USGS Science⌕ Search

SEARCH · USGS Science

Results for “Bulletin of Volcanology”

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.

1,661 records · Page 86Linked to original sources
USGS2025

Groundwater-level elevations in the Denver Basin bedrock aquifers and Upper Black Squirrel Creek alluvial aquifer, El Paso County, Colorado, 2021–24

El Paso County is the second-most populous county in Colorado and is projected to grow another 15 percent by 2030. Within El Paso County is the Upper Black Squirrel Creek Designated Groundwater Basin (Black Squirrel Basin), an area where surface water is scarce and water users rely primarily on groundwater from five different aquifers (the Upper Black Squirrel Creek alluvial aquifer and four bedrock aquifers within the Denver Basin aquifer system: the lower Dawson, Denver, Arapahoe, and Laramie-Fox Hills aquifers) to meet their needs. Currently (2024), land within the Upper Black Squirrel Creek Basin is primarily used for rural grazing and agriculture; however, municipal development is ongoing. In 2021, the U.S. Geological Survey, in cooperation with the Upper Black Squirrel Creek Ground Water Management District, began a study to establish a baseline dataset and assess the groundwater resources of the aquifers within the Black Squirrel Basin. A network of 39 wells was established in 2021; discrete groundwater-level measurements were made bimonthly. Nine of the 39 wells were equipped with pressure transducers to record hourly groundwater-level data. Seven wells had statistically significant seasonal trends, and trends at 3 wells were negative. For the discrete data, 16 wells had a significant trend for the study period, and 4 wells had negative trends. For the time-series data, 8 wells had significant trends, and 3 wells had negative trends. Potentiometric surface maps were created for this study using discrete, static groundwater levels measured in April 2023. These maps showed the estimated groundwater flow direction from the north-northwest to the south-southeast in the alluvial aquifer and from the northwest to the east-southeast for the lower Dawson and Denver aquifer wells. This study indicates the potential benefit of monitoring wells in the areas near municipal pumping. Additional monitoring could lead to a better understanding of connectivity between aquifers and be an important tool for assessing long-term sustainability of groundwater use.

Colorado↗
USGS1996

Pesticides in surface water from three agricultural basins in south-central Georgia, 1993-97

Twenty-two of 43 pesticides analyzed were detected in 128 water samples collected from the Tucsawhatchee Creek, the Little River, and the Withlacoochee River. These streams drain agricultural basins in south-central Georgia and were sampled from March 1993 through June 1995. Herbicides were detected more frequently than insecticides. The most frequently detected herbicides were atrazine and metolachlor and the most frequently detected insecticide was carbaryl. Pesticide concentrations in the three streams were low and did not exceed U.S. Environmental Protection Agency drinking water standards. The maximum pesticide concentration was 2.6 mg/L (micrograms per liter) for propargite, a miticide detected in only one sample. The maximum concentrations of the remaining 21 pesticides were less than 0.25 mg/L. The median concentrations were equal to the method detection limit for all pesticides except atrazine (0.008 mg/L) and metolachlor (0.012 mg/L). The ratio of herbicide detections to nondetections was largest in the planting season, smaller in the harvest season and smallest in the fallow season for the three basins. The same pattern existed for the insecticide ratios in the Little River and the Withlacoochee River. Pairwise correlations between concentrations of atrazine and metolachlor, and four parameters (discharge, and concentrations of dissolved organic carbon, suspended organic carbon, and suspended sediment) were evaluated for each stream. The strongest correlations existed between metolachlor and mean daily discharge, and metolachlor and sediment in the Withlacoochee River. The only significant correlation for the Little River was between atrazine and suspended sediment.

Georgia↗
USGS2024

Participatory engagement to reduce communication gaps

Underserved communities, especially those in coastal areas in Puerto Rico, face significant threats from natural hazards such as hurricanes and rising sea levels. Limited funding hinders the investment in costly mitigation measures, increasing exposure to natural disasters. Providing coastal resources and data products through effective communication mechanisms is fundamental to improving the well-being of these underserved coastal communities. The overall objectives of the pilot effort to engage and connect with underserved coastal communities in Puerto Rico were the following: (1) compile a comprehensive database of the projects and resources relevant to natural hazards in Puerto Rico; (2) foster connections with Puerto Rican interested parties to better understand their priorities regarding coastal hazards and provide them with pertinent U.S. Geological Survey (USGS) resources; and (3) identify knowledge gaps to guide future USGS projects in Puerto Rico. Here we outline our participatory engagement framework and process, along with two specific resources developed with the information collected from this effort. These resources are available in English and Spanish and consist of user-friendly, non-technical information products. Among them are: (1) a website where users can learn about USGS research on landslides, hurricanes, earthquakes, water resources, coastal hazards, tsunamis, and ecosystem hazards and environmental contaminants, and (2) a geonarrative highlighting shoreline changes in Puerto Rico with sections on historical shoreline trends, hurricane impacts, and potential solutions that could help protect communities and mitigate coastal hazards. Continuing participatory engagement in future projects could enhance the accessibility and usability of natural hazards resources within the community.

