USE OF AN ANALYTIC ELEMENT MODEL (GFLOW) COUPLED TO A PARAMETER ESTIMATION MODEL (UCODE) TO ESTIMATE ANNUAL GROUNDWATER RECHARGE RATES IN NORTHERN WISCONSIN

Understanding the spatial and temporal distribution of groundwater recharge is a basic prerequisite for effective groundwater resource management and modeling. We coupled a two-dimensional analytic element flow model (GFLOW) to a nonlinear parameter estimation code (UCODE) to estimate annual recharge rates for 1996 - 2000 for an undeveloped basin in northern Wisconsin. GFLOW is a steady state model; the model was thus optimized for each year independently.

UCODE computes values for a set of model parameters that provide a least square minimum for an objective function which the modeler defines in terms of sets of field observations and assigned weights. We used median daily discharge data from four stream gauges and bimonthly water level measurements from wells within the basin as flux and head targets in the annual optimizations. Weights were assigned based on an uncertainty analysis of the field data.

We first optimized the model on an annual basis assuming a single value of recharge for the entire basin and, in general, obtained reasonable matches to the majority of the measured baseflows. Between 1996 and 2000, the model results showed more than a two-fold difference in annual recharge; during this same period there was an observed 1.7-fold difference in annual precipitation. These results underscore the potential significance of temporal recharge variability for annual regional water budget planning for even undeveloped watersheds.

Discrepancies between the modeled and measured baseflows at certain gauges suggested that a single value of recharge for the entire basin may be inappropriate and that recharge varies across the basin. We thus incorporated various potential recharge patterns based on the vegetation and soil distributions and re-optimized the flow model in an attempt to improve the model fit at the subbasin level. Optimizations in which recharge was zoned based on vegetation type consistently provided the best fit for the five years and suggest that vegetation is the dominant control on recharge variability in this undeveloped basin. The final recharge values and distributions compared favorably with field measurements of recharge and results from soil water balance and terrestrial biosphere models that we have also run in the basin.