loading page

Augmenting Sparse Groundwater Level Data with Earth Observations vis Machine Learning
  • +2
  • Norm Jones,
  • Steven Evans,
  • Gustavious Williams,
  • Jim Nelson,
  • Daniel Ames
Norm Jones
Brigham Young University

Corresponding Author:njones@byu.edu

Author Profile
Steven Evans
Brigham Young University
Author Profile
Gustavious Williams
Brigham Young University
Author Profile
Jim Nelson
Brigham Young University
Author Profile
Daniel Ames
Brigham Young University
Author Profile

Abstract

Groundwater development will provide a more stable water source and enhance food security. Sustainable groundwater development requires collecting and analyzing data produced at global and national levels and disseminating that data and knowledge to end users such as States, NGOs, municipalities, businesses, and agropastoralists in a format that is useful for planning and decision-making. In developing countries, analyzing in situ measurements to de can be challenging due to sparsity of data and lack of tools and expertise. To address these problems we have developed a web-based geospatial tool that ingests in situ water level measurements and performs temporal and spatial interpolation to build interactive animated maps, time series plots, and long-term aquifer depletion curves. We use machine learning to find correlations among Earth observation data, such as precipitation or soil moisture, with water level data and perform more accurate interpolation. This approach ensures that scarce in situ data are used as effectively and accurately as possible. This tool helps water managers gain a better understanding of groundwater resources and determine how aquifers are responding to groundwater development, droughts, and climate change.