Bagged Fuzzy k-nearest Neighbors for Identifying Anomalous Propagation in Radar Images

Hansoo Lee, Jonggeun Kim, Suryo Wibowo, Sungshin Kim
Several advanced observation devices, such as ra-diosondes, satellites, and radars, are utilized in practical weather prediction. The weather radar is an essential device because of its broad coverage with excellent resolution. However, the radar inevitably observes meteorologically irrelevant signals. An anomalous propagation echo is a nonprecipitating echo generated by significantly refracted radar beam towards ground or sea surface. In the case, the radar misrecognizes the surface as a
more » ... surface as a meteorological phenomenon. The false observation results may decrease the accuracy of weather prediction result. Therefore, we propose a novel classification method for identifying anomalous propagation echoes in the radar data by combining fuzzy k-nearest neighbors and Hamamoto's bootstrapping algorithm. By using actual occurrence cases of anomalous propagation, we confirm that the proposed method provides good classification results.