Geology Reports⌕ Search

SEARCH · Geology Reports

Results for “Geological Society of London Special Publications”

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.

1,664 records · Page 81Linked to original sources
USGS2025

Do bighorn sheep use desert tortoise burrow spoil piles as mineral licks in southern California?

In a previous study, bighorn sheep ( Ovis canadensis ) were periodically photographed by trail cameras when they visited desert tortoise ( Gopherus agassizii ) burrows in the San Bernardino Mountains near Palm Springs, California, USA. The authors suggested that bighorn sheep may utilize the excavated spoil material from burrows as a mineral lick. To test that supposition, we collected soil samples from the spoil piles and undisturbed surface soil up to 1 m away at two desert tortoise study sites with bighorn sheep at the aforementioned site and another nearby site. We hypothesized that sodium concentrations would be greater in spoil than in nearby surface soil samples. Common major cations were quantified in soil samples using inductively coupled plasma-mass spectrometry following microwave-assisted digestion to solubilize the fraction of elements that could become biologically or environmentally available. Mean sodium concentrations were greater in burrow spoil samples than in undisturbed soil samples at both sites, likely due to aqueous leaching of sodium from surface to subsurface soil horizons. Although we did not collect evidence of geophagy by bighorn sheep, the behavior is well known. However, trail camera photographs of bighorn sheep with their head next to the entrance of tortoise burrows suggests that they may periodically utilize sodium and other elements brought to the surface by desert tortoises that are otherwise unavailable to bighorn sheep on the surface.

California↗
USGS2025

Earthquake stress-drop values delineate spatial variations in maximum shear stress in the Japanese forearc lithosphere

Earthquake stress drop (Δσ) may increase with depth and stress in the brittle lithosphere. However, the range of uncertainty in Δσ and the lack of constraints on absolute stress make it difficult to establish whether they are correlated. Here, we investigate Δσ dependence on depth and maximum shear stress ( τ max ) based on ~11 years of seismicity in the northeastern Japanese forearc following the 2011 Tohoku-Oki megathrust earthquake. We interpret Δσ estimates computed using both individual spectra and spectral-ratio methods and find that Δσ exhibits a clear depth dependence within the seismically active upper ~60 km of the forearc lithosphere ( ~ 0.8 MPa per 10 km). We further compare Δσ values with quantitative τ max estimates from finite-element models of force balance. We find that median Δσ values increase with τ max in the brittle forearc lithosphere and that earthquake stress release is proportional to τ max . The dependence of Δσ on τ max explains the apparent depth dependence of Δσ and suggests that average Δσ values provide a relative measure of the stress at failure. In the northeastern Japanese forearc, Δσ values remained roughly constant in the decade following the Tohoku-Oki earthquake, suggesting negligible changes in failure stress in the forearc since the mainshock.

Communications Earth and Environment↗
USGS2025

Assessing flood water infiltration and storage in a restored floodplain

In urban areas, floodplain restoration is gaining prominence as a strategy for restoring the natural functions of floodplain ecosystems and reducing flood risk. This has spurred research into potential interactions between floodwaters, the hyporheic zone, and the floodplain aquifer. An urban restored stream in Wisconsin, USA, was used as a case study to examine four methods to estimate floodplain infiltration and storage during overbank floods. We characterised flood-related infiltration over a 4-year period from 2018 through 2021 by simultaneously and continuously measuring groundwater levels and vertical temperature profiles with stream water levels linked to high-resolution flood inundation maps. High-resolution topographic data helped to quantify surface floodplain storage and the unsaturated soil volume relative to flood stage. Infiltration estimates from the simple methods align well with those from the more complex methods; however, the complex methods provide additional insights about the factors influencing infiltration. Results from all methods indicate that the volume of water that vertically infiltrates during floods is likely small relative to the total volume of the flood, with 0.08%–0.52% of flood water infiltrating into the floodplain, on average. Spatially variable vertical hydraulic gradients, driven by flood depth, groundwater level, and permeability, imply heterogeneous patterns of infiltration across the floodplain. Gradients favourable for infiltration typically occurred during the onset of flooding but, over the study period, were mostly (98% of the time) favourable for groundwater discharge to the channel (non-flood periods). These findings highlight the importance of considering surface-groundwater dynamics, floodplain soils, and unsaturated floodplain volume in defining the benefits of floodplain infiltration for flood attenuation.

