Discover the latest academic research and innovations during this interactive poster session. Engage with students directly to discuss new ideas and emerging technologies shaping the field.
10:00 AM – 10:05 AM – Spatiotemporal Patterns of Tidal Creek Expansion and Riparian Mangrove Encroachment in the Southern Everglades
Sea-level rise is driving significant landward expansion of tidal creeks in low-relief coastal plains, reshaping hydrologic connectivity, shifting salinity gradients, and promoting the encroachment of halophytic communities into freshwater wetlands. This study examines these processes in the Shark-Harney River system located in the western coastal plain of Everglades National Park within the Florida Coastal Everglades Long Term Ecological Research (FCE LTER) domain. We quantified the spatiotemporal landward expansion of the tidal creek network and modelled the spatiotemporal change in creek expansion metrics, and riparian mangrove cover as a function of distance from the coastline.
Tidal creek expansion was mapped by progressively segmenting the current extent of the creek system, using historic aerial stereo and ortho photography (AP) mosaics covering six time intervals from 1952 to 2018. Creek segmentation and digitizing was performed utilizing the DAT/EM Stereo Plotter and ArcGIS Pro clipping tool to map the evolution of the creek network. Creek density was quantified using a spatial line density analysis and modeled as a function of distance to the coast using generalized additive models (GAMS) for each time interval. Creek width and riparian mangrove cover changes were evaluated by generating 500-m transects along visible creeks and segmented using the aerial photography as reference. Linear models (LM) were used to quantify the relationship between proportional creek width change and riparian mangrove change as a function of distance from the coastline.
Preliminary results showed the creek system lengthened and branched landward by ~ 118 kilometers since 1960, with density increased (0.14-0.31 meters/hectare) most pronounced at intermediate to long distances (12-20 km) from the coastline between 1952 and 1984, followed by a slower rate of expansion thereafter. From 1999 to 2021, creek width decreased by 8.4% per kilometer inland, while riparian mangrove cover increased by 8.2 m per kilometer. These findings highlight the coupled dynamics of hydrologic change, vegetation succession, and geomorphic processes in response to sea-level rise and saltwater intrusion. The transect-based methodology proved more effective for detecting mangrove cover encroachment than for creek width changes, due to image resolution limitations and canopy obstruction of visible creek shores.
Jessika Reyes, Florida International University
10:05 AM – 10:10 AM – Monitoring Soil Subsidence Using Airborne Lidar Data in the Everglades Agricultural Area, Florida
Soil subsidence in the Everglades agricultural area (EAA) has been a persistent issue since the region was drained for cultivation. Historically, the highest rates of subsidence occurred due to microbial oxidation of organic soils following exposure to atmospheric oxygen. While subsidence rates have declined in recent decades, largely due to the adoption of best management practices, the process continues, and comprehensive temporal monitoring remains limited.
We utilized publicly available airborne laser scanning (ALS) data from the USGS 3D Elevation Program (3DEP), with datasets from 2007 (3 m resolution) and 2016 (1 m resolution), to estimate average elevation loss across the study area. To improve temporal resolution and validate our findings, we conducted a new ALS survey in June 2025 and collected high-accuracy RTK GPS measurements for ground truthing.
This study highlights the potential of integrating historical and modern elevation datasets for long-term monitoring of land surface dynamics in the EAA, with implications for land management and environmental conservation.
Junaid Lone, University of Florida
10:10 AM – 10:15 AM – Subpixel Monitoring of Land-Cover Changes Using Spectral Mixture Analysis and Trend Analysis Techniques
Monitoring land use and land cover change is essential for effective conservation and management in national parks such as Addo Elephant National Park in South Africa. Traditional remote sensing methods use discrete classification, assigning each pixel in a satellite image to a single land cover class. This approach often misses gradual changes and mixed signals, particularly in areas with complex or heterogeneous landscapes.
In this research, a continuous approach is used by applying multi-endmember spectral mixture analysis (MESMA) to Landsat imagery from 1986 to 2024. Pure spectral endmembers for main land cover types such as thicket, grass, and bare ground are identified from high-resolution reference data. Each Landsat image is unmixed to create fraction maps that show the proportion of each cover type within every pixel. Validation is conducted using PlanetScope 3 meter data, with RMSE calculated between the fraction estimates and actual values.
The time series of subpixel fractions is analyzed with the Breaks for Additive Season and Trend (BFAST) algorithm to detect trends and classify areas as having increasing, decreasing, or stable land cover. The results of this continuous approach are then compared with discrete approaches using machine learning and deep learning methods, including Random Forests and UNet++, to evaluate differences in detecting and characterizing land cover changes. This comparison highlights how continuous, subpixel monitoring can capture subtle landscape changes often missed by traditional discrete methods, improving our understanding of landscape dynamics and supporting better conservation strategies for national parks.
