This paper presents a modeling study conducted to evaluate the uncertainty of a regional model in simulating hurricane wind and pressure fields, and the feasibility of driving coastal storm surge simulation using an ensemble of region model outputs produced by 18 combinations of 3 convection schemes and 6 microphysics parameterizations, using Hurricane Katrina as a test case. Simulated wind and pressure fields were compared to observed H*Wind data for Hurricane Katrina, and simulated storm surge was compared to observed high-water marks on the northern coast of the Gulf of Mexico. The ensemble modeling analysis demonstrated that the regional model was able to reproduce the characteristics of Hurricane Katrina with reasonable accuracy and can be used to drive the coastal ocean model for simulating coastal storm surge. Results indicated that the regional model is sensitive to both convection and microphysics parameterizations that simulate moist processes closely linked to the tropical cyclone dynamics that influence hurricane development and intensification. The Zhang and McFarlane (ZM) convection scheme and the Lim and Hong (WDM6) microphysics parameterization are the most skillful in simulating Hurricane Katrina maximum wind speed and central pressure, among the three convection and the six microphysics parameterizations. Error statistics of simulated maximum water levels were calculated for a baseline simulation with H*Wind forcing and the 18 ensemble simulations driven by the regional model outputs. The storm surge model produced the overall best results in simulating the maximum water levels using wind and pressure fields generated with the ZM convection scheme and the WDM6 microphysics parameterization.