USGS Science⌕ Search

subject

Article

Article: explore 974 source-linked works published from 1974 to 2026, with original documents and citations.

This collection is a preview while coverage and quality are evaluated.

Search within this collection

Coverage and selection

Includes records with this source-supplied label or an explicit phrase match in their metadata. Matches indicate a mention, not proof that a paper uses a method or tests a material. Source versions are consolidated by DOI.

Sources: usgs. Collection updated 2026-09-15. Counts describe this index, not the complete source archives.

USGS2025

Fluid inclusion constraints on the geometry of the magmatic plumbing system beneath Mauna Loa – Part 2: Xenoliths

Mauna Loa volcano erupts crystal-poor material at its summit and more crystal-rich material on its rift zones. Some of the more olivine-rich lava flows contain xenoliths with diverse mineralogy, including cumulate harzburgites with high-Mg# orthopyroxenes and high-Fo olivines (both > 84). Previous experimental work and thermodynamic modelling has proposed that high-Mg# orthopyroxenes only crystallize from Mauna Loa melts at high pressures (> 6 kbar, > 20 km), leading to suggestions that there is a region of sub-Moho magma storage at Mauna Loa in addition to the geophysically imaged magma reservoir at 2–5 km depth below the summit. We use melt and fluid inclusion barometry combined with thermodynamic models to further investigate this suggestion. Fluid inclusion data from harzburgites and dunitic xenoliths yield storage depths remarkably similar to those found in non-xenolithic crystals from lavas and tephras, with a clear peak at ~ 2–3 km (below the summit). Depths from melt inclusions in these xenoliths overlap with fluid inclusion pressures, ruling out the possibility of fluid inclusion re-equilibration during a period of stalling in a shallower reservoir. We examine five different thermodynamic models and find that the minimum pressure of olivine-orthopyroxene co-saturation varies by ~ 4 kbar (~ 12 km). These models also fail to predict that orthopyroxene is stable in ~ 15–80% of compositionally relevant experimental charges which grew orthopyroxene. Overall, this shows that phase stability modelling is an unreliable method of determining magma storage depth at Mauna Loa. We suggest that model discrepancies reflect a lack of experimental constraints on orthopyroxene stability at > 1200 ℃ and 0.01–5 kbar. Based on the presence of large oikocrystic orthopyroxenes completely enclosing rounded olivine chadacrysts, we suggest that these harzburgitic xenoliths formed through the reaction of intruding melts with olivine mush piles within the Mauna Loa edifice at ~ 3 km depth below the summit, with no need for a deeper storage reservoir. The predominance of pre-eruptive shallow storage means that there is more chance of detecting reservoir destabilization with geophysical monitoring techniques compared to a scenario where melts are supplied from sub-Moho reservoirs.

Hawaii↗
USGS2025

A regional model comparison between MODPATH and MT3D of groundwater travel time distributions

Groundwater quality changes in wells and streams lag behind changes to land use due to groundwater travel times. Two contaminant transport methods were compared to assess differences in their simulated travel time distributions (TTDs) to streams and wells in the Wisconsin Central Sands. MODPATH simulates advective groundwater flow with particle tracking, while MT3D simulates age-mass using a finite difference solution without dispersion to allow for direct comparison of the two methods. MODPATH appropriately simulates groundwater TTDs from the water table to surface discharge but is subject to inaccuracies at weak-sink well cells due to the flow-model grid discretization and imprecise location of well discharge within well cells. MT3D better represents weak-sink well cells since it removes mass in proportion to the prescribed pumping rate, although travel time within well cells is neglected. Conversely, MT3D's treatment of surface water boundary cells is not as accurate as MODPATH because mass should be removed from the water table rather than the full cell volume. MT3D simulations of TTDs can also be confounded by the instantaneous vertical distribution of mass introduced throughout recharge cells instead of at the water table, which initiates mass along deeper flow paths. We evaluated 9 MODPATH and 13 MT3D implementations, generating differences in median travel times of up to 18 years. Both methods have strengths and weaknesses, with MT3D better representing weak-sink well cell behavior and MODPATH better representing surficial recharge and discharge. The effect of these characteristics on simulated TTDs, along with ideas for ameliorating method weaknesses, is discussed.

