Search USGS⌕ Search

SEARCH · Search USGS

Results for “Machine Learning with Applications”

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.

At least 55 records · Page 3Linked to original sources
USGS2025

Evaluating hydrologic data products for scientific and management applications related to potential future streamflow conditions in the Upper Mississippi and Illinois Rivers

The hydrology of the Upper Mississippi and Illinois Rivers is a fundamental driver of ecosystem patterns and processes across a large portion of the United States. Quantitative hydrologic data for the main stems of these rivers underlie numerous scientific investigations, statistical models, and decision-making processes for local, State, and Federal agencies involved in the Upper Mississippi River Restoration program. Although historical hydrologic data exist, data representing potential future conditions of the Upper Mississippi and Illinois Rivers lack the resolution necessary to anticipate biotic and abiotic responses to altered hydrology and to determine resilient management actions. A source of future hydrologic scenarios is the readily available LOCA–VIC–mizuRoute hydrologic data products (named for the chain of models the data are produced from—localized constructed analogs, Variable Infiltration Capacity macroscale hydrological model, and the mizuRoute hydrologic routing model—that we shorten further to LVM in this report) that include simulated discharges for historic and future timeframes. The objective of this study is to assess the reliability of the hydrologic data products for their use in Upper Mississippi River Restoration program applications. Key study questions are (1) do the hydrologic data products reproduce characteristics of hydrology necessary to support ecological modeling and restoration decision-making applications within the Upper Mississippi River Restoration program? and (2) are there geographic differences in the reliability of the hydrologic data products? Seven characteristics of river hydrology were selected related to flow magnitude, seasonality, and regime for evaluation. The seven characteristics were calculated using observed and historical simulated hydrologic data at 19 U.S. Geological Survey streamgages throughout the basins of the Upper Mississippi and Illinois Rivers; two streamgages are located on the main stem of the Mississippi River and two streamgages are located on the main stem of the Illinois River. Statistical comparisons between observed and historical simulated characteristics indicated that the hydrologic data products did not reliably represent historical hydrologic conditions in the basin or main stem. The hydrologic data products we evaluated could not reliably capture the overall hydrologic regime or flow magnitudes; the latter is evidenced by substantial underestimates of discharge at most streamgages. Seasonal hydrologic characteristics were captured more reliably than flow magnitude, but overall correspondence was low for most streamgages. A weak latitudinal pattern in seasonal characteristics indicated the hydrologic data products poorly represent streamflow timing in snow-affected regions of the basin. Discrepancies in magnitude, seasonality, and regime indicate the potential for multiple sources of error. Because poor correspondence was present across all 19 streamgages, it was not possible to identify specific drivers of poor performance (that is, drainage area or geography). The modeling chain should be evaluated for biases associated with meteorologic forcing data, as well as hydrologic model formulation and calibration. We conclude that the hydrologic data products we evaluated appear unsuitable for applications tied to habitat and ecosystem restoration and management in the Upper Mississippi and Illinois Rivers. Plans to develop a future hydrology dataset for the Upper Mississippi River Restoration program would benefit from ongoing work to improve global climate model output downscaling methods, to improve hydrologic models, to make use of innovations in machine-learning approaches for projecting hydrology, and other efforts. The framework developed herein to evaluate hydrometeorological outputs generated using global climate models for a specific water resources application is a transferrable approach that could be applied to other data products and river systems.

