Assessing Water Level Hindcasts from FVCOM in the New York Harbor Area

The increasing challenges in risk reduction and marine navigation associated with climate change require robust systems for predicting water levels. The Coastal Applications Team (CAT) for the Unified Forecast System (UFS), a community-based, coupled Earth modeling system, aims to evaluate the coastal ocean models available and provide recommendations for next generation coupled coastal ocean models. Our group at the University of Maryland is focused on evaluating the Finite Volume Community Ocean Model (FVCOM) through hindcasts, analyzing how hindcast accuracy is affected by aspects of the setup process around the New York Harbor area. FVCOM is a prognostic coastal ocean model that uses free surface 3-D primitive equations on an unstructured grid. While originally designed to simulate flooding/drying over estuaries and wetlands with circulations driven by wind and tides, FVCOM has expanded to a fully coupled current-ice-wave-sediment-ecosystem model with parallelization implemented. Its use of a finite volume method allows for geometric flexibility while having computational efficiency comparable to finite difference methods. Each group in CAT generates its own mesh and sets up hindcast simulations to assess FVCOM’s ability to predict tidal constituents (first round) and water levels and currents (second round).

In this presentation, we focus on the sensitivity of water level and timing of peak water height to mesh characteristics and tidal forcing. Initial hindcast tests have been conducted for two periods: one from July 1 to September 30, 2021, and one from January 1 to March 30, 2022. First, baseline 2-D barotropic simulations with only tidal forcing are performed and compared with observations. Next, the models are refined with wind and river forcing as well as temperature and salinity observations. Then, a 3-D run is performed and the water levels and currents output by the model are compared with observations. The accuracy of the simulations are promising and show the trade-offs that occur in the choice of tidal forcing as well as the balance between accuracy and computation time that results from mesh resolution.