Paper No. 6
Presentation Time: 9:30 AM


LACKEY, Greg, Department of Civil, Environmental, and Architectural Engineering, University of Colorado, Boulder, CO 80310, NEUPAUER, Roseanna, Department of Civil, Environmental, and Architectural Engineering, University of Colorado, Boulder, Boulder, CO 80301 and PITLICK, John, Department of Geography, University of Colorado, Boulder, CO 80310,

Extensive extraction of groundwater from an aquifer has the potential to reduce the flow of a hydraulically connected stream or river. This phenomenon, known as stream depletion, negatively affects the availability of water and the surrounding environment. In the arid west and southwest of the United States, impacting streams through groundwater extraction often has legal consequences. Numerical groundwater flow models are typically used to assess the legality of pumping well locations in an aquifer. Thus, the process of siting new pumping wells in an aquifer relies on the development of groundwater models that can accurately estimate stream depletion through simulating surface and groundwater interactions. The assumed hydraulic properties of the streambed, in particular the streambed hydraulic conductivity (Kr), control flow across the streambed. Currently, it is standard practice for stream depletion modelers to assume or calibrate for a single representative value of Kr. In natural systems, the transport of sediment along a streambed results in a Kr that is spatially and temporally heterogeneous. We use the stream package (STR) in MODFLOW-2000 to investigate how variations in Kr impact stream depletion estimations. We create 29 variations of a simple stream depletion model that are identical with the exception of Kr. Each of the model variations is assigned a different homogeneous Kr within a Kr spectrum of 10-9 m s-1 to 8.64 x 10-2 m s-1­. The difference between each model streambed Kr is 0.25 log units. We estimate stream depletion for each scenario and demonstrate that stream depletion models are sensitive to a range of streambed Kr that is dependent upon the model input parameters. We also investigate methods for predicting the range of Kr to which a stream depletion model is sensitive using only the model input parameters.