Removal of River‐Stage Fluctuations from Well Response Using Multiple Regression

Many contaminated unconfined aquifers are located in proximity to river systems. In groundwater studies, the physical presence of a river is commonly represented as a transient‐head boundary that imposes hydrologic responses within the intersected unconfined aquifer. The periodic fluctuation of river‐stage height at the boundary produces associated responses within the adjacent aquifer system, the magnitude of which is a function of the existing well, aquifer, boundary conditions, and characteristics of river‐stage fluctuations. The presence of well responses induced by the river stage can significantly limit characterization and monitoring of remedial activities within the stress‐impacted area. This article demonstrates the use of a time‐domain, multiple‐regression, convolution (superposition) method to develop well/aquifer river response function (RRF) relationships. Following RRF development, a multiple‐regression deconvolution correction approach can be applied to remove river‐stage effects from well water‐level responses. Corrected well responses can then be analyzed to improve local aquifer characterization activities in support of optimizing remedial actions, assessing the area‐of‐influence of remediation activities, and determining mean groundwater flow and contaminant flux to the river system.