San Juan↗
USGS2024

Separating signals in elevation data improves supervised machine learning predictions for hydrothermal favorability

A recent study identified topography (land surface elevation above sea level) as an important input dataset (feature) for predicting the location of hydrothermal systems in the Great Basin in Nevada. Yet, topography is generally a result of more than one geological process and may consequently contain multiple distinct signals. For example, the geologic evolution of the Great Basin has produced both crustal thickening (i.e., regional-scale trends in elevation) and thinning via Basin and Range extensional faulting (i.e., valley-scale topographic relief). We postulate that these geologic processes may affect the occurrence of hydrothermal systems differently. Therefore, we separate the regional trend from the valley-scale signal in the Great Basin, and then use them separately to evaluate the importance of each as predictors for hydrothermal favorability. Our prior work applying supervised machine learning (ML) using the data from the Nevada Machine Learning Project demonstrated that employing a training strategy that randomly selects negative training sites produces better performing models for predicting hydrothermal favorability than a training strategy that uses expert-selected negatives. The models created using both training strategies exhibited a west-east geographic trend in the predictions for the favorability of hydrothermal resources. These models generally predicted higher favorability in western Nevada and lower favorability in eastern Nevada. This west-east trend in predicted favorability correlates with elevation across the Great Basin, which trends higher from west to east. By separating the original elevation feature into distinct features for elevation trend (i.e., regional-scale topography) and detrended elevation (i.e., valley-scale or local relative topography), we find that models using the separated topographic signals consistently outperform competing models that use the original elevation feature. Although western Nevada still exhibits higher favorability than eastern Nevada, using separated signals for regional elevation and local structure reduces the west-east prediction trend in the region and emphasizes structures associated with hydrothermal upflow. This work emphasizes how carefully engineering features to represent geological conditions relevant to hydrothermal systems allows ML algorithms to detect important patterns for predicting hydrothermal resource favorability and leads to better model performance.

Conference Paper↗
USGS2025

Methods for estimating selected low-flow statistics at gaged and ungaged stream sites in Massachusetts

The U.S. Geological Survey, in cooperation with the Massachusetts Department of Conservation and Recreation, Office of Water Resources, computed selected at-site streamflow statistics at U.S. Geological Survey streamgages in and near Massachusetts and developed regional regression equations for estimating selected streamflows at ungaged stream sites in Massachusetts. Two sets of regional regression equations were developed: (1) the “mainland” equations, for mainland Massachusetts excluding the area covered by the second set, and (2) the “southeastern” equations, for the Plymouth-Carver-Kingston-Duxbury aquifer area in southeastern Massachusetts and for Cape Cod. The regression equations and at-site statistics may be used by Federal, State, and local water managers in addressing water-resources issues relevant in Massachusetts. Regional regression analyses for the mainland equations were developed to estimate the following 27 streamflow statistics: 99-, 98-, 95-, 90-, 85-, 80-, 75-, 70-, 60-, and 50-percent flow durations; monthly June, July, August, and September 90- and 50-percent flow durations; February, June, and August median of the monthly means; harmonic mean; and medians of the following annual low-flow frequency statistics: 7-day; 7-day, 2-year; 7-day, 10-year; 30-day, 2-year; and 30-day, 10-year. The analyses used 81 streamgages with minimal to no regulations in and near Massachusetts. The regression analyses determined that four basin characteristics—drainage area, combined hydrologic soils A and B, streamflow variability index, and annual mean temperature—were the only significant explanatory variables for the different mainland equations. Regional regression equations were developed for the Plymouth-Carver-Kingston-Duxbury aquifer area in southeastern Massachusetts and Cape Cod, because surface-water drainage areas and groundwater contributing areas do not always coincide in this area of the State. The regression analyses to estimate 10 flow durations from the 99th to 50th percentiles used 18 streamflow sites with some occasional minor regulations—because there are few unregulated streams in southeastern Massachusetts. The analyses determined that groundwater contributing area and storage (combined water bodies and wetlands) were the only significant explanatory variables in the southeastern equations.

