Data-Driven Boundary Correction and Optimization of a Nearshore Wave and Hydrodynamic Model to Enable Rapid Environmental Assessment

The present project is part of a comprehensive effort by the PI, his students, and collaborators at the Naval Research Laboratory to increase the robustness and viability of the Delft3D model suite as an operational forecasting tool, and aid its continued transition to Navy forecasting centers. Prior projects have focused on determining the model s response to characteristics and sample sizes of bathymetric information. The present project focuses on determining the effect of boundary errors on model response, and the development of methods to ameliorate these issues. These boundary errors arise when the numerical grid for the forcing condition (wave model) is is insufficiently extended in the lateral (longshore) direction relative to the numerical grid for the current (flow model). The primary objective of this work is to investigate the effect of boundary forcing errors on the model response. This has a direct impact on the use of Neumann lateral boundary conditions, since these errors arise from insufficient extension of the WAVE (SWAN) grid beyond the lateral boundaries of the FLOW grid. There is a balance between accuracy and computational efficiency that must be struck; the computational bottleneck in the Delft3D system is SWAN, the wave model, and the run time increases with lateral extent of the WAVE grid. While point-by-point comparisons between model and data for various WAVE grid extensions can offer a view of the dependence of accuracy on grid extension, it does not yield any information on the spatial characteristics of the solution; poor datamodel comparison could be the result of slight spatial mismatches of highly variable solution fields, or oversmoothed solutions which have little relevance to the physics at hand.
The original document contains color images.