Wisconsin↗
USGS2024

Self-guided decision support groundwater modelling with Python

The GMDSI tutorial notebooks repository provides learners with a comprehensive set of tutorials for self-guided training on decision-support groundwater modelling using Python-based tools. Although targeted at groundwater modelling, they are based around model-agnostic tools and readily transferable to other environmental modelling workflows. The tutorials are divided into three parts. The first covers fundamental theoretical concepts. These are intended as background reading for reference on an as-needed basis. Tutorials in the second part introduce learners to some of the core concepts parameter estimation in a groundwater modelling context, as well as providing a gentle introduction to the PEST, PEST++ and pyEMU software. Lastly, the third part demonstrates how to implement highly-parameterized applied decision-support modelling workflows. The tutorials aim to provide examples of both “how to use” the software as well as “how to think” about using the software. A key advantage to using notebooks in this context is that the workflows described run the same code as practitioners would run on a large-scale real- world application. Using a small synthetic model facilitates rapid progression through the workflow.

Journal of Open Source Education↗
USGS2024

Slope Unit Maker (SUMak): An efficient and parameter-free algorithm for delineating slope units to improve landslide modeling

Slope units are terrain partitions bounded by drainage and divide lines. In landslide modeling, including susceptibility modeling and event-specific modeling of landslide occurrence, slope units provide several advantages over gridded units, such as better capturing terrain geometry, improved incorporation of geospatial landslide-occurrence data in different formats (e.g., point and polygon), and better accommodating the varying data accuracy and precision in landslide inventories. However, the use of slope units in regional ( > 100 km 2 ) landslide studies remains limited due, in part, to the large computational costs and/or poor reproducibility with current delineation methods. We introduce a computationally efficient algorithm for the parameter-free delineation of slope units that leverages tools from within TauDEM and GRASS, using an R interface. The algorithm uses geomorphic laws to define the appropriate scaling of the slope units representative of hillslope processes, avoiding the often ambiguous determination of slope unit size. We then demonstrate how slope units enable more robust regional-scale landslide susceptibility and event-specific landslide occurrence maps.

Natural Hazards and Earth Systems Sciences (NHESS)↗
USGS2026

Family life is critical for migration and survival in a long-lived social bird

Extended parental care is common among birds and particularly well known among long-lived migratory species. Yet, factors influencing the duration of parental care are poorly understood, and empirical evidence for the functional importance of family life for offspring development is scarce. Using unique tracking and genetic data of 55 adult and juvenile migratory Bewick’s swans, among which 18 parent-offspring pairs and 13 sibling pairs, we quantify variation in family bond duration and identify the cause and consequences of their dissolution. Families generally broke up during the second half of the first northward migration of the offspring, when offspring were nearly one year old. Earlier separation occurred on the wintering grounds in response to disturbance ( n = 10 juveniles) and led to delayed migration and, critically, lower survival (40 vs. 86%). Our results provide rare empirical evidence that family bonds, as a pathway for social learning, are not just important, but vital to the migration and survival of young family-living birds.

Behavioral Ecology and Sociobiology↗
USGS2025

An early Holocene wet period in the southwestern United States

Multiple generations of spring-fed streams traversed ∼800 km 2 of the Las Vegas Valley in southern Nevada between ca. 10.9 ka and 8.5 ka, depositing an extensive tufa network. The scale of this network and diversity of tufa morphologies is novel in North America and offers an opportunity to obtain quantitative paleoclimate data for the region during the early Holocene. We determined isotopic compositions and estimated past temperatures using clumped isotope data from early Holocene tufa on the valley floor (698 m) as well as tufa forming today at higher elevation in the nearby Spring Mountains at Cold Creek Spring (1856 m). Modern and fossil tufa yielded comparably low δ 18 O values, implying that source waters for both were derived from high-elevation winter precipitation. Clumped isotope temperatures of modern tufa average 15.8 ± 2.5 °C, aligning with mean summer temperatures of the emergent spring water, and indicate equilibrium conditions of tufa formation. The early Holocene tufa yielded similar clumped isotope temperatures, averaging 15.2 ± 3.9 °C, meaning it precipitated at temperatures that occur at much higher elevations today. The Las Vegas tufa record, combined with nearby and temporally correlative paleospring and lacustrine records, suggest that cool/wet conditions prevailed throughout the Mojave Desert during the early Holocene. These records also demonstrate that spring ecosystems responded to millennial-scale hydroclimate variations that supersede climate change driven solely by insolation. The previously unrecognized pattern of ecosystem response to hydroclimate documented here may assist in understanding climate drivers for the early Holocene and provide critical information for the fate of groundwater-dependent ecosystems in the southwestern United States.

