USGS Science⌕ Search

SEARCH · USGS Science

Results for “International Journal for Disaster Risk Reduction”

Search indexed USGS publications on groundwater, aquifers, geologic maps, mineral resources and earthquakes. Explore source records by subject and place.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

1,443 records · Page 57Linked to original sources
USGS2026

Origin of the Pd/Pt ratio of the J-M Reef, Stillwater Complex, Montana, USA

The J-M reef of the Stillwater Complex is characterized by a high Pd/Pt ratio (mean ~3.8 with a standard error of 0.03) with a homogeneous geospatial distribution at the deposit scale. In this contribution, we demonstrate that the Pd/Pt ratio of the reef is the product of equilibration of an immiscible sulfide liquid with a silicate melt rich in Pd relative to Pt. Despite the high tenors of the J-M reef sulfides (avg 2,700 ppm Pt and 770 ppm Pt), numerical modeling shows that the parental melts did not have extraordinary Pd and Pt concentrations. Instead, the initial composition of a plausible parental silicate melt can have Pd and Pt contents well within the expected range of a normal, mantle-derived partial melt (i.e., ~10–20 ppb for both Pd and Pt with Pd/Pt of ~1). The relative differences in the partitioning behavior of Pt and Pd between sulfide liquid and silicate melt are unlikely to produce a consistent Pd/Pt ratio across a wide range of silicate melt to sulfide liquid mass ratios (i.e., R factors). Instead, the pre-emplacement fractionation of Pt alloy from S-undersaturated silicate magma accounts for the homogeneous and high Pd/Pt ratio of the J-M reef. We show that batch equilibration of sulfide liquid with silicate melt can produce the high Pd/Pt ratios of the reef if the partition coefficients between sulfide liquid and silicate melt for Pd and Pt are extremely high (>10 6 ). In an alternative model, Pd enrichment could be achieved by sulfide upgrading in resident footwall mush even if the partition coefficients between sulfide liquid and silicate melt are relatively small (between 10 4 and 10 6 ) because the instantaneous mass ratio of silicate melt to sulfide liquid is small (R ≈ 100–700), so the partitioning behavior of Pt and Pd has little impact on the composition of sulfide liquid.

Montana↗
USGS2025

Uppermost Oligocene and Miocene diatom biostratigraphy of Ocean Drilling Program Sites 682 and 688 from the Peru Margin

The diatom biochronology of ODP (Ocean Drilling Program) Holes 682A and 688E provides a detailed framework for refiningMiocene diatom zonation in the East Pisco Basin of southern Peru, establishing both a nearly complete offshore reference section and a correlation tool for the fragmentary onshore vertebrate-bearing deposits. This new biostratigraphic record documents a complete succession of low latitude and/or northeastern Pacific Miocene diatom zones, with two notable exceptions: a dissolution and/or hiatus interval (*16.5–14 Ma) during the Middle Miocene Climatic Optimum and a likely earliest Miocene hiatus (*23.4–21.8 Ma). Although eastern equatorial Pacific diatom zones characterize the Upper Oligocene and Lower Miocene strata, an increased abundance of cool-water diatoms that lived during the Middle and Late Miocene allows better application of northeast Pacific diatom zones, except during the Messinian (7–6 Ma) when warm-water diatoms predominate. The effects of eustatic sea level and tectonics on depositional sequences in the EPB and in offshore cores off central Peru are discussed.

Stratigraphy↗
USGS2025

Gulf Coast Basin CORE-CM initiative final report

The Bureau of Economic Geology at the University of Texas at Austin (UT-BEG) is leading the Gulf Coast Carbon Ore, Rare Earth, and Critical Minerals (CORE-CM) Initiative to assess the potential to produce critical minerals (CMs), including rare earth elements (REEs) from coal, coal ash, and produced water related to oil and gas production, and related materials (alumina processing waste [red mud], heavy mineral sands, graphite, and zeolite) within the Gulf Coast Basin. This project represents the first phase in a long-term program and provides reconnaissance data that will be foundational for future work by assessing resources and suggesting plans to be conducted in future work and expanding stakeholder engagement. The project includes several tasks designed to identify, characterize, and assess several necessary aspects for development of CMs and REEs in the Gulf Coast Basin.

