Using an Object-Based Approach to Quantify the Influence of Cumulus Parameterization in the Spatial Structure of Precipitation in Hurricane Isabel (2003)

We present a modeling study of Hurricane Isabel (2003) to showcase the utility of a spatial metric approach in uncovering TC structural differences that can be traced to the modeling framework, in this case the handling of mass, momentum and temperature by the cumulus parameterization (CP). This study employs the Advanced Research Weather Research and Forecasting (WRF-ARW) model version 3.6.1. Three simulations are performed using Kain-Fritsch (KFS) and modified Tiedtke (TS) schemes and an altered Tiedtke scheme with the momentum tendencies set to zero (zTS). First, by comparing the azimuthal average winds, temperature and moisture, we explore the CP-related physical processes and their influence on simulated TC structure. During the first 24 hours of the simulation, when only the coarse domain is employed, the TS-simulated TC uniquely develops a smaller inner core and an outer region of convection associated with a secondary wind maximum. Thus, the Tiedtke momentum tendencies are key to the development of a second radial wind maximum, which leads to an inner and outer rainband configuration as opposed to a single broader eyewall. These different rainband configurations persist after the inner nests are initialized, as evidenced by 24-hour averages of winds, temperature, and moisture taken 48-72 hours into the model simulation (around the time of landfall). Spatial metrics are then calculated using polygons with reflectivity > 20 dBZ to characterize the spatial distribution of TC rainbands. Building on previous research, we utilize five spatial metrics: circularity, solidity, closure, dispersion, and fragmentation. We demonstrate that these metrics effectively capture significant differences among the simulations. In particular, the TS-simulated TC displays lower solidity and higher dispersion due to a smaller inner core and a discrete primary rainband. In the zTS simulation, the broader TC inner core is more similar to the KFS simulation, as evidenced by significantly (p < 0.01) correlated area, solidity, and dispersion metrics. Based on these results, future research should explore object-based approaches for characterizing TC structure to investigate the performance of physical parameterizations across a broader spectrum of TCs in a variety of storm environments.