loading page

Toward Field Scale Groundwater Withdrawals in the Western U.S. using Remote Sensing and Climate Data
  • +2
  • Sayantan Majumdar,
  • Thomas Ott,
  • Justin Huntington,
  • Ryan Smith,
  • Venkataraman Lakshmi
Sayantan Majumdar
Desert Research Institute

Corresponding Author:[email protected]

Author Profile
Thomas Ott
Desert Research Institute
Justin Huntington
Desert Research Institute
Ryan Smith
Colorado State University, University of Virginia
Venkataraman Lakshmi
University of Virginia


In the Western U.S., the combination of increased and projected droughts, rising irrigation water demands, and population growth is expected to intensify groundwater consumption leading to adverse consequences like land subsidence and aquifer depletion. Despite the urgent need to address these challenges, there is limited local-scale monitoring of groundwater withdrawals in most of the groundwater basins in this region. Understanding the volume of groundwater being withdrawn is indispensable for implementing sustainable solutions to tackle water security issues. Hence, developing reliable and efficient groundwater withdrawal monitoring solutions is critical to address the pressing water management concerns in this region. The existing methods for estimating groundwater withdrawals are either costly and time-consuming, or they cannot generate dependable predictions at the scales required for effective local management. While our earlier works on integrating remote sensing and machine learning techniques to estimate gridded (1-5 km) groundwater use in Kansas, Arizona, and the Mississippi Alluvial Plain have been successful, field-scale estimation is still a challenge. Here, we use statistical and machine learning-based approaches to relate field-scale groundwater withdrawals with remote sensing-derived datasets, e.g., Landsat evapotranspiration (ET), downscaled SMAP surface soil moisture, and other hydrometeorological datasets for several Western U.S. states. We apply and test our approach by estimating and comparing groundwater pumping measurements at field- and regional-scales for multiple groundwater basins. Preliminary results using linear regression and machine learning-based approaches in Nevada and Arizona show promise (R2 of 0.5 to 0.7), additional in-situ pumping data actively being compiled will likely improve the models. While there are clear opportunities for model improvements, modeled withdrawal estimates are likely more accurate than common water right duties and potential crop ET-based estimates. We aim to enable water resource and user communities better understand water use, water budgets and support field-scale management practices for metered and unmetered groundwater basins.
26 Jan 2024Submitted to ESS Open Archive
02 Feb 2024Published in ESS Open Archive