Nevada↗
USGS2026

Earthquake ground-motion model adjustments for the San Francisco Bay area

We develop adjustments to ergodic ground‐motion models (GMMs) to improve their performance in the San Francisco Bay Area (SFBA). GMMs are widely used in hazard assessments to estimate characteristics of ground shaking based on known properties of the source, path, and site. Such models are often developed using datasets containing records from various regions, resulting in models that represent median ground‐motion behavior, which may not adequately represent ground motions within subregions. This is true for the SFBA, where ground motions attenuate more rapidly with distance than in many other parts of California that dominate GMM databases. To support improved seismic hazard estimates in the SFBA, we calculate regional constants and anelastic attenuation coefficient adjustments relative to two commonly used ergodic GMMs: BSSA14 ( Boore et al. , 2014 ) and ASK14 ( Abrahamson et al. , 2014 ). These adjustments are obtained for a suite of ground‐motion intensity measures (peak ground acceleration, peak ground velocity, and 5%‐damped pseudospectral acceleration at oscillator periods ranging from 0.075 to 10 s) using mixed‐effects regression. Use of the regionally adjusted models reduces the overall bias by up to 0.5 natural log units for BSSA14 and up to 0.6 natural log units for ASK14. We demonstrate one application of our attenuation adjustments and their implications in an earthquake early warning case study of the 2014 M 6.0 South Napa earthquake. The predicted extent of shaking using the adjusted models better matches observed shaking at large source‐to‐site distances, especially for lower shaking intensities, thus potentially reducing overalerting. We encourage the use of our model adjustments when ergodic models are considered for seismic hazard studies in the SFBA.

California↗
USGS2026

King Rails in southeastern Arkansas are majority resident

King Rails ( Rallus elegans ), a declining species of secretive marsh bird, have a mixed migratory strategy, with some individuals migrating and some being year-round residents. Determining the distribution of migratory versus resident King Rails is critical to guiding habitat management and assessing risks to populations. We examined the migratory status of adult King Rails at the Choctaw West Wildlife Management Area in southeastern Arkansas, the only known site in Arkansas with multiple breeding pairs. Of the 23 adult King Rails we tagged in 2022 and 2023, two birds migrated away from the local area, one in each year. One migrated 145 km, and the other 450 km. Our findings suggest that the line between the migratory and resident range of Arkansas King Rails could be north of our site.

Arkansas↗
USGS2024

Potentially toxic elements in wild Agassiz’s desert tortoises: Tissue concentrations and association with disease

Background: Desert tortoise ( Gopherus agassizii ) populations have continued to decline due to infectious and other diseases, predation, and habitat alteration. The potential contribution of minerals and heavy metals to tortoise health and susceptibility to disease remains uncertain. Objective: The objective of this study was to evaluate the results of elemental analysis of trace minerals and macrominerals in scute keratin, kidney, and liver tissue from ill and dying desert tortoises salvaged for necropsy between 1993 and 2000. Methods: Salvaged tortoises were categorized by age (adult, juvenile), geographic location, and primary disease based on necropsy findings. A subset of tortoises that were injured or killed by vehicular trauma or predation but with no notable pathologic abnormalities were used as controls. A panel of 21 trace minerals and 6 macrominerals was analyzed in scute keratin, kidney, and liver tissue samples by inductively-coupled plasma spectrometry and atomic absorption spectrophotometry. Results: Necropsies were done on 46 tortoises, including 9 juveniles salvaged from 5 regions in the Colorado and Mojave deserts of California. Primary diseases were cutaneous dyskeratosis (n=9), infection/ inflammation (n=8), malnutrition (n=7), mycoplasmosis (n=5), and urolithiasis (n=3); 14 tortoises were classified as controls. Concentrations of elements differed significantly by tissue, age, desert region, and disease (P < 0.05). Tortoises with cutaneous dyskeratosis had significantly higher Se concentrations, primarily in keratin and liver, than tortoises with other diseases (P < 0.001). Juveniles were more likely than adults to have high Pb, Sn, and Zn levels (P < 0.05). All tortoises had detectable levels of more than one potentially toxic heavy metal, including As, Cd, Cr, Hg, Ni, Pb, Sn and V. Conclusions: Potentially toxic elements are frequently found in tissue from tortoises in desert regions of California, with significantly higher concentrations in diseased tortoises. Metal exposure from soils, mining, historic and ongoing military activities, and other human activities could increase susceptibility to disease in desert tortoises.

