Remote Sensing-Based Estimates of Changes in Stored Groundwater at Local Scales: Case Study for Two Groundwater Subbasins in California’s Central Valley

Abstract

Sustainable groundwater management requires high-quality and low-latency estimates of changes in groundwater storage (∆Sgw). However, estimates of ∆Sgw produced using traditional methods, including groundwater models and well-based measurements, typically lag years behind the present because collecting the required on-the-ground data is a time consuming, expensive, and labor-intensive task. Satellite remote sensing measurements provide potential pathways to overcome these limitations by quantifying ∆Sgw through closing the water balance. However, the range of spatial scales over which ∆Sgw can be accurately estimated using remote sensing products remains unclear. To bridge this knowledge gap, this study quantified ∆Sgw for the period of 2002 through to 2021 using the water balance method and multiple remote sensing products in two subbasins (~2700 km2–3500 km2) within California’s Central Valley: (1) the Kaweah–Tule Subbasin, a region where the pumping of groundwater to support agriculture has resulted in decades of decline in head levels, resulting in land subsidence, damage to infrastructure, and contamination of drinking water and (2) the Butte Subbasin, which receives considerably more rainfall and surface water and has not experienced precipitous drops in groundwater. The remote sensing datasets which we utilized included multiple sources for key hydrologic components in the study area: precipitation, evapotranspiration, and soil moisture. To assess the fidelity of the remote sensing-based model, we compared estimates of ∆Sgw to alternative estimates of ∆Sgw derived from independent sources of data: groundwater wells as well as a widely used groundwater flow model. The results showed strong agreement in the Kaweah–Tule Subbasin in long-term ∆Sgw trends and shorter-term trends during droughts, and modest agreement in the Butte Subbasin with remote sensing datasets suggesting more seasonal variability than validation datasets. Importantly, our analysis shows that the timely availability of remote sensing data can potentially enable ∆Sgw estimates at sub-annual latencies, which is timelier than estimates derived through alternate methods, and thus can support adaptive management and decision making. The models developed herein can aid in assessing aquifer dynamics, and can guide the development of sustainable groundwater management practices at spatial scales relevant for management and decision making.