🌍 GeoGuard. Real-Time Earthquake Response Intelligence

Inspiration

Earthquakes don’t just cause destruction.
They cause chaos, confusion, and delayed response when every minute matters.

We already have live earthquake alerts.

But we still don’t have a system that turns a seismic event into real-time operational decisions:

Where should people go?
What resources are needed first?
Which zones are most dangerous?
What should responders prioritize in the first 30 minutes?

GeoGuard was built to bridge that gap.

What it does

GeoGuard is a real-time earthquake monitoring and emergency response intelligence platform that:

Streams live earthquakes from USGS
Visualizes epicenters on an interactive map
Overlays tectonic plate boundaries for context
Generates risk zones (high, medium, low) using AI + historical signals
Finds nearby emergency infrastructure (hospitals, shelters, fire, police)
Generates safe routes to critical locations
Produces an AI-generated emergency response plan and report
Provides a voice-based 911-style emergency agent
Includes a resource allocation layer for the first 0–48 hours
Uses Snowflake to load historical earthquake and risk context for smarter decisions

Key Features

1) Live Earthquake Map

Live USGS feed (latest quakes)
Epicenter marker + event list
Interactive map UI built for emergency dashboard clarity

2) Plate Boundaries + AI Risk Zones

PB2002 tectonic plate boundaries overlay
AI-generated zoning around the epicenter (red / orange / green)
Zones generated using:
- quake magnitude + depth
- tectonic context
- historical earthquake signals from Snowflake
Clear zone explanations for transparency

3) Emergency Infrastructure Discovery

GeoGuard identifies real-world nearby infrastructure using Overpass:

Hospitals
Shelters
Fire stations
Police stations
Staging areas

4) Safe Routes

Road routing via OSRM
Route selection + highlight on map
One-click open in Google Maps for real navigation

5) AI Response Plan + Report Generation

GeoGuard generates:

Situation summary
Priority action list
Safety-first citizen guidance
Responder-style decision support
A structured response report for the event

6) Chatbot for Immediate Response

A Gemini-powered assistant designed for crisis situations:

quick next steps
evacuation guidance
aftershock safety
first-aid direction
responder-style summaries

7) Voice 911-Style Agent (ElevenLabs)

GeoGuard includes a voice agent that acts like an automated 911 responder:

spoken emergency briefings
voice Q&A for citizens
voice guidance during high-stress situations

8) Resource Allocation Layer (AI)

GeoGuard estimates resource needs by time window:

0–30 minutes
30–120 minutes
2–12 hours
12–48 hours

Resources include: water, medical kits, shelter beds, radios, rescue teams, ambulances, and more.

APIs used in GeoGuard

USGS Earthquake API (no key)

Live earthquake events and event details.

Snowflake (historical context)

Used to load previous-year earthquake and risk context to support:

zoning
risk reasoning
smarter allocation logic

Google Gemini API (key)

Used for:

risk zone generation
emergency response planning
resource allocation reasoning
crisis chatbot responses
report generation
decision explanation

ElevenLabs API (key)

Used for:

voice-based 911-style agent
spoken emergency briefings
voice Q&A

OpenStreetMap tiles (no key)

Base map tiles.

Leaflet

Frontend mapping engine.

Overpass API (no key)

Used for nearby emergency infrastructure.

OSRM Routing API (no key)

Used for safe route generation.

PB2002 Plate Boundaries (public dataset)

Used for tectonic context.

How we built it

Frontend

Next.js + Tailwind CSS
Leaflet map with OpenStreetMap tiles
Map overlays: zones, infrastructure, routes
Clean dashboard UI optimized for crisis readability

Backend

FastAPI (Python)
USGS ingestion
Snowflake historical context loading
Overpass queries
OSRM routing
Gemini zoning + planning + reports
ElevenLabs voice agent integration

Deployment

Deployed on DigitalOcean using App Platform
Built to be production-ready and demo-ready

Challenges we faced

1) Balancing realism vs time

Earthquakes aren’t clean circles. We had to balance scientific realism with hackathon execution speed while still keeping the output structured and defensible.