Effect of MODIS Terra radiometric calibration improvements on Collection 6 Deep Blue aerosol products: Validation and Terra/Aqua consistency

Abstract

The Deep Blue (DB) algorithm's primary data product is midvisible aerosol optical depth (AOD). DB applied to Moderate Resolution Imaging Spectroradiometer (MODIS) measurements provides a data record since early 2000 for MODIS Terra and mid‐2002 for MODIS Aqua. In the previous data version (Collection 5, C5), DB production from Terra was halted in 2007 due to sensor degradation; the new Collection 6 (C6) has both improved science algorithms and sensor radiometric calibration. This includes additional calibration corrections developed by the Ocean Biology Processing Group to address MODIS Terra's gain, polarization sensitivity, and detector response versus scan angle, meaning DB can now be applied to the whole Terra record. Through validation with Aerosol Robotic Network (AERONET) data, it is shown that the C6 DB Terra AOD quality is stable throughout the mission to date. Compared to the C5 calibration, in recent years the RMS error compared to AERONET is smaller by ∼0.04 over bright (e.g., desert) and ∼0.01–0.02 over darker (e.g., vegetated) land surfaces, and the fraction of points in agreement with AERONET within expected retrieval uncertainty higher by ∼10% and ∼5%, respectively. Comparisons to the Aqua C6 time series reveal a high level of correspondence between the two MODIS DB data records, with a small positive (Terra‐Aqua) average AOD offset <0.01. The analysis demonstrates both the efficacy of the new radiometric calibration efforts and that the C6 MODIS Terra DB AOD data remain stable (to better than 0.01 AOD) throughout the mission to date, suitable for quantitative scientific analyses.

Two decades of satellite observations of AOD over mainland China using ATSR-2, AATSR and MODIS/Terra: data set evaluation and large-scale patterns

“…The C6 DT expected error is ±(0.05 + 0.15τ AERONET ) over land and +(0.04 + 0.1τ AERONET ), −(0.02 + 0.1τ AERONET ) over sea relative to the AERONET optical thickness (τ AERONET ) (Levy et al, 2013). The C6 DB expected error is ∼ ±0.03 + 0.2τ MODIS ) relative to the MODIS optical thickness (τ MODIS ) Sayer et al, 2015).…”

“…Details about these updates in C6 DT and DB data can be found in a number of recent studies (e.g., Levy et al, 2013;Tao et al, 2015;Sayer et al, 2015;Georgoulias et al, 2016). Corrections for polarization, gain and response-versus-scan corrections and detrending for MODIS/Terra degradation have been included in DB but not in DT.…”

Impact of MODIS sensor calibration updates on Greenland Ice Sheet surface reflectance and albedo trends

“…A challenge in using long-term satellite records to detect change is maintaining consistent instrument performance. Recent literature has discussed MODIS Terra sensor calibration degradation and its impacts on apparent data trends (e.g., Franz et al, 2008;Wang et al, 2012;Lyapustin et al, 2014;Polashenski et al, 2015;Sayer et al, 2015). Specifically related to GrIS trends, a recent work by Polashenski et al (2015) indicated that uncorrected sensor degradation in MODIS Collection 5 (C5) data, particularly on the Terra platform, was contributing significantly to an apparent albedo declining trend over the GrIS and that the albedo trend in large areas of the ice sheet which do not experience melt (the dry snow area) may disappear once Collection 6 (C6) calibrations are applied.…”

Novel Decomposition Scheme for Characterizing Urban Air Quality with MODIS

“…The uncertainty in the AOD (δx C ) from DB is ±(0.03 + 0.23τ) [30] as discussed before. We used the α uncertainty calculated from DT, which is 0.4 [45], since we could not find literature supporting uncertainty of α (δy C ) derived from DB.…”

“…Bilal and Nichol [29] found that DB retrieved AOD performed well compared to AERONET in the national capital region of China on both hazy and clear days, yet tended to overestimate during very polluted days (AOD > 1.5). The level of uncertainty of DB Collection 6 AOD retrievals (QA = 3) is approximately ±(0.03 + 0.23τ) [30], an improvement over Collection 5.1 uncertainty of ±(0.05 + 0.2τ) [31]. In Collection 6, a 'best of' AOD product also exists that combines AOD from DB and DT algorithms to estimate aerosol distribution over densely vegetated areas as well as bright surfaces.…”