Illinois, Indiana, Iowa, Minnesota, Missouri, Sout↗
USGS2023

A 1.2 billion pixel human-labeled dataset for data-driven classification of coastal environments

The world’s coastlines are spatially highly variable, coupled-human-natural systems that comprise a nested hierarchy of component landforms, ecosystems, and human interventions, each interacting over a range of space and time scales. Understanding and predicting coastline dynamics necessitates frequent observation from imaging sensors on remote sensing platforms. Machine Learning models that carry out supervised (i.e., human-guided) pixel-based classification, or image segmentation, have transformative applications in spatio-temporal mapping of dynamic environments, including transient coastal landforms, sediments, habitats, waterbodies, and water flows. However, these models require large and well-documented training and testing datasets consisting of labeled imagery. We describe “Coast Train,” a multi-labeler dataset of orthomosaic and satellite images of coastal environments and corresponding labels. These data include imagery that are diverse in space and time, and contain 1.2 billion labeled pixels, representing over 3.6 million hectares. We use a human-in-the-loop tool especially designed for rapid and reproducible Earth surface image segmentation. Our approach permits image labeling by multiple labelers, in turn enabling quantification of pixel-level agreement over individual and collections of images.

Scientific Data↗
USGS2025

A method to obtain remotely sensed grain size distributions from nonplanar granular deposits

Constraining the grain size distribution of granular deposits with complex surfaces is difficult with existing approaches. Field and laboratory techniques are time consuming and limited by the maximum grain size that laboratories can accommodate. In this study, we present a new method to identify the coarse fraction of the grain size distribution at a debris-flow fan deposit surveyed with terrestrial laser scanning (TLS) in Glenwood Canyon, Colorado, USA. This method is a novel grain segmentation algorithm developed for application to point cloud data of deposits with complex surfaces and angular grains ranging in size from centimeters to a meter. This approach combines an existing random forest machine learning method with a novel iterative clustering algorithm. We compared the grain size distribution from our algorithm with a Wolman pebble count conducted in the field, and found a root mean squared error of less than 2 cm from the 5th to 95th percentile of the grain size distribution of grains ranging from cobble to boulder sized (6.3–78 cm in our application). Finally, we compared our new algorithm with an existing open-source grain segregation algorithm, and our method outperformed the selected alternative when applied to the debris-flow deposit point cloud.

Colorado↗
USGS2020

Ecological interfaces between land and flowing water: Themes and trends in riparian research and management

This paper provides an overview of past, present and future themes for research and management of riparian zones, often relating to papers within this Wetlands Special Feature. Riparian research expanded in the United States around 1980 with themes that recognized (1) damage from excessive livestock, or (2) damage from river damming and diversion, and (3) the beneficial capacity of riparian buffers to intercept and assimilate nutrients and other water contaminants. Research expanded globally in the 1990s, with themes including (4) plant life history requirements and (5) reliance on fluvial geomorphic dynamics that enable riparian rejuvenation. Resource managers recognized that riparian areas provide (6) rich wildlife habitats (7) along with valued ecosystem services, (8) which encouraged conservation and restoration initiatives, (9) including environmental flow regimes. Floodplains are (10) vulnerable to invasive plants and management has included biocontrol such as for Tamarix in the American Southwest. Into the twenty-first century, (11) climate change is advancing, and riparian ecosystems may be especially impacted due to the compound challenges from increasing water demand and declining summer flows. As an emerging opportunity, (12) while reservoirs submerge floodplain vegetation, reservoir deltas may support compensatory riparian wetlands. (13) Studies increasingly utilize remote sensing tools including satellite imagery, LiDAR and unmanned aircraft systems, and (14) the coordination of large data sets invites digital ecology, including artificial intelligence and machine learning. Since riparian zones are centres for human activities, (15) there are opportunities for citizen science, social media and internet applications, which will increasingly democratize riparian research and management.

Wetlands↗
USGS2026

Benchmark dataset of historical annual peak floods classified by causal mechanisms for select US river basins

Considering the causal mechanisms of floods can improve estimates of flood recurrence intervals given that certain flood types can be associated with higher magnitude and more damaging floods. However, few verified datasets of flood types are available to validate the semiautomated and automated classification algorithms needed to apply flood-typing across large hydrologically diverse regions. To address this gap, a benchmark dataset of manually classified flood types was compiled for 1,763 annual maximum flood peaks from 18 stream gauges in six different river basins across the conterminous United States from 1851 to 2022. Within each basin, three representative stream gauges were selected for manual flood typing. A flexible classification framework is introduced that facilitates flood typing across hydrologically diverse regions and accommodates unique combinations of weather and antecedent watershed conditions specific to each region. Floods were manually typed by domain experts using multiple lines of evidence to identify a primary surface water input of each flood (rainfall, snowmelt, or both) and, if relevant, associated storm type and secondary causal mechanisms characterizing antecedent watershed conditions. Across all the study basins, 49% of historical annual maximum flood peaks were attributable to rainfall, 28% to snowmelt, 22% to mixed precipitation, and 1% could not be assigned to a mechanism due to missing or incomplete data. The proposed flood-typing schema supports varying levels of flood typing specificity required for mixed population flood-frequency analysis, flood-type-specific design hydrographs, water quality response studies, and additional applications. This detailed, manually determined benchmark dataset serves as a resource that can be used developing and validating automated or machine learning-based algorithms capable of operationalizing expanded flood peak information.

conterminous United States↗
USGS2018

A distributed pipeline for DIDSON data processing

Technological advances in the field of ecology allow data on ecological systems to be collected at high resolution, both temporally and spatially. Devices such as Dual-frequency Identification Sonar (DIDSON) can be deployed in aquatic environments for extended periods and easily generate several terabytes of underwater surveillance data which may need to be processed multiple times. Due to the large amount of data generated and need for flexibility in processing, a distributed pipeline was constructed for DIDSON data making use of the Hadoop ecosystem. The pipeline is capable of ingesting raw DIDSON data, transforming the acoustic data to images, filtering the images, detecting and extracting motion, and generating feature data for machine learning and classification. All of the tasks in the pipeline can be run in parallel and the framework allows for custom processing. Applications of the pipeline include monitoring migration times, determining the presence of a particular species, estimating population size and other fishery management tasks.

Conference Paper↗
USGS2024

Rapid estimation of single-station earthquake magnitudes with machine learning on a global scale

The foundation of earthquake monitoring is the ability to rapidly detect, locate, and estimate the size of seismic sources. Earthquake magnitudes are particularly difficult to rapidly characterize because magnitude types are only applicable to specific magnitude ranges, and location errors propagate to substantial magnitude errors. We developed a method for rapid estimation of single‐station earthquake magnitudes using raw three‐component P waveforms observed at local to teleseismic distances, independent of prior size or location information. We used the MagNet regression model architecture ( Mousavi and Beroza, 2020b ), which combines convolutional and recurrent neural networks. We trained our model using ∼2.4 million P ‐phase arrivals labeled by the authoritative magnitude assigned by the U.S. Geological Survey. We tested input data parameters (e.g., window length) that could affect the performance of our model in near‐real‐time monitoring applications. At the longest waveform window length of 114 s, our model (Artificial Intelligence Magnitude [AIMag]) is accurate (median estimated magnitude within ±0.5 magnitude units from catalog magnitude) between M 2.3 and 7.6. However, magnitudes above M ∼7 are more underestimated as true magnitude increases. As the windows are shortened down to 1 s, the point at which higher magnitudes begin to be underestimated moves toward lower magnitudes, and the degree of underestimation increases. The over and underestimation of magnitudes for the smallest and largest earthquakes, respectively, are potentially related to the limited number of events in these ranges within the training data, as well as magnitude saturation effects related to not capturing the full source time function of large earthquakes. Importantly, AIMag can determine earthquake magnitudes with individual stations’ waveforms without instrument response correction or knowledge of an earthquake’s source‐station distance. This work may enable monitoring agencies to more rapidly recognize large, potentially tsunamigenic global earthquakes from few stations, allowing for faster event processing and reporting. This is critical for timely warnings for seismic‐related hazards.

Bulletin of the Seismological Society of America↗
USGS2022

Automating the detection and classification of wildlife in aerial imagery by developing a framework to support machine learning techniques

This report presents a collected effort to construct a framework that supports machine learning techniques for wildlife surveys using digital aerial imagery. A robust foundation has been developed to advance both scientific and production-ready applications of Artificial Intelligence/Machine Learning (AI/ML) for wildlife surveys by constructing a data pipeline to handle large image datasets, building an annotation dataset of wildlife targets, and using those annotations within AI/ML processes. This report is a summary of those efforts.

OCS Study↗
USGS2022

Modeling subsurface performance of a geothermal reservoir using machine learning

Geothermal power plants typically show decreasing heat and power production rates over time. Mitigation strategies include optimizing the management of existing wells—increasing or decreasing the fluid flow rates across the wells—and drilling new wells at appropriate locations. The latter is expensive, time-consuming, and subject to many engineering constraints, but the former is a viable mechanism for periodic adjustment of the available fluid allocations. In this study, we describe a new approach combining reservoir modeling and machine learning to produce models that enable such a strategy. Our computational approach allows us, first, to translate sets of potential flow rates for the active wells into reservoir-wide estimates of produced energy, and second, to find optimal flow allocations among the studied sets. In our computational experiments, we utilize collections of simulations for a specific reservoir (which capture subsurface characterization and realize history matching) along with machine learning models that predict temperature and pressure timeseries for production wells. We evaluate this approach using an “open-source” reservoir we have constructed that captures many of the characteristics of Brady Hot Springs, a commercially operational geothermal field in Nevada, USA. Selected results from a reservoir model of Brady Hot Springs itself are presented to show successful application to an existing system. In both cases, energy predictions prove to be highly accurate: all observed prediction errors do not exceed 3.68% for temperatures and 4.75% for pressures. In a cumulative energy estimation, we observe prediction errors that are less than 4.04%. A typical reservoir simulation for Brady Hot Springs completes in approximately 4 h, whereas our machine learning models yield accurate 20-year predictions for temperatures, pressures, and produced energy in 0.9 s. This paper aims to demonstrate how the models and techniques from our study can be applied to achieve rapid exploration of controlled parameters and optimization of other geothermal reservoirs.

Energies↗
USGS2026

Fifteen years of WRTDS for advancing water-quality science: A critical review of methodological developments and global applications

Contamination by nutrients, major ions, and metals poses a major threat to global water sustainability. Understanding how these pollutants vary across time and space requires long-term monitoring and robust statistical approaches. Traditional methods, however, often struggle to account for streamflow variability, seasonality, and nonlinear responses. Introduced in 2010, the Weighted Regressions on Time, Discharge, and Season (WRTDS) method offers a flexible, data-driven framework that generates both observed and flow-normalized estimates of concentration and load. Over the past 15 years, WRTDS has become a state-of-the-art tool for water-quality science and management, with applications spanning a wide range of hydrologic, climatic, and policy contexts─including major watersheds across North America, Europe, Asia, Australia, and the Arctic. In this review of WRTDS, we document the method’s major advancements, examine its expanding geographic and thematic applications, and summarize its relevance to water-quality management programs and policies worldwide. We also discuss its performance relative to other regression and machine-learning approaches. Finally, we identify key priorities for future development to support the continued evolution of WRTDS as a trusted and practical tool for scientists and managers working to protect and sustain water resources.

Environmental Science and Technology↗
USGS2022

Can machine learning accelerate process understanding and decision-relevant predictions of river water quality?

The global decline of water quality in rivers and streams has resulted in a pressing need to design new watershed management strategies. Water quality can be affected by multiple stressors including population growth, land use change, global warming, and extreme events, with repercussions on human and ecosystem health. A scientific understanding of factors affecting riverine water quality and predictions at local to regional scales, and at sub-daily to decadal timescales are needed for optimal management of watersheds and river basins. Here, we discuss how machine learning (ML) can enable development of more accurate, computationally tractable, and scalable models for analysis and predictions of river water quality. We review relevant state-of-the art applications of ML for water quality models and discuss opportunities to improve the use of ML for emerging computational and mathematical methods for model selection, hyperparameter optimization, incorporating process knowledge into ML models, improving explainablity, uncertainty quantification, and model-data integration. We then present considerations for using ML to address water quality problems given their scale and complexity, available data and computational resources, and stakeholder needs. When combined with decades of process understanding, interdisciplinary advances in knowledge-guided ML, information theory, data integration, and analytics can help address fundamental science questions and enable decision-relevant predictions of riverine water quality.

Hydrological Processes↗
USGS2024

Using explainable machine learning methods to evaluate vulnerability and restoration potential of ecosystem state transitions

Ecosystem state transitions can be ecologically devastating or be a restoration success. State transitions are common within aquatic systems worldwide, especially considering human-mediated changes to land use and water use. We created a transferable conceptual framework to enable multiscale assessments of state resilience and early warnings of state transitions that can inform strategic restorations and avoid ecosystem collapse. The conceptual framework integrated machine learning predictions with ecosystem state concepts (e.g., state classification, gradients of vulnerability, and recovery potential leading to state transitions) and was devised to investigate possible environmental drivers. As an application of the framework, we generated prediction probabilities of submersed aquatic vegetation (SAV) presence at nearly 10,000 sites in the Upper Mississippi River (United States). Then, we used an interpretability method to explain model predictions to gain insights into possible environmental drivers and thresholds or linear responses of SAV presence and absence. Model accuracy was 89% without spatial bias. Average water depth, suspended solids, substrate, and distance to nearest SAV were the best predictors and likely environmental drivers of SAV habitat suitability. These environmental drivers exhibited nonlinear, threshold-type responses for SAV. All the results are also presented in an online dashboard to explore results at many spatial scales. The habitat suitability model outputs and prediction explanations from many spatial scales (4 m to 400 km of river reach) can inform research and restoration planning.

Conservation Biology↗
USGS2025

Using machine learning in Minnesota’s StreamStats to predict fluvial sediment

A thorough understanding of fluvial sediment transport is essential for addressing key environmental issues such as aquatic habitat degradation, flooding, excess nutrients, and challenges with river restoration. Fluvial sediment samples are valuable for addressing these concerns, but their collection is often impractical across all rivers and timeframes of interest. In addition, previously used analytical and numerical methods have not allowed for the transfer of knowledge from sites that have data to sites that do not have data. To overcome this limitation, the U.S. Geological Survey developed machine learning models to predict suspended-sediment concentrations and bedload transport in Minnesota rivers that lack physical sediment data and integrated them into the U.S. Geological Survey StreamStats web application.

Minnesota↗
USGS2026

SlideDetect: Spatio-temporal landslide detection using a three-dimensional convolutional neural network

Landslides pose a serious and ongoing threat to both human lives and infrastructure worldwide; therefore, it is of interest to predict where and when landslides are likely to occur. Advances in machine learning techniques have spurred numerous studies aimed at estimating relative landslide propensity, but are limited to spatial (as opposed to temporal) prediction due to the sparsity of landslide timing data. We address this data gap by training SlideDetect, a 3-dimensional convolutional neural network (3D CNN), to identify landslides based on their spatial and temporal occurrence within multitemporal image stacks. We use an inventory of landsides triggered by the 2018 Hokkaido earthquake and two years of monthly composite optical imagery spanning this event. The model can identify not only landslide location but also landslide date with an area under the precision-recall curve (PR-AUC) of 0.84. We further present a new standard for presenting PR curve results that explicitly compares model performance at different confidence thresholds, allowing for clearer model evaluation and comparison. Our new approach to constraining landslide timing paired with this more consistent and objective method for evaluating model performance shows considerable promise, and with further application and testing, SlideDetect could enhance the data availability and tools needed to advance landslide hazard and risk assessments.