Massachusetts↗
USGS2026

PFAS mixture composition and internal exposure profiles shape biological responses under field-realistic exposure

Per- and polyfluoroalkyl substances (PFAS) occur as complex mixtures, yet mixture-dependent biological effects under environmentally realistic exposure conditions remain poorly understood. We conducted multiyear (2018, 2019, 2021) continuous-flow, field-based exposures of male fathead minnows ( Pimephales promelas ) using a low-PFAS reference well (REF; sum of measured PFAS (∑PFAS) 0.1–0.2 μg L –1 ) and PFAS-contaminated groundwater from a fire-training area (FTA) at Joint Base Cape Cod, Massachusetts. Dilution treatments enabled separation of concentration and mixture effects. Groundwater from well FTA1 was perfluorooctanesulfonate (PFOS) dominated (∑PFAS 10–31 μg L –1 ), whereas groundwater from well FTA2 had higher concentrations and was enriched in perfluorooctanoate (PFOA) and diverse precursors (∑PFAS ∼ 80 μg L –1 ). Despite comparable plasma ∑PFAS and PFOS in FTA1–100% and FTA2–50% on day-7, cumulative mortality reached approximately 35% in FTA2–50% and 17% in FTA2–100%, and secondary sex trait expression was reduced by approximately 65–80% relative to REF, whereas sperm motility effects were mixture- and time-dependent. Plasma PFAS profiles were dominated by perfluorohexanesulfonate (PFHxS), PFOS, and sulfonamide precursors. Liver transcriptomics from REF-acclimated fish revealed robust disruption of metabolic, mitochondrial, and endocrine pathways that link PFAS mixture chemistry and internal exposure profiles to organismal outcomes, and testis transcriptomics provided complementary insight into reproductive impairment. These results indicate that PFAS mixture composition and internal exposure profiles, including precursor-associated differences, are important determinants of ecotoxicological outcomes under field-realistic conditions.

Massachusetts↗
USGS2025

“Leaky weirs” capture alluvial deposition and enhance seasonal mountain-front recharge in dryland streams

“Leaky weirs” are rock structures installed in dryland streams, which are anchored into exposed bedrock, loosely cemented, and designed to allow water to slowly pass through. They are being tested at a ranch in southeastern Arizona, USA, to restore and conserve the historic range and desert wetlands. Data are collected to assess how leaky weirs impact surface water, subsurface water, and groundwater recharge—including stream discharge, timing, and depth of infiltration, and groundwater elevations. Three adjacent watersheds, two with outlets just below leaky weirs and one with leaky weirs farther upstream, were instrumented with water-level loggers, wildlife cameras, and crest stage instruments with temperature sensors in the soil. As most groundwater recharge is assumed to be focused along the mountain fronts in this region, mountain-block recharge is also evaluated to differentiate between the two using isotope analyses. Finally, a single, late-season flood event is scrutinized to consider the leaky weir effect on all monitored components in the water budget. Results indicated groundwater flow is primarily from the mountains to the east via older, regional mountain-block recharge. However, the development of shallow alluvial aquifers is supported by the leaky weirs, that slow flows, capture permeable sediments, and allow infiltration, thus enhancing mountain-front recharge. In turn, these new pockets of water help support the restoration of historic wetlands. Sediment accumulates where leaky weirs are installed, reducing flashy peak flows, and resulting in a series of infiltration ponds along the channel that support vegetation during growing seasons and recharge the shallow aquifer during non-growing seasons.

Arizona↗
USGS2026

Measuring stress In high pressure deformation experiments with high speed fiber-optics

High‐pressure, high‐temperature rock deformation experiments are essential for understanding deep Earth processes, but accurately measuring stress is challenging due to the inaccuracy introduced by seal friction within the apparatus and large inertia, which affects earthquake process measurements made far from the experimental fault. To overcome these limitations, we developed and implemented a simple, compact fiber‐optic sensor based on an External Cavity Fabry–Perot Interferometer for in situ load measurement on a piston inside the pressure vessel of a Griggs‐type high‐pressure apparatus. The sensor can be used at sample temperatures up to 800°C for both slow creep and fast rupture testing with bandwidth capability from DC to 6 MHz. Two important experimental results are described: (1) the first direct measurements of seal friction in this apparatus type; these measurements record seal friction approximately one‐tenth of the confining pressure under both low (150 MPa) and high (1 GPa) pressure conditions; and (2) high‐temporal resolution capture of dynamic stress drops during stick‐slip faulting events at high‐confining pressure, revealing near‐fault wave propagation details, high‐frequency oscillations, and implying high‐transient slip rates (6–12 m/s) previously inaccessible with external sensors.

