iHorse: improving air quality and health risk by data-driven modelling of traffic and atmospheric environment
Stockholm city implements a new forecasting system for air pollution and pollen. Based on air pollution measurement at a number of stations, an air quality health index (AQHI) is calculated in order to inform the population on the health risks associated with outdoor activities in the next few days. The AQHI captures the combined effects associated with exposures to pollutants (NOx, O3 and PM10) and birch pollen. The AQHI is presented in a mobile phone app and via the web and will be available from 2021. The same system will also be used by the Swedish Transport Administration for predicting the impact on air pollutant concentrations of environmental variable speed limit control on E4/E20. The objective of our project is to increase the accuracy of the air pollution and health risk forecasts. The current system relies on deterministic meteorological dispersion modelling to forecast impacts of emissions on concentrations. One of the main uncertainties is to forecast emissions from road traffic. Road traffic is a dominant source of air pollution in urban environment. Dynamic data-driven traffic information has earlier mainly been used for estimating the impact on the emissions, but not widely applied to predict air pollution concentrations and associated health risk impacts. The project combines big traffic data, air pollution data and meteological forecasting information for predicting air pollution and AQHI in several future days.
Keywords: Big data; Air quality forecasting; Data-driven modelling; Environment monitoring; IoT sensors.
Recent publications:
Z. Zhang, X. Ma∗, C. Johansson, J. Jin, M. Engardt, "A Meta-graph Deep Learning Framework for Forecasting Air Pollutants in Stockholm," 2023 IEEE 9th World Forum on Internet of Things (WF-IoT), Aveiro, Portugal, 2023.
Z. Zhang et al., Improving 3-day deterministic air pollution forecasts using machine learning algorithms, Atmos. Chem. Phys., 24, 807–851, https://doi.org/10.5194/acp-24-807-2024, 2024
Recent presentations:
