Search USGS⌕ Search

SEARCH · Search USGS

Results for “JGR Machine Learning and Computation”

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.

1,657 records · Page 5Linked to original sources
USGS2024

An orbital comparison of a late mantling unit on Aeolis Mons with other erosion-resistant strata explored by MSL in Gale crater, Mars

Previous investigations along Curiosity 's traverse in Gale crater have explored the relationship between orbital and in situ observations. This work aims to better understand the geologic environment of units only observable from orbit and compare them to the properties of units examined by Curiosity . Here, we map an erosion-resistant dark-toned mantling unit that overlies the modern topography of Aeolis Mons (informally known as Mt. Sharp) and compare this unit to two other previously mapped dark-toned resistant units, the marker band and the mound skirting unit (MSU), that have been inferred to represent different geologic environments (lacustrine and aeolian, respectively). Visible to short wave infrared spectra from the Compact Reconnaissance Imaging Spectrometer for Mars and visual images from the High Resolution Imaging Science Experiment and Context Cameras aboard the Mars Reconnaissance Orbiter are used for this comparison. Spectral data suggest a mafic composition with minor alteration, although the composition varies more with location around Mt. Sharp rather than between units. Morphologically, the mantling unit has strong similarities to the marker band based on their consistent low-albedo, erosion-resistance, and smooth appearance, contrasting with the highly variable surface texture of the MSU. We hypothesize that all three units had a similar sediment source but experienced aqueous alteration at different times: early ubiquitous cementation in a surface aqueous environment in the mantling unit and marker band versus patchy late diagenesis in the MSU. If true, these results suggest that water activity continued within the Gale crater long after the erosion of Mt. Sharp.

JGR Planets↗
USGS2026

Structural evolution and slip rate variations through time of the Puente-Hills blind-thrust fault beneath Los Angeles: Implications for seismic hazard and folding kinematics

Using seismic reflection profiles, historical well logging data, and luminescence and radiocarbon ages, we determine a Pleistocene-Holocene slip history for the central, Santa Fe Springs segment of the Puente Hills blind-thrust fault (PHT), a major seismogenic fault situated beneath the urbanized Los Angeles metropolitan region. We analyze the geometry of correlative stratigraphic units in the forelimb and backlimb of the overlying growth-fold, the Santa Fe Springs anticline, and determine the uplift of seven age-correlative markers. Uplift measurements are converted to thrust displacements on the underlying PHT using a structural method laid out by Don et al. (2022, https://doi.org/10.1785/0120220048 ) that accounts for the geometry of the fault. These data indicate that deep thrust displacement on the PHT is partially consumed updip in the creation of a hanging-wall fault-bend fold, with forelimb growth strata recording <80% of the slip documented within the backlimb. Chronological data from growth strata yield age constraints for folding and faulting on the underlying PHT, providing a detailed incremental slip history derived from both the forelimb and backlimb folding for seven discrete growth horizons, spanning the past 1.4 million years. The resulting six incremental slip rates demonstrate that fault slip has varied through time from the middle Pleistocene to the Holocene. Moreover, these results reveal synchronous acceleration of both the central, Santa Fe Springs and western, Los Angeles segments of the fault system since late Pleistocene time (after 200 ka) and slip rates of greater than 2 mm/yr on the downdip, backlimb, fault-ramp below the anticline.

California↗
USGS2025

Evidence of mineral alteration in a salt marsh subterranean estuary: Implications for carbon and trace element cycling

Subterranean estuaries (STE) in salt marshes are biogeochemically active zones where interactions between terrestrial groundwater and seawater drive complex cycling of carbon and trace elements, influenced by mineral dissolution. These systems, characterized by fine-grained organic-rich peat overlying permeable coastal aquifers, play a crucial role as a blue carbon sink, yet their geochemical dynamics remain poorly understood. We investigated dissolved trace elements, carbon, silica, and radium isotopes in a salt marsh STE (Sage Lot Pond, Waquoit Bay, MA) over seasonal and annual cycles. Our results reveal that groundwater and estuarine water circulation through marsh peat and aquifer sediments leads to enrichments of dissolved organic and inorganic carbon (DOC and DIC), Si, Ba, and Mn, with variable source/sink behavior of Fe and net removal of U. Submarine groundwater discharge dominated Ba fluxes, whereas pore water drainage from marsh peat acted as the main sink for U and source of Si. Fe cycling was variable, with terrestrial Fe largely removed as groundwater passed through the STE, consistent with Fe-sulfide and amorphous phase formation. Radium isotope ratios identified two distinct subsurface flow pathways, influenced by metal-oxide cycling and organic matter breakdown. Si production was decoupled from DIC, suggesting Si originates from mineral alteration, whereas DIC results from both mineral weathering and microbial respiration. Silicate mineral alteration, coupled with marsh pore water drainage, accounts for up to 16% of annual DIC exports (66 g C m −2 y −1 ), highlighting the importance of STEs in coastal carbon and trace element cycling, especially as marshes face environmental change.

