2-DIMENSIONAL ADVECTIVE TRANSPORT IN GROUNDWATER-FLOW PARAMETER-ESTIMATION

Nonlinear regression is useful in ground-water flow parameter estimati on, but problems of parameter insensitivity and correlation often exis t given commonly available hydraulic-head and head-dependent flow (for example, stream and lake gain or loss) observations. To address this problem, advective-transport observations are added to the ground-wate r flow, parameter-estimation model MODFLOWP using particle-tracking me thods. The resulting model is used to investigate the importance of ad vective-transport observations relative to head-dependent now observat ions when either or both are used in conjunction with hydraulic-head o bservations in a simulation of the sewage-discharge plume at Otis Air Force Base, Cape Cod, Massachusetts, USA. The analysis procedure for e valuating the probable effect of new observations on the regression re sults consists of two steps: (1) parameter sensitivities and correlati ons calculated at initial parameter values are used to assess the mode l parameterization and expected relative contributions of different ty pes of observations to the regression; and (2) optimal parameter value s are estimated by nonlinear regression and evaluated. In the Cape Cod parameter-estimation model, advective-transport observations did not significantly increase the overall parameter sensitivity; however: (1) inclusion of advective-transport observations decreased parameter cor relation enough for more unique parameter values to be estimated by th e regression; (2) realistic uncertainties in advective-transport obser vations had a small effect on parameter estimates relative to the prec ision with which the parameters were estimated; and (3) the regression results and sensitivity analysis provided insight into the dynamics o f the ground-water flow system, especially the importance of accurate boundary conditions. In this work, advective-transport observations im proved the calibration of the model and the estimation of ground-water flow parameters, and use of regression and related techniques produce d significant insight into the physical system.