Wisconsin↗
USGS2025

A glimpse into the future of tectonic tremor monitoring

Tectonic tremor is a weak, long-duration seismic signal often observed in subduction zones and on some other plate-bounding faults. Because of tremor's characteristically low amplitude (and low signal-to-noise) and lack of clear phase arrivals, detecting and locating tremor usually requires techniques distinct from those applied to typical earthquakes. Major advances in detection and understanding of tremor have derived in the past from a powerful combination of new data and new analysis techniques. In a recent study, Sagae et al. (2025, https://doi.org/10.1029/2025jb031348 ) exploit that combination again, developing a new machine-learning based workflow and applying it to the S-net cabled seismic network in the Japan trench offshore northern Honshu. Their approach, although complex, succeeds in detecting several times more tremor activity than earlier studies, resulting in new insights and providing a blueprint for similar approaches that could be applied elsewhere. As real-time earthquake monitoring adopts similar tools, it may present an opportunity to bring tremor monitoring into operational workflows. In turn, this could solidify tremor monitoring as a component of future operational earthquake forecasting.

JGR Solid Earth↗
USGS2025

Updating regional‐scale geospatial liquefaction models with locally available geotechnical data

We present a method to update the geospatial liquefaction model used by the U.S. Geological Survey’s near‐real‐time ground failure product with subsurface geotechnical data. The geospatial model estimates liquefaction probability from peak ground velocity (via ShakeMap) and geospatial susceptibility proxies. In many regions, additional information relevant to constraining liquefaction likelihood is also available, including surface geology maps and subsurface geotechnical measurements. There is currently no mechanism to use these data in the ground failure product liquefaction model, even though these data could provide more precise constraints on spatial variations in the lithologic character of the soil (surface geology) and direct measurements of the subsurface mechanical properties that affect liquefaction occurrence and severity (geotechnical measurements). In this study, we develop a method to integrate these data with the geospatial model and assess how these data can improve regional‐scale predictions. We develop a Bayesian updating framework and apply it to the 1989 magnitude 6.9 Loma Prieta, California, earthquake, for which mapped observations are available to evaluate performance. We constrain the Bayesian framework with 373 Northern California cone penetration tests and liquefaction susceptibility classes based on the mapped surface geology. This Bayesian model incorporates geotechnical information into the geospatial model and more accurately predicts liquefaction occurrences than the geospatial model, while sacrificing less accuracy in terms of predicting the absence of liquefaction than the geotechnical model. In future applications, this approach could be adapted to update other geospatial models using locally available subsurface data.

California↗
USGS2025

To heal or not to heal?: 2. The moment-recurrence time behavior of repeating earthquakes in the 2011 Prague, Oklahoma aftershock sequence is consistent with laboratory healing rates

The timing and failure conditions of an earthquake are governed by the interplay between fault reloading and restrengthening. The moment-recurrence time behavior of repeating earthquakes can give observational estimates of fault healing rates; however, it is difficult to link these observed healing rates to laboratory studies of frictional healing in part because of uncertainty in lithology. Here, we study the 2011 Prague earthquake sequence, which includes repeating earthquakes in the Arbuckle group and the granitic basement, and compare them to laboratory experiments on samples of the Arbuckle and Troy granite (representative of the basement rock) (Okamoto et al., 2025, https://doi.org/10.1029/2024JB030573 ). We find three spatially distinct groups of repeating earthquakes with different moment-recurrence behavior: (a) constant moment-recurrence time in the Arbuckle group, (b) scattered moment-recurrence time at the intersection of the foreshock-mainshock fault in the granitic basement, and (c) moment-predictable behavior outside of the foreshock-mainshock fault intersection also in the granitic basement. Our observation of stagnant healing for repeating sequences in the Arbuckle group is consistent with laboratory observations of low healing rates for moderately high pore fluid pressures in Arbuckle samples. For the moment-predictable group, the source radius that is required in order to match healing rates is consistent with source radius estimations when taking into account reasonable attenuation of the P- pulse width. Overall, we observe diverse healing behaviors in the seismic families that are consistent with laboratory healing rates, providing seismic evidence that contact-scale frictional mechanisms are relevant to large-scale earthquake dynamics.

