Forecast track and intensity sensitivities of tropical cyclones to various parameterizations using the WRF-ARW model

The goal of this study was to determine potential biases in tropical cyclone tracks or intensities amongst various cumulus or microphysics parameterizations. To achieve this goal, three common cumulus parameterizations in the WRF-ARW (Kain-Fritsch, Betts-Miller-Janjic, and Grell-Devenyi) as well as three microphysics parameterizations (Kessler-warm rain, Eta-Ferrier, and WSM 6-class) representing a range of complexity were used in all possible combinations for a number of different cases. All other parameterizations were held constant. Four tropical cyclones were studied (Hurricane Ernesto (2006), Hurricane Ioke (2006), Typhoon Cimaron (2006), and Hurricane Lenny (1999)) using various initialization times for each case. These cyclones were chosen to represent a range of basins as well as a range of real-time forecast uncertainty for both track and intensity. After approximately two hundred simulations, most of which are 180 hour forecasts, pronounced track biases become evident. Certain parameterizations appear to have poleward biases while others display equatorward biases. To a somewhat lesser extent, intensity biases were also noted with certain parameterizations. This research will attempt to indicate both (1) which track and intensity biases are associated with certain parameterizations as well as (2) some possible reasons these are observed.