PM2.5, PM10, sulfur dioxide (SO2), nitrogen dioxide (NO2), carbon monoxide (CO), and ozone (O3) are six fundamental pollutants in atmospheric pollution, posing significant threats to human health and the ecological environment. Given the high spatiotemporal heterogeneity of atmospheric pollutants, there is a lack of in-depth exploration into the fine spatial variations and interactions among multiple atmospheric pollutants, at ultra-high spatiotemporal resolution. In addition, most of the current studies are single-pollutant predictions, which are deficient in time and resource consumption compared with multi-pollutant synergistic predictions. To address these issues, we integrated ground-measured pollutant data, top-of-atmosphere (TOA) radiation remote sensing data, Baidu Street View data, reanalysis data, and other relevant spatiotemporal data. Using a multi-output extremely randomized trees model, we collaboratively predicted six atmospheric pollutants, generating an ultra-high-spatiotemporal-resolution (temporal resolution: hourly; spatial resolution: 100 meters) dataset of atmospheric pollutants in Wuhan, where air quality still suffers from concentration exceedances in the year 2023. The ten-fold cross-validation R² for PM2.5, PM10, O3, NO2, CO, and SO2 models were 0.71-0.95, respectively. Synergistic prediction models consume only one-fifth the time of single prediction models. The spatiotemporal analysis revealed that among them, the annual average values of PM2.5 and PM10 exceeded the first-level concentration limits in China. In addition, the annual average values of pollutants have obvious spatial and temporal heterogeneity, showing a distinct spatial pattern of higher concentrations in urban centers and decreasing outwards. Correlation analyses on annual and hourly scales showed some correlation between atmospheric pollutants. In addition, the correlation between them was found to be dynamic over time at the hourly scale. These findings provide a comprehensive, high-resolution dataset of atmospheric pollutants in Wuhan, offering valuable insights into their spatiotemporal distribution and interactions.
Previous Article in event
Next Article in event
Next Article in session
Research on Synchronous Estimation of Ultra-High Spatiotemporal Resolution Concentrations for Six Standard Air Pollutants Using Satellite Remote Sensing and Street View Data
Published:
30 May 2025
by MDPI
in The 7th International Electronic Conference on Atmospheric Sciences
session Air Pollution Control
Abstract:
Keywords: Street View Imagery; Multi-output Extremely Randomized Trees; Spatiotemporal Characteristics of Atmospheric Pollutants; Remote Sensing Data; Ultra-high Spatiotemporal Resolution
Comments on this paper