Oklahoma↗
USGS2025

To heal or not to heal?: 1. The effect of pore fluid pressure on the frictional healing behavior of lithologies in Oklahoma

The competition between fault healing (i.e., re-strengthening) and fault loading determines the timing and magnitude of fault failure within the seismic cycle. Repeating earthquakes can give observational estimates of fault healing rates, however, it is difficult to link laboratory studies of frictional healing and observed healing rates from repeating earthquakes in part because of uncertainty in lithology at depth. Due to well-constrained and relatively simple geology, earthquakes in Oklahoma can be linked to the granitic basement rock and to the Arbuckle Group, which is primarily composed of dolomite at earthquake depths. Here, we conduct friction experiments to measure healing rates of the two earthquake-bearing lithologies at confining pressures representative of earthquake depths and pore pressures ranging from 0% to 80% of the confining pressure. We measure frictional healing by executing slide-hold-slide tests with hold times ranging from 3 s to 3000 s. The friction experiments on the Troy Granite indicate that pore fluid pressure does not greatly affect healing rate. On the other hand, the dolomite of the Arbuckle Group exhibits decreased healing with increased pore fluid pressure, with weakening at the highest pore pressure. We hypothesize that this is due to an increase in dissolution of dolomite at high pore pressures/low effective normal stress. These healing rates are used in the companion paper to understand the moment-recurrence time behavior of repeating earthquakes in Prague, Oklahoma. This work has implications for possible enhanced dissolution and weakening behavior of the Arbuckle Group during wastewater injection activities.

Oklahoma↗
USGS2025

Scientific opportunities in the National Landscape Conservation System

The National Landscape Conservation System consists of unique and beautiful places across America’s landscapes where identified resources and values are protected and science is highlighted. The mission of the National Landscape Conservation System (NLCS), which is managed by the Bureau of Land Management and is often referred to as the agency’s National Conservation Lands, is to conserve, protect, and restore nationally significant landscapes for their cultural, ecological, and scientific values. This clear inclusion of science in the NLCS mission sets the stage for individual units to serve as places of learning, teaching, discovery, and innovation. Science is an integral part of managing the National Conservation Lands, and science conducted within and across the more than 900 units that make up the NLCS can inform and influence conservation and public land management well beyond its boundaries. Here, we highlight seven core aspects of National Conservation Lands that present valuable science opportunities: (1) the scientific values for which individual units are designated; (2) the many other resources, objects, and values within units; (3) the value of units as “control” sites for understanding the effects of activities such as mineral extraction that commonly occur elsewhere on multiple-use public lands but are often prohibited within National Conservation Lands; (4) the value of units for studying the effects of activities such as recreation that regularly occur and may be intensified on National Conservation Lands; (5) the high visibility of units, which draws strong interest and engagement from scientists, partners, and the public; (6) the functioning of the units as a network managed for a common purpose, which provides an opportunity to explore cross-cutting science questions across widely varying contexts and geographies; and (7) the opportunities units provide to promote and apply Indigenous Knowledge to scientific research to manage natural and cultural resources. Because of all of these characteristics, National Conservation Lands can serve as hubs for basic and applied science that can inform management of all public lands and resources into the future. We highlight these science opportunities through examples from existing units and suggest two actions that could help further science activities and impact on National Conservation Lands.

Parks Stewardship Forum↗
USGS2025

A soil velocity model for improved ground motion simulations in the U. S. Pacific Northwest

Near-surface seismic velocity structure may significantly impact the intensity, duration, and frequency content of ground shaking during an earthquake. In this study, we compile 649 shear wave velocity (Vs) profiles throughout the U.S. Pacific Northwest and southern British Columbia (PNW) and use these measured profiles to develop a representative soil velocity model for four major Holocene soil provinces: Puget Lowlands, Willamette Valley, fill and alluvium, and `other' soils. The resulting soil velocity model shows good agreement to measured data for a wide range of site conditions, with variability between different geologic domains reflecting fundamental differences in depositional environments. We then show that using this regional soil velocity model in simulations of the 2001 M6.8 Nisqually, Washington earthquake improves the fit to observed high-frequency (≥ 0.5 Hz) ground motions in the Puget Sound region compared to simulations that do not incorporate shallow (≤ 200 m) seismic velocity structure. Overall, this work shows that incorporating localized soil velocity profiles into seismic velocity models is important for accurately estimating high-frequency ground motion and regional seismic hazard in earthquake simulations. Future earthquake simulations and hazard studies in the PNW could incorporate these soil velocity profiles to capture the region's distinct site response characteristics.

