TY - JOUR
T1 - A novel multi-pollutant space-time learning network for air pollution inference
AU - Song, Jun
AU - Stettler, Marc E.J.
N1 - Funding Information:
Jun Song was supported by the China Scholarship Council (CSC). The air pollutant monitoring data and meteorological data were collected form Chengdu Ecological Environment Bureau (http://sthj.chengdu.gov.cn/); the land use data and traffic data were collected from Gaode Map (https://www.amap.com/); and the public vitality data was provided by Cloud-Guizhou Research Institute.
Publisher Copyright:
© 2021 Elsevier B.V.
PY - 2022/3/10
Y1 - 2022/3/10
N2 - Detailed information about air pollution in space and time is essential to manage risks to public health. In this paper we propose a multi-pollutant space-time learning network (Multi-AP learning network), which estimates pixel-wise (grid-level) concentrations of multiple air pollutant species based on fixed-station measurements and multi-source urban features, including land use information, traffic data, and meteorological conditions. We infer concentrations of multiple pollutants within one integrated learning network, which is applied to and evaluated on a case study in Chengdu (4900 km2, 26 April - 12 June 2019), where air pollutant (PM2.5, PM10 and O3) measurements from 40 monitoring sites are used to train the network to estimate pollutant concentrations in 4900 grid-cells (1 km2). The Multi-AP learning network allows us to estimate highly-resolved (1 km × 1 km, hourly) air pollution maps based on pollutant measurements which cover less than 1% of the grid-cells with better accuracy compared to other approaches, and with significant computational efficiency improvements. The time-cost is 1/3 of the time-cost of modelling each pollutant individually. Furthermore, we evaluate the relative importance of features and find that the meteorological feature set is the most important followed the land use features. The proposed Multi-AP method could be used to estimate air pollution exposure across a city using a limited set of air pollution monitoring sites.
AB - Detailed information about air pollution in space and time is essential to manage risks to public health. In this paper we propose a multi-pollutant space-time learning network (Multi-AP learning network), which estimates pixel-wise (grid-level) concentrations of multiple air pollutant species based on fixed-station measurements and multi-source urban features, including land use information, traffic data, and meteorological conditions. We infer concentrations of multiple pollutants within one integrated learning network, which is applied to and evaluated on a case study in Chengdu (4900 km2, 26 April - 12 June 2019), where air pollutant (PM2.5, PM10 and O3) measurements from 40 monitoring sites are used to train the network to estimate pollutant concentrations in 4900 grid-cells (1 km2). The Multi-AP learning network allows us to estimate highly-resolved (1 km × 1 km, hourly) air pollution maps based on pollutant measurements which cover less than 1% of the grid-cells with better accuracy compared to other approaches, and with significant computational efficiency improvements. The time-cost is 1/3 of the time-cost of modelling each pollutant individually. Furthermore, we evaluate the relative importance of features and find that the meteorological feature set is the most important followed the land use features. The proposed Multi-AP method could be used to estimate air pollution exposure across a city using a limited set of air pollution monitoring sites.
KW - Air pollution
KW - Air pollution modelling
KW - Learning network
KW - Particulate matter
UR - http://www.scopus.com/inward/record.url?scp=85121609311&partnerID=8YFLogxK
U2 - 10.1016/j.scitotenv.2021.152254
DO - 10.1016/j.scitotenv.2021.152254
M3 - Journal article
SN - 0048-9697
VL - 811
JO - Science of the Total Environment
JF - Science of the Total Environment
M1 - 152254
ER -