JGR Machine Learning and Computation↗
USGS2025

RUSH: Rapid remote sensing Updates of land cover for Storm and Hurricane forecast models

Coastal vegetated ecosystems, including tidal marshes, vegetated dunes, and shrub- and forest-dominated wetlands, can mitigate hurricane impacts such as coastal flooding and erosion by increasing surface roughness and reducing wave energy. Land cover maps can be used as input to improve simulations of surface roughness in advanced hydro-morphological models. Consequently, there is a need for efficient tools to develop up-to-date land cover maps that include the accurate distribution of vegetation types prior to an extreme storm. In response, we developed the RUSH tool (Rapid remote sensing Updates of land cover for Storm and Hurricane forecast models). RUSH delivers high-resolution maps of coastal vegetation for near-real-time or historical conditions via a Jupyter Notebook application and a graphical user interface (GUI). The application generates 3 m spatial resolution land cover maps with classes relevant to coastal settings, especially along mainland beaches, headlands, and barrier islands, as follows: (1) open water; (2) emergent wetlands; (3) dune grass; (4) woody wetlands; and (5) bare ground. These maps are developed by applying one of two seasonal random-forest machine learning models to Planet Labs SuperDove multispectral imagery. Cool Season and Warm Season Models were trained on 665 and 594 reference points, respectively, located across study regions in the North Carolina Outer Banks, the Mississippi Delta in Louisiana, and a portion of the Florida Gulf Coast near Apalachicola. Cool Season and Warm Season Models were tested with 666 and 595 independent points, with an overall accuracy of 93% and 94%, respectively. The Jupyter Notebook application provides users with a flexible platform for customization for advanced users, whereas the GUI, designed with user-experience feedback, provides non-experts access to remote sensing capabilities. This application can also be used for long-term coastal geomorphic and ecosystem change assessments.

Florida, Louisiana, North Carolina↗
USGS2021

Hyperspectral narrowband data propel gigantic leap in the earth remote sensing

Hyperspectral narrowbands (HNBs) capture data as nearly continuous “spectral signatures” rather than a “few spectral data points” along the electromagnetic spectrum as with multispectral broadbands (MBBs). Almost all of satellite remote sensing of the Earth in the twentieth century was conducted using MBB data from sensors such as the Landsat-series, Advanced Very High-Resolution Radiometer (AVHRR), SPOT (Système Pour l’Observation de la Terre), and the Indian Remote Sensing (IRS) satellites. These systems typically provide 4 to 9 broad spectral wavebands spread from 400 to 2500 nm, often with one or two additional bands in the thermal range. Significant advances in the study of the Earth have been made based on these data [Thenkabail et al., 2018a,b,c,d; Thenkabail et al., 2015a,b,c]. Possibilities of great advances that can be made using HNB data over MBB data are well established based on studies conducted using hyperspectral sensors such as the hand-held spectroradiometers, the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS), and spaceborne Earth Observing -1 (EO-1) Hyperion [Thenkabail 2018a,b,c,d]. The twenty-first century is already seeing the dawn of hyperspectral imaging data from sensors such as the German Aerospace Center’s (DLR’s) DESIS (DLR Earth Sensing Imaging Spectrometer) onboard the MUSES (Multi-User System for Earth Sensing) platform on the International Space Station (ISS), the polar-orbiting Italian Space Agency’s (ASI) PRISMA (PRecursore IperSpettrale della Missione Applicativa), and many other upcoming sensors such as the NASA Surface Biology and Geology (SBG) [Thenkabail et al., 2018a,b,c,d]. These satellites acquire data in hundreds of narrow spectral bands of 1 to 10 nm width, typically between 400 to 2500 nm; also future planned missions will be extending HNBs to the thermal (9,000 to 14,000 nm) electromagnetic spectrum. This expansion creates a quantum leap in new data, new information, and myriad possible new applications in the study of the Earth in addition to great advances in existing applications. Given the above, the objective of this article is to provide insights on the gigantic leap in our understanding, modeling, mapping, and monitoring of the Earth that can be made using HNB relative to MBB by focusing on agricultural and vegetation applications. We will address this in four aspects: 1. Comparison between HNB and MBB data; 2. Spectral libraries of agricultural crops; 3. HNB data analysis in general; and 4. HNB analysis using machine learning (ML) and cloud computing.

