Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Filters applied. Clear all
. 2017 Aug;197:125-140.
doi: 10.1016/j.rse.2016.11.015. Epub 2016 Dec 10.

Retrieval of Aerosol Optical Properties Using MERIS Observations: Algorithm and Some First Results

Affiliations
Free PMC article

Retrieval of Aerosol Optical Properties Using MERIS Observations: Algorithm and Some First Results

Linlu Mei et al. Remote Sens Environ. .
Free PMC article

Abstract

The MEdium Resolution Imaging Spectrometer (MERIS) instrument on board ESA Envisat made measurements from 2002 to 2012. Although MERIS was limited in spectral coverage, accurate Aerosol Optical Thickness (AOT) from MERIS data are retrieved by using appropriate additional information. We introduce a new AOT retrieval algorithm for MERIS over land surfaces, referred to as eXtensible Bremen AErosol Retrieval (XBAER). XBAER is similar to the "dark-target" (DT) retrieval algorithm used for Moderate-resolution Imaging Spectroradiometer (MODIS), in that it uses a lookup table (LUT) to match to satellite-observed reflectance and derive the AOT. Instead of a global parameterization of surface spectral reflectance, XBAER uses a set of spectral coefficients to prescribe surface properties. In this manner, XBAER is not limited to dark surfaces (vegetation) and retrieves AOT over bright surface (desert, semiarid, and urban areas). Preliminary validation of the MERIS-derived AOT and the ground-based Aerosol Robotic Network (AERONET) measurements yield good agreement, the resulting regression equation is y = (0.92 × ± 0.07) + (0.05 ± 0.01) and Pearson correlation coefficient of R = 0.78. Global monthly means of AOT have been compared from XBAER, MODIS and other satellite-derived datasets.

Keywords: AOT; MERIS; Retrieval.

Figures

Fig. 1
Fig. 1
Comparison of different cloud mask algorithms. In the upper left panel is the RGB composite figure for the selected cloud scene; the upper right panel is the “true” cloud mask created by Gómez-Chova et al. (2007); the lower left panel shows BAER cloud mask; the lower right panel shows XBAER cloud mask. The XBAER cloud mask uses also the 3 × 3 adjacency pixels to minimize the “twilight zone” effect. If a pixel is detected as cloud, the surrounding 3 × 3 block will be automatically marked as cloud.
Fig. 2
Fig. 2
Aerosol types over land used in the XBAER algorithm designated at 1° × 1° grid for different seasons. The four sub-figures represent four seasons. Upper row: left – December, January and February (DJF), right - March, April and May (MAM). Lower row: left - June, July and August (JJA), right - September, October and November (SON).
Fig. 3
Fig. 3
Plots of the relationship between SSR at 412 nm, SSR at 885 nm and the ratio of SSR at 412 nm and 885 nm and the value for SAVI for three different AERONET stations (Rio_Branco, Paris and Bodele) having different surface types (Vegetation, city and desert). “cc” refers to correlation coefficient.
Fig. 4
Fig. 4
Plots of the spectral coefficients dataset available for the three different AERONET stations (Rio_Branco, Paris and Bodele) having different surface types (vegetation, city and desert).
Fig. 5
Fig. 5
Plots of the global spectral coefficients dataset components for July. Left panels – MERIS channels 1. Right panels – MERIS channel 14 (Please note a14) over desert (dark blue) can be negative). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 6
Fig. 6
Plots of the approximated spectra for the reference surface spectra for the three different AERONET stations shown in Fig. 3 The upper row is for the 16th July 2008. The lower row is for the 2nd July 2009. The red, green and blue circles represent the SSR from land-cover cci (Ref), SSR fitted using Eq. (16) (Ref_f) and surface reflectance using standard SAVI (Ref_s). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 7
Fig. 7
Global 7 days composite SSR for MERIS channel 1 and channel 14 and the difference between XBAER estimated MERIS SSR and land-cover-cci MERIS SSR product for the period of the 24th to 30th of July 2009 for. Upper panels: left – XBAER estimated MERIS SSR for channel 1, right-XBAER estimated MERIS SSR for channel 14. Lower panels: left-the difference between XBAER estimated MERIS SSR and reference SSR for channel 1, right- the difference between XBAER estimated MERIS SSR and reference SSR for channel 14.
Fig. 8
Fig. 8
Flowchart, which pictorially describes the steps in the XBAER aerosol retrieval algorithm over land.
Fig. 9
Fig. 9
Plot of AOTs (550 nm) uncertainty versus AOT (550 nm) obtained from the sensitivity study of the XBAER algorithm and their dependence on the surface reflection type and aerosol model. (Red line - Lambertian surface; green line - BRDF effect for three different days; blue line - Moderately absorbing aerosol; black line - Strongly absorbing aerosol). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 10
Fig. 10
Comparison of XBAER AOT with AERONET observation for 2009 July. R and N refer to the Pearson correlation coefficient and the number of locations used in the validation respectively. The dash lines are ± 20%τ ± 0.05.
Fig. 11
Fig. 11
Histogram of difference between XBAER AOT and AERONET observations for 2009 July separated into the following: global, Southern Hemisphere and Northern Hemisphere.
Fig. 12
Fig. 12
Comparison of retrieved global monthly mean AOT at 550 nm for July 2009. Upper row: left – MODIS, right – MISR. Lower row: left – AATSR (SU), right – MERIS (XBAER).

Similar articles

See all similar articles

Cited by 1 article

LinkOut - more resources

Feedback