GIS Day 2026

Hyper-scale data centers are expanding across the U.S., yet few tools show how their locations relate to communities with high social need. This project tests whether a GeoAI approach can screen for such communities nationwide using only two reference cities using the GeoAI embeddings tool. Using two cities and their vulnerability indexes as input data - Los Angeles (CDC/ATSDR Social Vulnerability Index) and Pittsburgh (Allegheny County Community Need Index) - the tool extracts high need areas and expands them across the entire country. The created dashboard assesses the proximity of the high-need cells to the nearest data center locations in order to prioritize areas, which might need greatest support in policy implementations.

Goal: Document the economic, social, and environmental value of various community-led
development projects in southwestern Pennsylvania
The Problem: Traditional impact reporting often focuses too narrowly on direct economic returns rather than indirect and long-term social, environmental and quality-of-life benefits to the community.
Scope: Covers a wide range of sectors including food systems, agricultural workforce development, arts, solar energy, transportation, and neighborhood revitalization.
Target Audience/Outcome: Designed to provide funders, policymakers, and the public with evidence to support expanded investment in community-led initiatives.

Where do impaired waterways and environmental justice concerns intersect? Developed through my volunteer GIS work with Three Rivers Waterkeeper, this project explores this question across the organization’s southwestern Pennsylvania jurisdiction.

The analysis combines Pennsylvania Department of Environmental Protection water quality assessments, environmental justice boundaries, and watershed boundaries in ArcGIS. Approximately 720 miles of impaired waterways intersect environmental justice areas, with overlaps occurring in 68 of 94 small watersheds. Aquatic life accounts for the largest share of impaired stream-use miles.

These spatial findings add community context to water quality assessments and can inform monitoring priorities, outreach, and watershed protection. For an organization with limited resources, mapping these intersections can help identify locations for closer investigation and support discussions about monitoring coverage and resource allocation.

The poster communicates these patterns through maps and charts for a broad audience. Findings demonstrate spatial overlap but do not establish causation, residents’ exposure, or higher impairment rates within environmental justice areas. Further investigation can combine these results with monitoring data and community knowledge to guide action.

Navix starts with a simple spatial question: is the shortest route always the right one? For the 1 in 4 U.S. adults living with a disability, one missing curb ramp, a steep block or a closed sidewalk can make a six-minute walk impossible, and today’s maps give no warning. Navix is a GeoAI pedestrian layer that reads the city the way different bodies experience it. Piloted in Downtown Mountain View, it fuses the city’s GIS sidewalk network with deep-learning semantic segmentation of 0.30 m aerial imagery, text-prompted SAM 3 segmentation of Mapillary street-level photos, USGS elevation for running grade, and live city closure and 311 data. All of this goes into one edge table where every attribute carries a source and a confidence. Profile-specific cost functions for wheelchair, low vision and stroller users then re-weight the same network, and a least-cost routing engine returns a different route for each, labelled verified, unverified or blocked. Early results show 30 of 38 aerial checks correct with no false positives, and 83–100% precision on street-level detection. The result is not one map for everyone, but one city with different paths.

My submission uses maps to identify the effects that the Home Owner's Loan Corporation's (HOLC) neighborhood grades from 1937 have had on current neighborhood racial distribution and current Neighborhood Median Property Value (NMPV).

What I found was that gentrification and white flight have led neighborhood racial demographics to diverge from the historical distribution determined by HOLC Grades (A, B, C, or D). A Grades were the best and were given to wealthy and White neighborhoods, while D Grades were the worst and were given to poor Black neighborhoods. People in A Grade neighborhoods were able to easily get favorable mortgages while those in D Grade neighborhoods were almost never able to get a mortgage.

Additionally, I found that the NMPVs for neighborhoods of each grade have ranges hundreds of thousands of dollars wide, and that where a neighborhood's NMPV falls within the range of their grade generally corresponds to the neighborhood's current racial majority.

My conclusions from my maps and findings were that HOLC neighborhood Grades no longer strongly indicate current neighborhood demographics or NMPV. Instead, current neighborhood demographics have a much stronger effect on NMPV.

How can geospatial analysis inform disaster recovery policies and more realistic budget coverage? SEACyclone 2025 explores this question by mapping potential flood exposure in Indonesia, Malaysia, and Thailand following extreme rainfall in November 2025.

The project combines rainfall, population, and land cover data in ArcGIS. Approximately 2,006 km² met the model’s exposure criteria, with the largest share in Indonesia. Recovery cost scenarios explore how different land cover types and reconstruction assumptions affect potential funding needs.

The poster connects spatial findings with policy recommendations and budgeting: identifying locations for closer assessment, distinguishing agricultural restoration from infrastructure reconstruction, and considering how limited funds could cover different recovery needs. This approach illustrates how geographically targeted budgets could better reflect local conditions and support more realistic service and recovery coverage.

Motivated partly by flooding in my home region of North Sumatra, the project demonstrates how GIS can communicate disaster information to nontechnical audiences. Results represent potential exposure, while flood extent, damage, and cost assumptions require further investigation before guiding funding allocations.

Three-fourths of Americans have companion animals (pets). These animals require ongoing veterinary checkups and vaccinations which improve quality of life and reduce zoonosis (transmission of disease from animal to human). However, veterinary care is difficult to access due to financial models of care (private equity), lack of transportation, as well as veterinary deserts where no care exists. Animal Friends community outreach offers vaccination clinics in the summer and fall to meet the need for vaccinations. This research poster asks the questions: (1) where are people coming from in relation to the clinic location; (2) is there a relationship between clinic locations --. are people coming across the county for vaccinations according to distance theory (distance rather than need driving access)? This research poster displays six maps of each clinic as well as the flow diagram and a map of the veterinary landscape of the county in order to answer these questions. All maps were created using ArcGIS through the University of Pittsburgh ULS and the School of Social Work, and Animal Friends. The maps suggest that veterinary need is the motivation for owners driving long distances, and in particular the need for a repeat (booster) vaccination. Suggestions for future research as well as clinic management are included in this poster.