QUALARIA: artificial intelligence based system for sub-urban scale air quality prediction

• Grant number: 23/00755-3
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Start date: May 01, 2024
• End date: April 30, 2026
• Field of knowledge: Physical Sciences and Mathematics - Geosciences - Meteorology
• Agreement: EUREKA Network
• Principal Investigator: Gabriel Martins Palma Perez
• Grantee: Gabriel Martins Palma Perez
• Company: Meteoia Datascience Ltda
• CNAE: Pesquisa e desenvolvimento experimental em ciências físicas e naturais
• City: São Paulo
• Associated researchers: Maria de Fátima Andrade ; Thiago Nogueira ; Thomas Christian Marcel Martin
• Associated scholarship(s): 24/10214-2 - Configuration of chemical transport models to support air quality AI-based models, BP.TT; 24/06438-2 - QUALARIA: artificial intelligence based system for sub-urban scale air quality prediction, BP.PIPE

Abstract

We propose to develop a prototype system at the technology readiness level (TRL) 7 for the analysis, forecasting and attribution of air quality in the urban area of São Paulo, Brazil and in the surrounding region. Exploitation and business models will be developed with the purpose of commercialising the system as a user-friendly toolkit (product or service) aimed at decision-makers and lawmakers at the city level. Over the past decade, the air quality modelling community has been putting substantial effort to increase the model resolution and to integrate the interaction of air quality with other aspects in the Earth System and the society. Most of these models are chemical-transport models (CTMs) that numerically solve equations describing the physical and chemical processes that affect the atmospheric composition. However, these models are computationally expensive and cannot effectively provide accurate estimates at the sub-urban scale. New Artificial Intelligence (AI) based downscaling approaches that take advantage of remote sensing, coarse atmospheric data and observational stations networks have recently shown good performance to simulate sub-grid atmospheric processes (Martin et al., 2020; Martin et al., 2019; Perez and Silva Dias, 2018). This approach will be adapted to provide air pollution concentration forecasts at very high spatio-temporal resolution in cities. The raw prediction will then serve to derive relevant indicators for decision-makers and policy makers, which will be accessed through interactive visualisation on an online dashboard. The system prototype at TRL level 7 will be demonstrated in the operational environment of São Paulo before it undergoes a transition to the market. Identified "prime users" with interest in air pollution questions will be invited to express their specific needs, to help improve product specification and test developing versions of the prototype system (AU)