Massachusetts↗
USGS2025

Beyond the mangroves: A global synthesis of tidal forested wetland types, drivers and future information opportunities

There is increasing awareness of the global diversity of tidal forested wetlands (TFWs) and their significance in the provision of ecosystem services. These ecosystems, including mangrove forests, tidal freshwater forested wetlands, supratidal forests and transitional forests together span multiple climatic zones, geomorphic settings, and inundation and salinity regimes. We utilise case studies across five continents to demonstrate the state of knowledge among TFWs. Intertidal mangroves are the best-defined of the TFWs thanks to decades of research on their geomorphology, hydrology and ecology across their broad distribution. Non-mangrove forest settings, however, demonstrate more diverse hydrological, biochemical and vegetation conditions. In many cases, non-mangrove forests are situated at upper intertidal or supratidal elevations, where surface waters and groundwater are subject to interactions between tides freshwater inputs. Salinity datasets show variations ranging from tidal freshwater forested wetlands and ‘low-salinity mangroves’ to mesohaline or marine salinities, often with high temporal variability. While the floristic composition of non-mangrove forests vary among biogeographic regions, locally dominant TFW species are commonly distributed beyond the tidal niche into non-tidal wetland and upland forests. This presents challenges for traditional remote sensing approaches to ecosystem mapping, which are mostly lacking for non-mangrove forests. Geomorphic approaches and developments in machine learning offer opportunities to address this.

Preprint↗
USGS2025

The δ13C signature of dissolved organic and inorganic carbon reveals complex carbon transformations within a salt marsh

Coastal wetlands have high rates of atmospheric CO 2 uptake, which is subsequently respired back to the atmosphere, stored as organic matter within flooded, anoxic soils, or exported to the coastal ocean. Transformation of fixed carbon occurs through a variety of subsurface aerobic and anaerobic microbial processes, and results in a large inventory of dissolved carbon. Carbon source and the roles of aerobic respiration, sulfate reduction, and methane cycling were evaluated within salt marsh peat and the underlying sandy subterranean estuary. There is a large increase in dissolved inorganic carbon (DIC, 7,350 ± 3,900 μmol L −1 ), dissolved organic carbon (DOC, 1,040 ± 1,480 μmol L −1 ) and CH 4 (14.5 ± 33.3 μmol L −1 ) within the marsh porewaters compared to creek waters. Alkalinity production (5,730 ± 2,170 μeq L −1 ) and sulfate removal (1,810 ± 1,970 μmol L −1 ) indicate anaerobic respiration, however, relative contributions from the various decomposition pathways cannot be identified due to overlapping geochemical signatures. The δ 13 C of the DOC (−29.0 ± 3.7‰) and DIC (−11.2 ± 1.1‰) produced within the marsh differed from the bulk soil organic matter δ 13 C (−14.5 ± 0.2‰). We explore a variety of mechanisms that could result in co-occurring depleted δ 13 C-DOC and enriched δ 13 C-DIC compared to the bulk soil organic carbon pool and salt marsh vegetation, including selective mineralization, production of δ 13 C-depleted bacterial biomass, and methane-derived DOC. While important questions remain about carbon cycling pathways, we found evidence of a cryptic methane cycle. Alteration of the δ 13 C of carbon species complicates source attribution in solid and dissolved phases and careful consideration should be used when carbon is partitioned between in situ salt marsh production and external marine and terrestrial sources.

