🌬️ Air Quality Intelligence System (AQIS)

AI-Powered Pollution Source Attribution & Geospatial Monitoring

AQIS is a comprehensive environmental monitoring solution that transforms raw particulate matter data into actionable intelligence. By using a custom AI Attribution Engine, the system identifies potential industrial sources of pollution spikes, enabling data-driven accountability.

---

Key Features

• Real-Time Monitoring: Live dashboard for PM2.5, PM10, and AQI across multiple urban centers (Ernakulam, Delhi, Mumbai).
• AI Source Attribution: Proprietary scoring engine that identifies "Suspect Sites" (factories, construction zones) based on pollution levels and proximity.
• Crowd-Sourced Reporting: Any user can report a potential pollution source (construction site, brick kiln, quarry, waste dump) with a map pin, photo-less description and optional distance.
• Admin Moderation: Reports stay invisible on the public site until an administrator verifies them — approved reports are published to the suspect list, the live map and the AI verdict with a "Verified" badge.
• Live Incident Map: Interactive Leaflet.js integration showing monitoring stations and flagged suspect sites with dynamic "Fly-To" navigation.
• Admin Governance Panel: Secure portal for environmental administrators to review crowdsourced reports and manage the suspect database.
• Shared Site Navigation: One services/nav.js drives the header, active-page highlighting, mobile hamburger menu and footer for every public page.
• Incident History & Log: Persistent storage of past pollution violations for long-term trend analysis.

---

System Architecture

The project utilizes a Client-Server Architecture optimized for low-latency updates:

1. Data Layer: Fetches real-time sensor feeds via the OpenWeatherMap API.
2. Logic Layer (Next.js API routes): Processes raw data through an AI-scoring algorithm to calculate suspect confidence levels.
3. Persistence Layer: Reports are stored server-side in data/submissions.json (or Upstash Redis via REST when configured); incident history and the active incident selection stay in localStorage.
4. Presentation Layer: Responsive UI built with Tailwind CSS and dynamic mapping via Leaflet.js.

---

Tech Stack

• Frontend: HTML5, Tailwind CSS, JavaScript (ES6+), Leaflet.js — served as static assets from public/
• Backend: Next.js 16 (App Router, route handlers)
• Data Source: OpenWeatherMap API
• Development AI: Google Gemini (Logic Optimization & Data Structuring)

---

⚙️ Installation

Requires Node.js 20.9+ (developed on 24.x).

1. Clone the repository

git clone https://github.com/ASP-31/AirPro
cd AirPro

2. Install dependencies — the Next.js app lives at the repo root

npm install

3. Configure the API key (optional — the app falls back to simulated data without it)

cp .env.example .env.local
# then set OPENWEATHER_API_KEY in .env.local
# get a key at https://home.openweathermap.org/api_keys

4. Configure admin moderation (required for approving reports)

# in .env.local
ADMIN_USERNAME=your_admin
ADMIN_PASSWORD=a_strong_password
AUTH_SECRET=<any long random string>   # optional, signs session tokens

Without these the portal falls back to admin / admin123 for local development only.

5. Start the dev server

npm run dev

6. Launch the Dashboard

Open http://localhost:3000 — this serves the landing page at /index.html.

Page	Path
Landing page	/index.html (/ serves this)
Dashboard	/dashboard.html
Live incident map	/map.html
City-wide spike scanner	/incident.html
AI suspect analysis	/suspected.html
Printable report	/report.html
Incident history	/history.html
Report a pollution source	/reporting.html
Admin login	/admin.html
Admin moderation queue	/data.html

Site navigation

All public pages render their header and footer from public/services/nav.js — one array of links, so navigation can never drift between pages. Each page only needs a <div data-aqis-header></div> and <div data-aqis-footer></div> placeholder plus the script tag.

<script src="services/nav.js"></script>
...
<div data-aqis-header></div>
...
<div data-aqis-footer></div>

The component highlights the current page from location.pathname, collapses into a hamburger menu below lg, and injects the site-wide footer. The admin console (admin.html, data.html) intentionally keeps its own minimal chrome.

Production build

npm run build
npm start

Report moderation flow

1. Anyone opens /reporting.html, picks the monitoring zone, drops a pin (or types the distance from the station) and submits the site.
2. POST /api/submissions stores the report as pending. The browser keeps a receipt so the reporter can poll their own report's status.
3. An admin signs in at /admin.html, reviews the queue in /data.html, optionally fixes the coordinates on a map, then approves.
4. Approval flips the record to approved; it is immediately merged into GET /api/report (crowdSources) and therefore renders on /map.html, /suspected.html and /report.html for every visitor, marked "Admin Approved".

Unpublishing (approved → rejected) removes it from the public site again. Duplicate reports for the same site within 0.5 km are flagged in the admin queue instead of being published twice.

API

Method	Path	Auth	Description
GET	/api/report?zone=<station>	public	Pollution report + scored suspects + crowdSources (admin-approved community reports) for the matched station. Falls back to scenarios.json data with "source": "MOCK_DATA" if OpenWeather is unreachable or no API key is set.
GET	/api/report?stations=1	public	List of monitoring stations (id, name, coordinates) used to populate dropdowns.
POST	/api/submissions	public	Submit a potential pollution source. Returns { id, receipt, status: "pending" }. Rate limited to 10 per 10 minutes per IP.
GET	/api/submissions?status=<all|pending|approved|rejected>&zone=<station>	public for approved, admin token for everything else	List submissions. Anonymous responses strip reporter contact details.
GET	/api/submissions?id=<id>&receipt=<receipt>	receipt	Status of one of your own reports, including the admin note.
PATCH	/api/submissions/<id>	admin	{ status, adminNote?, verifiedLat?, verifiedLng? } to approve, reject, reopen, or correct coordinates.
DELETE	/api/submissions/<id>	admin	Permanently delete a report.
POST	/api/auth/admin	public	{ username, password } → signed 8-hour admin token.

Station matching is case-insensitive and fuzzy; an unknown zone falls back to the first station in data/scenarios.json.

Storage

Submissions are written to data/submissions.json with atomic writes. On read-only hosts (Vercel, most serverless platforms) set UPSTASH_REDIS_REST_URL and UPSTASH_REDIS_REST_TOKEN and the same API will store them in Redis over REST instead — no extra npm dependency.

---

📈 Future Roadmap

• Wind Correlation: Integrating real-time wind speed/direction to track plume trajectories.
• IoT Integration: Connecting custom hardware sensor meshes (ESP32/Arduino).
• Predictive AI: Implementation of LSTM models to forecast AQI trends 24 hours in advance.