The Atmospheric Waves Experiment (AWE) utilizes the Advanced Mesospheric Temperature Mapper (AMTM), built by the Space Dynamics Laboratory (SDL) and mounted on the International Space Station (ISS), to observe OH airglow emissions at approximately 87 km altitude, enabling the study of gravity wave dynamics in the mesosphere. A critical component of AWE data processing is the geolocation algorithm, which determines the latitude and longitude coordinates where each pixel’s line-of-sight (LOS) vector intersects the Earth’s atmosphere. This algorithm is essential for producing accurate data products for the corresponding spatial radiance swaths generated by the science team at Utah State University (USU). The algorithm employs a series of transformations between the AMTM, ISS, Local Vertical Local Horizontal (LVLH), and Earth-Centered Earth-Fixed (ECEF) coordinate systems. These conversions rely on ISS position, velocity, and attitude data, which are interpolated to match AMTM image timestamps. Interpolation is performed using linear and spherical linear interpolation (SLERP) depending on the magnitude of attitude changes between samples. The geolocation algorithm also accounts for the Earth’s self-rotation rate to improve geolocation accuracy. To refine the rotation matrix between the AMTM and ISS coordinate systems, an on-orbit calibration process is implemented using ground-truth tie points. An iterative adjustment minimizes residuals between observed and predicted locations, optimizing the geolocation accuracy of the AMTM imagery. An additional set of tie points are set aside for validating the accuracy of the geolocation process after the on-orbit calibration. These show position deviations well within program requirements for the AMTM’s ~0.35° instantaneous field-of-view (IFOV) and 2 km pixel resolution at ~420 km orbit altitude. This ensures robust SDL geolocation support for the NASA-funded AWE mission to investigate atmospheric wave structures and their global propagation characteristics.

General circulation models (GCMs) exhibit systemic precipitation biases over much of South America during its wet season (December to March), undermining their utility in projecting future mean and extreme precipitation shifts and Amazon carbon cycle feedbacks in response to rising greenhouse gases. Though these biases, including a dry bias over the Amazon, have persisted across generations of model development, their origins are an ongoing topic of debate. Here, we explore whether the underestimation of Andes peak heights—and thus their orographic blocking effect—due to overly-smooth resolved topographic boundary conditions in GCMs underlies the biases. We performed a suite of idealized experiments forced by elevated topography over the Andes (and Central America) to better capture orographic blocking with the NASA GISS ModelE version 2.1 (E2.1) and 3 (E3). In the E2.1 experiments conducted with a fully-coupled atmosphere-ocean, we find precipitation over the central Amazon and South Atlantic Convergence Zone (SACZ) is increased relative to a control simulation forced with the default topography, reflecting a reduction of the model’s precipitation biases in these regions. The former results from enhanced blocking of low-level tropical easterly winds and thus moisture convergence east of the mountains, while the latter is driven by shifting stationary wave patterns that enhance moisture convergence in the SACZ. Yet, the wet bias in the eastern Amazon is worsened in the elevated topography experiment due to an intensification of the ITCZ bias in the southern tropical Atlantic. Analysis of analogous experiments run with E2.1 and E3 in atmosphere-only mode and prescribed sea surface temperatures reveals that the high pressure stationary wave pattern generated in the south Atlantic relied on coupled atmosphere-ocean mechanisms. Coupled processes are shown to be important to generating stationary wave patterns arising from orographic blocking and sea surface temperature interactions, enhancing moisture convergence and thus reducing model biases over the central Amazon and SACZ, though this is balanced by increased model biases influenced by the ITCZ.

Wetlands are among the world’s most powerful ecosystems. In Michigan, they improve water quality, protect against erosion, and provide habitats for dozens of species; however, Michigan has lost fifty percent of its historic wetlands that existed prior to European settlement. Remote sensing has become a crucial tool to protect Michigan’s remaining wetlands. Wetlands can be seasonal, causing variations in satellite imagery from month to month. Existing studies show that certain months are better for classification, but have been limited to small-scale field studies. However, the impact of using imagery from different seasons over a larger geographic area for wetland classification remains unexplored. This study analyzed the effectiveness of Sentinel-2 and Landsat 9 imagery across seasonal and annual time frames in wetland classification, focusing on the lower peninsula of Michigan. I used a supervised classification approach to build a machine learning model, incorporating standard variables used to identify wetlands such as computed texture, indices, and elevation. I used composited imagery from 2022 and 2023 to assess across four time periods (spring, summer, fall, annual) for wetland classification using accuracy metrics. Notably, the spring composite image attained the highest accuracy among Sentinel-2, while the summer and annual composite images performed best among Landsat 9. This study reveals the potential of using spring imagery from Sentinel-2 and summer imagery from Landsat 9 to enhance wetland classification, underscoring how seasonality can affect classification accuracy. This can further guide scientists into selecting optimal data for monitoring wetlands within Michigan and areas with similar climate and geographical features.