Massachusetts↗
USGS2026

A roadmap for identifying and interpreting physical processes and national water model prediction bias associated with baseflow index regimes across the contiguous United States

Understanding how groundwater–surface water interactions shape streamflow variability is critical for diagnosing low flow behavior and prediction bias in continental scale hydrologic models. We present a process informed framework that links observed baseflow (BF) dynamics, watershed attributes, and National Water Model (NWM) performance across the contiguous United States. Using daily observed streamflow from 797 reference quality streamgages, we developed monthly baseflow index (BFI) signatures using a streamgage specific, calibrated digital filter. Hierarchical clustering of these signatures identified seven distinct BFI regimes capturing regional and seasonal variability. We evaluated NWM v3.0 retrospective streamflow performance within each regime using multiple hydrograph and flow duration curve-based metrics. Model skill varied systematically across regimes: mixed flow systems were simulated most accurately, while predominantly BF dominated and quickflow dominated regimes exhibited substantially poorer performance. Across nearly all regimes, the NWM underestimated observed BFI magnitude and frequently failed to reproduce seasonal BF patterns, indicating systematic biases in simulated low flow contributions. To relate these regimes to potential process controls, we trained a Random Forest classifier using static watershed attributes and applied Shapley Additive Explanations to identify features most strongly associated with each regime. Results highlight regionally varying influences, including the dominant role of snow fraction and seasonal runoff timing in snow dominated basins and the importance of evapotranspiration and aridity in quickflow dominated systems. Collectively, these findings demonstrate how hydrologic signatures combined with interpretable machine learning can diagnose regime specific model biases and generate process-based hypotheses about limitations in large scale hydrologic prediction systems.

contiguous United States↗
USGS2025

Favorability mapping for hydrothermal power resource assessments of the Great Basin, USA

The U.S. Geological Survey (USGS) is updating the 2008 assessment of conventional hydrothermal resources for the Great Basin in the western United States. As part of this work, the workflow for hydrothermal resource favorability maps is being modified to integrate modern data-driven machine learning (ML) methods. Improvements include: [1] using new and refined evidence layers (features); [2] using an order of magnitude more training sites (labeled examples); [3] utilizing simple but non-linear supervised ML algorithms; [4] representing positive training sites (wells with measured heat flow) with their ordinal value proportional to the magnitude of convective upflow (i.e., low, high, or very high convective signals instead of past strategies using positive-negative labels); [5] supplementing training sites with additional sites with low convective signals to represent diverse under-sampled areas where hydrothermal systems are unlikely to exist; [6] comparing with competing approaches; and [7] utilizing Monte Carlo cross-validation to estimate and evaluate prediction uncertainty. For the new favorability map, over half of the power-producing systems (i.e., 15 of 28) are predicted in the 99th percentile of most favorable locations (i.e., the highest 1 % of favorability, corresponding to 1 % of the map area), exceeding the performance of past models that have explicitly used power plants as training sites. Previous favorability maps predicted approximately half of the power-producing hydrothermal systems above the 80th percentile (i.e., 20 % of the map area). For the new favorability map, 93 % of power-producing systems (i.e., 26 of 28) are above the 80th percentile. The power-producing systems for which the new model does not perform well are either comparatively small, low-temperature systems or systems also not predicted well by prior modeling approaches, suggesting that these few systems are unusual when compared with most power-producing systems. Focusing research on these known, seemingly different systems may yield new insights and subsequent discovery of new prospects.

California, Idaho, Nevada, Oregon, Utah↗
USGS2026

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

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

Washington↗
USGS2025

Case study of deep learning image segmentation for the purposes of rapid 2D petrographic analysis in volcanic rocks

Automation using deep learning methods is a useful alternative to manual methods of petrographic segmentation, but often requires user familiarity with coding and/or algorithms. We examine the Dragonfly TM program's deep learning tools for application by users with a variety of skill levels as a method for petrographic image segmentation. An image processing methodology, bimodal image stacking, was created for low-input-data, high-efficacy training of models which can then be applied to varied samples. Using backscatter electron images we show that the resulting model segmentations agree with manual segmentation total and modal crystallinity values within 5%, and calculated plagioclase crystal size distribution (CSD) values within 2σ, despite limitations in discriminating mafic phases. Model creation and training takes <24 hours, 1–3 hours of which are supervised, and the resultant model can then be applied to new uncharacterized samples in <15 minutes per image. This allows for non-experts to create and utilize deep learning models to segment images of variable brightness and texture, at low user-time cost and resulting in size and shape data which are within uncertainty of manual segmentation. While some limitations are noted (for example, sieve-textured phases may need manual correction, and different minerals with similar BSE intensity may not be resolved as separate phases), this methodology can be utilized for general application of models to wide ranges of volcanic crystalline and bubble textures, and to create a library of models for rapid petrological analysis during volcanic eruptions.

