USING A 30-YEAR RECONSTRUCTION OF GROUNDWATER LEVELS IN CALIFORNIA’S CENTRAL VALLEY TO EVALUATE REGIONAL LINKAGES BETWEEN WATER USE AND QUALITY DURING DROUGHT

Assessment of dynamic relations between groundwater levels and water quality is often limited by the lack of monitoring data that are co-located in time and space. This study leveraged over 160,000 groundwater-level measurements from monitoring wells throughout California’s Central Valley to characterize relative hydrologic change at over 6,000 public-supply wells where only water-quality (nitrate) data were available over the past 30 years. We used a novel, iterative algorithm to impute gaps in annual groundwater-level time series using multiple linear regression with a user-defined error threshold. The imputation routine regularized control points used for spatial interpolation of groundwater-levels across the Central Valley, which optimized quantification of relative head change on an annual timestep. Modeled groundwater levels approximated independent validation data (n = 1,752) with a mean absolute error of 5 m (70% of residuals < 5 m and 26% < 1 m) and were highly correlated with observed interannual fluctuations in validation time series (Pearson’s r > 0.95), underscoring the efficacy of these methods in determining the relative direction and magnitude of head change over areas encompassing heterogenous aquifer systems. Relation of model results to water-quality data showed increasing nitrate concentrations were more prevalent during drought where groundwater levels declined rapidly due to pumping stress. Study results highlight the need to account for transient hydrologic factors in regional water-quality trend assessments.