Floodlines

When Disaster Funding Follows the Past, Not the Future

Floodlines is a geospatial analysis and interactive dashboard exploring how flood risk, social vulnerability, and FEMA mitigation funding align across Vermont’s 250+ towns. By combining GIS, Census demographics, and federal disaster datasets, the project investigates a growing climate-policy question: are vulnerable communities actually receiving the support they need?

Project Links

Live Website

Data Analysis

GitHub Repository

I grew up in Vermont, where rivers only jumped their banks in a freak January thaw. After Irene in 2011, there was a sense that the state had survived extraordinary disaster together — “Vermont Strong”. When the floods returned in 2023, that feeling curdled into something more sober: no once-a-century disaster, just a recurring condition. Floodlines grew out of that vibe shift. And I got to indulge my interest in maps, public policy, systems thinking, and a pinch of institutional skepticism to explore a question that’s only going to become more urgent: how do institutions shape climate resilience and disaster preparedness?

Project Overview

After Tropical Storm Irene and Vermont’s historic 2023 floods, it became increasingly difficult to view flooding as a rare disaster rather than a structural reality. At the same time, new floodplain remapping and climate adaptation debates raised larger questions about how mitigation money is distributed — and whether funding systems are responding to future risk or simply reacting to past damage.

Using Python, GeoPandas, FEMA datasets, ACS demographics, and spatial analysis tools, I built a reproducible pipeline measuring flood exposure, social vulnerability, FEMA mitigation investment, and funding gaps across Vermont towns. The dashboard combines choropleth mapping, interactive rankings, quadrant analysis, and model comparisons to make complex spatial patterns explorable to both technical and non-technical audiences.

The analysis found that FEMA mitigation funding was only weakly correlated with modeled structural need, while past insurance claims were a much stronger predictor of investment. Roughly 55–60% of Vermont towns appeared underfunded relative to measured need, suggesting disaster funding systems may operate more reactively than proactively.

Technically, the project demonstrates geospatial ETL workflows, composite index construction, sensitivity analysis, spatial statistics, and interactive dashboard development using Leaflet and D3. More importantly, it reflects the kind of work I most enjoy: combining environmental systems, public policy, spatial data, and public-facing storytelling to translate complicated problems into something people can actually understand and use.

Tools & Technologies:

  • Frontend: JavaScript, Leaflet.js, D3.js, HTML/CSS
  • Backend & Analysis: Python, Pandas, GeoPandas, NumPy, Jupyter Notebook
  • Spatial Analysis: Shapely, libpysal, ESDA (Moran’s I)
  • Statistics & Modeling: scipy, scikit-learn, statsmodels
  • Data Sources: FEMA, OpenFEMA, NFIP, U.S. Census ACS, Vermont ANR, FRED CPI
  • Hosting: GitHub Pages

Gallery



Default view — Quadrant map All Vermont towns classified into funding quadrants (high/low funding, high/low need, zero-funding)

Need index choropleth Measuring Need: A rank-based composite of flood exposure and social vulnerability

Funding gap choropleth Towns where need substantially exceeds FEMA HMA investment, highlighting the most underserved communities

Scatter plot — Need vs. Funding Loose cloud confirming weak alignment between structural need and federal mitigation dollars

Town detail panel Per-town statistics, percentile rankings, and quadrant classification for a selected community

Relative toggle — EAL risk Risk map rescaled to show deviation from the Vermont statewide average rather than absolute percentile rank

NFIP claims overlay Reactive benchmark layer showing where insured losses occurred vs. where modeled risk is highest

References

Back to Top Home