Alaska↗
USGS2024

Stream nitrate dynamics driven primarily by discharge and watershed physical and soil characteristics at intensively monitored sites: Insights from deep learning

We developed a suite of models using deep learning to make hindcast predictions of the 7‐day average backward‐looking nitrate concentration at 46 predominantly agricultural sites across the midwestern and eastern United States. The models used daily observations of discharge and meteorological variables and watershed attributes describing anthropogenic modification to hydrology, nitrogen application, climate, groundwater, land use, watershed physiographic attributes, and soils. Across all sites, discharge and watershed soil and physiographic attributes showed a strong influence on model performance. Analysis of drivers across sites revealed considerable regional differences related to controlling processes such as groundwater contributions. We tested several ways to pool data across sites to develop accurate models and make the most effective use of available data. Single‐site models, in which models are trained and tested at a single location, showed generally strong predictive performance (median Kling‐Gupta Efficiency = 0.66), and accuracy at poorly performing sites could be improved by grouping sites with similar characteristics. Developing a single model for all sites reduced performance at several locations with distinct characteristics, suggesting that there is a threshold of dissimilarity beyond which more data does not improve the model. While many deep learning studies have shown that national or even global models can outperform local models, it is not clear that this is true for water quality constituents. This study demonstrates how data can be combined effectively, using deep learning to develop accurate and interpretable models of instream nitrate at sites where varying processes are responsible for changes in nitrate concentration.

Water Resources Research↗
USGS2025

Groundwater structures fish growth and production across a riverscape

Landscapes are composed of habitat patches and conditions that vary across space and time. While habitat variability and complexity can support important ecological processes and ecosystem services, the dynamic nature of habitats can also constrain organismal growth and production as optimal conditions are fleeting. In riverine ecosystems, groundwater discharge to streams stabilises water temperature and flow regimes, thus mediating how habitat complexity is expressed. Yet, how stable habitats structure growth and production within the broader landscape matrix is not well understood. In this study, we explored the effects of groundwater on spatiotemporal variation in growth and production for juvenile Yellowstone cutthroat trout ( Oncorhynchus virginalis bouvieri ) across the upper Snake River catchment, Wyoming, USA. We combined machine learning techniques and remotely sensed landscape data to estimate groundwater availability across the river network, which we linked to stream temperature regimes and conspecific density. We then used Bayesian hierarchical models to quantify the effects of temperature, density and groundwater on spatiotemporal variation in fish growth and production in 52 focal reaches. Finally, we predicted body size trajectories and trends in total production continuously over both space and time to understand the effect of groundwater at the riverscape scale. Groundwater discharged to streams where topography changes abruptly in valley-bottom areas underlain by coarse glacial deposits. Groundwater stabilised temperature regimes and was associated with high trout densities. Temperature and density, in turn, interacted to influence growth rates: growth increased strongly with temperature, but this effect was reduced when density was high. Accordingly, variation in groundwater availability among stream reaches diversified growth and production regimes. In reaches with low groundwater availability, growth and production declined over time from summer maxima. In contrast, in reaches with high groundwater availability, temporal trends in growth and production were hump-shaped—peaking in autumn—and mean production was greater. At the riverscape scale, temporal asynchrony in growth rates generated convergent spatial variation in growth capacity, but—when combined with density—led to the formation of distinct hotspots of production. Our results demonstrate how groundwater, an important driver of aquatic ecosystem heterogeneity, structures trout growth and production across space and time. Importantly, rare, but stable habitats may disproportionately affect ecological processes and serve as key sources of population diversity at larger spatial scales.

