VIIRS Deep Blue Aerosol Products Over Land: Extending the EOS Long‐Term Aerosol Data Records

Abstract

A primary goal of the Deep Blue (DB) project is to create consistent long‐term aerosol data records, suitable for climate studies, using multiple satellite instruments. In order to continue Earth Observing System (EOS)‐era aerosol products into the Joint Polar Satellite System era, we have successfully ported the DB algorithm to process data from the Visible Infrared Imaging Radiometer Suite (VIIRS). Although the basic structure of the VIIRS algorithm is similar to that for the Moderate Resolution Imaging Spectroradiometer (MODIS), many enhancements have been made compared to the MODIS collection 6 (C6) version. Most have also been implemented in the latest MODIS Collection 6.1 (C6.1). For example, a new smoke mask was developed based on the spectral curvature of measured reflectance to distinguish biomass burning smoke from weakly absorbing urban/industrial aerosols. Consequently, a new aerosol‐type flag was added into the VIIRS DB data set. In addition, new dust models have been developed to account for the nonsphericity of mineral dust. As a result, a discontinuity in the retrieved aerosol optical depth (AOD) of Saharan dust plumes seen in MODIS C6 products near the boundary between North Africa and the Atlantic has been much reduced. We have also evaluated the VIIRS and MODIS Terra/Aqua C6.1 AOD against Aerosol Robotic Network data. VIIRS and MODIS retrievals show similar performance; around 80% of matchups agree with Aerosol Robotic Network within the expected error of ±(0.05 + 20)%, indicating that DB can provide consistent AOD through the historical EOS and present Joint Polar Satellite System eras.

AERONET Remotely Sensed Measurements and Retrievals of Biomass Burning Aerosol Optical Properties During the 2015 Indonesian Burning Season

“…The VIIRS DB algorithm provides retrievals on every day except the thickest few when the AERONET daily mean AOD at 550 nm was above 5. Although VIIRS does have a broader swath than MODIS, the primary reason for this extra coverage is the additional relaxation of cloud masking tests in cases of heavy smoke, designed to retain such events (Hsu et al, ). However, for this case, the whole of September–October is masked out by the QA flag, which is intentional as these extreme conditions are expected to be inherently more uncertain and more likely to suffer from cloud contamination.…”

“…2.2.2. MODIS/VIIRS DB The C6.1 MODIS DB retrieval algorithm (Hsu et al, 2019) is a refinement of the previous C6. Like DT, DB provides AOD at 550 nm with a nominal 10-km nadir pixel size.…”

“…Journal of Geophysical Research: Atmospheres designed to retain such events (Hsu et al, 2019). However, for this case, the whole of September-October is masked out by the QA flag, which is intentional as these extreme conditions are expected to be inherently more uncertain and more likely to suffer from cloud contamination.…”

Climate Models and Remote Sensing Retrievals Neglect Substantial Desert Dust Asphericity

“…In (c) and (e), Chou et al (2008) is not used in regressions due to the insignificant number of analyzed particles for HWR. In (e), values of the ratio of dust length to its height for climate models and retrieval algorithms are based on Mahowald et al (2014), Dubovik et al (2006), Hsu et al (2019), and Kalashnikova and Kahn (2006), respectively. MISR v23 is unity for diameter <0.16 μm and offset for clarity.…”

How should we aggregate data? Methods accounting for the numerical distributions, with an assessment of aerosol optical depth

“…O'Neill et al (2000) showed that AOD derived from Sun photometer measurements at a variety of individual Aerosol Robotic Network (AERONET) sites spread around the world tends to have frequency distributions which statistically resemble a lognormal distribution to a much stronger degree than normal. All these previous studies were of data aggregated over time summarised at individual locations; around the same time, providing an early satellite example, Ignatov and Stowe (2000) found approximately log-normal AOD (and normal α) in aerosol retrievals over ocean scenes. This indicated that log-normal tendencies might be found in AOD data also aggregated spatially as opposed to just temporally.…”