Toward Sustainable Groundwater Management: Harnessing Remote Sensing and Climate Data to Estimate Field-Scale Groundwater Pumping
Abstract
Groundwater overdraft in western U.S. states has prompted water managers to start the development of groundwater management plans that include mandatory reporting of groundwater pumping (GP) to track water use. Most irrigation systems in the western U.S. are not equipped with irrigation water flow meters to record GP. Of those that do, performing quality assurance and quality control (QAQC) of the metered GP data is difficult due to the lack of reliable secondary GP estimates. We hypothesize that satellite (Landsat)-based actual evapotranspiration (ET) estimates from OpenET can be used to predict GP and aid in QAQC of the metered GP data. For this purpose, the objectives of this study are: 1) to pair OpenET estimates of consumptive use (Net ET, i.e., actual ET less effective precipitation) and metered annual GP data from Diamond Valley (DV), Nevada, and Harney Basin (HB), Oregon; 2) to evaluate linear regression and ensemble machine learning (ML) models (e.g., Random Forests) to establish the GP vs Net ET relationship; and 3) to compare GP estimates at the field- and basin-scales. Results from using a bootstrapping technique showed that the mean absolute errors (MAEs) for field-scale GP depth are 12% and 11% for DV and HB, respectively, and the corresponding root mean square errors (RMSEs) are 15% and 14%. Moreover, the regression models explained 50%-60% variance in GP depth and ~90% variance in GP volumes. The estimated average irrigation efficiency of 88% (92% and 83% for DV and HB, respectively) aligns with known center pivot system efficiencies. Additionally, OpenET proves to be useful for identifying discrepancies in the metered GP data, which are subsequently removed prior to model fitting. Results from this study illustrate the usefulness of satellite-based ET estimates for estimating GP, QAQC metered GP data and have the potential to help estimate historical GP.