This session consists of lightning talks highlighting research being performed by members of ASPRS Student Chapters around the country, under the guidance of their faculty advisors.
2:00 PM – 2:12 PM – Lidar Derived Road Profiles to Identify Low Clearance Railroad Crossings
Automatic evaluation of roadway and rail vertical geometry is essential for identifying grade crossings where low ground clearance vehicles are at risk of becoming immobilized. Despite long-standing design guidance, most crossing inventories do not include detailed vertical profile information, and research on low clearance or hump crossings has relied mainly on manual inspection or limited field studies. As a result, many hazardous crossings remain unclassified, and the use of modern high-resolution LiDAR for this purpose has received little systematic attention.
To address this gap, we develop an automated workflow that uses classified LiDAR, vector road and rail data, and dense transect sampling to derive elevation profiles at road and rail intersections. A unified clearance metric based on AASHTO (American Association of State Highway and Transportation Officials) tolerances at the rail, at 2 feet, and at 30 feet provides consistent categorization of crossings as at grade, low clearance, or superelevated without requiring incident data.
Results show that LiDAR derived profiles reliably detect geometric conditions associated with vehicle hang ups, conditions that are not captured in current inventories. This work establishes a scalable foundation for evaluating grade crossing vertical geometry and supports future comparison as automated methods continue to evolve.
Meghan Touat, University of Maryland
2:12 PM – 2:24 PM – Capturing the King Tide Impact: A Case Study on Quantifying Coastal Erosion Using Terrestrial Laser Scanning
As sea levels rise and extreme weather events become more frequent, the ability to rapidly quantify coastal erosion is important for hazard resilience planning. This case study presents a high-resolution change analysis of the bluff and beach systems at Beverly Beach, Oregon, capturing the immediate impacts of the November 2025 King Tide event. Erosion at Beverly Beach is of immediate concern to the stability of adjacent Hwy 101, making these surveys a key factor in vulnerability assessments.
Data was acquired in two discrete epochs: a baseline scan on November 2, 2025, and a post-event scan on November 11, 2025, capturing morphological change from the November 5–7 King Tide. Utilizing the Leica RTC360 terrestrial laser scanner, we captured dense 3D geometry of the unstable bluff face and intertidal zone. The study details the field-to-finish workflow, emphasizing the efficacy of the RTC360 for rapid data collection in dynamic, hazardous coastal environments.
By performing a cloud-to-cloud comparison between the pre and post datasets, we identified discrete areas of bluff and beach erosion and accretion. The results demonstrate how terrestrial scanning can provide a precise dataset to assess structural integrity and predict change rates of the bluff after a high-energy tidal event.”
Sheng Tan and Heather Maran, Oregon State University
2:24 PM – 2:36 PM – Event-Based Spatio-Temporal Graph Convolutional Networks for Rainfall–Runoff Prediction in HUC12 Watersheds
This research develops an event-based Spatio-Temporal Graph Convolutional Network (E-STGCN) designed to improve short-term flood forecasting by combining watershed connectivity with recurrent sequence modeling. The approach represents the river network through a directed topological graph, allowing the model to follow true upstream–downstream flow paths, while an LSTM backbone captures the temporal evolution of storm events. Training the model on individual rainfall–runoff events proved especially effective, enabling it to learn the nonlinear rise and recession characteristics that dominate flood response. Among the tested configurations, the combination of event-based training, a physically informed graph structure, and an MSE loss function delivered the most stable and accurate outcomes, reaching an R2 of 0.9973 across the evaluation basins. The workflow is computationally efficient and supports near-instantaneous inference, demonstrating clear potential for operational use in real-time flood warning and watershed management systems.
Anil Mandal, Florida Atlantic University
2:36 PM – 2:48 PM – Building a Database of Land Subsidence Targets Under SGMA to Help Evaluate California Central Valley’s Future Sustainability
The San Joaquin Valley in Central California plays an integral role in the United States’ agricultural industry, providing about 25% of food in the U.S. alone. The Central Valley’s agricultural industry is largely dependent on groundwater and has led to severe over-pumping in most parts of the basin, especially southern regions. The consequential passage of California’s Sustainable Groundwater Management Act (SGMA) in 2014 introduced a long-term plan aimed at protecting groundwater resources. SGMA, however, has yet to deter unsustainable management practices from being utilized, these practices having led to excessive groundwater pumping, a principal cause of land subsidence. Without effective management strategies and the regulation of groundwater pumping, subsidence could damage key infrastructure in the valley, including the California Aqueduct. This report will analyze SGMA reports of various groundwater subbasins and include a constructed database of SGMA subsidence monitoring stations that will eventually be used to model projected subsidence levels due to groundwater depletion up to the year 2070. The subbasins analyzed will be Delta-Mendota, Chowchilla, Madera, Tule, Turlock, and Modesto respectively. Subbasins will be ranked in order of highest priority and subsidence severity to lowest. A total of 178 subsidence monitoring sites comprises the constructed subsidence database, and 107 of these sites, more than half, have sustainability targets that use groundwater levels as a proxy for subsidence. This research aims to determine the regions of most severe subsidence and evaluate the effectiveness and sustainability of current SGMA groundwater sustainability plans.