Photogrammetric Engineering and Remote Sensing.↗
USGS2019

Modeling barrier island habitats using landscape position information

Barrier islands are dynamic environments because of their position along the marine–estuarine interface. Geomorphology influences habitat distribution on barrier islands by regulating exposure to harsh abiotic conditions. Researchers have identified linkages between habitat and landscape position, such as elevation and distance from shore, yet these linkages have not been fully leveraged to develop predictive models. Our aim was to evaluate the performance of commonly used machine learning algorithms, including K-nearest neighbor, support vector machine, and random forest, for predicting barrier island habitats using landscape position for Dauphin Island, Alabama, USA. Landscape position predictors were extracted from topobathymetric data. Models were developed for three tidal zones: subtidal, intertidal, and supratidal/upland. We used a contemporary habitat map to identify landscape position linkages for habitats, such as beach, dune, woody vegetation, and marsh. Deterministic accuracy, fuzzy accuracy, and hindcasting were used for validation. The random forest algorithm performed best for intertidal and supratidal/upland habitats, while the K-nearest neighbor algorithm performed best for subtidal habitats. A posteriori application of expert rules based on theoretical understanding of barrier island habitats enhanced model results. For the contemporary model, deterministic overall accuracy was nearly 70%, and fuzzy overall accuracy was over 80%. For the hindcast model, deterministic overall accuracy was nearly 80%, and fuzzy overall accuracy was over 90%. We found machine learning algorithms were well-suited for predicting barrier island habitats using landscape position. Our model framework could be coupled with hydrodynamic geomorphologic models for forecasting habitats with accelerated sea-level rise, simulated storms, and restoration actions.

Remote Sensing↗
USGS2024

Mapping and classification of volcanic deposits using multi-sensor unoccupied aerial systems

The deposits from volcanic eruptions represent the record of activity at a volcano. Identification, classification, and interpretation of these deposits are crucial to the understanding of volcanic processes and assessing hazards. However, deposits often cover large areas and can be difficult or dangerous to access, making field mapping dangerous and time-consuming. Remote sensing techniques are often used to map and identify the deposits of volcanic eruptions, though these techniques present their own trade-offs in terms of image resolution, wavelength, and observation frequency. Here, we present a new approach for mapping and classifying volcanic deposits using a multi-sensor unoccupied aerial system (UAS), and demonstrate its application on lava and tephra deposits associated with the 2018 eruption of Sierra Negra volcano (Galápagos Archipelago, Ecuador). We surveyed the study area and collected visible and thermal infrared (TIR) images. We used structure-from-motion photogrammetry to create a digital elevation model (DEM) from the visual images and calculated the solar heating rate of the surface from temperature maps based on the TIR images. We find that the solar heating rate is highest for tephra deposits and lowest for ʻaʻā lava, with pāhoehoe lava having intermediate values. This is consistent with the solar heating rate correlating to the density and particle size of the surface. The solar heating rate for the lava flow also decreases with increasing distance from the vent, consistent with an increase in density as the lava degasses. We combined the surface roughness (calculated from the DEM) and the solar heating rate of the surface to remotely classify tephra deposits and different lava morphologies. We applied both supervised and unsupervised machine learning algorithms and demonstrate that supervised methods can replicate the manual classification while the unsupervised method can identify major surface units with no ground truth information. These methods allow for remote mapping and classification at high spatial resolution (< 1 meter) of a variety of volcanic deposits, with potential for application to deposits from other processes (e.g., fluvial, glacial) and deposits on other planetary bodies.

Galápagos Archipelago, Sierra Negra volcano↗