Assessing the trafficability of Arctic sea ice is vital for Arctic communities, researchers, and operational users who depend on over-ice mobility. While synthetic aperture radar (SAR)-based methods have shown promise for evaluating sea ice roughness, they are limited in resolving surface conditions influenced by snow. In this study, we explore the potential of high-resolution optical satellite imagery for characterizing sea ice surface roughness and estimating trafficability through texture analysis.Using the Grey-Level Co-occurrence Matrix (GLCM) method, we calculate texture measures (contrast, homogeneity, and energy) from optical imagery over landfast sea ice near Utqiaġvik, Alaska. Parameter sensitivity tests across multiple window sizes, orientations, and displacements inform optimal settings for extracting meaningful spatial texture metrics. The texture measures are compared against snowmachine travel speed data collected during field surveys near Utqiaġvik, Alaska. Our results show that GLCM-derived contrast metrics strongly correlate with observed travel speeds. Based on these correlations, we generate speed maps and apply an optimal pathfinding algorithm to identify efficient travel corridors across the ice. This remote sensing–driven approach enables fast, scalable assessments of sea ice trafficability that incorporate snow-related surface complexity. By linking texture measures from high-resolution optical satellite imagery to in situ travel velocity data, we demonstrate a transferable methodology for sea ice condition assessment at spatial scales relevant to both local and regional applications. This framework supports community travel planning and remote logistics and is increasingly valuable with the growing availability of high-resolution, high-temporal-frequency optical imagery across polar environments.

DL-based earthquake detection can detect substantially more induced microseismic events than conventional detection methods, revealing finer-scale patterns of hydraulic fracturing-induced seismicity (HFIS). We implement a novel workflow for the Preston New Road-1z (PNR-1z) shale gas site, UK, that integrates DL phase picking, Oct-Tree-based phase association, and probabilistic non-linear location to generate a high-resolution “deep” seismic catalogue from a single high-frequency (2000 Hz) borehole array. PhaseNet has detected 11,021 previously undetected events (49,472 events total), compared to the previous catalogue, in which 38,452 events were identified by a beamforming method. We find that the enhanced catalogue increases event cluster densities by an average of 42% and lowers the magnitude of completeness (Mc) from -0.2 to -0.5, adding thousands of additional events to Gutenberg-Richter b-value analyses. The spatial patterns suggest that HFIS interacted with a fault or fracture zone north of the PNR wells, contrasting with previous event locations and their interpretations. Combining the largest and best recorded events, using locations from our independently derived catalogue, with 3D seismic reflection data reveals clustering along previously unmapped seismic discontinuities ~300 m north of the injection wells, indicating strong structural controls on HFIS. In addition, we re-estimate moment magnitudes (Mw) using continuous downhole waveforms, finding that magnitudes reported in the previous operational catalogue were systematically underestimated by an average of 0.93 units. The temporal b-values revealed higher b-values during active injection, consistent with tensile HF-dominated microseismicity, and relatively lower b-values during pauses and certain mid-stages, indicating increased shear failure and interaction with a pre-existing fracture zone. The deep catalogue allows for more robust b-value estimation and temporal tracking, potentially reflecting evolving source characteristics and fluid-driven processes during stimulation. This study demonstrates that DL-enhanced catalogues can refine subsurface interpretations, identify re-activated structures efficiently, and contribute to improved hazard assessments in geo-energy systems.

