Open AccessA novel approach of Landsat 8 imagery to predict PM2.5 concentrations in a south-eastern coastal city of China
Author(s) -
Lijuan Yang
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/619/1/012046
Subject(s) - environmental science , remote sensing , satellite , aerosol , vegetation (pathology) , scale (ratio) , satellite imagery , reflectivity , geography , meteorology , cartography , medicine , physics , optics , pathology , aerospace engineering , engineering
Satellite remote sensing data with moderate-to high-resolution has been commonly used in deriving spatial coverage of PM 2.5 concentrations in urban areas. Previous studies focusing on city-scale PM 2.5 estimation mainly retrieved aerosol optical depth from moderate-to high-resolution remote sensing data. In this study, the spectral response experiment was carried out to explore the sensitivity of spectral wavelengths to PM 2.5 concentrations in Fuzhou, China. The results showed that the near-infrared reflectance was much more sensitive to PM 2.5 than other wavelengths. We also found that the difference vegetation index (DVI) presented higher correlation with PM 2.5 than other indexes. A linear mixed effects (LME) model was then developed to explain the variability of PM 2.5 , and the results showed that the overall R 2 of LME model using DVI and meteorological parameters reached 0.80 with RMSE of 7.82 μg/m 3 . The results suggested that the proposed LME model using DVI from Landsat 8 OLI could be effectively used for predicting PM 2.5 in a city scale.