Comparison of hourly aerosol retrievals from JAXA Himawari/AHI in version 3.0 and a simple customized method

Abstract

Advanced Himawari imager (AHI) carried on the new-generation geostationary meteorological Himawari-8 satellite of Japan has been generating aerosol observations with a high temporal resolution since 7 July 2015. However, the previous studies lack a comprehensive quality assessment and spatial coverage analysis of AHI hourly aerosol products (level 3 version 3.0) across the full disk scan. The monitoring accuracy of different AHI aerosol products (AODpure and AODmerged) and a simple customized product (AODmean) was evaluated against Aerosol Robotic Network (AERONET) and Maritime Aerosol Network (MAN) observations from May 2016 to February 2019 in this study. Results showed that AHI AODmean demonstrates a better agreement to AERONET AOD measurements than AODpure and AODmerged over land (R = 0.81, bias = − 0.011) and all the AHI land retrievals present a significant regional performance differences, while the relatively better performance is observed in AODmerged over the coastal regions (R = 0.89, bias = 0.053). Over ocean, AHI exhibited overall overestimation in retrieving AOD against MAN observations and the relatively lower uncertainties were found in AODpure retrievals (R = 0.96, bias = 0.057). The hourly comparisons in different AHI products demonstrated a robust performance in the late afternoon (16:00–17:00 LT) over land and around the noon (10:00–13:00 LT) over coast. AHI AOD products indicated an obvious underestimation when compared to MODIS AOD retrievals over both land and ocean. Furthermore, the performance differences of AHI AOD products have also affected by the vegetation cover, pollution levels and relative humidity. For spatiotemporal coverage, the results of different AHI products demonstrated that AODmean can achieve relatively higher coverage than AODpure and AODmerged, and AHI retrievals present significant regional differences in coverage capability.

Spatiotemporal analysis of PM2.5 estimated using machine learning over Greater Bangkok: Variability, trends, and persistence

“…Sub-hourly to the hourly conversion of AERONET AOD was done to match AHI AOD. For spatial collocation, AHI AOD was averaged for the spatial box of 5 × 5 grid cells centered around AERONET sites to ensure enough non-missing sample points for both datasets as done in other studies (Xu et al 2020;Jiang et al 2019). Four statistical metrics were considered for AOD evaluation, which are correlation coe cient (ρ), mean bias (MB), mean absolute error (ME), and root mean square error (RMSE).…”

A Study of a Severe Spring Dust Event in 2021 over East Asia with WRF-Chem and Multiple Platforms of Observations

“…In order to facilitate comparison, the observed AOD is interpolated to the same resolution as the model in the follow-up study. Due to the removal of cloud-contaminated observations, the quality of Himawari-8 level 3 data is higher, but the coverage is reduced [55]. Although some observations are missing due to quality control, the AOD products of Himawari-8 are reliable.…”