Convective cell identification using multi-source data

Identification of convective cells is an important issue for detecting severe meteorological phenomena and precipitation nowcasting. The proposed model that classifies each individual radar pixel as convective or stratiform was developed based on multi-source data and applying a fuzzy logic approach. For both classes (stratiform or convective), membership functions for all investigated parameters were defined and aggregated as weighted sums. Comparison of the weighted sums decides which category a considered radar pixel belongs to. Each membership function was determined for selected parameters from: weather radar network, satellite Meteosat 8, lightning detection system, and numerical weather prediction (NWP) model. Then convective pixels were clustered to obtain individual cells, assuming that cells with a small distance between their maxima are joined.