Satellite-Based Estimation of Hourly PM2.5 Concentrations Using a Vertical-Humidity Correction Method from Himawari-AOD in Hebei

Qiaolin Zeng, Liangfu Chen, Hao Zhu, Zifeng Wang, Xinhui Wang, Liang Zhang, Tianyu Gu, Guiyan Zhu, Yang Zhang2018 · DOI

TL;DR

This paper introduces a novel method to estimate hourly PM2.5 concentrations using satellite-derived Aerosol Optical Depth (AOD) by incorporating vertical and humidity corrections. This approach significantly improves the accuracy of PM2.5 estimation compared to traditional methods, offering a more robust tool for monitoring air quality in regions like Hebei, China.

What Problem Does This Paper Address?

Accurate monitoring of fine particulate matter (PM2.5) is crucial due to its significant health impacts. While ground-based measurements provide high accuracy, their sparse and uneven distribution limits our understanding of spatial-temporal variations. Satellite remote sensing offers broad coverage, but directly correlating satellite AOD with ground-level PM2.5 is challenging due to factors like atmospheric humidity and the vertical distribution of aerosols. Existing methods often struggle with temporal resolution and fail to account for these crucial atmospheric conditions, leading to less accurate PM2.5 estimations. This study aims to bridge this gap by developing a more sophisticated satellite-based estimation method that accounts for these complexities.

Methods

The core of this research lies in a new "vertical-humidity correction" method applied to Himawari-8 AOD data. The methodology involves several key steps:

  • Vertical Correction: This step aims to convert the column-integrated AOD into a surface aerosol extinction coefficient. It assumes a homogeneous atmosphere and models the vertical distribution of extinction as an exponential decay, using the boundary layer height as a scale factor.
  • Humidity Correction: Aerosol particles can absorb water vapor, changing their optical properties. This method accounts for this by introducing a "hygroscopic growth factor" (f(RH)). This factor, which describes how aerosol extinction changes with relative humidity (RH), is fitted at each ground-based monitoring site using observed data.
  • Spatial Interpolation of f(RH): To apply the humidity correction across the entire satellite image, the study assumes that f(RH) doesn't change drastically over small scales. Therefore, the nearest fitted f(RH) value from a ground site is used for each satellite pixel.
  • PM2.5 Estimation: Finally, the corrected extinction coefficient is used to estimate PM2.5 concentrations using a derived relationship.

This approach differs from previous studies by fitting the hygroscopic growth function at individual ground sites and then spatially interpolating these values, rather than using a single, generalized correction factor.

Datasets

The study utilizes the following datasets:

  • Himawari-8 AHI AOD: High spatial (1 km) and temporal (every 30 minutes) resolution AOD data retrieved using a new dark target algorithm.
  • Ground-based PM2.5 Concentrations: Hourly PM2.5 measurements from January to June 2017, collected from 137 matched environmental monitoring sites in Hebei province.
  • Meteorological Data: Hourly data including visibility (VIS), relative humidity (RH), and wind direction from 142 meteorological sites in Hebei province, also from January to June 2017.

Key Insights and Ideas

  • Humidity is Key: The study strongly emphasizes that relative humidity significantly impacts the relationship between AOD and PM2.5. Ignoring this leads to substantial errors in estimation.
  • Site-Specific Corrections Matter: Fitting the hygroscopic growth factor (f(RH)) at individual ground sites, rather than using a regional average, is crucial for improving accuracy. This acknowledges the spatial variability in aerosol composition and its hygroscopic behavior.
  • Coastal vs. Inland Aerosols: The research highlights distinct hygroscopic growth characteristics between coastal sites (like Qinhuangdao-Changli, with sea salt components) and inland sites (like Xingtai-Nanhe, with black carbon). This underscores the need for localized correction factors.
  • "Dry" Extinction is More Reliable: Correcting for humidity and vertical distribution results in a "dry" extinction coefficient that shows a significantly stronger correlation with ground-measured PM2.5 than raw AOD.

Key Results

  • Improved Correlation: The correlation between the "dry" extinction coefficient and ground-measured PM2.5 improved substantially, with 'r' values increasing from 0.19-0.47 to 0.61-0.76.
  • High Accuracy Estimates: The satellite-estimated hourly PM2.5 concentrations showed a high correlation (r = 0.8 ± 0.07) and a low RMSE (30.4 ± 5.5 µg/m³) when compared to ground measurements.
  • Temporal Accuracy: Accuracy was found to be higher in the afternoon (13:00-16:00) than in the morning (09:00-12:00).
  • Daily Averages: For daily average PM2.5 concentrations at 11 sites, the correlation was approximately 0.91 ± 0.03, with RMSEs ranging from 13.94 to 31.44 µg/m³.
  • Pollution Dynamics: The high spatial-temporal resolution data effectively captured pollution diffusion and accumulation processes, demonstrating its utility for regional air quality assessment.

Limitations and Future Work

  • Hygroscopic Factor Spatial Extent: The study acknowledges that obtaining the hygroscopic growth factor far from ground monitoring sites can impact accuracy, especially when meteorological conditions change rapidly.
  • Particle Composition Ignored: The research did not consider the influence of specific particle compositions on PM2.5 estimation, which could be a source of error.
  • Future Work: The authors suggest that future research should focus on obtaining more ground-based data and investigating the impact of particle composition to further improve PM2.5 inversion accuracy. They also propose that obtaining the hygroscopic growth factor from more ground-based data and researching particle composition will help improve the PM2.5 inversion accuracy in future research.

Summary updated

BibTeX
@misc{zeng2018satellite,
  author = {Qiaolin Zeng and Liangfu Chen and Hao Zhu and Zifeng Wang and Xinhui Wang and Liang Zhang and Tianyu Gu and Guiyan Zhu and Yang Zhang},
  title = {{Satellite-Based Estimation of Hourly PM2.5 Concentrations Using a Vertical-Humidity Correction Method from Himawari-AOD in Hebei}},
  year = {2018},
  doi = {10.3390/s18103456},
}
Figure