Washington↗
USGS2025

Toward a new framework to evaluate process-based model configurations and quantify data worth prior to calibration

Model criticism, discrimination, and selection methods often rely on calibrated model outputs. Because calibration can be computationally expensive, model criticism can first be undertaken by assessing model outputs obtained from limited prior parameter ensembles. However, such prior-based methods are often heuristic and do not formalize the notion of balancing model consistency with data and model complexity (i.e., model adequacy). We present a new framework to discriminate among candidate models prior to calibration that formalizes prior-to-calibration model adequacy into a metric to implicitly balance prior model output data coverage with model complexity represented by prior output (co)variance. The prior model adequacy metric “Mahalanobis distance deviation” quantifies the deviation of (a) the set of squared Mahalanobis distances of data from a prior model output distribution from (b) the set of squared Mahalanobis distances of data from their own distribution. A new data worth metric “discernment value” is also presented which quantifies the value of data for screening less-adequate models prior to calibration. Discernment value is calculated from the change in variance of a weighted average of prior model outputs from all candidate models due to less-adequate model outputs receiving lower weight. The framework is demonstrated using a one-dimensional groundwater flow model with eight possible configurations. A synthetic data network is used to test the framework. Results show the framework identifies the candidate models most similar to the true model used to create the synthetic data. Discernment values show variation in the value of different data types and locations for screening less-adequate models.

Water Resources Research↗
USGS2025

Did they feel it? Legacy maroseismic data illuminates an engimatic 20th century earthquake

The challenges and the importance of preserving legacy instrumental records of earthquakes are now well-recognized (e.g., Richards & Hellweg, 2020, https://doi.org/10.1785/0220200053 ). Seismologists may not be aware of parallel challenges and opportunities with legacy macroseismic data for earthquakes in the United States. For much of the 20th century, macroseismic data were collected by a series of U.S. government agencies using a standard questionnaire distributed on postcards. Published summaries of postcards provide macroseismic data akin to modern Did You Feel It? questionnaire responses. In this paper we focus on the M 6.5 Fickle Hill, California earthquake, on 21 December 1954 (Hellweg et al., 2025) as a proof-of-concept, illustrating the potential of what we dub Did They Feel It? (DTFI) data to improve our understanding of significant 20th century U.S. earthquakes for which instrumental data are sparse. Legacy macroseismic data interpreted following modern conventions can potentially constrain traditional ShakeMaps at a level of detail and accuracy that in some respects rival maps for modern earthquakes. The updated ShakeMap for the 1954 Fickle Hill earthquake, also drawing from recently published media and first-person accounts, supports the location, depth, and stress drop value estimated from available instrumental data (Hellweg et al., 2025).

California↗
USGS2025

Mechanics and statistics of postseismic shaking

Analysis of two weeks of continuous post-seismic shaking after the 2019 M7.1 Ridgecrest, CA earthquake sequence using 4 nearby borehole seismometers reveals that continuous ground motions decay as Omori’s law in time and follow the Gutenberg-Richter distribution in logarithmic amplitude. The measured temporal decay in amplitudes agrees with predictions of the rate-and-state framework and indicates shaking amplitudes are proportional to the velocity of afterslip. Our ground motion-based statistical framework provides a basis to forecast shaking intensity in the minutes to hours after a large earthquake.

California↗
USGS2025

Constraining landslide frequency across the United States to inform county-level risk reduction

