Empirical techniques for a priori parameter estimation

Inverse methods have allowed for the development of modeling strategies to solve for flow variables which satisfy the governing equations of complex hydrodynamic models while remaining as close as possible to observed data. Insufficiencies in observational data can be overcome through the use of a priori parameter estimates. The estimated parameters are systematically adjusted through fitting techniques until the model results approach the measured data. An alternative analysis method for determining the relative contribution of input data parameters for use in models employing inverse methods is empirical modeling. Empirical modeling does not make a priori assumptions about the data but instead allows the data itself to determine the form of the relationships within constraints determined by the structure of the selected empirical relations. A multivariable polynomial autoregressive analysis with exogenous variables (PARX model) is shown to produce effective empirical fits to measured nearshore current data. A nonlinear model developed to predict the observed current structure was able to explain 82% of the measured variability of the time series record and reproduce observed complex frequency distributions of tide and current energy densities.