Wyoming↗
USGS2026

A framework for integrating spatiotemporal deep learning methods with landsat for annual land cover and impervious surface mapping

Land cover information is essential for understanding Earth’s surface dynamics and how vegetation, water, soil, climate, and terrain interact. The National Land Cover Database (NLCD) has been the authoritative source for consistent U.S. land cover mapping. To extend NLCD’s temporal resolution and reduce production latency, we developed the Land Cover Artificial Mapping System (LCAMS)—a prototype spatiotemporal deep learning framework piloted as the foundation for the new Annual NLCD. LCAMS builds on concepts from legacy NLCD and the U.S. Geological Survey Land Change Monitoring, Assessment, and Projection (LCMAP) initiatives. It employs a loosely coupled two-stage architecture consisting of independent but functionally interdependent spatial and temporal models. Spatial models extract per-year information from Landsat data, while the temporal models refine the spatial outputs to enforce inter-annual consistency—critical for reliable land change monitoring. LCAMS produces annual 30 m resolution land cover and impervious surface outputs, with region-specific fine-tuning to generalize across diverse landscapes and temporal dynamics. Validation was conducted using an independent dataset of 1925 randomly sampled plots from five U.S. Landsat Analysis Ready Data (ARD) tiles spanning 1985-2021, selected for spatial and temporal variability. This dataset was used consistently to evaluate LCAMS, Legacy NLCD, and LCMAP. Using the NLCD legend, LCAMS achieved 72.1 ± 1.60% overall agreement, compared to 71.1 ± 1.7% agreement for Legacy NLCD. Using the LCMAP legend, LCAMS achieved 83.4 ± 1.22% agreement, compared to 84.6 ± 1.11% agreement for LCMAP. Overall, LCAMS delivers comparable accuracy while offering higher thematic resolution, longer temporal coverage, and automated production of annual 30 m CONUS land cover.

Remote Sensing of Environment↗
USGS2026

Detecting earthquakes in noisy real-time GNSS data with deep learning for improved PGD magnitude estimation

To disseminate accurate and useful warnings, earthquake early warning (EEW) systems must quickly determine the size and location of an earthquake to estimate expected shaking. Traditional seismic‐based algorithms tend to underestimate the true magnitudes of large earthquakes, a phenomenon known as magnitude saturation. This limitation motivated the recent inclusion of Global Navigation Satellite Systems (GNSS) data into the U.S. Geological Survey’s ShakeAlert EEW system with the Geodetic First Approximation of Size and Time (GFAST) algorithm because GNSS data do not saturate with large ground motions. However, the noise levels of GNSS data are very high compared with traditional seismic data, which obscures P ‐wave arrivals and can result in less accurate magnitude estimations if displacement amplitudes are low, such as for lower magnitude earthquakes or large source–station distances. In this study, we develop a deep‐learning model that detects earthquakes in GNSS data and use the Ridgecrest, California, earthquake sequence as a case study to demonstrate how the model could act as a filter to reduce the amount of low‐quality data that enters an algorithm like GFAST. To preserve our limited real earthquake data for model inference, we generated a training dataset composed of >700,000 synthetic displacement waveforms. We combined the synthetic waveforms with real‐time GNSS noise to produce realistically noisy training waveforms and then tested our model on additional synthetic data and performed inference using the real data that were held back. We discuss the performance of our trained model on both the unseen synthetic data and real inference data. Our model can be used to selectively filter only high‐quality data where an earthquake signal is observed for input into an algorithm like GFAST (outperforming a simple signal‐to‐noise ratio–based filter) to reduce the error in GFAST’s real‐time earthquake magnitude estimations.

California↗
USGS2025

Identifying precursors and tracking pulses of magma ascent in multidisciplinary data during the 2018–2023 phreatomagmatic eruption at Semisopochnoi Island, Alaska