Zoe De Buhr, University of Wyoming
2:48 PM – 3:00 PM – Tracking Mining and Deforestation in SE Ghana Using Landsat Imagery
This project examined deforestation linked to illegal gold and bauxite mining (galamsey) in Atewa Forest between 2015 and 2024 using multi-temporal remote sensing analysis. Landsat 8 imagery was employed to detect vegetation loss and land-cover change associated with mining activities over time. Image processing and analysis were conducted using ERDAS Imagine, including preprocessing, band combinations, and change detection techniques to identify spatial patterns of forest degradation. To improve spatial accuracy and ground relevance, Google Earth Engine (GEE) was used to visually correlate satellite-derived outputs with high-resolution imagery, allowing for the verification of disturbed areas and the alignment of detected changes with known or observable mining locations. The analysis revealed expanding patches of deforestation and landscape disturbance consistent with illegal mining encroachment within and around the forest reserve. By combining satellite-based analysis with cloud-based visualization and validation, the project demonstrates the value of remote sensing as an empirical tool for documenting environmentally destructive extractive activities in protected forest landscapes and supporting evidence-based environmental governance in Ghana.
Ransford Tege, University of Wyoming
3:00 PM – 3:12 PM – Evaluation of 4D Alignment Techniques for Repeat UAS-SfM and UAS-Lidar Surveys
Uncrewed aircraft systems (UASs) provide cost-effective solutions for surveying and mapping of roadway corridors, thereby supporting tasks such as progress monitoring, deformation identification, and asset inventory and management. In the transportation sector, performing accurate surface change quantification from repeat UAS surveys remains a challenging task due to errors such as dataset misalignment, scene dynamics, and variability in global navigation satellite system (GNSS) quality in the survey area. This study stems from a project with the Texas Department of Transportation (TxDOT) that evaluated UAS-based structure-from-motion / multi-view stereo (SfM/MVS) photogrammetry, or UAS-SfM, and light detection and ranging (lidar), or UAS-Lidar, technologies in support of land surveying activities. Specifically, the research evaluates the efficiency of four-dimensional (4D) alignment techniques to improve the relative accuracy of multi-epoch datasets and surface change detection analysis, with focus on the vertical component. Experiments relied on repeat acquisition of UAS-SfM and UAS-Lidar point cloud datasets over a concrete runway surface mimicking a typical two-lane Texas State right-of-way (ROW) corridor. The data was georeferenced using post-processed kinematic (PPK) GNSS, and the 4D alignment workflow was compared against standard three-dimensional (3D) processing both with and without ground control points (GCPs). Compared to PPK-only 3D processing, 4D alignment reduced surface change detection errors (i.e., vertical root means square errors (RMSEz) of DTMs of difference) from 2.62 cm to 0.70 cm for UAS-SfM and from 1.62 cm to 0.52 cm for UAS-Lidar. Compared to 3D processing using PPK corrections + 4 GCPs, 4D alignment yielded error reductions of 0.72 cm (UAS-SfM) and 0.36 cm (UAS-Lidar). Standard deviation errors and mean errors also decreased consistently across all scenarios when using 4D alignment. In addition, cloud-to-cloud (C2C) distance also underscored these improvements, with corresponding C2C distance errors reducing by 1.92 cm for UAS-SfM and 1.10 cm for UAS-Lidar. These findings demonstrate the efficacy of 4D alignment for high-precision monitoring of ROW corridors, particularly in scenarios where GCP deployment is impractical or unsafe.
José Pilartes-Congo, Texas A&M University-Corpus Christi
3:12 PM – 3:24 PM – Analysis of Satellite-Based Kd490 Products for Quantifying ICESat-2 Bathymetry Retrievability
The space-based photon-counting laser altimeter (ATLAS) aboard the ICESat-2 provides new capacities for global shallow-water bathymetry. In April 2025, a new global ICESat-2 bathymetric data product, ATL24, was released. However, the depth to which ICESat-2 can reliably detect the seafloor is strongly dependent on water clarity. The basic model used in airborne bathymetric lidar assumes that the maximum depth for the detection of the bottom is an inverse function of the diffuse attenuation coefficient of downwelling irradiance, Kd. It is therefore reasonable to assume that global Kd490 (diffuse attenuation coefficient at 490 nm) datasets should be useful in predicting ICESat-2 bathymetric measurement success and maximum depths. However, with a number of Kd490 data products available, each using different source data and/or algorithms, several questions remain to be answered: 1) do different Kd490 data products differ in their ability to predict successful bathymetric measurement? 2) Does including a rough estimate of seafloor reflectance improve the predictions? 3) Using the best data and techniques, how well can we predict maximum depth retrievals for ICESat-2? In this study, we investigate these questions using ATL24 and three well-known global Kd490 datasets.
Owusuah Osei-Kwakye, Oregon State University
