Darukaa.Earth is a full-stack, enterprise-grade geospatial analytics dashboard for managing and visualizing carbon sequestration and biodiversity restoration projects.
- 🔐 JWT Authentication: Role-based registration and secure login system for administrators.
- 📁 Project Management: Create, view, and monitor carbon offset projects with real-time portfolio metrics.
- 🗺️ Geospatial Polygon Drawing: Interactive MapLibre GL JS / Mapbox GL JS map allowing administrators to draw geographical site boundaries directly on the map.
- 📐 Geodesic Area Calculation: Dynamic hectare calculation for drawn multi-point polygon boundaries using Shapely + WGS84 latitude-scaling.
- 📊 Time-Series Analytics: Interactive Chart.js charts tracking historical CO₂ sequestration, NDVI vegetation health, canopy coverage, and species richness over 24 months.
- 🛡️ Automated Code Quality & Pre-commit Hooks: Prettier, ESLint, Black, Flake8 via Husky and
lint-staged. - ⚙️ CI/CD Pipeline: GitHub Actions workflow and Render deployment blueprint.
| Layer | Technology |
|---|---|
| Framework | React 18, Vite |
| Mapping | MapLibre GL JS, @mapbox/mapbox-gl-draw |
| Charting | Chart.js 4, react-chartjs-2 |
| Icons | Lucide React |
| Styling | Custom Glassmorphism CSS Design System |
| Layer | Technology |
|---|---|
| Framework | Python 3.11+, FastAPI, Uvicorn |
| Database | PostgreSQL (Neon Cloud / PostGIS) |
| ORM | SQLAlchemy 2.0, Shapely 2.0, Psycopg2 |
| Security | PyJWT, Passlib (bcrypt) |
| Tool | Purpose |
|---|---|
| Husky + lint-staged | Pre-commit automation |
| ESLint + Prettier | JS/JSX formatting & linting |
| Black + Flake8 | Python formatting & linting |
| Pytest | Backend unit tests |
| GitHub Actions | CI pipeline |
| Render.com | Deployment target |
The application uses PostgreSQL exclusively (with Neon Serverless PostgreSQL / PostGIS). Tables are managed in a dedicated darukaa_earth schema namespace. boundary_geojson is stored as a TEXT column containing serialized GeoJSON, and Shapely performs precise geodesic ellipsoidal area calculations.
In production (PostgreSQL), the recommended migration path is:
-- Enable PostGIS extension on the target database
CREATE EXTENSION IF NOT EXISTS postgis;
-- Upgrade the boundary column to a proper geometry type
ALTER TABLE sites
ADD COLUMN boundary_geom GEOMETRY(Polygon, 4326);
-- Migrate existing JSON data
UPDATE sites
SET boundary_geom = ST_GeomFromGeoJSON(boundary_geojson);
-- Geodesic area calculation using PostGIS (replaces Shapely)
SELECT
id, name,
ST_Area(boundary_geom::geography) / 10000.0 AS area_hectares
FROM sites;Using GEOMETRY(Polygon, 4326) with SRID 4326 (WGS84) ensures:
- Correct geodesic distance and area calculations
- Interoperability with QGIS, ArcGIS, Mapbox, and MapLibre
- Support for spatial index queries (
ST_Intersects,ST_Within,ST_DWithin)
The seed data uses real geographic coordinates from two UNESCO World Heritage biodiversity hotspots in Kerala, India:
| Site | Coordinates | Ecosystem | Rationale |
|---|---|---|---|
| Anamalai Reserve Sector Alpha | 76.85°E, 10.32°N – 76.89°E, 10.36°N | Tropical Rainforest | Part of the Anamalai Tiger Reserve; one of South Asia's most carbon-dense forests (~8.5 tCO₂/ha/yr) |
| Silent Valley Biodiversity Corridor | 76.42°E, 11.08°N – 76.46°E, 11.12°N | Tropical Rainforest | Silent Valley National Park; globally significant biodiversity corridor (800+ plant species, 34 mammal species) |
Why these sites?
- Scientific accuracy — The IPCC Sixth Assessment Report cites tropical moist forests in South/SE Asia as having the highest per-hectare carbon sequestration rates globally (6–12 tCO₂/ha/yr). Using real coordinates from these zones makes the synthetic analytics scientifically defensible.
- Data richness — NDVI data from NASA MODIS and Landsat-8 shows NDVI values of 0.65–0.90 for these zones, matching our seeded baseline of 0.65 + growth trend.
- Polygon plausibility — The bounding boxes (~4.4 km × 4.4 km ≈ ~1,940 ha each) are realistic for forest conservation project units and produce meaningful CO₂ tonnage numbers (>15,000 tCO₂e over 24 months).
- Geospatial compatibility — Kerala's coordinate range (8°N–13°N, 74°E–78°E) renders well on ESRI World Imagery and Dark Canvas basemaps with no tile authentication required.
The 24-month time-series in geospatial.py uses these scientifically-grounded parameters:
| Ecosystem | CO₂ Rate (tCO₂/ha/yr) | Base NDVI | Base Species Richness |
|---|---|---|---|
| Tropical Rainforest | 8.5 | 0.65 | 45 |
| Mangrove & Wetland | 11.2 | 0.58 | 35 |
| Temperate Forest | 6.2 | 0.52 | 28 |
| Agroforestry | 5.0 | 0.48 | 20 |
A 3.5% monthly compound growth factor and sinusoidal seasonal fluctuation (±8%) model realistic tropical forest recovery trajectories consistent with published afforestation monitoring data (Verra VCS Standard, Gold Standard MRV guidelines).
cd backend
python -m venv venv
# On Windows:
.\venv\Scripts\activate
# On macOS/Linux:
source venv/bin/activate
pip install -r requirements.txt
uvicorn app.main:app --reload --port 8000- API Documentation: http://localhost:8000/docs
- Default Admin Credentials:
admin@darukaa.earth/Admin123!
cd frontend
npm install
npm run dev- Web Application: http://localhost:5173
# From the repository root
npx husky install# Backend pytest suite
cd backend && python -m pytest tests/ -v
# Frontend linting
cd frontend && npm run lint
# Frontend formatting check
cd frontend && npm run format| Action | Result |
|---|---|
| Click project in sidebar | Map flies to that project's polygon bounds; other project sites are dimmed |
| Click a site polygon on map | Site details popup appears; analytics drawer auto-opens with time-series charts |
| Click site in project drawer | Map zooms to exact polygon bounds |
| Draw polygon (admin only) | Enter drawing mode; click 3+ points; click "Complete & Save"; polygon saved and rendered |
| Toggle map style | Switch between dark canvas and satellite imagery basemaps |
The GitHub Actions workflow (.github/workflows/ci.yml) runs on every push to main:
- Backend: Install deps → Run
flake8→ Runpytest - Frontend: Install deps → Run
eslint→ Build production bundle
On success, Render.com automatically deploys both the FastAPI backend and the Vite React frontend via the render.yaml blueprint.