Informative landslide hazard estimates are needed to support landslide mitigation strategies to reduce landslide risk across the United States. Whereas existing national-scale landslide susceptibility products assess where landslides are likely to occur, they do not address how often , which is a critical element of landslide hazard and risk assessments. In particular, the U.S. Federal Emergency Management Agency's National Risk Index (NRI) requires landslide frequency estimates to inform expected annual loss estimates. We present county-level landslide frequency (landslides per area per year) estimates for the 50 US states. We applied Bayesian negative binomial regression to estimate both the expected (average) reported landslide frequency and full distribution of annual landslide counts for each county. We compared a suite of models that used combinations of landslide-susceptible area, probability of potentially triggering earthquakes, frequency of potentially triggering precipitation, and ecological region as predictors. We trained our models with landslide inventory data from counties with the most comprehensive records available nationwide and used zero-inflated negative binomial distributions as an incompleteness model to correct for temporal reporting gaps. We selected a preferred frequency model to inform the NRI based on information criteria and physically plausible parameter estimates. The model showed that average annual reported landslide frequencies vary by 5 orders of magnitude across US counties, ranging from 0.002 (0.00015–0.05) landslides 1000 km −2 yr −1 in Kusilvak Census Area, Alaska, to 29 (19–46) landslides 1000 km −2 yr −1 in Lake County, California, reflecting the country's strong variations in landslide susceptibility, earthquake probability, and other factors for which ecological region serves as a proxy. Counties with estimated frequencies in the top 20 % of all counties are predominately along the West Coast of the continental United States, in mountainous regions of the Pacific Northwest and Intermountain West, in locally steep or earthquake-prone regions of the Midwest and Southeast, along the Appalachians, in southern and southeastern Alaska, and on some Hawaiian islands. By examining the number of landslides predicted in 99th percentile years for each county, we identified that 26 % of US counties likely have potential for widespread landsliding with more than 10 landslides 1000 km −2 yr −1 , even when such large events have not been reported in the training data for that county. Overall, our results better represent the range of possible landslide frequencies and spatial variations than previous national-scale estimates reported in the NRI, and our approach can inform other risk-reduction and loss-mitigation efforts across the United States and globally.

Natural Hazards and Earth System Sciences↗
USGS2025

Understanding market sensitivity: Estimation of supply and demand elasticities for non-fuel minerals

In today’s rapidly changing economic landscape, understanding market responsiveness to price changes and the factors influencing commodity prices has become increasingly relevant. Price elasticities serve as indicators of how variations in market conditions affect supply and demand, providing insights into the sensitivity of commodity markets to price fluctuations. This paper presents a comprehensive analysis of price elasticities of supply and demand for 74 non-fuel mineral commodities including precious metals, base metals, minor metals, and industrial minerals that are utilized across various industries. We employ various econometric techniques, including fixed effects models for panel data and two-stage dynamic ordinary least squares (2S-DOLS) alongside autoregressive distributed lag (ARDL) models for time series analysis, to derive robust estimates of price elasticities. Our findings reveal variability in elasticities among different commodities and indicate that all studied mineral commodities exhibit price inelastic supply and demand in the short run, which we define as one year for the purposes of our analysis, given that the data is all annual. This research provides original estimates of price responsiveness for a wide range of commodities that have not been previously addressed in the literature, thereby enhancing the understanding of market dynamics in the mineral sector. Given that price elasticities can be influenced by factors such as market structure, technological advancements, mining costs, and industry-specific demand drivers, we use variables that serve as proxies for these factors.

Mineral Economics↗
USGS2025

An exploration of the relative influence of physical models for Omori’s law

Omori’s law states that the rate of aftershocks decays as a function of inverse time. There are multiple physical explanations that we reduce into a nonlinear mixed effects relation of three terms: (1) a Rate/State expression that can account for static/dynamic and viscoelastic triggering caused directly by the mainshock, (2) a fluid diffusion triggering term, and (3) a randomized secondary triggering (cascade) term. We fit free physical-model parameters to an observed aftershock sequence through two nonlinear regression methods to find the relative contributions of physics-based models in an observed aftershock sequence. Results from both methods show that Rate/State models overpredict aftershock rates by ∼0–30%. Secondary aftershocks cause a net negative contribution (seismicity rate reduction that corrects overprediction by other terms) ranging between ∼0 and 30%. All regression solutions yield negative secondary triggering contributions without being guided to do so. A physical explanation for this is that aftershock occurrence relieves stress from the crust, ultimately causing the sequence to extinguish itself. Fluid diffusion triggering contributions range from ∼0 to 20%. Diffusion processes are observed to be shorter in time than the full duration of an aftershock sequence and they are also spatially limited, diminishing their influence. Our results apply to an aftershock decay curve from the 2016 Central Apennines earthquake sequence, meaning that our specific results may not be general. Our primary conclusion is that any one physical model cannot alone fit the observed sequence as well as the combination of three we investigated.