Gulf Coast basin↗
USGS2025

Magnitude, depth and methodological variations of spectral stress drop within the SCEC/USGS Community Stress Drop Validation Study using the 2019 Ridgecrest Earthquake Sequence

We present the first ensemble analysis of the 56 different sets of results submitted to the ongoing Community Stress Drop Validation Study using the 2019 Ridgecrest, California, earthquake sequence. Different assumptions and methods result in different estimation of the source contribution to recorded seismograms, and hence to the source parameters (principally corner frequency, f c ⁠ , spectral stress drop, Δσ, and seismic moment, M 0 ⁠ ) obtained from modeling calculated source spectra. For earthquakes smaller than magnitude (M) 2.5 there is negligible correlation between the f c values obtained by different studies, implying that no present method is reliable using available data. For larger magnitude events, correlation between f c measurements of different studies, within even a small M range is always higher than spectral ⁠Δσ , because the f c measurements simply reflect the underlying physical decrease in f c with increasing M. We model the observed trends of submitted f c with both magnitude and depth. Most methods report an increase in spectral Δσ with M, although a magnitude‐invariant spectral Δσ is within the confidence limits. The depth dependence is smaller and depends on whether a study allows attenuation to vary with source depth; a combination of depth‐dependent attenuation correction, and depth‐dependent shear‐wave velocity can compensate for reported depth trends. We model the submitted values to remove differing M and depth variation to investigate the relative interevent variability. We find consistent relative variation between individual events, and also lower relative spectral Δσ in the northwest of the aftershock sequence, and higher on the cross fault and in the region of main fault intersection. This large‐scale comparison implies that absolute spectral Δσ estimates are dependent on the methods used; studies of different regions or using different methods should not be directly compared and improved constraints on path and site corrections are needed to resolve these absolute spectral Δσ differences.

California↗
USGS2024

Vulnerability assessment of groundwater influenced ecosystems in the Northeastern United States

Groundwater-influenced ecosystems (GIEs) are increasingly vulnerable due to groundwater extraction, land-use practices, and climate change. These ecosystems receive groundwater inflow as a portion of their baseflow or water budget, which can maintain water levels, water temperature, and chemistry necessary to sustain the biodiversity that they support. In some systems (e.g., springs, seeps, fens), this connection with groundwater is central to the system’s integrity and persistence. Groundwater management decisions for human use often do not consider the ecological effects of those actions on GIEs. This disparity can be attributed, in part, to a lack of information regarding the physical relationships these systems have with the surrounding landscape and climate, which may influence the environmental conditions and associated biodiversity. We estimate the vulnerability of areas predicted to be highly suitable for the presence of GIEs based on watershed (U.S. Geological Survey Hydrologic Unit Code 12 watersheds: 24–100 km 2 ) and pixel (30 m × 30 m pixels) resolution in the Atlantic Highlands and Mixed Wood Plains EPA Level II Ecoregions in the northeastern United States. We represent vulnerability with variables describing adaptive capacity (topographic wetness index, hydric soil, physiographic diversity), exposure (climatic niche), and sensitivity (aquatic barriers, proportion urbanized or agriculture). Vulnerability scores indicate that ~26% of GIEs were within 30 m of areas with moderate vulnerability. Within these GIEs, climate exposure is an important contributor to vulnerability of 40% of the areas, followed by land use (19%, agriculture or urbanized). There are few areas predicted to be suitable for GIEs that are also predicted to be highly vulnerable, and of those, climate exposure is the most important contributor to their vulnerability. Persistence of GIEs in the northeastern United States may be challenged as changes in the amount and timing of precipitation and increasing air temperatures attributed to climate change affect the groundwater that sustains these systems.