BioSCape (Biodiversity Survey of the Cape) is NASA’s first biodiversity-focused airborne and field campaign, designed to advance understanding of ecosystem structure, function, and composition, and how these are changing over time and space in South Africa’s Greater Cape Floristic Region. This international collaboration involving ~160 US and South African scientists across 19 sub-projects produced an unprecedented, open-access dataset spanning 45,000 km², integrating imaging spectroscopy (AVIRIS-NG, PRISM, HyTES), LiDAR (LVIS, ELMAP-V), and extensive ground data from ~2,600 plots. Data are publicly archived at ORNL DAAC and accessible via a cloud computing environment, fostering open science and reproducibility. In this presentation, we will summarize the results to date across the BioSCape project, highlighting key scientific outcomes from both terrestrial and aquatic landscapes. We will also discuss some of the broader implications of the project, including laying the groundwork for achieving ethical and equitable benefit-sharing via compliance with the Nagoya Protocol. A core principle of BioSCape is ”use-inspired science,” which involves co-developing research questions with local partners to directly address pressing conservation and resource management challenges. This ensures that our scientific findings translate into actionable data products for decision-makers in South Africa and globally. For example, BioSCape’s data and products are already informing conservation and resource management efforts, including kelp range maps for ecosystem classification and woody invasive alien plant maps that feed into national ecosystem condition assessments. The project’s contributions have already been recognized in the 2024 State of Environment Outlook Report for the Western Cape Province.

Data and scientific methods are becoming more complex in today’s landscape of affordable sensors, open data, and machine learning. In addition, collaboration and team science are growing to meet the increasing demand for interdisciplinary expertise. With larger teams, more data, and higher complexity analyses, making science reproducible is difficult. Skills in data and code management are often those most needed to meet these growing challenges, but scientists often lack formal training in these areas. In this presentation, we will introduce concepts and techniques to iteratively improve the reproducibility of scientific analyses using a published water data workflow shared publicly on GitHub. The reproducibility improvements will be presented in a stepwise fashion to introduce a set of recommendations for different skillsets: 1) beginner, focused on improved documentation and organization; 2) intermediate, focused on code modularity; and 3) advanced, focused on automation. Furthermore, the recommended improvements for each skill level will be tracked in GitHub version control, enabling transparent review of changes via pull requests and commits. Injecting good data and code practices to support reproducibility throughout a project can enhance a team’s ability to pivot and test new hypotheses, have greater confidence in results, and speed up the publication process. Demystifying how to improve scientific reproducibility is key to having these skills and approaches valued, practiced, and taught.

Extreme weather poses significant threats to the reliability of diversified electricity generation portfolios across the United States. For example, ice accumulation disables wind turbines, heavy snow disrupts fuel transportation, floods damage infrastructure, and extreme heat reduces thermal plant efficiency. In this work, we examine the impacts of an extensive list of extreme weather events—including heavy snow, blizzards, ice storms, winter storms, extreme cold/wind chill, hurricanes, tornadoes, floods, droughts, heat waves, and wildfires—on wind, solar, natural gas, coal, and other generation sources covering nearly all power plants in the United States, analyzing impacts at temporal resolutions varying from hourly to monthly depending on data availability. For instance, for wind power plants, we integrated monthly reported generation from 1202 facilities across the nation (2008-2022) with physics-informed modeled generation derived from reanalysis-driven simulations, finding that extreme weather events cause substantial reductions in actual generation relative to model predictions for that month: hurricanes reduce generation by 71%, cold/wind chill events by 14%, blizzards by 15%, and ice storms by 13%, demonstrating the severe operational impacts of extreme weather on wind power output. Our comprehensive assessment across all generation technologies reveals systematic vulnerabilities in our power infrastructure and can support development of proactive planning strategies to enhance grid resilience under future climate conditions.

The rapid and unexpected nature of flash floods leaves communities with minimal time to respond. Accurate forecasting of such events is crucial, as it allows for timely warnings, early evacuations, and preparations, significantly reducing potential losses. This study applies a transformer-based Deep Learning (DL) with Remote Sensing (RS) validation to predict floods with at different lead times. The Guadalupe River basin in Texas was selected as a case study due to the 2025 catastrophic flash flood, one of the deadliest floods in U.S. history. For forecasting purposes, discharge (Q) data from a downstream USGS station were chosen as the target variables. Precipitation, air temperature and moisture and vegetation indices were extracted for upstream station from RS sources. These RS products were validated against corresponding ground-based observations to ensure quality as dynamic inputs for flood prediction. Input climatic variables were structured based on different times windows (t-1, t-3, t-5, t-7, t-15, and t-30), representing conditions at 1, 3, 5, 7, 15 and 30 days before the current downstream state (t). different input sequences were utilized in a Multivariate Predictive Transformer Network, a DL model designed to make predictions from a temporal multivariate sequential dataset. The flood events were identified by exceedance from the 90th, 95th and 99th percentile threshold for Q within the 2003 to 2025 period. The predictive performance of the model was assessed regarding both the accuracy in predicting the flooding events and timing of impact. Furthermore, the model's capability to forecast floods with annual exceedance probabilities of 2, 5, 10, 25, 50, and 100 was thoroughly evaluated. Prediction results show that developed model can accurately detect the flood occurrence up to 5 days notice (Accuracy: 0.96). While the accuracy drops in next time lags, the metric shows comparable results with similar studies. The outcomes of this study, which integrates advanced transformer-based predictive modeling with remote sensing data, can substantially improve early forecasting and management strategies for floods as a detrimental natural hazard phenomenon.