California↗
USGS2025

Tracing metal sources and groundwater flow paths in the Upper Animas River watershed using rare earth elements and stable isotopes

Groundwater flow paths and processes that govern metal mobility and transport are difficult to characterize in mountainous bedrock watersheds. Despite the difficulty in holistic characterization, conceptual understanding of subsurface hydrologic and geochemical processes is key to developing remediation plans for locations affected by acid mine drainage, such as the Upper Animas River watershed in southwestern Colorado, USA. Stable isotopes of water and rare earth elements were utilized to evaluate groundwater flow and metal sources within this complex catchment. Stable isotope samples collected from draining mine adits and springs display systematic spatial variation wherein sample sites at higher elevations have greater seasonal variability than sites at lower elevations. The Upper Cement Creek watershed, where multiple draining mines are present, displays the lowest seasonal variation in stable isotopic signatures, potentially indicating the presence of a large, well-mixed volume of groundwater storage or interbasin groundwater flow. Rare earth elements display statistically significant variation between different alteration styles in the catchment. Overprinting of regional propylitic alteration is evident based on enrichment of middle rare earth elements in acidic springs and mines that are not spatially associated with surficial exposures of acid generating alteration styles. Europium anomaly and middle rare earth enrichment signatures from two flooded mine tunnels on opposite sides of a watershed divide indicate connections to the same subsurface flooded mine workings.

Colorado↗
USGS1993

Ground-water quality near an inactive landfill and sludge-spreading area, Tallahassee, Florida

Ground-water quality of a 120-acre area encompassing an inactive landfill southwest of Tallahassee, Florida, was assessed by installing and sampling 21 monitoring wells. Sludge from a municipal sewage-treatment plant was also applied to the land surface near the landfill site. Water samples were collected from the 21 monitoring wells from June 1987 through September 1990 and analyzed for concentrations of major ions, nitrogen species, trace metals, field measurements, and more than 140 organic constituents. Water quality in the Upper Floridan aquifer at the study site was compared to background water quality in other areas of Leon County. Ground-water quality of the study site was evaluated by the location of wells sampled in relation to the landfill and the sludge-spreading areas. Results of nonparametric statistical analysis indicated that potassium and nitrite plus nitrate as nitrogen concentrations in water samples from wells completed in the Upper Floridan aquifer at the study site were significantly different at the 5 percent probability level from those in samples from background wells completed in this aquifer in Leon County. Median potassium concentrations were 0.4 mg/L (milligrams per liter) in water from wells at the study site and 0.7 mg/L in water from background wells in Leon County. Median nitrite plus nitrate concentrations were 6.48 mg/L in water samples from wells at the study site and 0.51 mg/L in water samples from background wells. Graphical comparison of the water quality characteristics in water from six categories of wells; upgradient, landfill, adjacent to the landfill, downgradient onsite, downgradient offsite, and background Leon County, indicated that sodium, bicarbonate, sulfate, iron, manganese, and dissolved solids concentrations and specific conductance values were highest within the landfill. Upgradient and downgradient well categories include wells in areas affected by sludge spreading. Localized reducing conditions were indicated in a downgradient area by median iron concentrations of 930 and 30 ug/L (micrograms per liter) in samples from wells only 5 feet apart. Nitrite plus nitrate concentrations were lowest in water from wells in the landfill compared to water from the other categories. Concentrations of most trace elements and organic compounds were mostly below detection limits, although State maximum contaminant levels of 1 ug/L for benzene and vinyl chloride and 3 ug/L for tetrachloroethene were exceeded in 14 samples. Concentrations of sodium, potassium, chloride, sulfate, nitrite plus nitrate, total organic and ammonia nitrogen, and ammonia nitrogen in ground water were statistically analyzed in areas affected by sludge-spreading and other areas. Only nitrite plus nitrate and chloride concentrations were significantly different between the two areas. Median nitrite plus nitrate and chloride concentrations in sludge-spreading areas were 6.9 mg/L and 2.9 mg/L, respectively, compared to 1.1 mg/L and 1.8 mg/L in other areas.

