Evaluation of LJ1-01 Nighttime Light Imagery for Estimating Monthly PM<sub>2.5</sub> Concentration: A Comparison With NPP-VIIRS Nighttime Light Data
Guo Zhang, Yingrui Shi, Miaozhong Xu
Abstract
Air quality degradations caused by fine particulate matter (PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> ) can lead to various health problems, and accurate PM2.5 data are critical for managing the environment and ensuring public health. Radiation signals collected by nighttime light (NTL) remote sensing satellites are influenced by PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> concentrations, and thus, incorporating NTL imagery in statistical models has been widely used to predict PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> concentrations. However, scarce work has been carried out with new-generation NTL data from the LJ1-01 satellite, which has a fine spatial resolution and wide measurement range. In this study, we integrated satellite observation data and meteorological data to construct five models based on the geographically weighted regression to validate the feasibility of LJ1-01/NPP-VIIRS in Moderate Resolution Imaging Spectroradiometer AOD-based PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> prediction in the Beijing-Tianjin-Hebei region. The models were validated by the cross-validation method. The results showed that the addition of NTL information could improve the performance of the PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> prediction model. The seasonal R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> with NTL in AOD-PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> model have improved by 5.07%, 4.50%, 2.95%, and 2.56% in model fitting and 1.20%, 1.75%, 2.20%, and 4.41% in cross-validation. Furthermore, the LJ1-01 NTL data revealed additional details and improved the prediction accuracy, compared with the NPP-VIIRS in AOD-PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> model, the seasonal R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> with LJ1-01 in AOD-PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> model increased by 1.16%, 1.79%, 0.76%, and 1.15% in model fitting and 1.04%, 0.85%, 0.78%, 1.37% in cross-validation. Thus, our findings indicate that LJ1-01 and NTL data have the potential for predicting PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> and that they could constitute a useful supplemental data source for estimating ground-level PM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</sub> distributions.