Frontiers in Earth Science↗
USGS2025

Extremophile hotspots linked to containerized industrial waste dumping in a deep-sea basin

Decaying barrels on the seafloor linked to DDT contamination have raised concerns about the public health implications of decades old industrial waste dumped off the coast of Los Angeles. To explore their contents, we collected sediment cores perpendicular to five deep-sea barrels. The concentration of DDT and its breakdown products were highly elevated relative to control sites yet did not vary with distance from the barrels, suggesting that they were not associated with the contamination. Sediment cores collected through white halos surrounding three barrels were enriched in calcite and had elevated pH. The associated microbial communities were low diversity and dominated by alkalophilic bacteria with metagenome-assembled genomes adapted to high pH. A solid concretion sampled between a white halo and barrel was composed of brucite, a magnesium hydroxide mineral that forms at high pH. Based on these findings, we postulate that leakage of containerized alkaline waste triggered the formation of mineral concretions that are slowly dissolving and raising the pH of the surrounding sediment pore water. This selects for taxa adapted to extreme alkalinity and drives the precipitation of “anthropogenic” carbonates forming white halos, which serve as a visual identifier of barrels that contained alkaline waste. Remarkably, containerized alkaline waste discarded >50 years ago represents a persistent pollutant creating localized mineral formations and microbial communities that resemble those observed at some hydrothermal systems. These formations were observed at one-third of the visually identified barrels in the San Pedro Basin and have unforeseen, long-term consequences for benthic communities in the region.

California↗
USGS2025

Evaluating mass flow meter measurements from chambers for greenhouse gas emissions from orphan wells and other point sources

This study evaluates the performance of a rigid gas flux chamber equipped with a mass flow meter (MFM) for measuring gas emissions from leaking orphan wells and similar pressure-driven gas point sources. We conducted a series of laboratory and field experiments to evaluate the sensitivity, stability, and dynamic range of an MFM chamber system and found an optimal method for sealing the chamber to the ground to isolate the emission source. From these results, we estimate the effects of different soil gas permeabilities on measurements and identify the uncertainty of environmental processes that can impact measurements. Simulations of an MFM chamber are compared to those of a dynamic flux chamber to contrast the data derived with both methodologies and illustrate the potential for measuring high variability leaks with the MFM chamber. Using a low flow resistance MFM and a chamber well-sealed to the ground, it is possible to measure leaks down to 1.08 x 10 -3 cubic meters per hour (m 3 h −1 ) (refenced to 25°/1 atm), corresponding to 0.77 grams per hour (g h −1 ) methane or 2.11 g h −1 carbon dioxide, with a mean uncertainty of 0.89 % relative standard deviation. Environmental processes such as heated gas inside the chamber from solar gain, wind blowing across the chamber vent, and changing humidity in the chamber, can cause variation in MFM measurements. Over 11 d of continuous monitoring under varying weather conditions, the standard deviation of the environmentally sourced signals was found to be 7.40 x 10 -3 m 3 h −1 (equivalent to or 5.27 g h −1 methane or 14.45 g h −1 carbon dioxide). Strategies to obtain the highest quality data from MFM chambers include burying the edges of the chamber below the surface sufficiently deep to seal the chamber edges against gas flow and soaking the dirt with water to lower the chances of escaping gases, while monitoring the gas flow and adjusting the chamber seal to achieve a maximum flow rate.

Atmospheric Measurement Techniques↗
USGS2025

Speleothem evidence for Late Miocene extreme Arctic amplification – An analogue for near-future anthropogenic climate change?

