the artificial neural network shows good performance in the classification process.