The Seismic Record↗
USGS2026

Season and antecedent conditions impact concentration-discharge relationships for dissolved organic carbon and alkalinity in southeast Alaskan watershed

Fluvial export of dissolved carbon plays an important role in watershed-scale biogeochemistry. Predicted changes in climate are expected to impact watershed hydrologic regimes, and in turn, the sources and export of dissolved carbon from watersheds. Here, we utilize high resolution measurements of discharge and dissolved carbon concentration to examine how concentration-discharge (CQ) relationships vary seasonally and during high flow events over the main runoff season (May–October) in a temperate forested watershed in Southeast Alaska. Concentration-discharge relationships for dissolved organic carbon (DOC) and alkalinity demonstrated strong seasonal patterns, with more linear relationships in May and June versus other months. Changing power law model slopes ( b values; the exponent in a power law regression between runoff and carbon yields) indicated potentially shifting watershed sources (biogenic vs. geologic) and contrasting dominant flowpaths (shallow vs. deeper groundwater) for DOC and alkalinity over the sampling period. During the largest storm event of the study, DOC and alkalinity b values shifted from an overall pattern of transport (mean b = 1.58 values >1.0 indicate transport limitation) and source limitation (mean b = 0.48, values <1.0 indicate source limitation) to chemostatic (DOC, b = 0.99; alkalinity, b = 1.019). In June through August, patterns in hysteresis index suggest that CQ relationships were altered when storms followed in close succession to each other. Together, these findings indicate that seasonal and antecedent flow conditions play a role in dissolved carbon export from forested watersheds. Understanding these dynamics, particularly during winter months, will become increasingly important as changes to hydroclimate impact riverine carbon export.

Alaska↗
USGS2026

pySATSI: A Python package for computing focal mechanism stress inversions

We introduce pySATSI, a Python package for computing earthquake focal mechanism stress inversions. This algorithm can handle a wide variety of types of stress inversion problems with a single script and can duplicate many capabilities of preceding methodologies. We also add new capabilities that include spatiotemporally variable inversion grids, damped stress estimates for clusters with few or no focal mechanisms, and variable fault‐plane ambiguities that the user can assign to individual events. In addition, we added the ability to use damped stress inversions with fault‐plane ambiguity probabilities that are weighted by fault instabilities. Our algorithm is computationally efficient with faster runtimes than previous algorithms, scales well for large datasets, and can be easily parallelized.

Seismological Research Letters↗
USGS2025

Ambient field seismology in critical zone hydrological sciences

Passive ambient noise monitoring is an emerging tool in environmental seismology, leveraging the ambient seismic field to assess temporal variations in shallow subsurface properties. This review focuses on the potential and challenges of using scattered coda waves from noise correlation functions to monitor critical zone dynamics. The sensitivity of seismic velocities to various environmental factors, including precipitation, snowmelt, atmospheric pressure, and groundwater fluctuations, underscores the method’s versatility. While coda waves excel in detecting subtle changes due to their scattered nature, ballistic waves provide higher spatial resolution, albeit with challenges in source stability. Advances in seismic sensing, including distributed acoustic sensing and low-cost geophone networks, have enabled high-resolution monitoring of hydrological processes, subsurface deformation, and seismic hazards. Integrating seismic data with hydrological models provides insights into water storage, pore pressure changes, and soil moisture dynamics. However, limitations in spatial resolution, calibration with ground truth data, and coupled effects between environmental factors remain key challenges. This review emphasizes the importance of interdisciplinary approaches in refining methodologies, enhancing sensor deployments, and addressing data gaps. Passive seismic monitoring offers opportunities to understand critical zone processes and their broader impacts on seismic hazards and environmental sustainability.

