The article outlines the initial findings of the "APEMAIA" project, funded by the Italian Space Agency in support to NASA JPL’s Multi-Angle Imager for Aerosols (MAIA) mission. In response to mounting evidence of particulate matter’s (PM) adverse health effects, the study aims to improve spatio-temporal variability representation of PM10 concentrations within urban areas. While air quality monitors offer high accuracy, they often fall short of comprehensive area monitoring. The study suggests employing Machine Learning (ML) techniques to model daily intra-urban PM10 using AOD satellite data from sources like Sentinel-3, MODIS and future MAIA data. To better understand spatio-temporal PM changes, the approach integrates supplementary factors, including meteorological data, land cover, urban morphology, socio-economic factors, and variables from auxiliary layers offering insights into vehicle traffic. Testing various Random Forest model configurations on Bari (southern Italy) using ARPA Puglia 2019-2022 PM10 reference data yielded an R 2 value exceeding 0.77
Advancing Intra-Urban PM10 Concentration Patterns in Bari City: Insights from the Apemaia Project
Aquilino, Mariella;De Lucia, Marica;Fuina, Silvana;Carbone, Francesco;Bellotti, Roberto;Monaco, Alfonso;Cilli, Roberto;Fania, Alessandro;Pantaleo, Ester;Adamo, Maria
2024-01-01
Abstract
The article outlines the initial findings of the "APEMAIA" project, funded by the Italian Space Agency in support to NASA JPL’s Multi-Angle Imager for Aerosols (MAIA) mission. In response to mounting evidence of particulate matter’s (PM) adverse health effects, the study aims to improve spatio-temporal variability representation of PM10 concentrations within urban areas. While air quality monitors offer high accuracy, they often fall short of comprehensive area monitoring. The study suggests employing Machine Learning (ML) techniques to model daily intra-urban PM10 using AOD satellite data from sources like Sentinel-3, MODIS and future MAIA data. To better understand spatio-temporal PM changes, the approach integrates supplementary factors, including meteorological data, land cover, urban morphology, socio-economic factors, and variables from auxiliary layers offering insights into vehicle traffic. Testing various Random Forest model configurations on Bari (southern Italy) using ARPA Puglia 2019-2022 PM10 reference data yielded an R 2 value exceeding 0.77I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.