VarunaWatch — Intelligence Before Impact

Pan-India AI-Powered Cyclone Impact & Infrastructure Resilience Command Platform

Developed by Team MetaMinds

---

Overview

VarunaWatch is an enterprise-grade parametric cyclone intelligence and infrastructure resilience command platform. Built to safeguard all 9 Indian Coastal States, National Grid, and Island Union Territories across 7,516+ km of high-vulnerability coastline (Gujarat, Maharashtra, Goa, Karnataka, Kerala, Tamil Nadu, Andhra Pradesh, Odisha, West Bengal, Puducherry & Island UTs), VarunaWatch unifies real-time storm tracking, high-resolution geospatial hazard modeling, digital twin scenario simulation, parametric SDRF/NDRF insurance triggers, and explainable AI operational advisories under one single roof.

No tab-hopping. No disconnected spreadsheets. No manual GIS synthesis during crisis hours.

"Storm Telemetry & Forecasts" ──▶ Deterministic Spatial Engine ──▶ ImpactGrid (0–100 Vulnerability)
                                             │
"What-If Category 4 Landfall" ──▶ StormTwin Digital Simulation ──▶ Live Cascading Delta Analysis
                                             │
"Draft State Evacuation SOP"  ──▶ Varuna AI Copilot (Gemini) ──▶ Evidence-Grounded Response Plan

Core Architecture Rule

"The Backend computes every number; AI explains, prioritizes, and advises."
Every vulnerability score (0–100), inundation depth, wind field, road cutoff, and trigger confidence is deterministically calculated by backend physics and geospatial models. Google Gemini is grounded strictly in raw telemetry and tool calls, completely eliminating hallucinations in disaster response.

---

Application Walkthrough

┌────────────────────────────────────────────────────────────────────────────────────────┐
│                                 VARUNAWATCH PLATFORM                                   │
├─────────────────┬──────────────────┬─────────────────┬────────────────┬────────────────┤
│ 🛰️ Live Tracker │ 🌐 ImpactGrid    │ 🌪️ StormTwin    │ 🤖 Varuna AI   │ 🛡️ Sentinel   │
│ Dual Basin Met  │ Critical Assets  │ What-If Digital │ Grounded Multi-│ Parametric     │
│ 7,516 km Grid   │ 0-100 Risk Index │ Twin Simulation │ Tool Advisory  │ SDRF Triggers  │
└─────────────────┴──────────────────┴─────────────────┴────────────────┴────────────────┘

1. Landing Page (/): Interactive command hero featuring live fluid cyclone & wind breeze particle motion, Pan-India state breakdown badges, and quick module launchers.
2. Command Center (/dashboard): Unified situational dashboard with dynamic Coastal State / Lane Switcher, live surveillance status bar, active KPI metrics, zero-lag Carto Dark Matter GIS map, and live alert feeds.
3. Storm Intelligence (/storm): Real-time dual-basin cyclone trajectory (Bay of Bengal & Arabian Sea), 72-hour cone of uncertainty, landfall coordinates, and time-series wind/rainfall forecasts.
4. ImpactGrid (/infrastructure): Multi-state spatial risk register scoring every critical asset (hospitals, substations, port terminals, relief shelters, bridges, and coastal highways like NH-16 & NH-66) with 0–100 vulnerability indexes and multi-factor inspection modals.
5. StormTwin Scenarios (/stormtwin): Interactive digital twin simulator allowing disaster commanders to adjust cyclone intensity (Categories 1–5), rainfall multipliers, and surge heights across any Indian coastal state to calculate live cascading delta impacts.
6. Varuna AI Copilot (/ai): Grounded generative AI workspace utilizing Gemini function-calling with quick corridors (Sundarbans, Paradeep, Godavari Delta, Chennai Coromandel, Kochi Malabar, Mumbai MMR, Kutch & Saurashtra) to generate formal district-level evacuation advisories and multi-section response reports.
7. VarunaVision (/satellite): Multi-temporal satellite earth observation layers for pre-storm baseline, current radar reflectivity, and projected flood inundation extents.
8. Sentinel & Advisories (/advisories): Parametric trigger engine monitoring wind thresholds, storm surge levels, and rain accumulations with pre-approved SDRF/NDRF liquidity allocations and automated emergency call dispatch.

