Spatial Metrics that Facilitate the Comparison of Radar Reflectivity Values within Landfalling Tropical Cyclones

Characterizing the spatial patterns of rainfall regions within tropical cyclones as they move over land can help identify where high rain rates are likely to occur and can be utilized to compare observed and modeled storms. In this study, we calculate metrics to compare the spatial patterns of radar reflectivity for varying reflectivity levels observed by the Weather Surveillance Radar 1988 Doppler (WSR-88D) network during tropical cyclone landfall over the U.S. These metrics should reveal how quickly and in what ways the rainband configurations change due to interaction with middle latitude weather systems and the relatively dry continental air mass. A mosaic of Level II reflectivity values observed by the WSR-88D network is created and the spatial analysis is performed at a constant altitude of 3.5 km. Every 30 minutes, the mosaicked data are contoured and converted into polygons that represent reflectivity regions equal to or greater than 20, 25, 30, 35, and 40 dBZ. The spatial metrics are calculated at each time step for each reflectivity level. The circularity and solidity are measured for the largest polygon in each image. As dry air encircles the storm, we expect the main reflectivity region to become less circular and less solid. Higher reflectivity values correspond to convective cells that cover smaller regions. Therefore, there are likely to be multiple regions of similar size so that emphasizing changes to the largest region does not capture what is happening to the majority of the regions with convective rainfall. Therefore, three metrics are calculated that encompass all polygons whose centroids fall within 600 km of the circulation center. Dispersion measures the radial distribution of polygons, with higher values indicating a more dispersed pattern that should occur as convection erodes near the circulation center due to dry air entrainment and strong vertical wind shear. Closure measures the azimuthal distribution of polygons, which should cover less of the 360° around the circulation center over time due to dry air advection and strong vertical wind shear. Fragmentation combines the solidity and connectivity of all polygons and should increase due to dry air advection and erosion of the storm’s core. Results during the landfall of Hurricane Isabel (2003) show strong similarities for 20 and 25 dBZ and 35 and 40 dBZ regions, but that the spatial patterns of areas of lower and higher reflectivity are not correlated with one another. Closure has the largest difference among the reflectivity levels when compared with dispersion and fragmentation. Closure decreases over time at a rate of 16.5 degrees per hour for 20 dBZ regions and 7 degrees per hour for 40 dBZ regions. For dispersion, values are similar among reflectivity levels with a slow increase over time. Fragmentation increases for 20 and 25 dBZ regions, but remains high for 30, 35, and 40 dBZ regions throughout the study period. All metrics save for 35 and 40 dBZ fragmentation exhibit a strong correlation with time. Thus, rainfall regions become more dispersed away from the storm center, the center becomes more exposed, and regions of lighter rainfall become more fragmented as Isabel experiences extratropical transition while moving inland. As expected, solidity and circularity were most useful for 20 and 25 dBZ regions, demonstrating that as dry air encircled the storm, the main rainfall region became less solid and circular as it eroded. Future work will examine additional tropical cyclones to determine if changes in reflectivity regions occur at a similar rate.