ANALYTIC ELEMENT MODELING OF A SLOPING BASE AQUIFER: NON-LINEAR MODEL OPTIMIZATION USING A LEVENBERG-MARQUARDT ALGORITHM

This presentation studies the impact of a sloping aquifer base on a large aquifer. Our case study is the Ogallala Aquifer in the central plains of the USA. Modeling of groundwater is accomplished through the use of numeric and analytic techniques. MODFLOW, the most widely used numeric model, has been utilized to model the Ogallala Aquifer, but problems often arise in the models due to steep slopes, small saturated thicknesses, and high hydraulic conductivities. These aquifer properties frequently result in cells going dry unexpectedly, as well as failure to meet model convergence criteria. The sloping base model is an analytic method that has also been used to simulate groundwater flow in the Ogallala in previous studies by the authors. This approach breaks the aquifer into a set of rectangular cells, solves for the 1-D flow and iterates across cells. The initial solution is achieved without taking into account groundwater to surface water interactions resulting in large errors in the final results. An optimization algorithm is used to locate areas where water is discharging from groundwater to surface water as well as the areas where surface water is percolating to the groundwater table. Results are shown for the two modeling approaches, and the strengths of each are illustrated.