Comptes Rendus. Géoscience↗
USGS2026

Ductile and brittle Rio Grande Rift deformation in Oligocene granite records a two-stage rift history in southern Colorado

The timing and nature of early deformation in the Rio Grande Rift remains poorly constrained. We present evidence for the earliest structural signature of rift extension in the Sangre de Cristo Range, southern Colorado, based on new geologic mapping, structural analysis, rock magnetic data, and thermochronology. These analyses focus on the ~30.0 Ma granite of Chokecherry Canyon, which hosts discrete low-angle mylonitic shear zones and a distributed, gently SW-dipping protomylonitic fabric. Incremental stretching axes, stretching lineations, and Kmax magnetic lineations plunge gently WSW. Quartz microstructures and crystallographic orientations indicate dominantly coaxial strain in the protomylonite and general shear in the discrete shear zones. Quartz c-axis opening-angle thermometry suggests deformation at ~420–540°C. Thermal modeling of ⁴⁰Ar/³⁹Ar K-feldspar data indicates rapid post magmatic cooling below the brittle–plastic transition, supporting shear-zone formation immediately after emplacement. Slow cooling from ~20–13 Ma was followed by renewed rapid cooling at ~13 Ma, interpreted as the onset of extensional exhumation along the Sangre de Cristo Fault System. These results show that extension in the northern Rio Grande Rift was active by ~30 Ma, earlier than previously recognized. We propose a two-stage model for northern Rio Grande Rift evolution: Stage I (30–23 Ma) records ENE–WSW extension localized in low-angle mylonitic shear zones associated with mid-crustal intrusions; Stage II (≤18 Ma) reflects brittle high-angle normal faulting, focused exhumation, and rift narrowing. Stage I magmatism and deformation along the western range front likely established crustal weaknesses that guided later fault development.

Colorado↗
USGS2025

Permafrost–wildfire interactions: active layer thickness estimates for paired burned and unburned sites in northern high latitudes

As the northern high-latitude permafrost zone experiences accelerated warming, permafrost has become vulnerable to widespread thaw. Simultaneously, wildfire activity across northern boreal forest and Arctic/subarctic tundra regions impacts permafrost stability through the combustion of insulating organic matter, vegetation, and post-fire changes in albedo. Efforts to synthesis the impacts of wildfire on permafrost are limited and are typically reliant on antecedent pre-fire conditions. To address this, we created the FireALT dataset by soliciting data contributions that included thaw depth measurements, site conditions, and fire event details with paired measurements at environmentally comparable burned and unburned sites. The solicitation resulted in 52 466 thaw depth measurements from 18 contributors across North America and Russia. Because thaw depths were taken at various times throughout the thawing season, we also estimated end-of-season active layer thickness (ALT) for each measurement using a modified version of the Stefan equation. Here, we describe our methods for collecting and quality-checking the data, estimating ALT, the data structure, strengths and limitations, and future research opportunities. The final dataset includes 48 669 ALT estimates with 32 attributes across 9446 plots and 157 burned–unburned pairs spanning Canada, Russia, and the United States. The data span fire events from 1900 to 2022 with measurements collected from 2001 to 2023. The time since fire ranges from 0 to 114 years. The FireALT dataset addresses a key challenge: the ability to assess impacts of wildfire on ALT when measurements are taken at various times throughout the thaw season depending on the time of field campaigns (typically June through August) by estimating ALT at the end-of-season maximum. This dataset can be used to address understudied research areas, particularly algorithm development, calibration, and validation for evolving process-based models as well as extrapolating across space and time, which could elucidate permafrost–wildfire interactions under accelerated warming across the high-northern-latitude permafrost zone. The FireALT dataset is available through the Arctic Data Center ( https://doi.org/10.18739/A2RN3092P , Talucci et al., 2024).

Arctic↗
USGS2026

Extreme precipitation variability and soil texture controls on water-table response

Extreme precipitation events (EPEs), a key class of hydrometeorological extremes, are intensifying globally under climate change; however, their effects on water-table dynamics across varying soil textures remain poorly understood. To better understand the impacts of EPEs, we conducted one-dimensional modeling to evaluate water-table response time, displacement, recession time, and total recharge under EPEs of 0.20 m, 0.40 m, and 0.60 m amounts, applied over 1-, 7-, and 20-day durations across twelve soil textures. The results show that coarse soils (i.e., sand) respond within days, while fine soils (i.e., clay) may take over 200 days. Water-table displacement ranged from 0.30 to 1.64 m and increased with EPE magnitude. The time it took for water tables to recede ranged from 1.2 to 3.0 years. A first-order estimate of total possible recharge, calculated from porosity and displacement, ranged from 17% (clay) to 97% (sand), averaging ~63% across soil textures. These findings highlight that recharge is primarily governed by EPE magnitude and soil properties, not event duration. This modeling effort provides new insight into how soil texture modulates groundwater response to extreme precipitation, informing future water budget and resilience assessments.

