India Temperature Map

An open hyperlocal India temperature platform. Pulls real-time temperature data from every public/freemium source we could reach and overlays them on a single Leaflet map — three browser-live models per city, a 4,647-cell heat grid covering India's full claim, plus ground-sensor networks where they exist.

Live site: https://indiatemperaturemap.pages.dev/

On the map

• 39 city points — three independent models browser-live on every page load: Open-Meteo (ICON/IFS/GFS blend, batched), MET Norway (ECMWF-backed), and 7Timer (GFS). Popups show all three values plus the model-spread Δ; if the spread exceeds 3 °C, it's flagged red.
• 0.25° heat grid · 4,647 cells (~27 × 25 km each) covering India's full territorial claim — mainland + Andaman & Nicobar + Lakshadweep + J&K (incl. PoK) + Ladakh (incl. Aksai Chin + Karakoram). Polygon: simplified from datameet/maps composite. Range typically –22 °C in Karakoram to +41 °C in Rajasthan.
• OpenAQ ground stations — ~570 India stations (509 CPCB regulatory + AirGradient + Clean Air Catalyst). ~340 report current temperature.
• Weather Union ground sensors — Zomato's hyperlocal fleet, ~37 discovered points across Bengaluru / Mumbai / Delhi-NCR / Pune / Hyderabad / Chennai / Kolkata / Ahmedabad / Jaipur / Lucknow / Chandigarh / Indore / Coimbatore.
• Multi-source comparison sample — all 39 cities × up to 12 source readings, side-by-side in the Sources tab.

UI

Four tabs:

Tab	What it shows
Live Map	Leaflet map with overlays. Layer control starts collapsed; hover to expand, mouse-out to dismiss. Pan-locked to India's bbox with a translucent mask fading out neighbouring countries.
Trends	Today's 39-city distribution, six-band histogram, and the All-India 1980-2025 annual-mean trend (Open-Meteo ERA5 archive).
Sources	All 12 providers with status badges (ok · key needed · IP gated · idle), plus the per-city × per-source comparison matrix.
About	Project description, data sources, what's gated, and License.

Responsive: sidebar stacks above map on ≤ 768 px, cards collapse to single column on ≤ 480 px, Leaflet zoom buttons enlarge to 36 × 36 px for touch.

Data freshness

Layer	Mode	Refresh
39 cities · 3 models	browser-live	every page load
0.25° heat grid	server snapshot	every 3 h via GitHub Actions
OpenAQ stations	server snapshot	every 3 h via GitHub Actions
Weather Union sensors	server snapshot	every 3 h via GitHub Actions
Multi-source comparison	server snapshot	every 3 h via GitHub Actions

The cron lives in .github/workflows/refresh-data.yml. It runs every 3 hours UTC (and on manual trigger via gh workflow run refresh-data.yml), commits the refreshed JSONs, and GitHub Pages rebuilds within ~1 minute. Repo secrets OPENAQ_API_KEY and WEATHER_UNION_KEY are required.

Source matrix

Source	Key	Status	Browser-live?	Notes
Open-Meteo	none	✅	✅	ICON/IFS/GFS blend, ~11 km grid; batched call
MET Norway	none	✅	✅ (streamed)	ECMWF-backed, ~10 km
7Timer	none	✅	✅ (streamed)	GFS forecast, first 3-hour slot
METAR (Iowa State)	none	✅	❌ (no CORS)	Real airport instruments, 16 Indian ICAOs
NASA POWER	none	⚠	❌	~24 h lag; often empty for today
Weather Union	WEATHER_UNION_KEY	✅	❌ (key)	Zomato hyperlocal ground sensors; lat/lon auto-resolved
OpenAQ	OPENAQ_API_KEY	✅	❌ (key)	CPCB + community AQ sensors with temperature
IMD	—	❌	—	api.imd.gov.in IP-gated; formal request only
CPCB realtime · data.gov.in	data.gov.in key	⚠	—	AQ pollutants only, no temperature field
OpenWeatherMap	OPENWEATHER_API_KEY	—	❌ (key)	Wire when key provided
Visual Crossing	VISUAL_CROSSING_KEY	—	❌ (key)	Wire when key provided
Tomorrow.io	TOMORROW_IO_KEY	—	❌ (key)	Wire when key provided
Ambee	AMBEE_KEY	—	❌ (key)	India-based provider
WeatherAPI	WEATHERAPI_KEY	—	❌ (key)	Wire when key provided

Keys are read from env vars locally and from repo secrets in CI — never checked in.

