Sunday, 23 January 2011
Hydrodynamic Models are a sophisticated approach to simulating and visualizing water flow. Based off the momentum and mass conservation principle, these models can use inputs of oceanic and atmospheric forcings to simulate water levels, sediment transport, salinity and temperature within the model area. Variables like currents, bathymetry, wind and geographic features influence the way the model behaves. Corpus Christi Bay is primarily a shallow bay (~3 meters) located off the Gulf of Mexico in South Texas. The difficulty in modeling this bay comes from a deep ship channel (~15 meters) running across the bay's length. The Corpus Christi ship channel is the primary source of water inflow from the Gulf of Mexico and has a significant impact on water levels. Another important variable affecting water levels in Corpus Christi Bay is the wind, typically directed from the Southeast or North, perpendicular to the ship channel. To implement a successful prediction model, the model grid has to accurately represent the physical features and depth of the study area. Other inputs to the model include forcings such as surface winds, tidal input or water level time series at the model boundaries. Two different hydrodynamic models were implemented for the Corpus Christi Bay area based on the same coastline and bathymetry information. The first model is the Finite Volume Coastal Ocean Model (FVCOM), a 3-D triangular unstructured grid model, which produces calculations based on a sigma layer water column. The second model, the Coastal Modeling System (CMS), is a 2-D structured square grid model. To understand which model performs best for the study area, the models were analyzed and compared based on their accuracy in predicting water levels and currents. Their computational efficiency relative to grid resolution was compared as well. The models were compared over a 21 day test case (October 27 to November 16, 2002) using similar grids of varying resolution and identical inputs i.e. water levels and wind at local coastal observation stations. The results show that average absolute errors for water level predictions are between 4 and 7 cm with the best model predictions varying with location. Overall water level maps show a similar dynamic. Differences are observed in average current maps with eddy formations occurring at similar locations but not always simultaneously. Differences are also observed in the predicted currents along the ship channel. FVCOM runs substantially slower than CMS on a similar computer but its computational efficiency can be considerably improved by running the model on a multiprocessor machine.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner