Publications

Su, X; Wang, LC; Zhang, M; Qin, WM; Bilal, M (2021). A High-Precision Aerosol Retrieval Algorithm (HiPARA) for Advanced Himawari Imager (AHI) data: Development and verification. REMOTE SENSING OF ENVIRONMENT, 253, 112221.

Abstract
Due to the complexity of land cover and aerosol types, the high-precision retrieval of land aerosol properties is challenging. A land general aerosol (LaGA) algorithm called the High-Precision Aerosol Retrieval Algorithm (HiPARA) is proposed for the Advanced Himawari Imager (AHI) sensor over East Asia. In this algorithm, a monthly spectral base reflectance ratio library was constructed to obtain a pixel-by-pixel spectral reflectance relationship model. Statistical methods were used to obtain aerosol types in China, and a linearization scheme for the aerosol types was proposed based on sensitivity analysis. Based on these techniques, HiPARA achieved a completely dynamic determination of the surface reflectance and aerosol types. The new multiband aerosol characteristic retrieval strategy can return two parameters, aerosol optical depth (AOD) and single scattering albedo (SSA). The AOD retrieved from HiPARA showed high consistency with AERONET AOD measurements, with correlation coefficients (R) of 0.939, a mean absolute error (MAE) of 0.082, a root mean square error (RMSE) of 0.113, ratios that meet an expected error (EE) of 0.825, and ratios that meet a Global Climate Observing System (GCOS) error of 0.339. Comparison of the HiPARA retrieved AOD with other operational aerosol products revealed that the accuracy of the HiPARA product was better than those of the Japan Aerospace Exploration Agency (JAXA) product (R < 0.9, RMSE > 0.175), Moderate-resolution Imaging Spectro-radiometer (MODIS) products (Dark Target (DT) algorithm, R = 0.907, RMSE = 0.203; Deep Blue (DB) algorithm, R = 0.909, RMSE = 0.139) and Visible/Infrared Imager Radiometer Suite (VIIRS) AERDB product (R = 0.932, RMSE = 0.117). The HiPARA fitting line was close to 1:1 (i.e., y = x). The aerosol products were further evaluated for four extreme aerosol events: a smoke aerosol event, a haze aerosol event, a dust aerosol event, and a continuous aerosol variation event. Observation of the true-color images and AOD retrievals showed that the HiPARA AOD distribution matched the true-color images very well. The JAXA product had abnormal values and spatial discontinuities. The MODIS and VIIRS products were not as good as the HiPARA product in terms of spatial coverage. In the application to continuous monitoring of a dust event, the HiPARA AOD captured the variations in and intensity of the dust aerosols very well. These results suggested the robustness of HiPARA and its potential for monitoring extreme pollution events with high precision and high temporal resolution.

DOI:
10.1016/j.rse.2020.112221

ISSN:
0034-4257