Validation of Model Output versus ADCP Observations on the PR Insular Shelf, Part 2: Are all Sites the Same?
Validation of Model Output versus ADCP Observations on the PR Insular Shelf, Part 2: Are all Sites the Same?
Abstract:
We have previously compared the output from three oceanographic models against observed data from an ADCP at a common grid point location on the zonally oriented, southwestern Puerto Rico shelf that extends into the northern Caribbean Sea. The three models were: 1) AMSEAS (NCOM), 2) Regional ROMS and 3) a higher resolution version of ROMS nested within the Regional ROMS. These models faced great difficulty in accurately depicting the bathymetry of the ocean in the PR-USVI archipelago which is characterized by small islands, narrow insular shelves, steep slopes and deep water beyond. The resulting validations of the three models versus the ADCP at the selected location were poor. However, the insight we gained into the behavior of the models during the validation process suggested that models might do a better job at simulating currents across the inter-island straits that connect the Atlantic Ocean with the Caribbean Sea than along the insular Caribbean or Atlantic coastlines. We therefore focused our attention on expanding our previous research by performing a similar analysis using the ROMS model against ADCP observations in the Mona Passage, west of PR. This new ADCP location exhibits bathymetric features that are smoother, less complex, and better represented in the Regional ROMS model while flows at the site are stronger than at the previous ADCP site at La Parguera. Statistical time-series analyses are performed on model and ADCP flow velocity time series to quantify the model’s skill. Results indicate that ROMS does a much better job at simulating ocean currents at the Mona Passage site than at La Parguera. Dynamical and numerical differences that might explain the spatially varying model skill are considered. In summary: model skill validation sites around PR are not all the same.