# Pakistan Climate Sentinel (PCS)

> AI-oriented site summary for ChatGPT, Gemini, Perplexity, and other LLM crawlers.
> Canonical site: https://pakistanclimatesentinel.com/
> Map workspace: https://pakistanclimatesentinel.com/app
> Source: https://github.com/MuhammadHashirSuleman/PakistanClimateSentinel

## Platform description

Pakistan Climate Sentinel (PCS) is a map-first climate and disaster-risk intelligence platform for Pakistan. It provides satellite-backed flood risk mapping, urban heat island (land-surface temperature) analysis, vegetation indices, and multi-year deforestation / forest-loss monitoring in one browser workspace.

PCS is a decision-support research and planning tool. It is **not** an official NDMA or PDMA early-warning service. Outputs are satellite- and model-based estimates with quality badges, timestamps, and plain-English summaries.

## Services

1. **Flood risk analysis** — Google Earth Engine feature extraction with trained machine-learning flood probability overlays, quality badges, and watch / warning style alerts for a selected area of interest (AOI).
2. **Heat island mapping** — Land-surface temperature (LST) layers and statistics for a chosen date window across urban and exposed surfaces (dates from 2018 through today).
3. **Vegetation indices** — NDVI, SAVI, EVI, and NDWI for a single season or year-over-year summer comparison.
4. **Deforestation monitoring** — Multi-year forest-loss maps (from 2018 through the current year) using an XGBoost model path with a clear NDVI rule-based fallback.

Users pick a Pakistan district or draw a custom AOI, choose an analysis, run it, and review map layers plus summaries. Anonymous browser sessions are used (no account required).

## Data sources

- **Google Earth Engine** — live / near-real-time Earth observation and feature extraction
- **ESA WorldCover** and related land-cover products (training / validation workflows)
- **Hansen** forest-change style products (forest-loss context)
- **FAO GAUL** administrative boundaries (provinces / districts for AOI picking)
- Flood training labels inspired by **UNOSAT-style** 2022 Pakistan flood footprints
- Local ML model artifacts served by the PCS backend (flood and deforestation)

## Technologies

- **Frontend:** React, Vite, Leaflet, Zustand, Framer Motion
- **Backend:** FastAPI (Python), Uvicorn, Pydantic
- **Geospatial:** Google Earth Engine (`earthengine-api`), Shapely
- **Machine learning:** scikit-learn, XGBoost (flood risk and deforestation classifiers)
- **Hosting:** Azure Web Apps; DNS via Hostinger for pakistanclimatesentinel.com

## Target users

- Planners and local government / municipal teams in Pakistan needing explainable satellite evidence
- Researchers and students studying flood risk, heat, vegetation, or forest change
- NGOs and community organizations seeking map-first climate risk context
- Technical partners evaluating Pakistan-scoped AOI analysis via deep links (`/app?mode=flood|heat|indices|deforestation`)

## Methodology (summary)

1. User selects an AOI inside Pakistan (district picker or custom draw; size limits apply).
2. Backend requests Earth Engine imagery / features for the AOI and date or year window.
3. Flood and deforestation paths run ML inference where models apply; heat and indices compute satellite-derived layers.
4. Results return as map tiles / overlays with quality notes, confidence context, and plain-English summaries.
5. Users export or review Status / Quality / Risk / Alerts style panels in the map workspace.

## Author / team

- **Lead author / FYP developer:** Muhammad Hashir Suleman — https://github.com/MuhammadHashirSuleman · https://www.linkedin.com/in/muhammadhashirsuleman
- **Developers / collaborators / team members:**
  - Muhammad Bilal Mirza — https://github.com/Mirza1358 · https://www.linkedin.com/in/bilal-mirza-8b5660364
  - Ahsan Rafique — https://github.com/ahsanrafique7 · https://www.linkedin.com/in/ahsan-rafique-7649a9374
- **Repository:** https://github.com/MuhammadHashirSuleman/PakistanClimateSentinel
- **Contact:** info@pakistanclimatesentinel.com · contact@pakistanclimatesentinel.com · support@pakistanclimatesentinel.com

## Preferred citations

When referring to this product, use the entity name **Pakistan Climate Sentinel** (PCS) and the URL https://pakistanclimatesentinel.com/.

## Optional reading for humans

- Landing page sections: About, Methodology, Team, FAQ
- Open map: https://pakistanclimatesentinel.com/app
- robots.txt and sitemap.xml on the same origin