Real-time wildfire monitoring powered by satellite intelligence — bringing live fire detection, weather conditions, and region analysis onto an interactive 3D globe for first responders.

Built for the Upstart Competition (GCES Concordia) under the themes of Health & Wellness and Smart Systems.

SparkGuard pulls live data from 5 free public APIs and displays it on an interactive 3D globe. Click any fire hotspot on Earth and instantly get:

• Where it is — real-time satellite-detected fire location, coordinates, and nearest populated area
• What's burning — biome, vegetation type, fuel type, and land cover pulled from OpenStreetMap
• How dangerous it is — fire intensity (MW), dryness index, population risk, and protected areas nearby
• Current weather on scene — temperature, wind speed & direction, gusts, humidity, and precipitation from MET Norway
• How the fire is contained — official fire perimeter boundaries and containment status for US and Canadian fires

Everything is live. No mock data, no historical datasets, no simulations.

Every piece of information displayed comes from a real API queried in real time:

API keys: Only NASA FIRMS requires a free API key. All other services are completely open.

• Full-viewport Three.js globe with NASA Blue Marble textures
• Fire hotspots colored by intensity — green (low) → yellow → orange → red (extreme)
• Click any hotspot to select it — the globe stops rotating, zooms in, and highlights the point with a bright pulsing ring
• Fire perimeter polygons overlaid for US (red) and Canadian (blue) fires
• Pulsing rings on high-intensity fires (FRP > 50 MW) for visual emphasis
• Smart downsampling — renders up to 2,000 points for smooth performance while preserving the most intense fires

• Key facts — date, intensity (MW), confidence (%), severity rating, brightness temperature, day/night
• Region Intelligence (live) — biome, vegetation, fuel type, land cover, dryness index (0–100 bar), population risk level, nearest place, elevation, protected areas
• Weather (live) — current conditions + 24-hour forecast in 3-hour intervals + wind risk assessment
• Responder Brief — one-click generated summary combining all live data into a structured briefing document
• Export — download full detection data as JSON

• Time range selector: 24h / 48h / 7 days of fire data
• Fire perimeter toggle with source filter (US / Canada / Both)
• Detection count and last-updated timestamp
• Manual refresh

NASA FIRMS API ──→ VIIRS CSV ──→ Parse & downsample ──→ 3D Globe points
                                                          │
User clicks a point ──→ Globe stops + highlights ─────────┤
                                                          │
                    ┌─────────────────────────────────────┘
                    ▼
              InfoPanel loads (in parallel):
                ├─ MET Norway ──→ Weather + 24h forecast
                ├─ OSM Overpass ──→ Land use, biome, protected areas
                ├─ OSM Nominatim ──→ Country, city, population
                └─ Open-Meteo ──→ Humidity, elevation → dryness index
                    │
                    ▼
              Responder view with all live data

All API responses are cached in memory (weather: 15 min, region data: 30 min, FIRMS: 5 min) so repeated queries are instant and rate limits are respected.

• Faster situational awareness — responders see fire conditions before arriving on scene
• Reduced smoke exposure — earlier detection enables smaller, faster interventions
• Responder safety — wind, humidity, and terrain data help assess risk before deployment

• Prevention-oriented — early satellite detection enables containment while fires are small
• Reduced burned area — faster response protects ecosystems and reduces carbon emissions
• Efficient resource allocation — severity and region data help prioritize response

• Node.js 18+ and npm
• A modern browser with WebGL support (Chrome, Firefox, Safari, Edge)

# Clone
git clone https://github.com/Lekerj/SparkGuard.git
cd SparkGuard

# Install dependencies
npm install

# Start development server
npm run dev

The app opens at http://localhost:5173.

You need a free NASA API key to load live fire data:

1. Create a free account at NASA Earthdata
2. Request a MAP_KEY at FIRMS API
3. Create a .env file in the project root:

   VITE_FIRMS_API_KEY=your_key_here
   

4. Restart the dev server

All other APIs (MET Norway, OSM, Open-Meteo, WFIGS, CWFIS) require no keys — they're completely open.

npm run build
npm run preview

src/
├── components/
│   ├── FireGlobe.tsx          # 3D globe with points, rings, polygons, selection
│   ├── InfoPanel.tsx          # Right panel — responder view, weather, region intel
│   ├── GlobeControls.tsx      # Time range, perimeter toggles, search
│   ├── FireDetectionPanel.tsx # Detection list for the Home page
│   ├── ErrorBoundary.tsx      # React error boundary wrapper
│   ├── FutureSection.tsx      # Placeholder section for future features
│   ├── layout/                # Header, Footer, AppLayout, Container
│   └── ui/                    # Reusable UI primitives (Badge, Button, etc.)
├── services/
│   ├── firms.ts               # NASA FIRMS VIIRS fire detection API
│   ├── weatherMetNo.ts        # MET Norway Locationforecast 2.0
│   ├── regionIntelligence.ts  # OSM Overpass + Nominatim + Open-Meteo
│   ├── wfigs.ts               # US fire perimeters (NIFC ArcGIS)
│   ├── cwfis.ts               # Canada fire perimeters (GeoServer WFS)
│   └── cache.ts               # TTL cache + bbox utilities
├── data/
│   ├── fireDataService.ts     # FIRMS adapter — parse, confidence map, downsample
│   ├── perimeterService.ts    # WFIGS + CWFIS fetching with pagination
│   └── teamMembers.ts         # Team member data
├── hooks/
│   ├── useFireData.ts         # React hook — live FIRMS data + auto-refresh
│   └── useFirePerimeters.ts   # React hook — live perimeters + source filter
├── pages/
│   ├── GlobeExplorer.tsx      # Main page — globe + controls + info panel
│   ├── Home.tsx               # Landing page with hero section
│   ├── Comparison.tsx         # Imagery comparison page
│   └── Team.tsx               # Team profiles
├── types/
│   ├── fireData.ts            # FirePoint type
│   ├── firePerimeter.ts       # FirePerimeter type
│   └── wildfireEvent.ts       # WildfireEvent + RegionIntelligence types
└── App.tsx                    # Router (/, /comparison, /team)

MIT

SparkGuard is designed around existing, accessible infrastructure:

1. Public satellite data is available today — NASA FIRMS, Sentinel, and others provide free or low-cost fire detection feeds.
2. Modular integrations — the platform connects to external systems via standard APIs and export formats, minimizing custom development.
3. AI is assistive, not autonomous — we use proven classification techniques with human oversight, avoiding regulatory and liability complexity of fully autonomous systems.

The core innovation is integration and interpretation — bringing fragmented data into a unified view and translating it into actionable guidance for responders.

SparkGuard · Built for Upstart @ GCES Concordia · Health & Wellness · Smart Systems