Florida↗
USGS2025

Water supply in the conterminous United States, Alaska, Hawaii, and Puerto Rico, water years 2010–20

We present an assessment of water supply across the conterminous United States (CONUS), Alaska, Hawaii, and Puerto Rico covering water years 2010–20. Our analysis drew on two national hydrologic models, the National Hydrologic Model Precipitation-Runoff Modeling System and the Weather Research and Forecasting model hydrologic modeling system. Both models produced estimates of streamflow, evapotranspiration, soil moisture, snow water equivalent, and other hydrologic states and fluxes. The models were driven by the bias-adjusted 4-kilometer-resolution, long-term regional hydroclimate simulation over the conterminous United States dataset (CONUS404). We assessed spatial and temporal error distributions by comparing monthly simulations at the 12-digit hydrologic unit code and regional scale from both models against external benchmarking datasets. Results showed that average annual rainfall across the CONUS was 857 millimeters per year for the period of analysis, with water year 2012 the driest year (729 millimeters) and water year 2019 the wettest year (995 millimeters). Key interannual variability results included the following: (1) the California–Nevada hydrologic region had the highest variability in precipitation and snow accumulation, and (2) the Texas hydrologic region was among hydrologic regions with the highest variability in precipitation. We related interannual variability in precipitation to storage volumes in soil moisture, snow water equivalent, and lakes and reservoirs to highlight areas with little storage and large year-to-year variability in precipitation. These areas included the Southern High Plains, Central High Plains, Texas, Souris–Red–Rainy, Mississippi Embayment, and Midwest regions. Our analysis of groundwater-level data showed that several of these areas overlap aquifers where groundwater levels were considerably lower than historical averages, including the Colorado Plateaus aquifers, the Rio Grande aquifer system, and the Central and Southern regions of the High Plains aquifer. Many of these lowered groundwater levels are continuations of decades-long declines from overpumping that started well before the assessment period. The resulting water budgets and their analyses provide a high-resolution foundational assessment of the mean state and variability of the terrestrial hydrologic cycle across the CONUS and Alaska, Hawaii, and Puerto Rico to support a wide range of water resource management applications.

Professional Paper↗
USGS2025

Spatiotemporal variations in strain release and seismic rupture in multifault systems: An example from Panamint Valley, southeastern California

Geometrically complex, multifault ruptures have been observed in recent, damaging earthquakes in southeastern California, sparking renewed efforts to identify physical conditions that promote or inhibit fault discontinuity-spanning coseismic ruptures. The likelihood of ruptures propagating across fault discontinuities is thought to be partly controlled by fault geometries, rupture direction, and the history of strain release. However, these parameters vary in space and time over multiple earthquake cycles, making it difficult to forecast the likelihood that an earthquake on one fault will trigger rupture on a nearby fault. Here we use tectono-geomorphic mapping of a geometrically complex fault zone in Panamint Valley, southeastern California, to assess spatiotemporal variations of paleo-rupture patterns and geometries of fault discontinuities over multiple earthquake cycles. First, we identify ten generations of late Pleistocene to Holocene alluvium using geomorphic parameters and luminescence dating to constrain ages of alluvium and bracket late Holocene earthquake timing. Then, we quantify slip kinematics using high-resolution structure from motion digital surface models. We find the Panamint Valley transtensional relay (PVTR) hosted four late Holocene earthquakes, bracketed to ~5.8–3.4 ka, ~3.8–2.2 ka, ~2.4–0.6 ka, and ~0.64–0.16 ka, with ~0.6–1.1 m of slip per event, correlative to Mw ≈ 6.7–6.9 earthquakes. Additionally, we find similarities in earthquake timing on the Ash Hill, PVTR, and Panamint Valley faults and similarities in the slip magnitude and slip kinematics between the Ash Hill and PVTR faults, implying that the PVTR may co-rupture with nearby faults. Paleo-rupture patterns indicate that seismogenic strain transfer may occur through the PVTR, along different combinations of fault segments and jump distances, over multiple earthquake cycles. These data highlight the utility of tectono-geomorphic mapping in evaluating paleo-rupture patterns and suggest that the PVTR may act to propagate and/or arrest rupture between the Ash Hill and Panamint Valley faults.