The Miocene provides an excellent climatic analogue for near-future runaway anthropogenic warming, with atmospheric CO 2 concentrations and global average temperatures similar to those projected for the coming century under extreme-emissions scenarios. However, the magnitude of Miocene Arctic warming remains unclear due to the scarcity of reliable proxy data. Here we use stable oxygen isotope and trace element analyses, alongside clumped isotope and fluid inclusion palaeothermometry of speleothems to reconstruct palaeo-environmental conditions near the Siberian Arctic coast during the Tortonian (8.68 ± 0.09 Ma ). Stable oxygen isotope records suggest warmer-than-present temperatures. This is supported by temperature estimates based on clumped isotopes and fluid inclusions giving mean annual air temperatures between + 6.6 and + 11.1 °C , compared with − 12.3 °C today. Trace elements records reveal a highly seasonal hydrological environment. Our estimate of > 18 °C of Arctic warming supports the wider consensus of a warmer-than-present Miocene and provides a rare palaeo-analogue for future Arctic amplification under high-emissions scenarios. The reconstructed increase in mean surface temperature far exceeds temperatures projected in fully coupled global climate models, even under extreme-emissions scenarios. Given that climate models have consistently underestimated the extent of recent Arctic amplification, our proxy data suggest Arctic warming may exceed current projections.

Lena River delta↗
USGS2025

Total uncertainty quantification in inverse solutions with deep learning surrogate models

We propose an approximate Bayesian method for quantifying the total uncertainty in inverse partial differential equation (PDE) solutions obtained with machine learning surrogate models, including operator learning models. The proposed method accounts for uncertainty in the observations, PDE, and surrogate models. First, we use the surrogate model to formulate a minimization problem in the reduced space for the maximum a posteriori (MAP) inverse solution. Then, we randomize the MAP objective function and obtain samples of the posterior distribution by minimizing different realizations of the objective function. We test the proposed framework by comparing it with the iterative ensemble smoother and deep ensembling methods for a nonlinear diffusion equation with an unknown space-dependent diffusion coefficient. Among other applications, this equation describes the flow of groundwater in an unconfined aquifer. Depending on the training dataset and ensemble sizes, the proposed method provides similar or more descriptive posteriors of the parameters and states than the iterative ensemble smoother method. Deep ensembling underestimates uncertainty and provides less-informative posteriors than the other two methods. Our results show that, despite inherent uncertainty, surrogate models can be used for parameter and state estimation as an alternative to the inverse methods relying on (more accurate) numerical PDE solvers.

Journal of Computational Physics↗
Compare source metadata on this page
WorkPublishedSource identifierSource
Fluid inclusion constraints on the geometry of the magmatic plumbing system beneath Mauna Loa – Part 2: Xenoliths2025-09-2370271918usgs
A regional model comparison between MODPATH and MT3D of groundwater travel time distributions2025-09-2270272029usgs
A glimpse into the future of tectonic tremor monitoring2025-09-1770273507usgs
Updating regional‐scale geospatial liquefaction models with locally available geotechnical data2025-09-1770275721usgs
To heal or not to heal?: 2. The moment-recurrence time behavior of repeating earthquakes in the 2011 Prague, Oklahoma aftershock sequence is consistent with laboratory healing rates2025-09-1670271504usgs
To heal or not to heal?: 1. The effect of pore fluid pressure on the frictional healing behavior of lithologies in Oklahoma2025-09-1670277742usgs
Scientific opportunities in the National Landscape Conservation System2025-09-1570271683usgs
A soil velocity model for improved ground motion simulations in the U. S. Pacific Northwest2025-09-1570271695usgs
Toward a new framework to evaluate process-based model configurations and quantify data worth prior to calibration2025-09-1370273430usgs
Did they feel it? Legacy maroseismic data illuminates an engimatic 20th century earthquake2025-09-1270271911usgs
Mechanics and statistics of postseismic shaking2025-09-1270271954usgs
Constraining landslide frequency across the United States to inform county-level risk reduction2025-09-1170265227usgs
Understanding market sensitivity: Estimation of supply and demand elasticities for non-fuel minerals2025-09-1170271433usgs
An exploration of the relative influence of physical models for Omori’s law2025-09-1170271914usgs
Extremophile hotspots linked to containerized industrial waste dumping in a deep-sea basin2025-09-0970271364usgs
Evaluating mass flow meter measurements from chambers for greenhouse gas emissions from orphan wells and other point sources2025-09-0970271439usgs
Speleothem evidence for Late Miocene extreme Arctic amplification – An analogue for near-future anthropogenic climate change?2025-09-0870271463usgs
Total uncertainty quantification in inverse solutions with deep learning surrogate models2025-09-0870274693usgs

These are bibliographic comparisons, not experimental rankings. Follow the original document for methods and conditions.