38-4 Predicting Groundwater Hydrochemical Facies in Three Dimensions across the Conterminous United States with Random Forest Classification

Session: Geoscience and Hydrogeology in the AI Era: From Predictive Models to Real-Time Applications

Presenting Author:

Paul Stackelberg

Authors:

Stackelberg, Paul E.¹, Knierim, Katherine J.², Belitz, Kenneth³, Cravotta III, Charles A.⁴, McCleskey, R.Blaine⁵, Killian, Courtney D.⁶

(1) U.S. Geological Survey, Water Resources Mission Area, Troy, NY, USA, (2) U.S. Geological Survey, Water Resources Mission Area, Little Rock, AR, USA, (3) U.S. Geological Survey, Water Resources Mission Area, Taghkanic, NY, USA, (4) U.S. Geological Survey, Water Resources Mission Area, New Cumberland, PA, USA, (5) U.S. Geological Survey, Water Resources Mission Area, Boulder, CO, USA, (6) U.S. Geological Survey, Water Resources Mission Area, Bridgeville, PA, USA,

Abstract:

A random forest classification (RFC) model was developed to predict hydrochemical facies (HCFs) of groundwater in three dimensions across the conterminous United States (CONUS). Major-ion data from 152,673 sites were used to categorize groundwater into one of six HCFs based on predominant cations and anions (CaMg-HCO₃, NaK-HCO₃, CaMg-SO₄, NaK-SO₄, Cl, or Mixed). These six HCFs were used as prediction targets for RFC modeling. Model features that represent relevant geochemical processes and(or) physical conditions were derived from readily available, CONUS-scale data. Additional model features were specifically engineered to support this analysis and include the elevation of well bottoms relative to the base of drinking water (ERDW) and six binary flags that relate the prevalence of each HCF to units of the geologic map of North America. The most important model feature overall was ERDW. The model was used to map HCFs at a 1-km² resolution across the CONUS and to depths of 400 m below the base of drinking water (which varies from 22 m to 2 km). Model predictions are consistent with the expected evolution of HCFs along groundwater flow paths. CaMg-HCO₃ is predicted to occur near the water table in the humid east, and near the water table in areas of the west underlain by crystalline rocks, volcanics, and unconsolidated sands and gravels. At depths below the base of drinking-water supplies, the model predicts a rapid transition from HCO₃ HCFs to Cl. Model predictions are reliable based on point data, and data averaged across hydrogeologic regions and with depth. Model predictions of HCFs could be used for mapping salinity and other groundwater characteristics across the CONUS. Predictions of HCFs can also be used – in conjunction with measurements of specific conductance – to provide reliable estimates of salinity or total dissolved solids.

Geological Society of America Abstracts with Program. Vol. 57, No. 6, 2025

doi: 10.1130/abs/2025AM-7885

© Copyright 2025 The Geological Society of America (GSA), all rights reserved.