Airvoice is developing a novel street-scale-resolution continuous air quality map

Today, similar services already exist, but most of them provide unverified data. Such services are useful for obtaining an overview of the air quality in a city but are not suitable for making responsible decisions. City managers need reliable data. To get that reliable data, it is necessary to identify or deploy reliable real-time monitoring networks in the area of interest, adapt numerical models to local conditions, and conduct an inventory of emission sources (and then to keep it up-to-date combining automatic and manual updates).

Here’s how our model works.

Our fusion model combines a regional chemical transport model with a quasi-stationary building-height-aware Gaussian model. This predictive pollutant dispersion model is driven by either real-time traffic data or traffic forecasts provided by a powerful ML model; other pollution source data; the city’s built environment data; and more. Meteorological inputs are drawn from up-to-date regional weather forecasts produced by an expertly tuned WRF-ARW model.