AI for soil & Groundwater Contamination

Monitoring for groundwater contamination–particularly groundwater sampling and analysis–is currently done primarily for compliance purposes, with data often used simply to flag any regulatory exceedance then archived without further analytics. We are developing various ML tools for advanced analysis of soil and groundwater data, including unsupervised ML for identifying the spatiotemporal patterns of contaminant concentrations, or quantifying their responses to climate perturbations such as flooding or drought. We are also developing supervised ML for evaluating the ability to estimate contaminant concentrations based on in situ measurable parameters, as well as the effectiveness of well configuration to capture contaminant concentration distributions. Our main objective here is to develop python/R packages to perform various machine-learning tasks for groundwater data, which is transferable among the DOE sites.

Correlation heat map among multiple analytes at one well.