Connecticut, Delaware, District of Columbia, Maine↗
USGS2026

ShakeAlert Earthquake Early Warning System performance during the Mw 7.0 offshore Cape Mendocino earthquake

The 5 December 2024 M w 7.0 Offshore Cape Mendocino earthquake was a challenging test of the U.S. West Coast ShakeAlert earthquake early warning system due to its offshore epicenter and limited near‐source station coverage. We analyzed real‐time performance of all components of the ShakeAlert system, including the seismic algorithms (earthquake point‐source integrated code [EPIC] and Finite‐fault rupture Detector [FinDer]), the geodetic algorithm (Geodetic First Approximation of Size and Time–peak ground displacement [GFAST‐PGD]), and network telemetry during the event. EPIC created the first solution for this earthquake 15 s after origin time with an initial magnitude estimate of M 5.6 and location error of 10 km from the Advanced National Seismic System epicenter. An early spurious trigger from station CE.89101 fortuitously maintained location accuracy and, correspondingly, magnitude accuracy. FinDer contributed its first solution at 18 s with a location estimate closer to the seismic network and produced two distinct rupture geometries, leading to minor fluctuations in estimated intensity contours. GFAST‐PGD did not meet alerting thresholds but otherwise performed as expected. Network latencies were <2 s for most stations, supporting the rapid detection of this earthquake by the system. Roughly five million alerts were delivered to cell phone devices in California and Oregon during this event. This was also the first instance of a school district‐wide ShakeAlert‐powered system being activated. Comparisons to recorded seismograms demonstrate that the maximum warning times before potentially damaging shaking (intensity 6+) were in the range of 5–55 s. Although the ShakeAlert system provided accurate solutions and useful alert delivery, this earthquake raised awareness of potential issues within the system, including the need for improved offshore location estimates, a combination of solutions from ShakeAlert servers, and handling of spurious triggers.

California↗
USGS1999

Effects of snowmobile use on snowpack chemistry in Yellowstone National Park, 1998

Snowmobile use in Yellowstone National Park has increased substantially in the past three decades. In areas of greatest snowmobile use, elevated levels of by-products of gasoline combustion such as ammonium and benzene have been detected in snowpack samples. Annual snowpacks and snow-covered roadways trap deposition from local and regional atmospheric emissions. Snowpack samples representing most of the winter precipitation were collected at about the time of maximum annual snow accumulation at a variety of locations in the park to observe effects of a range of snowmobile traffic levels. Concentrations of organic and inorganic compounds in snow samples from pairs of sites located directly in and off snow-packed roadways used by snowmobiles were compared to concentrations in samples collected at nearby off-road sites. Concentrations of ammonium were 2 to 5 times higher for the in-road snow compared to off-road snow for each pair of sites. Thus, concentrations decreased rapidly with distance from roadways. In addition, concentrations of ammonium, nitrate, sulfate, benzene, and toluene in snow were positively correlated with snowmobile use.

Wyoming↗
USGS2026

Vegetation cover and composition in environments surrounding uranium mines in the Grand Canyon ecosystem, Northern Arizona

