This project combined a high-resolution Digital Elevation Model (DEM) with avalanche susceptibility data to identify terrain with elevated avalanche hazard potential. Terrain characteristics were visualized using shaded relief to emphasize elevation and surface morphology, while avalanche risk levels were classified from low to high susceptibility using a green-to-red color gradient.
The final map highlights how terrain analysis and spatial modeling can be used to support hazard assessment and improve understanding of avalanche-prone landscapes.
This project focused on analyzing and visualizing the progression of the Mauna Loa eruption lava flows in Hawaii using GIS techniques. By comparing flow rates over time, the study evaluated the potential timeline for when the advancing lava could impact the highway located north of the flow path.
The analysis revealed that the lava flow gradually decreased in speed over time, providing insight into eruption behavior and potential infrastructure impacts.
This project applied deep learning techniques in GIS to detect and estimate prairie dog burrow locations from aerial imagery. Training samples were manually labeled to teach the model how to recognize burrows under varying soil textures and surface conditions. Through iterative testing, the model achieved better performance when labels included both the burrow opening and the surrounding disturbed soil, providing additional spatial context for detection.
The model was trained and optimized using multiple epoch and batch size configurations while monitoring training and validation loss curves to evaluate performance and stability. The final model successfully identified 206 predicted burrow locations across the study area using object detection workflows in ArcGIS.
Results demonstrated the effectiveness of deep learning for wildlife habitat analysis while also highlighting common challenges such as false positives and missed detections. This project showcases experience with image classification, object detection, model evaluation, and spatial analysis using GIS-based deep learning tools.
Assessing Changes in Sagebrush Density Using UAS Imagery — Castle Rocks, Idaho
This project used high-resolution RGB and multispectral UAS imagery to analyze changes in sagebrush density within the Castle Rocks region of Idaho between 2022 and 2025. Supervised classification techniques in ArcGIS Pro were applied to identify and compare sagebrush cover, bare ground, non-sagebrush vegetation, and tree cover across both years.
Spatial analysis and change detection workflows revealed relatively stable overall sagebrush coverage, while exposed ground and non-sagebrush vegetation increased in several portions of the study area. These patterns may indicate localized vegetation thinning, environmental stress, or shifts in plant community composition. The project also evaluated the effectiveness of multispectral imagery for vegetation monitoring in semi-arid ecosystems and highlighted challenges associated with shadows, mixed pixels, and spectral overlap in remote sensing classification.
This study demonstrates experience with UAS data processing, supervised raster classification, multispectral analysis, land cover change detection, and environmental GIS workflows for ecosystem monitoring.
As part of a Beaver Dam Analog (BDA) restoration project near Wayan, Idaho, I collected baseline aerial data to document stream conditions before restoration. Using a DJI Matrice drone, I planned and conducted 10 flight missions at approximately 40 feet above ground level to capture high-resolution imagery of the stream and surrounding area. I processed the collected imagery into an orthomosaic, stitching individual images together to create a continuous, georeferenced aerial dataset. I then brought the processed imagery into ArcGIS Pro for further analysis, including generating a Digital Elevation Model (DEM) and examining stream morphology. The resulting datasets provide a baseline representation of the stream's existing conditions and can be used to help guide BDA placement and evaluate changes to the stream following restoration. This project allowed me to apply drone-based remote sensing, photogrammetry, GIS processing, and spatial analysis to a hands-on stream restoration effort.