---

Pan-India Coastal Lanes Grid

Coastal Lane	Basin	Coastline	Key Districts & Ports	State Response Force
All Coastal India (National Grid)	Pan-India	7,516 km	All Major Ports (JNPT, Paradip, Chennai, Vizag, Kandla, Cochin)	NDRF & Joint SDRF Pool
Gujarat	Arabian Sea	1,600 km	Deendayal Port (Kandla), Mundra, Porbandar, Dwarka, Surat	Gujarat SDRF & GSDMA
Tamil Nadu	Bay of Bengal	1,076 km	Chennai Port, Kamarajar (Ennore), Cuddalore, Nagapattinam, VOC	Tamil Nadu SDRF Taskforce
Andhra Pradesh	Bay of Bengal	974 km	Visakhapatnam Port, Kakinada Deepwater, Machilipatnam, Bapatla	AP SDRF Rapid Deployment
Maharashtra	Arabian Sea	720 km	JNPT, Mumbai Port Trust, Alibaug, Ratnagiri, Sindhudurg	Maharashtra SDRF & MCGM
Kerala	Arabian Sea	590 km	Cochin Port, Vizhinjam International, Alappuzha, Kozhikode	Kerala SDRF Mitigation Wing
Odisha	Bay of Bengal	480 km	Paradip Port, Dhamra Port, Gopalpur, Puri, Balasore	ODRAF (Odisha Rapid Action)
Karnataka	Arabian Sea	320 km	New Mangalore Port, Karwar Sea Bird, Udupi, Honnavar	Karnataka SDRF Marine Wing
West Bengal	Bay of Bengal	157 km	Sundarbans Biosphere Delta, Haldia Dock Complex, Digha	WB SDRF & Civil Defence
Goa	Arabian Sea	101 km	Mormugao Port, Panaji Seaport, Zuari Estuary	Goa SDRF Marine Division
Puducherry & Island UTs	Pan-India	1,000 km	Puducherry, Karaikal, Port Blair (A&N), Kavaratti (Lakshadweep)	UT Disaster Response Forces

---

Key Features

Feature	Description
Pan-India Multi-State Engine	Full coverage of all 9 coastal states & UTs with dynamic state filtering, regional bounds, and localized GIS centroids.
Realistic Cyclone Particle Motion	Interactive fluid canvas simulating atmospheric vortex streamlines, breeze drift, and cyclonic eye physics.
Deterministic Risk Engine	Geospatial multi-criteria algorithm scoring assets from 0 to 100 based on surge depth, wind exposure, elevation, criticality, and accessibility.
StormTwin Digital Simulation	Category 1–5 slider with real-time recalculation of flooded hospitals, threatened substations, and inundated road segments per state.
Grounded Varuna AI Copilot	Google Gemini integration with server-side tool calling to produce evidence-backed operational SOPs without hallucinations.
Parametric Sentinel Watch	Automated threshold monitoring (wind > 140 km/h, surge > 2.5m) for parametric disaster payouts to state SDRF windows.
Zero-Lag MapLibre GIS	Dual Carto vector/raster tile switcher (Dark Matter & Voyager) with instant zero-lag dark/light theme switching.
Zero-Config Demo Mode	Runs immediately out of the box with comprehensive deterministic multi-state fallback data.

---

Tech Stack

Layer	Technology	Purpose / Notes
Frontend Framework	Next.js 15.5 (App Router)	Server Components, modern client layouts, dynamic client routing
UI & Language	React 19 · TypeScript 5.7	Strict mode typing, responsive state management
Styling	Tailwind CSS v4	Dark/light theme support, custom semantic UI tokens, smooth transitions
Mapping & Spatial UI	MapLibre GL & Carto	Zero-lag WebGL rendering, custom markers, GeoJSON storm tracks
Motion & Animation	HTML5 Canvas & Framer Motion	High-performance 60fps cyclonic wind breeze simulation
Icons & Visuals	Lucide React	Clean, consistent operational icon set
Backend API	FastAPI (Python 3.11)	High-performance asynchronous REST API with Pydantic validation
AI Advisory Engine	Google Gemini 2.0 / 3.7 Flash	LLM with native tool-calling and context-grounded synthesis
Database & GIS	PostgreSQL + PostGIS	Spatial querying, hazard geometry buffers, asset point-in-polygon
Deployment	Vercel & Docker	Native Vercel deployment with vercel.json and Docker Compose

---

System Architecture

flowchart TB
    subgraph Data_Sources ["Data & Ingestion Layer"]
        OSM["OpenStreetMap Ingestion"]
        POP["WorldPop / Census Data"]
        MET["IMD / NOAA Cyclone Telemetry"]
        INCOIS["INCOIS Ocean Buoy Feeds"]
    end

    subgraph Backend_Core ["VarunaWatch Engine (FastAPI)"]
        DB[(PostgreSQL + PostGIS)]
        RISK["Deterministic Risk Engine (0-100)"]
        TWIN["StormTwin Simulation Engine"]
        TRIG["Parametric TriggerWatch Engine"]
        GEMINI["Gemini Flash Agent (Multi-State Tool Calling)"]
    end

    subgraph Frontend_App ["Command Center (Next.js 15 / React 19)"]
        STATE["Coastal Lane Context (9 States + UTs)"]
        DASH["Live Command Dashboard"]
        MAP["MapLibre GL GIS View"]
        GRID["ImpactGrid Asset Matrix"]
        SIM["StormTwin Scenario Builder"]
        COPILOT["Varuna AI Advisory Console"]
        SENTINEL["Sentinel Parametric Feeds"]
    end

