Data interpretation
Modeled vs observed air quality: what each data type can tell you
A modeled forecast estimates conditions across a grid and future hours; an observation records conditions at a particular instrument and time. Both can be useful, but they are not interchangeable evidence.
Modeled grid
Point monitor
A model represents an area and a set of assumptions
OpenWeather atmospheric-composition systems combine scientific models, emissions inputs, weather, assimilation, and other data to estimate pollutant fields across a grid. OpenWeather distributes those forecasts through hourly API fields. AirAgora requests the provider’s automatic domain selection and keeps the modeled provenance visible.
The documented European grid is broad rather than street-level. Those resolutions describe broad cells, not a person’s street, room, or breathing zone. A model can still reveal regional timing patterns and provide coverage where a station is absent.
An observation belongs to one instrument and site
A regulatory or other monitor samples at a defined location using a particular instrument, method, and reporting process. Its value is tied to siting, quality control, averaging, and timestamp. It can describe that point more directly than a large grid cell.
A point measurement is not automatically representative of an entire city. Distance, terrain, local sources, sensor type, and data validation can matter. “Observed” means measured at a site; it does not mean every nearby person experienced exactly that number.
Why the two values can disagree honestly
The model cell and monitor may cover different space. Their timestamps or averaging periods may not align. A forecast run can have been issued hours before the observation. The model may smooth a local feature that the station detects, while the station may capture a site-specific feature that is not representative of the grid.
A productive comparison aligns location, valid time, pollutant, unit, averaging period, and OpenWeather scale. Without that alignment, the difference is descriptive but not a fair accuracy score.
Forecast verification requires storing the original prediction
It is incorrect to look at a later model analysis and call it the forecast that users saw. A proper verification saves the issued forecast, its run and valid time, then compares it with a reviewed observation after the period passes. AirAgora stores recommendation and source timestamps so this can be built later.
Until a verified observation adapter and matching methodology exist, AirAgora describes its output as modeled forecast data. It does not market a retrospective forecast as a measurement.
How to choose which source to use
For broad next-day planning, a forecast grid provides future hourly coverage. For an immediate local official decision, a nearby authoritative monitor or warning may be more relevant. During severe events, official instructions should be primary regardless of an app’s smoother forecast.
Using both sources can be valuable when their roles are explicit. AirAgora’s current product focuses on the modeled planner and links to methodology; it does not blend station data invisibly.
- Model: area coverage and future hours.
- Observation: one site and reporting method.
- Comparison: align time, place, pollutant, unit, and average.
- Urgency: follow official local warnings.
What AirAgora displays to prevent confusion
The result includes source attribution, modeled status, provider update time, OpenWeather provider scale, timezone, and completeness. Stale cache is labeled with the original timestamp. Missing provider data creates an unavailable state rather than a fabricated observation.
City pages repeat the spatial-resolution note because static factual context should not make a live card appear more precise. The goal is a useful decision with an explainable claim, not a false sense of measurement certainty.
Direct answers
Frequently asked questions
Is OpenWeather air quality modeled or observed?
AirAgora uses modeled OpenWeather forecasts distributed by OpenWeather. It labels that provenance rather than presenting the value as a local station measurement.
Are station observations always more accurate?
They measure one site more directly, but “accurate for the whole city” is a different question. Siting, instrument, validation, and representativeness matter.
Why is the model resolution important?
A broad forecast grid cannot resolve every road, building, or neighborhood. The value represents a modeled cell rather than an exact address.
Can a forecast be verified later?
Yes, if the original issued forecast and valid time are stored and compared with reviewed observations under a defined method. A later analysis must not be substituted for the original forecast.
Which source should I follow during a severe event?
Follow official local warnings and public-health instructions. AirAgora does not paywall urgent guidance or claim its modeled forecast overrides authorities.
Does AirAgora blend station and model data?
Not in the current product. It uses modeled OpenWeather forecast data and says so explicitly.
Evidence register
Sources and further reading
The links below identify the primary references used on this page. Selection, summary, and presentation are AirAgora modifications.
Keep exploring
AirAgora provides informational planning guidance, not medical advice. Local conditions can differ from modeled forecasts; follow official guidance during severe events.