STOCHASTIC NUMERICAL ANALYSIS OF UP-SCALED AQUIFER AND STREAMBED PROPERTIES FOR MODELING LAGGED RIVER DEPLETIONS DUE TO WELL PUMPING

This paper describes a stochastic-deterministic approach for development of upscaled streambed properties to use in numerical simulation of stream–aquifer interactions, specifically for analysis of delayed depletions to a partially penetrating stream due to pumping of a well close to a stream. Following a review of classical analytical methods used to compute lagged stream depletion, we develop a numerical model for a well pumping from an alluvial aquifer located near a partially penetrating stream. The model includes consideration of material heterogeneity such as one would expect in alluvial deposits. Using the model, we develop a method to estimate effective streambed conductance and aquifer conductivities for heterogeneous alluvial aquifers subjected to pumping by alluvial wells near streams. These effective properties can be used subsequently in coarse-grid numerical models to compute lagged river depletions due to the well pumping. Our stochastic-deterministic approach involves: (i) synthetically generating heterogeneous hydraulic conductivity fields based on field observations from alluvial / fluvial aquifer systems, (ii) Monte Carlo simulation of flow through 100 realizations of the random conductivity fields representing alluvial aquifer cross-sections in the vicinity of a stream influenced by well pumping, and (iii) inferring upscaled effective properties from the ensemble mean of the simulations. We apply this approach to results from large-scale aquifer pumping tests executed adjacent to the South Platte River in northeast Colorado.