    OSM --> DB
    POP --> DB
    MET --> DB
    INCOIS --> DB

    DB --> RISK
    RISK --> TWIN
    RISK --> TRIG
    
    RISK --> |Multi-State Spatial Hazard Data| GEMINI
    TWIN --> |Simulation Deltas| GEMINI
    TRIG --> |Trigger Telemetry| GEMINI

    STATE --> DASH
    STATE --> MAP
    STATE --> GRID
    STATE --> SIM
    STATE --> COPILOT
    STATE --> SENTINEL

    RISK --> |REST APIs| DASH
    RISK --> |GeoJSON Layers| MAP
    RISK --> |Asset Registry| GRID
    TWIN --> |Delta Scenarios| SIM
    GEMINI --> |Grounded Advisories| COPILOT
    TRIG --> |Alert Streams| SENTINEL

Loading

---

Vercel Deployment Configuration

The repository includes a ready-to-deploy vercel.json file:

{
  "$schema": "https://openapi.vercel.sh/vercel.json",
  "framework": "nextjs",
  "rootDirectory": "frontend",
  "buildCommand": "npm run build",
  "outputDirectory": ".next",
  "rewrites": [
    {
      "source": "/api/py/:path*",
      "destination": "http://localhost:8000/api/:path*"
    }
  ],
  "headers": [
    {
      "source": "/(.*)",
      "headers": [
        { "key": "X-Content-Type-Options", "value": "nosniff" },
        { "key": "X-Frame-Options", "value": "DENY" },
        { "key": "X-XSS-Protection", "value": "1; mode=block" }
      ]
    }
  ]
}

Deploying to Vercel

1. Import VarunaWatch into your Vercel Dashboard.
2. Set Root Directory to frontend (auto-detected via vercel.json).
3. Configure Environment Variables (NEXT_PUBLIC_MAPBOX_TOKEN, NEXT_PUBLIC_API_URL).
4. Click Deploy.

---

Getting Started

1. Quickstart with Docker Compose

# Clone the repository
git clone https://github.com/Premkumar1845/VarunaWatch.git
cd VarunaWatch

# Copy environment variables
cp .env.example .env

# Build and start all services (PostgreSQL/PostGIS, FastAPI backend, Next.js frontend)
docker compose up --build

• Frontend Application: http://localhost:3000
• Backend API Docs (Swagger): http://localhost:8000/docs
• PostGIS Database: localhost:5432

---

2. Manual Local Setup

# 1. Install Python dependencies
pip install -r requirements.txt

# 2. Run both servers
npm run backend   # Starts FastAPI on http://localhost:8000
npm run dev       # Starts Next.js on http://localhost:3000

---

API Reference

• GET /api/cyclone/current?state={ap|od|tn|mh|gj|wb|kl|ka|ga|ut|all} — Current cyclone status per state or national grid.
• GET /api/cyclone/forecast?state={state} — 72-hour forecast track, cone coordinates, and rainfall/wind forecasts.
• GET /api/hazards?category={1-5}&state={state} — Hazard zones with deterministic risk scores.
• GET /api/infrastructure?state={state}&asset_type={hospital|power|road|shelter|bridge} — Scored critical infrastructure register.
• GET /api/roads/affected?state={state} — Inundation risk across coastal highway segments.
• GET /api/risk/summary?state={state} — Aggregated risk summary, exposed population, and facility threat counters.
• POST /api/simulation?state={state} — Digital twin what-if scenario simulation.
• POST /api/ai/analyze — Conversational assistant with multi-state telemetry context.
• POST /api/ai/advisory — Formal operational SOP advisory generator for state collectors & SDRF units.

---

Security & Reliability

• 🔒 Server-Only API Secrets: Google Gemini and external API keys are kept strictly within backend environment configurations.
• 🛡️ Zero Hallucination Architecture: AI models cannot fabricate operational data; all figures are fetched via programmatic tool calls to backend models.
• 📐 Strict Typing & Schema Validation: Pydantic models on the backend and TypeScript interfaces on the frontend guarantee robust data contracts.
• ⚡ Graceful Offline Fallback: The platform operates seamlessly in low-connectivity scenarios via structured local telemetry stores.

---

Roadmap

• Core Parametric Risk Engine (Deterministic 0–100 scoring).
• Pan-India Multi-State Expansion (All 9 Indian Coastal States + Island UTs).
• Interactive Cyclone & Wind Breeze Motion Background.
• Official Brand Logo Integration across all modules.
• StormTwin Digital Simulation Engine (Categories 1–5 with cascading failure detection).
• Varuna AI Decision Copilot with tool calling and state advisory generation.
• Zero-Lag MapLibre GIS Mapping with dual Carto tile dark/light modes.
• Vercel Production Deployment Configuration.
• Automated WhatsApp / SMS Broadcast Relays for field disaster units.
• Offline PWA Support for emergency teams with low satellite bandwidth.

---

Team

Team MetaMinds
Building next-generation intelligent platforms for disaster resilience and humanitarian impact.

© MetaMinds · VarunaWatch — Intelligence Before Impact.