Mohammad Safaei, University of Florida
10:15 AM – 10:20 AM – Assessing Socio-Economic Influences on the Spatial Distribution of Building Damage From Hurricanes Helene and Milton
This study underscores the differential impact of disasters on communities with varying socio-economic conditions, demonstrating that disparities in damage levels are influenced not only by the physical intensity of hazards but also by communities’ capacity to prepare for and respond to such events. Specifically, this research examines the spatial distribution of building damage resulting from Hurricane Milton in 2024 and explores the socio-economic disparities associated with this damage. Utilizing FEMA damage assessment data, the study conducts a spatial analysis of building damage patterns. A statistical approach is then employed to evaluate the relationships between building damage, hurricane intensity, and socio-economic factors at both the building and census tract levels. Furthermore, geographically weighted regression (GWR) is applied to assess the spatial heterogeneity of factors influencing damage severity. The findings offer critical insights into the key determinants of building damage from events such as hurricanes and their spatial variability. By identifying the most vulnerable communities, this study provides valuable information for policymakers, emergency management officials, and community leaders to develop targeted strategies aimed at mitigating the impacts of future hurricanes.
Najiba Rashid, Oregon State University
10:20 AM – 10:25 AM – Landslide Detection via Change Detection Annotations and Deep Learning Segmentation
Landslides are a significant natural hazard capable of causing severe damage to life, infrastructure, and property. Accurate and timely detection is essential for mitigating their impact. While recent advancements in deep learning have shown considerable promise in landslide detection, challenges such as limited annotated data and poor model generalizability across different terrains and conditions continue to hinder progress.
This study addresses these challenges by leveraging annotated data generated through unsupervised change detection techniques applied to pre-event and post-event satellite imagery. Focusing on a case study in Western North Carolina, we use imagery from Hurricane Helene and existing landslide inventory data to create a training dataset for deep learning segmentation models. The approach includes data preparation, annotation using change detection outputs, model training, fine-tuning, and comprehensive evaluation.
Preliminary results indicate that the use of change detection-derived annotations enhances model performance, while fine-tuning further improves landslide detection accuracy. These findings highlight the value of combining automated annotation techniques with tailored deep learning models for more effective landslide identification.
This research contributes to the development of more accurate and reliable landslide detection systems with potential applications in early warning systems, disaster response, and risk mitigation planning.
Acknowledgement: This material is based on work supported by the North Carolina Department of Transportation, Project # RP 2023-04 and National Science Foundation, Project # 2401942.
Gazali Agboola, North Carolina A&T State University
10:25 AM – 10:30 AM – Analyzing Emergency Medical Service Response Times in Rural Vermont
The purpose of this project is to develop a model for ambulance response times in Vermont, allowing for response times to be predicted for any location in Vermont. This response time model will be validated using real ambulance response times from the past 5 years. A statistical analysis will be conducted to determine in what areas the model is most accurate. The nuance of this study comes from its ability to go beyond past studies that have simplified the current model that the Vermont Department of Health references. These previous models are based on 25-minute drive times from each station. Any incident that falls outside of these areas is deemed to be within an ambulance desert. This study will break down the 2 main aspects of an EMS response, chute times and drive times, to more accurately map ambulance deserts and predict response times to any area of the state.
Jack Foster, The University of Vermont
10:30 AM – 10:35 AM – Tracking Terrorism Trends Across Africa
This study evaluates terrorism trends across Africa from 2015–2021 using Global Terrorism Database (GTD) incident data to assess changes in attack frequency, perpetrator group activity, and civilian targeting. Through geocoding, spatial joins, and time-enabled visualizations—including animated point maps, choropleth layers, and statistical charts—the analysis reveals a broad decline in overall incidents and extremist group activity during the study period. These trends correspond with expanding African Union counterterrorism initiatives, though causal relationships remain unconfirmed. The results highlight both emerging improvements and persistent vulnerabilities within the Sahel, where displacement, underfunded humanitarian needs, and regional instability remain severe. The study emphasizes the need for coordinated intervention, strengthened judicial systems, and sustained humanitarian support to address ongoing security challenges.
Linnea Carlsson, Embry-Riddle Aeronautical University
10:35 AM – 10:40 AM – From Roots to Leaves: Understanding Multi-Scale Trait Variation in Freshwater Wetlands
Wetlands are ecologically complex systems shaped by spatio-temporal variation in hydrologic, edaphic, and disturbance regimes, yet critical gaps remain in understanding how functional trait variation governs effects of these drivers on ecosystem resilience and stability. Hydrologic modification, nutrient enrichment, and climate change have altered disturbance regimes in many wetland systems, diminishing their resilience and leading to persistent shifts in community structure. Restoration efforts aim to reverse these trajectories, but success depends on understanding how plant communities respond to environmental change across spatial and ecological scales. Functional traits—morphological and physiological features that influence species’ fitness—offer a mechanistic lens for evaluating ecosystem resilience and predicting transitions. Drawing from trait-based ecological theory, this project integrates the response-effect trait framework to investigate the role of functional diversity across spatial (plot, patch, and landscape) and ecological (individual, species, community) scales in freshwater wetlands of the Florida Everglades. My overarching objective is to examine how plant functional traits influence the stability, resilience, and transitions of freshwater wetland communities during active restoration, using a holistic approach that integrates trait-based ecology and remote sensing. WorldView-2 imagery will be used to delineate wetland patch types based on the vegetation structure and dominant species composition across six landscapes representative of key environmental gradients. This integration of field-based trait data with remote sensing aims to enhance our understanding of ecosystem responses to hydrologic restoration at multiple spatial scales.
Carlos Pulido, Florida International University
