AI Crisis Navigator automatically monitors global disaster events, analyzes their severity using AI, and provides real-time visualization and recommendations for emergency response teams. The system processes data from NASA EONET, ReliefWeb, and GDACS, then uses Google Gemini AI to generate severity assessments and response guidance.

• Real-time Data Ingestion: Automatically fetches disaster events from NASA EONET API (every 10 minutes)
• AI-Powered Analysis: Uses Google Gemini AI to analyze disasters and assign severity scores (1-10 scale)
• Interactive Map: Leaflet-based map with color-coded markers showing disaster locations and severity
• Live Updates: WebSocket-based real-time synchronization of disasters, statistics, and activities
• Response Recommendations: AI-generated evacuation guidance, resource priorities, and immediate actions
• Activity Feed: Real-time log of system events and disaster updates
• Filtering & Search: Filter disasters by processing status and severity level
• Bulk Processing: Process multiple unanalyzed disasters simultaneously

• Framework: React 18 with TypeScript
• Routing: Wouter for lightweight client-side routing
• State Management: TanStack Query for server state management
• UI Components: Radix UI primitives with shadcn/ui component library
• Styling: Tailwind CSS with custom CSS variables for theming
• Maps: Leaflet for interactive mapping with disaster visualization
• Real-time: WebSocket connection for live data synchronization

• Runtime: Node.js with Express.js server
• Language: TypeScript with ES modules
• API Design: RESTful endpoints with WebSocket support
• Data Ingestion: NASA EONET API, ReliefWeb API, GDACS API
• AI Processing: Google Gemini AI for disaster analysis
• Database: PostgreSQL (Neon) with Drizzle ORM (currently using in-memory storage)

• Frontend Build: Vite for fast development and optimized production builds
• Backend Build: esbuild for server-side bundling
• Development: tsx for TypeScript execution in development
• Database Migrations: Drizzle Kit for schema management

• Node.js 20 or higher
• npm or yarn package manager
• PostgreSQL database (optional, currently using in-memory storage)
• Google Gemini API key

1. Clone the repository

   git clone <repository-url>
   cd CrisisNavigator

2. Install dependencies

   npm install

3. Set up environment variables

   Create a .env file in the root directory with the following variables:

   # Required
   GEMINI_API_KEY=your_google_gemini_api_key
   GOOGLE_API_KEY=your_google_api_key  # Alternative to GEMINI_API_KEY
   
   # Optional - Database (currently using in-memory storage)
   DATABASE_URL=postgresql://user:password@host:port/database
   
   # Optional - Server Configuration
   PORT=5000
   NODE_ENV=development
   
   # Optional - AI Model Configuration
   GEMINI_MODEL=gemini-2.5-flash-lite
   
   # Optional - Data Ingestion
   DISASTER_MAX_AGE_DAYS=30
   ENABLE_RELIEFWEB=1
   ENABLE_GDACS=1
   RELIEFWEB_DAYS=30
   GDACS_DAYS=30
   RELIEFWEB_APPNAME=crisis-navigator-demo/0.1

4. Set up database (optional)

   If you want to use PostgreSQL instead of in-memory storage:

   npm run db:push

Start the development server with hot reload:

npm run dev

The application will be available at http://localhost:5000

Build the application for production:

npm run build

Start the production server:

npm start

Run TypeScript type checking:

npm run check

• GET /api/disasters - Get all disasters (filtered by max age)
• GET /api/disasters/:id - Get a specific disaster by ID
• GET /api/stats - Get system statistics
• GET /api/activities - Get recent activities (limit query param)
• POST /api/ingest - Manually trigger NASA data ingestion
• POST /api/analyze - Manually trigger AI analysis processing
• POST /api/process-all - Bulk process all unanalyzed disasters

Connect to /ws for real-time updates:

• stats - System statistics updates
• disasters - Disaster list updates
• activities - Activity feed updates

1. Data Ingestion: Periodic fetching from NASA EONET API (every 10 minutes)
2. Data Processing: Events are normalized and stored
3. AI Analysis: Gemini AI analyzes unprocessed events (every 30 seconds)
4. Storage: Processed data and analysis results are persisted
5. Real-time Distribution: WebSocket broadcasts updates to connected clients (every 15 seconds)
6. Client Rendering: React components display interactive maps and analysis

Disasters are assigned severity scores on a 1-10 scale:

• Low (1-3): Small, localized events with minimal impact
• Medium (4-6): Events with measurable impact but under control
• High (7-10): Major disasters requiring significant response (requires 2+ high-severity criteria)

High-severity criteria include:

• 1+ confirmed fatalities
• 1,000+ people displaced
• Major infrastructure failure
• Emergency declarations with external aid
• Rapid uncontrolled escalation
• Economic loss > $100M USD

*Either GEMINI_API_KEY or GOOGLE_API_KEY must be set.

https://youtu.be/1mn8ucRLJmI