Mining uranium from breccia-pipe deposits in the greater Grand Canyon region has occurred since the mid-1900s. However, possible ecosystem contamination with harmful levels of radionuclides may have occurred due to mining activities in the 21st century. In response, a 20-year Federal moratorium on new mining claims in the Grand Canyon watershed was initiated in 2012, to allow time to evaluate the potential effects of uranium exploration and mining on human health, wildlife, and water resources. This moratorium, nor the 2023 designation of the “Baaj Nwaavjo I’tah Kukveni–Ancestral Footprints of the Grand Canyon National Monument,” precludes operation or development of mining claims predating 2012. Vegetation is a core ecosystem component that may be affected by uranium mining (for instance, through uptake and storage of radionuclides from the air or soil) or may act as a vector of exposure to wildlife, livestock, and humans (for instance, via their consumption of contaminated plant tissues). To provide baseline information about the plant communities associated with uranium mines in the Grand Canyon region, the U.S. Geological Survey surveyed an approximately 200-meter-wide buffer surrounding four breccia-pipe deposits, each in a unique stage of mine development, and at one reference area (a livestock water tank) that underwent ground disturbance but contains no mineral deposits. We sectioned the buffer zones into 0.65–4.52 hectare plots, within which we (1) inventoried all plant species, (2) measured percent cover of plant species, plant functional groups, and ground surface types (dark cyanobacteria, lichen, moss, bedrock, rock, embedded litter, duff, plant bases, and bare soil) using line-point intercept, and (3) measured length and frequency of gaps between perennial plant canopies using canopy gap intercept. We found that plant composition at the mines and the reference area differed from one another but were all characteristic of expected regional vegetation patterns. We provide this data summary as potential baseline information for future research and management efforts.

Arizona↗
USGS2024

New occurrences of the rare, REE minerals daqingshanite, törnebohmite, biraite, sahamalite, and ferriperbøeite from the Sheep Creek area, Montana, USA

Over 30 small, discontinuous, tabular carbonatite bodies are located in the Sheep Creek area, Ravalli County, southwest Montana. The age and origin of these REE-Nb-rich deposits are currently being investigated. The purpose of this paper is to document the occurrence of several rare minerals, including daqingshanite, törnebohmite, biraite, sahamalite, and ferriperbøeite, in two of the carbonatite bodies. These minerals are found in association with monazite, hydroxylbastnäsite, ferriallanite, calcite, dolomite, baryte, quartz, actinolite, apatite, celsian, and Sr-rich aragonite. Automated SEM-EDS was used to target the areas of interest in polished specimens for more detailed spot SEM-EDS and electron probe microanalysis. Raman spectra were also acquired for each of the rare minerals. The complex mineralogy of the Sheep Creek carbonatites is most likely due to several overlapping thermal events, including primary magmatic, overprinting hydrothermal, and supergene weathering stages. The rare minerals described in this study are believed to be hydrothermal and/or carbothermal in origin, although no estimates of temperature are available at this time.

Montana↗
USGS2024

Rainfall intensification amplifies exposure of American Southwest to conditions that trigger postfire debris flows

Short-duration, high-intensity rainfall can initiate deadly and destructive debris flows after wildfire. Methods to estimate the conditions that can trigger debris flows exist and guidance to determine how often those thresholds will be exceeded under the present climate are available. However, the limited spatiotemporal resolution of climate models has hampered efforts to characterize how rainfall intensification driven by global warming may affect debris-flow hazards. We use novel, dynamically downscaled (3.75-km), convection-permitting simulations of short-duration (15-min) rainfall to evaluate threshold exceedance for late 21st-century climate scenarios in the American Southwest. We observe significant increases in the frequency and magnitude of exceedances for regions dominated by cool- and warm-season rainfall. We also observe an increased frequency of exceedance in regions where postfire debris flows have not been documented, and communities are unaccustomed to the hazard. Our findings can inform planning efforts to increase resiliency to debris flows under a changing climate.

Arizona, California, Nevada, New Mexico, Utah↗
USGS2025

The influence of nodule versus crust morphology on the composition of seamount-hosted ferromanganese minerals

Ferromanganese mineral precipitation in the global oceans is ubiquitous, occurring in the form of both crusts and nodules at a broad range of depths and seafloor terrains. Although ferromanganese crusts and nodules are both composed of ferromanganese minerals, mineralogy and mean element concentrations compiled for regional crust versus nodule occurrences differ. Notably, most published compilations compare nodules from abyssal plain sediments to crusts forming on seamounts, and do not address the question as to whether mineral morphology may affect the composition of crusts and nodules in addition to environmental factors. To address this, we have compiled a dataset of co-located hydrogenetic (seawater sourced) open-ocean ferromanganese crusts and seamount-hosted nodules. The preliminary data presented here support our hypothesis that co-located crusts and nodules exhibit similar compositions, both in terms of their mineralogy and mean element concentrations, as well as overall variance in concentrations for any given element. These data indicate that local processes influence sample-to-sample variation in both crusts and nodules, in contrast to broader ferromanganese variation reported between open-ocean seamounts- hosted crusts and abyssal plain-hosted nodules.

