Comparison of the sensitivity of hurricane simulations to convective parameterization and small-scale model noise

Understanding the relative importance of cumulus parameterization schemes for hurricane simulation is essential because convection drives the development and intensification process. Unlike previous studies which have focused on the sensitivity of hurricane track and intensity to cumulus parameterization, we examine whether sensitivity between convection schemes is more influential than small scale noise perturbations put back into the model with the stochastic kinetic-energy backscatter scheme (SKEBS).

Idealized 3D hurricanes were simulated with the Advanced Hurricane WRF model. Four cumulus schemes (Kain-Fritsch, Betts-Miller-Janjic, Grell 3D and Tiedtke) were tested. The model was configured to run the full physics real data setup, but for idealized initial conditions. The moist tropical sounding of Dunion provided an initial field. Two simulations were run for each scheme, one which used SKEBS and one which did not. The sensitivity of track and intensity was examined for three different initial conditions or model setups (i.e. the initial wind field strength, vortex size and model domain size).

Results show the use of SKEBS to introduce small scale noise back into the model creates more variation in simulated track than when changing the cumulus parameterization scheme. This has implications for modeling strategies for high-resolution dynamical simulations of hurricanes.