Curious for a glimpse of the mind behind the research?
If you are interested in joining the remote sensing hydrology research group, see the link below
Check out some of the other fantastic research conducted here at EMRGe
Integrating Remote Sensing Datasets to Estimate Groundwater Pumping and Subsidence
Multiple satellite datasets are sensitive to different aspects of the water balance, but are traditionally not integrated because they are challenging to relate to each other. In this project, we develop a machine learning framework to integrate these datasets to predict groundwater pumping and land subsidence at large spatial scales.
Estimating Plant Water Stress with Remote Sensing Data
Plant water stress is challenging to characterize without in-situ data. In this project, we combine satellite data (NDWI and ET) and in-situ measures of plant water stress (water content and potential), as well as yield and total water availability to develop estimates for plant water stress.
Groundwater Modeling with InSAR and AEM Data in California
Modeling Subsidence Using InSAR and Hydraulic Head
Land subsidence is linked to changes in hydraulic head due to groundwater extraction, but this relationship is challenging to model because it is non-linear. When the hydraulic head drops below its previously lowest level (preconsolidation head), inelastic, or plastic, deformation occurs. However, when the hydraulic head stays above the preconsolidation head, the deformation is elastic. It is challenging to estimate the preconsolidation head, because it varies spatially and over time. To further complicate the problem, subsidence is heavily influenced by the amount and thickness of clays, which is challenging to estimate.
In this project, we are developing approaches to model land subsidence using only hydraulic head and InSAR data. This model solves for the clay content, hydrologic properties, and spatio-temporal evolution of the preconsolidation head.
Joint Inversion of Subsidence and Electromagnetic Data
Modeling deformation data related to groundwater pumping is challenging, because it is a function of the subsurface geology, which is typically poorly known. In this study, we combined InSAR data, which is used to estimate deformation , with AEM data, which is used to estimate geology. We combined these two datasets in a joint inversion framework, solving for the subsurface geology and hydrologic properties of the geologic layers. This joint inversion resulted in a model that successfully predicted land deformation on our validation dataset.
Quantifying the Link between Groundwater Over-Pumping and Contamination
In the San Joaquin Valley, California, arsenic contamination affects over 10% of wells drilled in the San Joaquin Valley. Arsenic levels as low as 10 parts per billion have been linked to increased risk of cancer and numerous other health problems. Some have hypothesized that arsenic contamination is linked to groundwater pumping, but this had never been explored or quantified in California.
We processed subsidence data from satellites (interferometric synthetic aperture radar, or InSAR), which are a proxy for groundwater pumping. We then ran a machine learning model to explore the relationship between groundwater pumping (inferred by subsidence), geochemical conditions, and arsenic concentrations.
Our findings indicate that groundwater pumping is a significant factor in arsenic concentrations. This suggests that continued unsustainable pumping of groundwater in the San Joaquin Valley could raise arsenic levels throughout the Valley. However, our results also indicate that reductions in pumping are likely to improve water quality.
Estimating Permanent Groundwater Storage Loss Using Satellites
In the southern portion of the San Joaquin Valley, California, agricultural demands are so high that they account for roughly 10% of the pumped groundwater in the United States. Because of the high water demand of the crops that are grown, large volumes of groundwater are required to irrigate the crops. Groundwater pumping in this area has caused significant declines in groundwater levels over the past century, leading to subsidence as high as 10 m over this time period. Water managers are concerned that the subsidence represents a permanent loss of storage.
Using Interferometric Synthetic Aperture Radar (InSAR), we can measure subsidence with cm-scale accuracy over large areas. We have processed InSAR data over the study area from 2007-2010. Using groundwater level data, geologic models, and knowledge of the geomechanical properties of sediments in the area, in conjunction with InSAR, we estimated the permanent loss of groundwater storage.
Our results show that the vast majority (98%) of subsidence in our study area is related to the permanent loss of groundwater storage. By volume this is m3 of groundwater storage, or roughly 9% of the volume of groundwater used from 2007-2010.
The Use of Color Wheels to Communicate Uncertainty in the Interpretation of Geophysical Data
Water Resources Research, Estimating the Permanent Loss of Groundwater Storage in the Southern San Joaquin Valley, CA
Correlating Geological and Seismic Data with Unconventional Resource Production Curves Using Machines Learning
Groundwater Storage Loss Associated with Land Subsidence in Western US Mapped Using Machine Learning
Groundwater Withdrawal Prediction Using Integrated Multi-Temporal Remote Sensing Datasets and Machine Learning
Sanaz Vajedian - Postdoctoral Scholar
Sayantan Majumdar (he/him/his) - PhD Student
Jiawei Li - PhD Student
Rahel Pommerenke - MS Student
Fahim Hasan - MS Student
Dawit Asfaw - PhD Student
Lindi Oyler (they/them/theirs) - MS Student Lab Alumni