Conference Paper↗
USGS2026

Distributed faulting of the northern West Napa Fault Zone in Napa Valley, California

Mapped surface ruptures from the 24 August 2014 M w 6.0 South Napa earthquake in the Napa Valley, California, show a 2‐km‐wide zone of distributed faulting in the southern and central West Napa fault zone (WNFZ). In the northern WNFZ at Hendry Winery (HW), however, the mapped 2014 surface ruptures encompass an ∼100‐m‐wide zone, implying significant narrowing of the near‐surface fault zone to the north. We present a tomographic shear‐wave velocity ( ⁠⁠ V S ) model and guided‐wave data that indicate the northern WNFZ is at least 400‐m wide, with multiple near‐surface fault traces. Our V S model shows that the 2014 surface ruptures are underlain by discrete low‐velocity zones (LVZs), and coincident guided‐wave data show that the LVZs carry fault‐zone guided waves. If nearby (<500 m) mapped faults to the east of HW are part of the WNFZ, the entire WNFZ is more than 1 km wide in the northern Napa Valley. WNFZ guided waves travel up to 38% slower than S body waves, and low‐strain guided‐wave shaking is up to five times stronger than the associated body‐wave shaking. Our data suggest that guided waves, traveling along distributed faults, may result in an increased shaking hazard over a 1‐km‐wide area of the northern Napa Valley during future significant earthquakes. In places, the 2014 surface ruptures were difficult to find one year after the earthquake, and paleoseismic trenching showed only weak evidence for faulting, which may not have been identified in trenches if the locations of the 2014 surface ruptures had not been previously mapped ( Prentice et al. , 2015 ). Guided‐wave and V S tomography data, however, show strong evidence for faulting beneath the 2014 surface ruptures and at locations to the east. Although paleoseismic trenching is the gold standard for identifying near‐surface faulting, methods such as peak ground velocities of guided waves may better identify immature near‐surface fault traces.

California↗
USGS2026

Cultural benefits knowledge and social learning in Elwha River dam removal and ecosystem decision-making

In a retrospective case study of the seminal Elwha River dam removal in Washington State, USA, we explore pathways through which ecosystem services (ES) knowledge can inform environmental decision-making. We specifically investigate how knowledge of cultural benefits linked to the Elwha River ecosystem was made available. Through interviews and historical document review we identify diverse forms of cultural benefits knowledge (i.e., how individuals or groups know the contributions of an ecosystem to their identities, experiences, and capabilities), including both knowledge in the form of practices – embodied or enacted forms of knowledge, and knowledge in the form of products – quantitative or written documentation. Recognizing and learning from the diverse forms in which benefits knowledges are conveyed can support more accurate and equitable consideration of the diverse values, well-beings, and human-nature relationships linked to ecosystems. With this case study, we illuminate pathways for meaningful consideration of multiple knowledge forms, and the plural ecosystem values they convey, with relevance for environmental decision processes in the Puget Sound and beyond.