Understanding the basal geometry and thickness of a landslide is crucial for slope stability and risk assessment, design of mitigative measures, and the study of landscape evolution. However, estimating landslide thickness remains a major challenge. Existing methods primarily generate two-dimensional profiles of landslide thickness. Nevertheless, these methods often rely on oversimplified geometric assumptions and do not fully exploit the three-dimensional nature of surface displacement data. In this work, we propose TIKE (Thickness Inversion with Kinematic Elements), a novel method for estimating the location of slip surfaces of landslides based on mass conservation and the integration of surface kinematic and geomorphic units. Building upon the framework introduced by Booth et al. (2013), we present an inversion model that considers spatial inhomogeneity of landslides and explicitly accounts for the kinematic boundaries of landslide sections. This new approach leverages recent advances in inverse theory, utilizing domain-based regularization and the Balancing Principle to automatically determine the appropriate level of regularization. Preliminary results from the well-studied Slumgullion landslide support the validity of the proposed approach. Direct measurements of landslide thickness are used to validate the inversion results. Surface geomorphic features, such as the presence of mini-slides and the locations of stationary ponds, are also used to assess the model predictions. The estimated basal surface reveals a channel-like structure, consistent with prior studies, and pronounced localized depressions beneath stationary ponds that were previously underappreciated. These results indicate that TIKE can substantially improve landslide thickness estimation, especially in data-rich settings with mapped kinematic units, and provide a promising tool for advancing three-dimensional landslide characterization.

The top meter of the Earth’s crust holds nearly equal amounts of rock-derived organic carbon (OCpetro) and soil organic carbon. Due to this large reservoir size, the oxidation of OCpetro may be an important flux of CO2 to the atmosphere over geologic time. However, determining the magnitude and drivers of OCpetro oxidation remains difficult at the watershed scale. To understand the balance of OC sources eroded from a landscape and prone to breakdown, we measured the OC concentrations, stable isotope ratios, and the radiocarbon content of suspended sediment samples from the East River: a shale-dominated alpine watershed near Crested Butte, Colorado. Suspended sediment samples were studied because they integrate multiple sources of sediment across a watershed. With these OC measurements, we employ a Monte Carlo mass balance inversion to constrain inputs of carbon from plants, soil, and bedrock. We find that along the main stem of the river, δ13C values range from -27 to -26 ‰ VPDB and fraction modern OC (FmC) ranges from 0.73 to 0.95, suggesting millennial-scale floodplain residence times. In contrast, lighter δ13C values (-29 to -27 ‰) and older FmC values are observed within tributaries. Notably, samples collected at different river stages at the downstream-most locality exhibit the widest variability in carbon isotope data. A decrease in sediment OC abundances and FmC values with increasing discharge is likely the result of an increased proportion in bedrock carbon in transport. These data illuminate hydrologic and geomorphic controls on the balance of OC export, which has consequences for watershed-scale OC fluxes. Whereas the river main stem primarily transports young, labile, plant- and soil-derived organic carbon, small steeper tributaries appear to transport older, rock-derived organic carbon, likely due to erosion of bedrock. However, the relative contributions of bedrock and soil-derived carbon in floodplains can fluctuate in response to changing hydrologic states, which are influenced by climate. Ongoing measurements of rhenium and carbon isotopes on suspended sediment, bedrock, and soil will help refine endmembers and shed light on the role of bedrock OC in riverine carbon fluxes.