California↗
USGS2025

Constraining large magnitude event source and path effects using ground motion simulations

The purpose of this study is to use ground motion simulations to investigate ways in which source and path effects for large magnitude events can be represented in non-ergodic GMMs. While we initially developed computation techniques using CyberShake simulations, the range of magnitudes and source-site combinations is not adequate to replicate what is observed empirically. We therefore designed a new ground motion simulation study, which includes earthquakes with a large range of magnitudes distributed uniformly on a fault plane, and sites covering a large range of rupture distances and azimuths. After running a large suite of simulations (M4-M7), we then develop a non-ergodic GMM with the simulation data. We find that the within-site residuals are dominated by the radiation pattern, rupture directivity, and slip patterns. Next, we modify an existing rupture directivity model to fit and remove the observed radiation pattern and rupture directivity from the residuals. We also minimize the contributions of slip patterns by averaging the within-site residuals among multiple source realizations. Finally, after removing the source effects from the within-site residuals, we compare the path effects computed with different magnitude groups using two approaches. The first approach only considers the small events that have the same shortest path to a site with the large events, while the second approach considers all small events on the fault plane. The results indicate that the path effects of large events cannot be satisfactorily approximated with that of small events using either approach.

Conference Paper↗
USGS2025

Fiber-imaged supershear dynamics in the 2024 Mw 7 Mendocino Fault earthquake

Fault structure and rupture physics are deeply intertwined, and observations of this coupling are critical for understanding earthquake behavior. Rupture propagation is observable at fine scales using dense seismic networks. Fiber-optic sensing allows for long-term deployments of ultradense arrays that enable high-resolution measurements of infrequent, large earthquakes. We recorded the 2024 moment magnitude ( M w ) 7 Mendocino Fault earthquake with a nearby fiber-optic array and imaged its behavior with seismic beamforming. The rupture propagated to the east at subshear velocity; stagnated near the Mendocino Triple Junction, a zone of structural complexity; and subsequently transitioned to supershear velocity. The correlation between source physics and structure shows how lithospheric heterogeneity affects first-order characteristics of earthquake ruptures. Our results also demonstrate the potential for fiber-optic sensing to improve real-time estimation of key parameters for early warning.

California↗
USGS2025

Overcoming the data limitations in landslide susceptibility modelling

Data-driven models widely used for assessing landslide susceptibility are severely limited by the landslide and environmental data needed to create them. They rely on inventories of past landslide locations, which are difficult to collect and often nonrepresentative. Furthermore, susceptibility maps are most needed in regions without the means to assemble an inventory. To overcome these challenges, we develop a method for assessing shallow landslide susceptibility based on a probabilistic morphometric analysis of the landscape’s topography, rather than the characteristics of landslides. The model assumes that hillslopes with higher relief and gradient compared to the surrounding landscape are more prone to landslides. We demonstrate the superior performance of this approach over contrasting data-driven models across the northwestern United States. As our morphometric model only requires elevation data, it overcomes the major limitations of data-driven models and facilitates the creation of effective susceptibility models in areas where it was previously unfeasible.

Oregon, Washington↗
USGS2025

Divide and conquer: Separating the two probabilities in seismic phase picking

There are two fundamental probabilities in the seismic phase picking process – the probability of the existence of a seismic phase (detection probability) and the probability of correctly identifying the phase arrival time (timing probability). The nearly ubiquitous approach in developing deep learning phase picking models is to use a kernel, such as a truncated Gaussian, to mask the labeled phase arrival time, and train a segmentation model. Once a model is trained, the times of the peaks in the output are taken as phase arrival times (picks) and the height of the peaks are taken as “probability” of the picks. Here, we show that this “probability” represents neither the detection nor the timing probabilty because this approach forces the output to follow the shape of the kernel. We introduce an approach using two models to estimate these two distinct probabilities. We use a binary classifier with a calibrated confidence to address the detection probability and a multi-class classifier to obtain a probability mass function to address the timing probability. This new approach makes the deep learning-based phase picking process more interpretable and gives us options to logically control seismic monitoring workflows.

Geophysical Journal International↗