Air-sea interaction processes during hurricane Sandy: Coupled WRF-FVCOM model simulations

Abstract

A fully-coupled atmospheric-ocean model was developed by coupling WRF (Weather Research and Forecasting Model) with FVCOM (the unstructured-grid, Finite-Volume Community Ocean Model) through the Earth System Model Framework (ESMF). The coupled WRF-FVCOM is configured with either hydrostatic or non-hydrostatic oceanic dynamics and can run with wave-current interactions. We applied this model to simulate the 2012 Hurricane Sandy in the western Atlantic Ocean. The experiments examined the impact of air-sea interactions on Sandy’s intensity/path and oceanic responses under hydrostatic and non-hydrostatic conditions. The results showed that the increased storm wind rapidly deepened the mixed layer depth when ocean processes were included. Intense vertical mixing brought cold water in the deep ocean towards the surface, producing a cold wake within the maximum wind zone underneath the storm. This process led to a sizeable latent heat loss from the ocean within the storm, and hence rapid air temperature and vapor mixing ratio drop above the sea surface. The storm intensified as the central sea-level pressure dropped. Improving air pressure simulation with ocean processes tended to reduce the storm size and strengthen its intensity, providing a better simulation of hurricane path and landfall. Turning on the non-hydrostatic process slightly improved the hurricane central sea-level pressure simulation and intensified the winds on the right side of the hurricane center. Hydrostatic and non-hydrostatic coupled WRF-FVCOMs captured Sandy-induced rapidly-varying flow over the shelf and the wind-induced surge level at the coast. The coupled models predicted a higher water elevation around the coastal areas where Sandy made landfall than the uncoupled model. The uncoupled and coupled models both showed more significant oceanic responses on the right side of the hurricane center, with a maximum during the Sandy crossing period when the clockwise-rotating frequency of Sandy wind was close to the local inertial frequency. The area with a maximum response varied with Sandy’s translation speed, more prominent in the deep region than over the slope, and more substantial under the non-hydrostatic condition. The simulated ocean responses agreed with the theoretical work of Price (1981). The nonhydrostatic experiments suggest that to resolve a fully storm-induced convection process, the oceanic model grid configuration should meet the O(1) criterion for the ratio of local water depth to the model horizontal resolution.