QUALARIA Project: Air Quality Prediction Artificial Intelligence System in Street-Level Scale

The QUALARIA Project main objective is the development of a state-of-the-art Artificial Intelligence (AI) model, able to provide accurate predictions of air pollutant concentrations, air quality indexes and health risks in suburban scale (< 100 m). The inference capacity in high spatial resolution is enabled by the incorporation of a low-cost air pollutant sensor (Bernard Air Analysers, BAAs) observational network, spread in a miscellaneous urban configuration. In this way, the project pursuits overcome the limitation of current info by traditional methodologies, which provide regional scale estimates (for example, 40 km from the dynamic model Copernicus Atmosphere Monitoring Service/European Centre for Medium-Range Weather Forecasts (CAMS/ECMWF)), restricting the effective decision making in the urban context. The final product is an online dashboard with air pollution predictions and high-performance health indicators, allowing the implementation of risk quantification and actions for mitigation in cities. The features and predicted indicators of the dashboard are designed together with national and international stakeholders from the public and private sectors. The target audience are decision makers, public policy makers, urban planners, public health, climate and environment institutions, and air quality non-governmental organizations. A pilot is being implemented for the Metropolitan Area of São Paulo (MASP), with the intention of scaling to all Brazil and other countries later. The AI model is being trained by the air pollutant observational networks and BAAs deployed in 28 MASP spots, in addition to info of surface physical features derivative from satellite imagery, air pollutant prediction from CAMS/ECMWF, stationary and mobile sources, building height, populational density, simulations from the model Weather Research and Forecasting with Chemistry, among other inputs.