Water↗
USGS1998

Methods and guidelines for effective model calibration; with application to UCODE, a computer code for universal inverse modeling, and MODFLOWP, a computer code for inverse modeling with MODFLOW

This report documents methods and guidelines for model calibration using inverse modeling. The inverse modeling and statistical methods discussed are broadly applicable, but are presented as implemented in the computer programs UCODE, a universal inverse code that can be used with any application model, and MODFLOWP, an inverse code limited to one application model. UCODE and MODFLOWP perform inverse modeling, posed as a parameter-estimation problem, by calculating parameter values that minimize a weighted least-squares objective function using nonlinear regression. Minimization is accomplished using a modified Gauss-Newton method, and prior, or direct, information on estimated parameters can be included in the regression. Inverse modeling in many fields is plagued by problems of instability and nonuniqueness, and obtaining useful results depends on (1) defining a tractable inverse problem using simplifications appropriate to the system under investigation and (2) wise use of statistics generated using calculated sensitivities and the match between observed and simulated values, and associated graphical analyses. Fourteen guidelines presented in this work suggest ways of constructing and calibrating models of complex systems such that the resulting model is as accurate and useful as possible.

Water-Resources Investigations Report↗
USGS2024

Telecommunications fiber for sensing earthquake aftershocks: Progress and hurdles

Aftershocks offer valuable clues to earthquake behavior. The challenge: quickly deploying sensors to capture the early details of earthquake ruptures within the zone of aftershocks. Telecommunication fibers might be an answer, providing denser networks in otherwise difficult areas, potentially faster than traditional methods.

Conference Paper↗
USGS2026

CRESCENT earthquake dynamic rupture, earthquake cycle, and tsunami code verification platform

Physics-based simulations are critical for understanding natural hazards. The increasing complexity of numerical codes requires benchmark exercises to verify that different computational methods yield consistent results when solving the same governing equations. Here, we present an open-access web platform designed for the verification of earthquake dynamic rupture, seismic cycle, and tsunami simulations. The platform architecture utilizes a modular, serverless backend on Amazon Web Services (AWS) to provide scalable file processing and visualization. A lightweight static web application provides a secure interface for uploading and managing results, while the browser-based data visualization enables interactive analysis of time series and surface grid data. By using structured JavaScript Object Notation (JSON) text files to define benchmark structures, the system remains fully extensible, allowing the addition of new scenarios without modifying the underlying software logic. The platform hosts the "The Tsunami Problem Versions" (TTPV) 1 & 2, two benchmarks for 3D fully coupled earthquake dynamic rupture and tsunami generation, and provides a framework for earthquake cycle models. This community resource aims to build trust in numerical simulations and facilitate long-term collaborative code verification as modeling software continues to evolve.

Seismica↗
USGS2025

Preventing overfitting when using tree-based methods for mapping hydrothermal favorability

Ensemble tree-based algorithms are robust tools for estimating sparsely distributed resources with non-linear dependencies (e.g., hydrothermal systems). These algorithms naturally accommodate the threshold conditions necessary to enable and support hydrothermal systems (e.g., having sufficient heat and permeability) and are simpler than many other non-linear machine learning strategies (e.g., artificial neural networks), which is an advantage when working with few labeled examples from which to learn. In previous work, we used eXtreme Gradient Boosting (XGBoost) to produce regional prediction and uncertainty maps of hydrothermal favorability; however, recent studies suggest that, even when properly applied, XGBoost has some risk of overfitting when there are few labeled examples from which to learn. To evaluate overfitting when constructing hydrothermal favorability maps with tree-based methods, we compare XGBoost with Extremely Randomized Trees (ExtraTrees), another ensemble tree-based algorithm that has the potential to underfit when using few labeled examples. We hold all other modeling parameters constant, resulting in two contrasting favorability maps of conventional geothermal resources for the Great Basin. Our results indicate that ExtraTrees demonstrably reduces overfitting compared with XGBoost. After considering overall performance, we conclude that ExtraTrees provides a more suitable modeling approach than XGBoost for the purposes of conventional hydrothermal resource assessments.

Conference Paper↗