Katherine M. Ransom | Geology Reports

USGS2025

Select elements of concern in surface water of three hydrologic basins (Delaware River, Illinois River, and Upper Colorado River)—Data screening for the development of spatial and temporal models

The report focuses on the screening of previously published concentration data associated with 12 elements of concern (aluminum, arsenic, cadmium, chromium, copper, iron, mercury, manganese, lead, selenium, uranium, and zinc) measured in stream surface waters of three hydrologic basins (Delaware River Basin, Illinois River Basin, and the Upper Colorado River Basin). The purpose of this analysis is to determine what subsets of the original dataset (containing more than 1,500,000 observations) may be most suitable for each of two types of modeling efforts. The first type of modeling envisions a machine learning approach to determine which geospatial attributes are most significant in describing the spatial distribution of elemental concentrations within a basin. The second type of modeling envisions a stepwise regression approach to develop multivariable models that can be used to determine high resolution time-series estimates of elemental concentrations or loads at discrete U.S. Geological Survey real-time stream surface water sites. These site-specific temporal models are based on continuous measurements of available discharge and (or) in situ sensor data (temperature, pH, turbidity, dissolved oxygen, specific conductance, and (or) fluorescent dissolved organic matter) as the explanatory variables. The data screening for both model types considered historical trends in analytical methods and detection quantitation limits, the extent of censored data, data density, and environmental relevance with respect to three U.S. Environmental Protection Agency water quality thresholds (drinking water guidelines, human health criteria, and aquatic life criteria). The result of this analysis was the production of a final list of potential models deemed suitable for further development based upon the data exclusion (or inclusion) scheme developed herein for each model type. In both cases, the final models included mostly the three crustal elements (iron, manganese, and aluminum) that are found at comparatively high concentrations in surface water, whereas most of the more pernicious elements were excluded from the final model lists owing to various data limitations. The one exception to this was arsenic, for which the existing data were sufficient at three U.S. Geological Survey real-time sites for potential further development of time-series models.

Colorado, Delaware, Illinois, Indiana, Maryland, N↗

USGS2017

A hybrid machine learning model to predict and visualize nitrate concentration throughout the Central Valley aquifer, California, USA

Intense demand for water in the Central Valley of California and related increases in groundwater nitrate concentration threaten the sustainability of the groundwater resource. To assess contamination risk in the region, we developed a hybrid, non-linear, machine learning model within a statistical learning framework to predict nitrate contamination of groundwater to depths of approximately 500 m below ground surface. A database of 145 predictor variables representing well characteristics, historical and current field and landscape-scale nitrogen mass balances, historical and current land use, oxidation/reduction conditions, groundwater flow, climate, soil characteristics, depth to groundwater, and groundwater age were assigned to over 6000 private supply and public supply wells measured previously for nitrate and located throughout the study area. The boosted regression tree (BRT) method was used to screen and rank variables to predict nitrate concentration at the depths of domestic and public well supplies. The novel approach included as predictor variables outputs from existing physically based models of the Central Valley. The top five most important predictor variables included two oxidation/reduction variables (probability of manganese concentration to exceed 50 ppb and probability of dissolved oxygen concentration to be below 0.5 ppm), field-scale adjusted unsaturated zone nitrogen input for the 1975 time period, average difference between precipitation and evapotranspiration during the years 1971–2000, and 1992 total landscape nitrogen input. Twenty-five variables were selected for the final model for log-transformed nitrate. In general, increasing probability of anoxic conditions and increasing precipitation relative to potential evapotranspiration had a corresponding decrease in nitrate concentration predictions. Conversely, increasing 1975 unsaturated zone nitrogen leaching flux and 1992 total landscape nitrogen input had an increasing relative impact on nitrate predictions. Three-dimensional visualization indicates that nitrate predictions depend on the probability of anoxic conditions and other factors, and that nitrate predictions generally decreased with increasing groundwater age.

California↗