Washington↗
USGS2025

Hydrogeology, water budget, and simulated groundwater availability in the Salt Fork Arkansas River and Chikaskia River alluvial aquifers, northern Oklahoma, 1980–2020

The 1973 Oklahoma Groundwater Law (Oklahoma Statute §82–1020.5) requires that the Oklahoma Water Resources Board conduct hydrologic investigations of the State’s aquifers to determine the maximum annual yield for each groundwater basin. The U.S. Geological Survey, in cooperation with the Oklahoma Water Resources Board, conducted an updated hydrologic investigation of the Salt Fork Arkansas River and Chikaskia River alluvial aquifers in northern Oklahoma for the study period spanning 1980–2020 and evaluated the simulated effects of potential groundwater withdrawals on groundwater flow and availability in the Salt Fork Arkansas River alluvial aquifer. A hydrogeologic framework and conceptual model were developed to guide the development of a numerical model. Three groundwater-availability scenarios were evaluated by using the calibrated numerical model, which was focused on the Salt Fork Arkansas River alluvial aquifer. These scenarios were used to (1) estimate equal-proportionate-share groundwater withdrawal rates, (2) quantify the potential effects of projected well withdrawals on groundwater storage over a 50-year period, and (3) simulate the potential effects of a hypothetical 10-year drought. The 20-, 40-, and 50-year equal-proportionate-share groundwater withdrawal rates for the Salt Fork Arkansas River alluvial aquifer under normal recharge conditions were about 0.63, 0.58, and 0.57 acre-foot per acre per year, respectively. Projected 50-year groundwater withdrawal scenarios were used to simulate the effects of modified well withdrawal rates. Because well withdrawals were less than 2 percent of the calibrated numerical-model water budget, changes to the well groundwater withdrawal rates had little effect on simulated Salt Fork Arkansas River base flows and groundwater storage in the Salt Fork Arkansas River alluvial aquifer. A hypothetical 10-year drought scenario was used to simulate the potential effects of a prolonged period of reduced recharge on groundwater storage. Groundwater storage at the end of the hypothetical drought period was 14.5 percent less than the groundwater storage of the calibrated numerical model without the simulated drought.

Oklahoma↗
USGS2023

A comprehensive fault system inversion approach: Methods and application to NSHM23

We present updated inversion‐based fault‐system solutions for the 2023 update to the National Seismic Hazard Model (NSHM23), standardizing earthquake rate model calculations on crustal faults across the western United States. We build upon the inversion methodology used in the Third Uniform California Earthquake Rupture Forecast (UCERF3) to solve for time‐independent rates of earthquakes in an interconnected fault system. The updated model explicitly maps out a wide range of fault recurrence and segmentation behavior (epistemic uncertainty), more completely exploring the solution space of viable models beyond those of UCERF3. We also improve the simulated annealing implementation, greatly increasing computational efficiency (and thus inversion convergence), and introduce an adaptive constraint weight calculation algorithm that helps to mediate between competing constraints. Hazard calculations show that ingredient changes (especially fault and deformation models) are the primary driver of hazard changes between NSHM23 and UCERF3. Updates to the inversion methodology are also consequential near faults in which the slip rate in UCERF3 was poorly fit or was satisfied primarily using large multifault ruptures that are now restricted by explicit b ‐value and segmentation constraints.

Bulletin of the Seismological Society of America↗
USGS2025

Airborne radiometric data map alteration of porphyry copper systems in the Elkhorn district, MT

The Elkhorn district in southwestern Montana is known for past production of high-grade base and precious metals from skarn and carbonate replacement deposits. The district also hosts a porphyry Cu-Mo deposit. A radiometric survey flown over the district and surrounds provides maps of surficial concentrations of potassium, thorium, and uranium. Handheld gamma ray spectrometer measurements on altered and unaltered rocks in the district and surrounding region provide a means to ground truth the airborne data. We demonstrate that the airborne and ground-based measurements show similar map patterns. We calculate ratios of the radioelement concentrations and present a potassium enhancement map that combines potassium and ratios of potassium to thorium and uranium. The results highlight the rocks previously mapped as having K-feldspar-biotite and quartz-sericite alteration assemblages in the Elkhorn district and map additional hydrothermal systems in the region, including known alteration in the Radersburg district. The data enhancement techniques can be used as a screening tool for mapping additional porphyry copper systems.