Recent modeling and observational studies of the terrestrial hydrogen exosphere report that during a geomagnetic storm, exospheric density can increase by up to 60% in the 2 – 10RE radial distance range, driven by dynamic upper atmospheric heating. This density enhancement suggests an important role of the exosphere in inner magnetospheric dynamics, as neutral – ion charge exchange is a key energy loss mechanism for ring current ions. To investigate the interaction between the exosphere and the ring current, we conduct a geomagnetic storm simulation using the Model for Analyzing Terrestrial Exosphere (MATE), coupled with the Thermosphere – Ionosphere – Mesosphere – Electrodynamics General Circulation Model (TIME-GCM) and the Comprehensive Inner-Magnetosphere Ionosphere (CIMI) model. Strong high-latitude forcing during the storm heats the global upper atmosphere, increasing the number of ballistic and escaping hydrogen atoms entering the exosphere, thereby enhancing exospheric density in the inner magnetosphere. Subsequently, charge exchange between neutral hydrogens and ring current ions intensifies, accelerating energy decay in the ring current and reducing ion precipitation from the ring current to the upper atmosphere. These simulation results suggest a non-negligible role of the exosphere in magnetosphere – ionosphere – thermosphere coupling through neutral ion charge exchange. Upcoming NASA missions such as the Carruthers Geocoronal Observatory and the Storm Time O+ Ring Current Imaging Evolution (STORIE) will help illuminate these important physical processes.

Numerous cargo and tanker vessels every year transit the Bering Sea, and with decreasing sea ice in the Arctic, increasing levels of travel through the Bering Strait to new shipping routes is a contemporary reality. Increased shipping in the region not only intensifies the impact to remote communities but also causes concern for the safety of vessel operators and the environment. Analyzing existing policy within the Bering Sea is useful to determine policy effectiveness as maritime activity is amplified in the region. To examine changes in traffic patterns from 2015 to 2022, and therefore the effectiveness of existing policy, hot spot analysis (Getis-Ord Gi*) was conducted on the Bering Strait ‘Areas to be Avoided’ (ATBAs), implemented in 2018 as a part of the International Maritime Organization’s Polar Code. Additionally, hot spot analysis (Getis-Ord Gi*) was conducted on vessels transiting the region every year (8 of 8 years), frequently (4–7 years), or rarely (1–3 years) to investigate aspects of maritime activity that can impact adherence to the ATBA policy. The analysis conducted on vessel traffic in the Bering Strait from 2015 through 2022 shows that the ATBA policy influenced all vessel types. Determining the effectiveness of current policy in maritime activity in the Bering Strait will be necessary for future policymaking to protect this unique region of Alaska, as well as for the communities who have inhabited it before recorded history.

The Local Air Emissions Tracking Atlas (LOCAETA) is a tool to quantify and understand the co-benefits of industrial decarbonization. It includes data harmonization, emissions scenario development, and modeling. Its interactive platform was designed to help communities and decision-makers across the US understand the long-term air quality and public health impacts relative to various decarbonization approaches, such as carbon capture, electrification, and hydrogen fuel-switching. In this work, we explore the air dispersion modeling component of LOCAETA using HYSPLIT and InMAP for facility PM2.5 emissions. We present a case study of the Denver, Colorado region and connect “chronic-exposure” plumes from several large industrial emitters to available census and household demographic information, which supplements LOCAETA’s public health modeling and source apportionment results. This work can benefit decision-makers in all levels of government, as well as communities advocating on their own behalf, clean energy investors, and stakeholders interested in decarbonization.

Accurate hydrological simulation of the Laurentian Great Lakes-the world’s largest surface freshwater system-is critical for effective water resource management, yet remains challenging due to the basin’s vast size, international boundaries, and data heterogeneity. While deep learning models such as long-short-term memory (LSTM) networks have shown promise in hydrological forecasting, a unified, data-driven model for daily streamflow prediction across the entire Great Lakes basin has not been realized. This study addresses this gap by developing and evaluating an Entity-Aware LSTM (EA-LSTM) model, trained on a newly compiled dataset spanning 975 sub-basins across the U.S. and Canada from 1980 to 2023. The dataset integrates daily meteorological forcings, static catchment attributes, and observed streamflow. The EA-LSTM, using a temporal training split, achieved a median Nash-Sutcliffe Efficiency (NSE) of 0.685, significantly outperforming both a standard LSTM (median NSE 0.567) and the operational NOAA National Water Model (median NSE 0.208). Performance analysis reveals that the model’s predictive skill varies by catchment characteristics, with reduced performance observed in very small and highly urbanized basins. Our findings demonstrate that a single, data-driven model can effectively simulate streamflow across a vast, international watershed without basin-specific calibration, offering a robust and unified tool for transboundary water resource management. This work highlights the potential of advanced deep learning models to enhance hydrological prediction and support transboundary water resources management.

