Model parameter estimation for particle transport

A generic parameter estimation algorithm was developed for the simultaneous extraction of multiple parameters from experimental or field observations. The basis of the method lies in the minimization of the variation between model predictions and observations via the iterative numerical determination of the functional minima of the model with respect to the parameters being determined. A multivariable Newton technique, and derivatives thereof, are used for this purpose. By relying on numerical differentiation in the iterative process, the technique is not limited by the model complexity, solution methodology, or scale of its domain. Model parameters have been estimated successfully for analytical and numerical solutions to dispersive, and advective-dispersive particle transport model equations. The effect of sampling or measurement error and initial parameter estimates on algorithm performance has been evaluated. This approach has been successfully applied to determine data requirements and constraints (quantity, quality, and location) for measurement of hydrodynamic and transport characteristics of dye clouds and aquatic particles in laboratory-mixed settling columns.