Air quality forecasts as provided by different providers can create some confusion. It is important to understand what they actually indicate.
Forecasts of air quality are very similar to your daily weather forecast. We, the Copernicus Atmosphere Monitoring Service (CAMS) use computer models of the atmosphere that we initialise with as many observations (from satellites, and in Europe from measurement stations as well) as possible to provide a forecast for the next 5 days of various atmospheric pollutants. These computer models can only represent larger-scale patterns (the model that CAMS uses for its world-wide forecasts has a spatial resolution of 40 km) and will not be able to exactly match observed levels of for instance PM10. Those observations can be taken next to a road (high levels of PM10) or next to a park or lake (much lower levels of PM10), while the forecast model will represent the average of a larger area. What the forecast models can do and observations cannot is tell you about the expected changes for the coming days. These temporal changes are dominated by larger-scale patterns and therefore generally well captured by the forecast models.
As explained in the video "Why's the air quality forecast on my phone different from direct air quality monitoring stations?", this is a forecast that gives you an indication of the pollutant levels over a larger area. This is very similar to weather forecasting, where each day a weather forecast is generated combining observations and forecast models to give you the expected temperature (rainfall, wind, etc.) for the next few days. The reason I am using weather forecasts for comparison is that a weather forecast also won't give you the precise temperature of a certain measurement station. That does not mean that the weather forecast does not have value. It helps people to be aware of and prepare for the expected changes for the coming days. The same holds for air quality forecasts; they are meant to help people to be aware of upcoming changes. They are not meant to represent the actual conditions at individual measurement stations.
One could deploy a very dense measurement station network to capture the full variability of air quality around the world at any given time, but this is just not feasible. It also does not provide advance warning of episodes. Combining observations with forecast models is in our view a very valuable way of presenting information to the public. It does mean, though, that we indeed have to communicate this clearly to the general public, hence the video linked above.
Also, to be clear, while Copernicus as an European programme is developing and launching new satellite instruments to observe the Earth environment, we rely on individual countries around the world to give us access to measurements at the surface. We work with the many agencies in Europe, North America, and other parts of the world to have a good exchange of data to ensure the public in the end has access to the best information. Also, we do not control how companies like the Weather Underground use our forecasts to provide information about air quality, but we do talk to them to ensure they understand what information we can provide with our forecasts.
Finally, presenting information on air quality to the general public is usually done through a so-called Air Quality Index (AQI). This index combines the levels of specific pollutants to provide a number on a scale from 1 to 5 or sometimes 1 - 10. Unfortunately, these AQIs vary around the world. In Europe, we use the European Environment Agency AQIs, while in the USA the one from AirNow is more common. CAMS provides forecasts of the concentrations of the various species (O3, NO2, PM10), which are then converted to the relevant AQI by whomever provides the service.