Estimating historical groundwater levels based on relations with hydrologic and meteorological variables in the U.S. glacial aquifer system

Abstract

A number of previous studies have examined relations between groundwater levels and hydrologic and meteorological variables over parts of the glacial aquifer system, but systematic analyses across the entire U.S. glacial aquifer system are lacking. We tested correlations between monthly groundwater levels measured at 1043 wells in the U.S. glacial aquifer system considered to be minimally influenced by human disturbance and selected hydrologic and meteorological variables with the goal of extending historical groundwater records where there were strong correlations. Groundwater levels in the East region correlated most strongly with short-term (1 and 3 month) averages of hydrologic and meteorological variables, while those in the Central and West Central regions yielded stronger correlations with hydrologic and meteorological variables averaged over longer time intervals (6–12 months). Variables strongly correlated with high and low annual groundwater levels were identified as candidate records for use in statistical linear models as a means to fill in and extend historical high and low groundwater levels respectively. Overall, 37.4% of study wells meeting data criteria had successful models for high and (or) low groundwater levels; these wells shared characteristics of relatively higher local precipitation, higher local land-surface slope, lower amounts of clay within the surficial sediments, and higher base-flow index. Streamflow and base flow served as explanatory variables in about two thirds of both high- and low-groundwater-level models in all three regions, and generally yielded more and better models compared to precipitation and Palmer Drought Severity Index. The use of variables such as streamflow with substantially longer and more complete records than those of groundwater wells provide a means for placing contemporary groundwater levels in a longer historical context and can support site-specific analyses such as groundwater modeling.