Центр данных Atmy

Rankings use PM2.5 in µg/m³. The selected AQI standard, displayed above each chart, controls its colour scale. Daily tables sort measured PM2.5, not the colour category.

Sensor readings are averaged into hourly sensor means. City daily values average the available sensor-hour means with equal row weight, rounded to whole µg/m³; sensors reporting more hours can contribute more rows. The daily city endpoint averages stored city summaries for its requested date (by default yesterday in the server timezone), requiring more than two distinct contributing sensors (cnt > 2) and a non-null AQI. No minimum number of valid hours per sensor is enforced here. It returns up to 100 cities per ordering. Duplicate city rows are merged, keeping the row with the higher AQI. Where country city data are missing, available sensor averages are combined with equal sensor weight. Global daily rankings require at least three currently reporting sensors; the country overview has no additional sensor-count minimum.

Monthly charts use available monthly means, not a completed annual review. The frontend averages and rounds available positive monthly values with equal month weight. At least six PM2.5 months are required. Global monthly rankings also require three reporting sensors and an AQI above 50. The rolling chart requires six months and one reporting sensor. Missing or non-positive monthly values are gaps, not interpolated values. Cities affected by a reported PM2.5 value above 500 µg/m³ are excluded rather than clamped. No additional statistical outlier correction is applied here.

The daily chart date comes from the backend response and retains its calendar date; aggregation boundaries use the backend timezone. Other interface timestamps use the browser timezone. The backend does not return an aggregation timezone identifier. Statistical responses are cached for six hours; currently reporting sensor counts for five minutes. Data are refreshed on page activation or explicit refresh. This is not a guarantee of measurement frequency or full city coverage.

Visualization © Atmy. Underlying measurements retain their providers’ licences and attribution requirements. Aggregated city endpoints do not identify each contributing provider or supply per-row licence metadata, so an aggregate cannot be relabelled as Atmy CC BY 4.0.

• Sensor.Community: Database Contents License (DbCL) 1.0, with the applicable ODbL database obligations.
• AirGradient public API data: CC BY-SA 4.0. Data redistributed through another service may have different terms.
• Nebo, AirVisual, PurpleAir, Atmy and government sources are also supported by the application. Their applicable dataset terms must be verified with the provider; no blanket reuse licence is granted here. Test and unknown sources do not carry a verified licence.

Data use and attribution terms