Coperni
Air quality forecasts from Copernicus, readable at a glance.
Coperni shows the European air quality forecast of the
Copernicus Atmosphere Monitoring Service (CAMS) on a map
centred on an Italian municipality, a European city or any coordinates, plus a chart of how the
pollutants evolved over the past week and what is expected for the next 48 hours.
Live demo: https://coperni.azukibeans.dev
Available in English and Italian (/it/…).
Why
Air quality data in Europe is public and free, but it usually comes as technical files (NetCDF,
GRIB), expert portals or scattered monitoring stations. Coperni takes the CAMS forecast
and turns it into something anyone can read: a grid of ~10 km cells around the place you care
about, coloured against the European limit values.
It is a public, read-only site. There are no accounts and no tracking. The only cookie remembers
the last place you looked at.
What you see
- Map: the area around the chosen place (10, 20, 50 or 100 cells on the long side of the screen,
~150, ~390 or ~780 km at Milan's latitude), one cell every 0.1°; Max shows the whole CAMS
domain in 0.5° blocks. The colour follows
value / threshold: cyan to green below the EU
threshold, yellow at the threshold, red above it, purple above twice the threshold, drawn
as a halftone (dot size ∝ concentration) over a light basemap. Click a cell to read its value.
- Time navigation: step ±1 h / ±24 h through the past days and the next 48 h of the forecast.
- Ranking: the most populous places (Italian municipalities ≥ 50k, other cities ≥ 500k)
ranked by the daily average of each pollutant over the chart area, sortable and filterable.
- Charts: hourly average over the same area for the last 7 days and the latest 48 h forecast,
with the thresholds drawn as dashed lines.
- Search: Italian municipalities (including bilingual names such as Bolzano/Bozen) and
European cities above 15,000 inhabitants, or your current position.
Data
Source
CAMS European air quality forecasts (cams-europe-air-quality-forecasts), downloaded
daily from the Copernicus Atmosphere Data Store (ADS).
CAMS is part of the European Union's Copernicus programme and is operated by the European Centre
for Medium-Range Weather Forecasts (ECMWF).
| |
|---|
| Model | ensemble: the median of about ten European regional air quality models |
| Resolution | 0.1° × 0.1° (~11 km N–S, ~7–9 km E–W at Italian latitudes) |
| Domain | lat 30°–72° N, lon 25° W–45° E (North Africa to Scandinavia) |
| Level | surface, i.e. the air we breathe |
| Time step | hourly, forecast run at 00 UTC, lead times 0–48 h |
| Unit | µg/m³ |
Pollutants and thresholds
| Pollutant | Threshold (µg/m³) | Reference |
|---|
| PM2.5 | 25 | EU daily limit value from 2030 |
| PM10 | 45 | EU daily limit value from 2030 |
| NO₂ | 50 | EU daily limit value from 2030 |
| O₃ | 120 | EU target value (maximum daily 8-hour mean) |
| SO₂ | 50 | EU daily limit value from 2030 |
All thresholds are the strictest EU references of Directive 2024/2881 and apply to daily (O₃:
8-hour) means, so every pollutant is judged on the same footing. The hourly limits (NO₂ 200, SO₂
350) are meant for acute episodes and would leave the map almost always empty: the project aims
at awareness, not at signalling emergencies. Thresholds are configurable through environment
variables. The map compares each hourly value with the threshold: this is intentional (the
site always shows the latest hour) and the comparison is meant as indicative, not as a check
of legal compliance.
These are forecasts, not measurements. They come from a numerical model of the atmosphere,
not from monitoring stations. They describe the overall situation and its evolution well, but can
differ from what is measured at a specific spot (a busy street, the bottom of a valley). For
official data refer to your regional environmental agency.
Places
- Italian municipalities: names, codes and centroids from the ISTAT administrative
boundaries, altitude and population from ISTAT / GeoNames.
- European cities: GeoNames
cities15000, restricted to the
CAMS domain.
Both are small CSV files versioned in luoghi/dati/ and loaded into SQLite when the image is built.
How it works
Cloud Scheduler (08:30 UTC)
└─ Cloud Run Job → manage.py cams_export
└─ 1 ADS request: all of Europe, 5 pollutants, 49 h (~275 MB NetCDF)
└─ convert → gs://<bucket>/cams/YYYY-MM-DD.parquet (~54 MB) + indice.json
Browser → Cloud Run service (Django + DuckDB)
├─ reads only the requested window of the Parquet over HTTPS range requests
└─ luoghi.sqlite3 (municipalities + cities), built into the image
No database server, no queue, no GIS stack. The whole backend is Django, DuckDB and static
files in a bucket. See docs/architecture.md for the Google Cloud setup,
the reasoning behind it and how to deploy your own copy.
This is the beginning of the README. Read the rest at the original. Read the full README ↗