Montana↗
USGS2024

Geophysical characterization of mine influenced groundwater and surface water in the Mayflower section of the Animas River, Bonita Peak Mining District, Silverton Colorado

This report details findings from geophysical investigations to identify possible groundwater - surface water interactions near the Mayflower Section of the Animas River in Silverton, Colorado. The Mayflower Section is a mine influenced Superfund Site in Colorado. This investigation utilized electromagnetic induction (EM or EMI), magnetic, and fiber optic distributed temperature system (FODTS) geophysical methods to measure the bulk earth electrical conductivity, magnetic susceptibility, and temperature of specific surveyed volumes of the earth near the Mayflower Section of the Animas River. These physical parameters are used to understand the groundwater – surface water interactions, which can guide decision makers in their assessment of mine-impacted surface water. This report details the results from characterization and monitoring technologies to provide high data density and continuous monitoring of bulk earth electrical conductivity and temperature in the Mayflower section of the Animas River to identify zones of groundwater – surface water interactions and potential metal loading from mine influenced water. The investigation separated right and left bank characterization for each method and indicates more groundwater is entering from the right bank than the left bank and these predominantly right bank discharges potentially contain metal-rich water compared to the left bank. Results also indicate mineral veins facilitate preferential groundwater discharge to the river due to possible jointing, fractures, and permeability differences sometimes occurring along veins relative to host rock. For example, Boulder Gulch is likely groundwater dominated and may be receive water impacted by Mayflower Impoundments #1 and #2. Additionally, the beaver ponds near Blair Gulch may influence groundwater discharge and Mayflower Impoundment #4 is possibly impacting groundwater and surface water and may be connected to the wetlands to the west of the impoundment. These data could be further analyzed for smaller spatial scale analysis within areas of interest. The identification of these locations along sections of the river likely impacted by mine influenced groundwater potentially entering the Animas River and can be used by site investigators, decision makers, and stakeholders in mitigation decisions and strategies.

Colorado↗
USGS2026

Evaluating machine learning approaches to identify and predict oil and gas produced water lithium concentrations

Recently, the demand for battery-grade lithium has substantially increased, largely due to electrification of the transportation sector. The search for new lithium sources has turned to produced waters (frequently brines), a large-volume wastewater by-product of oil and gas extraction. Geochemical analysis indicates the presence of varying concentrations of lithium from produced water samples collected across the United States and represented in the U.S. Geological Survey’s National Produced Water Geochemical Database, as well as mixtures of Marcellus Shale produced water included in the Pennsylvania Department of Environmental Protection’s Oil and Gas Well Waste Reports. We first examined whether the geochemical signature of the lithium-bearing produced waters is sufficiently distinct so that machine learning (ML) can be used to correctly classify samples to the formation of origin. The produced water sample data used to assess classification accuracy were from the Marcellus Shale, Utica Shale and Point Pleasant Formation (Utica), and Smackover Formation oil and gas wells. Further, we evaluated the potential for ML to accurately classify Marcellus Shale produced water spatially (i.e., northeast versus southwest Pennsylvania). We then investigated whether ML algorithms applied to a suite of geochemical concentration data (i.e. Ba, Br, Cl, K, Mg, Sr) may be used to predict the lithium concentration of an unknown sample. Finally, we applied an estimated economic lithium grade cutoff of 150 milligrams per liter (mg/l) and assessed the utility of ML to predict whether a produced water sample would fall above or below the grade cutoff based on the suite of geochemical parameters. Four machine learning algorithms—Random Forest (RF), Gradient Boosting Trees (GBT), Extreme Boosting (XGBoost), and Deep Neural Networks (DNN) were assessed. This study successfully demonstrates that all four machine learning methods can precisely and accurately estimate lithium concentrations and geologic formation classification. The products of this study contribute to the growing body of knowledge aimed at expanding the lithium resource base within the United States.

Alabama, Arkansas, Florida, Georgia, Louisiana, Mi↗