The 2018–2023 phreatomagmatic eruptions at Semisopochnoi Island, Alaska produced abundant long-period (LP) seismicity, harmonic and broadband tremor, and explosion signals over several well-monitored periods of eruption and quiescence. The corresponding dataset provides an excellent opportunity to investigate precursory and syn -eruptive geophysical signals of long-lived phreatomagmatic eruptions using multiparameter observations. We generated explosion and LP event catalogs through novel implementations of the REDPy ( Hotovec-Ellis, 2024 ) repeating event detector in mid-2021 following a network upgrade and the onset of a new phase of the eruption. The hundreds of detected explosions show a high degree of infrasound waveform similarity over more than a year, indicating a repeating source mechanism likely associated with explosive magma-water interaction. The seismic LP catalog shows that events began over a month prior to renewed explosive activity at the beginning of August 2021, and that lower frequency index (FI) LPs were generated in the week prior to the onset of explosions. We applied a recently developed machine learning tool (VOISS-Net, Tan et al., 2024 ) to catalog abundant broadband and harmonic seismic tremor recorded before and during the renewed explosive activity, along with LPs and explosions. The tremor catalogs complement the LP and explosion catalogs by filling out the seismic sequence with the dominant signal types. Together, these catalogs reveal a seismic sequence of renewed unrest that started with several weeks of LP events, followed by LPs with lower FI values and harmonic tremor in the days prior to explosive activity, and finally the onset of discrete explosions and broadband eruption tremor. We interpret this sequence as the ascent of a new pulse of magma that first interacted with the hydrothermal/groundwater system to produce LPs, followed by harmonic tremor, and that ultimately drove explosive magma-water interactions and periods of continuous ash emissions. The 2021 seismic sequence, in combination with long-term records of satellite SO 2 emissions, deformation from interferometric synthetic aperture radar (InSAR) analysis, ash sample analysis, infrasound, and volcano tectonic seismicity, allows us to interpret the entire 9-year period of unrest and eruption that began with an intrusion and earthquake swarm in 2014.

Alaska↗
USGS2025

A spatiotemporal deep learning approach for predicting daily air-water temperature signal coupling and identification of key watershed physical parameters in a montane watershed

Seasonal shifts from runoff to groundwater dominance influence daily headwater stream temperatures, especially where local groundwater input is strong. This input buffers temperature during hot periods, supporting cold-water habitats. Recent studies use air–water temperature signal metrics to identify zones of strong stream–groundwater connectivity. While Previous studies used air–water signal ratios as proxies for groundwater influence but were limited to specific sites and periods, without dynamic forecasting. This study is the first to forecast daily A r as a spatiotemporal signal using a Graph Convolutional Network–Long Short-Term Memory (GCN-LSTM) model. The model was trained using hydroclimate data (air temperature, precipitation, shortwave radiation, streamflow) and watershed physical features (e.g., sand content, slope). Results showed high predictive skill, achieving R 2 (NSE, RMSE) of 0.86 (0.73, 0.0004) for one-day-ahead to 0.52 (0.50, 0.0009) for seven-days ahead forecasts. Prior studies often have not explicitly incorporated spatial hydrogeologic drivers, but this model explicitly incorporates them to assess their impact on A r forecasting and stream-groundwater connectivity. Feature analysis identified mean sand, elevation, slope, clay, and TWI as key predictors of A r . Stronger groundwater signals appeared in hillslopes, elevations, and tributaries, highlighting watershed influence on streamflow. However, limitations include reliance on historical air–water temperature patterns for training and limited representation of extreme climate conditions. Despite these limitations, unlike previous studies relying on measured in-situ stream and air temperature, this study forecasts A r directly from climate and physiographic features after training, avoiding in-situ data requirements. Findings aiding predictions of stream ecosystem resilience.

New York↗
USGS2025

Comparisons of shoreline positions from satellite-derived and traditional field- and remote-sensing techniques

Satellite-derived shorelines (SDS) have the potential to help researchers answer critical coastal science questions and support work to predict coastal change by filling in the spatial and temporal gaps present in current field-based and remote-sensing data collection methods. The U.S. Geological Survey conducted comparison analyses of traditionally sourced shorelines and SDS in diverse coastal landscapes to determine how SDS could be used in ongoing and future work across varied coastal environments and provided some initial findings that could be used for implementation. Using CoastSeg, a browser-based program for SDS detection and mapping, SDS for the period 1984–2023 for multiple locations across the United States were compared to shoreline positions from traditionally sourced shoreline data. In this report, the authors present these comparisons alongside lessons learned and challenges encountered when building SDS workflows in different coastal locations. Results show that individual SDS have larger uncertainty and yet produced similar linear trends to sparser, traditionally sourced shoreline data; because SDS methods provide orders of magnitude more data than traditional shoreline-detection methods, they can be used to evaluate shoreline behaviors. Refining average scalar slopes used in tidal corrections did not result in substantial decreases in uncertainty. Using lessons from this work to outline needs for regional implementation, initial setup time would be considerable, being on the order of weeks. However, once complete, shoreline detections and analyses are fast (on the order of minutes to hours) and achievable using a desktop computer.

Alaska, Florida, Massachusetts, Washington↗
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↗