Using Shape Metrics to Compare Observed and Simulated Reflectivity During the Landfall of Hurricane Isabel (2003)

Modeling of tropical cyclones (TCs) permits the investigation of processes leading to high rain rates and flooding. High resolution data produced by ground-based radars provide important observations of storm structure over land that can be used to verify modeled conditions. This study utilizes shape analysis to compare radar reflectivity values observed by the Weather Surveillance Radar – 1988 Doppler (WSR-88D) network and simulated reflectivity in the Advanced Research Weather Research and Forecasting (WRF-ARW) model for Hurricane Isabel (2003). Landfall occurred at 17 UTC on September 18 and forecast models predicted the system well due to its large size. Given the relative success of forecast models in predicting Isabel's track and intensity, this study determines whether a model simulation can accurately represent the spatial arrangement of its rainbands. We measure the size, position, and spatial attributes of Isabel's rainfall regions in simulations of reflectivity from the WRF model and compare with patterns observed using Level II radar data from the WSR-88D network that we mosaic to a 3D grid. To simulate radar reflectivity, we utilize the WRF-ARW model version 3.6.1. The model domain is triply nested through two-way nesting with a course domain of 27 km horizontal resolution and two inner nests of 9 and 3 km resolution, respectively. The coarse domain is initialized at 00 UTC September 16, and the inner nests are initialized 24 hours later. All simulations are integrated through 00 UTC September 20 to fully encompass the landfall period. As the calculation of simulated reflectivity depends on the model's microphysics scheme, we execute an ensemble of simulations with varying cumulus and microphysical parameterizations that are commonly employed in TC numerical modeling studies. These include two mass-flux convective schemes (Tiedtke and Kain Fritsch) and three microphysics schemes of varying complexities (WRF Single Moment-6 class, WRF Double Moment 6-class, and Thompson et. al.) For the analysis of observed data, we utilize a Map-reduce-based playback framework to interpolate large volumes of Level II reflectivity data onto 3D grids at 1 km resolution. Data are utilized from all radars within 600 km of the storm center. For both simulated and observed reflectivity, we then utilize a Geographic Information System (GIS) to interpolate values at grid points, contour these points to construct polygons, and calculate shape metrics to quantify the spatial distribution of reflectivity values at constant altitudes. For this study, we present results from the 3.5 km altitude slice every 30 minutes. The shape metrics calculated on the largest polygons include distance and bearing from storm center, compactness, elongation, orientation, solidity, convexity, and closure. At each time, we calculate the dispersion of polygons from the storm center. We focus our results on the simulation with Tiedtke convective scheme and WRF Single Moment-6 class microphysics scheme. The simulated TC makes landfall at approximately the same time as indicated in the best track dataset, but its position is ~90 km northeast of the official landfall point. As reported in previous studies, reflectivity values are approximately 4 dBZ higher in the simulation when compared to observations. A main difference is that the observed stratiform connecting rainband northwest of the eye is not apparent in the simulations; this region is instead occupied by small disconnected pockets of convective cells. The larger number of smaller polygons produced by the model leads to higher values of dispersion. Our closure metric indicates that the eyewall eroded more quickly in WRF, but the outer rainband located west of center retained a similar shape and size to the observed values.