GeoLatinas, a not-for-profit, international organization founded in 2019, promotes and supports Latina women in Earth and Planetary Sciences. Our organization has team-led initiatives focused on mentoring, education, and outreach for professional and personal development. It promotes sustainable training, and peers inspire individuals towards career advancement. GeoLatinas volunteers lead these efforts within the Leadership Council and Local Teams through co-leadership and peer mentoring practices, offering all members the opportunity to join existing or create new initiatives.Launched in 2020, the GeoLatinas Mentoring Team started with a one-to-one mentoring strategy. During this period, the team facilitated 50 mentor-mentee pairs matched according to the mentees’ aims and mentors’ intentions. Through this program, we retrieved insightful information that led us to understand our community’s needs and tailor our approach accordingly. Our priority was to build a safe space that supports our members and addresses the existing Latinx mentoring experiences, challenges, and opportunities literature gap. Hence, “GeoLatinas mentoring program: the process of creating a safe space to grow professionally by co-leadership” enclosed the results and lessons learned from this mentoring experience. This book’s chapter provides relevant and updated information on the status quo of Latinx in the geoscience field (Navarro–Pérez et al., 2025).More recently, between 2024 and 2025, we joined the Mentoring365 platform sponsored by AGU. Within this platform, we designed four different mentoring circles in Spanish, each lasting between two and six weeks. These sharing spaces boosted our community’s knowledge about mentoring, scientific writing and presentations, and the use of the platform itself. The platform closed permanently this year. Thus, our next steps include creating similar virtual places to discuss outstanding key topics related to authorship rights, life in academia as an immigrant, and mentoring at different career stages. This will benefit from the use of other free-sourced platforms (e.g., Slack) for accessibility of our international community. Thus, GeoLatinas remains committed to supporting and empowering Latinx communities in geosciences in response to recent challenging events.Launched in 2020, the GeoLatinas Mentoring Team started with a one-to-one mentoring strategy. During this period, the team facilitated 50 mentor-mentee pairs matched according to the mentees’ aims and mentors’ intentions. Through this program, we retrieved insightful information that led us to understand our community’s needs and tailor our approach accordingly. Our priority was to build a safe space that supports our members and addresses the existing Latinx mentoring experiences, challenges, and opportunities literature gap. Hence, “GeoLatinas mentoring program: the process of creating a safe space to grow professionally by co-leadership” enclosed the results and lessons learned from this mentoring experience. This book’s chapter provides relevant and updated information on the status quo of Latinx in the geoscience field (Navarro–Pérez et al., 2025).More recently, between 2024 and 2025, we joined the Mentoring365 platform sponsored by AGU. Within this platform, we designed four different mentoring circles in Spanish, each lasting between two and six weeks. These sharing spaces boosted our community’s knowledge about mentoring, scientific writing and presentations, and the use of the platform itself. The platform closed permanently this year. Thus, our next steps include creating similar virtual places to discuss outstanding key topics related to authorship rights, life in academia as an immigrant, and mentoring at different career stages. This will benefit from the use of other free-sourced platforms (e.g., Slack) for accessibility of our international community. Thus, GeoLatinas remains committed to supporting and empowering Latinx communities in geosciences in response to recent challenging events.

Low, mixed-phase clouds are common in polar regions and have a large impact on the surface radiative balance. One proposed method to mitigate warming during polar night is to add ice nucleating particles (INPs) to these clouds. The concept for this approach, known as mixed-phase cloud thinning (MCT), is that if liquid droplets are converted to ice particles, clouds might become optically thinner and could disappear faster, causing net cooling at both the surface and the top of the atmosphere (TOA) by decreasing downward longwave radiation to the surface and increasing longwave radiation escaping to space. However, this approach is still theoretical because modeled polar mixed-phase cloud processes are often not well-constrained.

The deployment of cameras and ground penetrating radar (GPR) on unmanned aerial vehicles (UAVs) enables the joint capture of high-resolution surface structure and subsurface features. Structure-from-Motion (SfM) photogrammetry generates dense 3D point clouds and surface models, while GPR provides depth-resolved information on buried features. When integrated, these complementary datasets may offer a powerful means to understand spatial relationships across surface and subsurface domains. However, SfM and GPR data can differ significantly in spatial density, resolution, and modality, posing challenges when being combined to support 3D modeling and interpretation. This work explores a sensor fusion framework that combines SfM-derived surface models and UAV-based GPR data to create unified 3D surface-subsurface representations. The approach can be used for cross-domain feature extraction, allowing surface features (e.g., stones placed over a historical site) to help inform subsurface interpretation. The proposed workflow addresses co-registration and alignment of SfM point clouds and GPR volumes, supporting AI detection and classification of features of interest (e.g., buried structures, voids, or boundaries). Applications include 3D representation that can be used to prepare systems to learn joint patterns across the two domains.