Improving Advanced Baseline Imager (ABI) Aerosol Optical Depth (AOD) Retrievals using a Bias Correction Algorithm

The Advanced Baseline Imager (ABI) on board the Geostationary Operational Environmental Satellite-R (GOES-R) series provides AOD retrievals using multi-band retrieval similar to polar-orbiting satellites sensors such as Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS). However, this work demonstrates that the current version of GOES-16 (GOES-East) ABI AOD has diurnally varying biases due to limitations in the land surface reflectance relationships between the 0.47 m band and the 2.2 m band and between the 0.64 m band and 2.2 m band used in the ABI AOD retrieval algorithm, which vary with the Sunsatellite geometry and normalized difference vegetation index (NDVI). To minimize these biases, a bias correction algorithm was developed: For any given time of the day and pixel, the lowest AOD c of a 30-day period at each time step and at each pixel is obtained from time series of ABI AOD and its difference with respect to the background AOD b is assumed to be the bias. b is obtained from a network of AErosol RObotic NETwork (AERONET) observations distributed around the globe. The bias correction improves the performance of ABI AOD when compared to AERONET AOD, especially for the high and medium (top 2) quality ABI AOD; the top 2 AODs have much larger spatial coverage in area than the high quality ABI AOD alone. The screening for quality of ABI AOD is done in the retrieval algorithm using several different criteria (e.g., snow/cloud adjacency, test for surface inhomogeneity, high solar zenith angle, etc.) and the high quality flag screens out many retrievals, sometimes even the good ones. After the bias correction, the correlation between ABI and AERONET AOD improved from 0.87 to 0.91, the mean bias improved from 0.04 to nearly zero bias, and root mean square error (RMSE) improved from 0.09 to 0.05. The improvements make top 2 quality AODs as good as high quality AOD and increase the AOD data coverage area by 100% rendering the product useful for near real time monitoring of changes in pollution on hourly to daily time scales.