Impact of Parameterized Physical Processes on Simulated Tropical Cyclone Characteristics in Community Atmosphere Model

This work studies the impact of parameterized physical processes on simulated tropical cyclones characteristics that are local to the storm in Community Atmosphere Model (CAM) using the Reed-Jablonowski TC test case. It targets on 24 parameters across the physical parameterization schemes that represent the convection, turbulence, precipitation and cloud processes in AGCMs. It is shown that TC intensity is mainly sensitive to the parcel fractional mass entrainment rate (dmpdz) in deep convection. A decrease in this parameter can lead to a change in simulated intensity from a tropical depression to a Category-4 storm. Precipitation and SWCF (shortwave cloud radiative forcing) are strongly affected by three parameters in deep convection: tau (time scale for consumption rate of convective available potential energy), dmpdz and C0_ocn (precipitation coefficient). Changes in physical parameters generally do not affect LWCF (Longwave cloud radiative forcing) much. In contrast, LWP (liquid water path) and IWP (ice water path) are very sensitive to changes in C0_ocn. The changes can be as large as 10 (5) times the control mean value for LWP (IWP).The nonlinear interactive effect among different physical parameters is negligible on simulated TC intensity and LWP. In contrast, this nonlinear interactive effect plays a significant role in other simulated tropical cyclone characteristics (precipitation, LWCF, SWCF, and IWP) and greatly enlarge their simulated uncertainties. The statistical emulator Extended Multivariate Adaptive Regression Splines (EMARS) is applied to characterize the response functions for nonlinear effect.