Performance

The page is engineered to render fast even on a cold cache:

• First paint uses one batched Open-Meteo call + small static JSONs (~10 KB).
• MET Norway + 7Timer (78 per-city calls) fire AFTER initial render and stream into popups in the background (concurrency-capped at 8 to avoid socket exhaustion).
• Grid weather uses a compact columnar format ({keys, cells} instead of array-of-objects) — 845 KB → 116 KB (–86%).
• All JSONs are emitted with separators=(',',':') (no indentation) and gzipped on the wire by GitHub Pages.
• Preconnect hints for api.open-meteo.com, api.met.no, www.7timer.info shave 200–400 ms off the first model fetch.

Repo layout

File	Purpose
index.html	Leaflet UI (Live Map · Trends · Sources · About tabs)
.github/workflows/refresh-data.yml	Cron: refresh all snapshots every 3 h
fetch.py	39-city Open-Meteo snapshot (CLI)
build_snapshot_cities.py	Reshape fetch.py output into snapshot_cities.json
grid.py	0.25° lattice fetch over India, rate-aware + resumable (JSONL partial); writes v2-columnar weather file
fetch_openaq.py	OpenAQ India temperature stations
fetch_datagovin.py	CPCB AQ via data.gov.in (kept for completeness; AQ only)
discover_weatherunion.py	Probe Weather Union sensors across 88 Indian neighborhoods
india_temp_pull.py	Multi-source comparison fetcher (12 sources × N locations)
build_india_trend.py	All-India 1980-2025 annual mean from Open-Meteo archive (ERA5)
cities.json	39-city coordinate list (state capitals + major metros)
all_cities.csv	Same 39 cities in CSV form (input to india_temp_pull.py in CI)
india_polygon.json	India full-claim MultiPolygon, simplified from datameet/maps
LICENSE · NOTICE	CC BY-NC-SA 4.0 license + third-party source attributions

Generated data files (refreshed by cron)

File	Description	Format
snapshot_cities.json	39 cities × current Open-Meteo readings	array of city objects
india_grid.json	Grid-cell geometry	array of {lat, lon}
india_grid_weather.json	Grid-cell current weather	v2-columnar {keys, cells}
india_temp_trend.json	1980-2025 All-India annual mean	array of {year, temperature_c}
stations_openaq.json	OpenAQ India temperature stations	array of station objects
stations_weatherunion.json	Weather Union discovered sensors	array of sensor objects
india_temp_readings.json	Multi-source comparison (12 × 39)	array of readings

Run locally

# 1. Serve the static site
python3 -m http.server 8765
open http://localhost:8765/

# 2. (Optional) Refresh any snapshot
python3 build_snapshot_cities.py
OPENAQ_API_KEY=...   python3 fetch_openaq.py
WEATHER_UNION_KEY=... python3 discover_weatherunion.py
python3 grid.py --resolution 0.25
WEATHER_UNION_KEY=... python3 india_temp_pull.py --locations all_cities.csv
python3 build_india_trend.py

Triggering a refresh in CI

gh workflow run refresh-data.yml --repo Ashwask/IndiaTemperatureMap

Acknowledgements

Open-Meteo · OpenAQ · Zomato Weather Union · MET Norway · Iowa State ASOS · 7Timer · NASA POWER · datameet/maps (India full-claim polygon).

License

This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0). See LICENSE for the full legal code and NOTICE for third-party data licenses.

In short: you may copy, share, and adapt the code and data freely for non-commercial purposes, provided you give attribution and license any derivative work under the same terms.