Fig 1.
Study site and the Rrs spectra.
The geographical locations of (a) Taihu Lake in East China and (b) the in situ measurements samples taken on Oct. 26, 2014 at Taihu Lake. (c) The Rrs spectra acquired from the 13 sites. The figure is for illustrative purposes only.
Fig 2.
Flow chart of the M6SV atmospheric correction.
L8/OLI L1B, the original L8/OLI L1B images; Lt, the TOA radiance; Date, the image acquisition date; Time, the image acquisition time; θs, the solar zenith angle; φs, the solar azimuth angle; θ0, the sensor zenith angle; φ0, the sensor azimuth angle; L8/OLI SRF, the L8/OLI spectral response function; AOD550, the aerosol optical depth at 550 nm; xa′, xb′, and xc′, atmospheric correction parameters.
Table 1.
M6SV configuration parameters for the L8/OLI image of Taihu Lake, China.
Fig 3.
Rrs (λ) images derived by the M6SV, VR-SWIR-F and SD-SWIR models.
Results for the atmospheric correction of an L8/OLI image (Taihu Lake, China, October 26, 2014) using the M6SV algorithm, for which (a) Rrs(483), (b) Rrs(561), (c) Rrs(655), and (d) Rrs(865) nm, the VR-SWIR-F algorithm, for which (e) Rrs(483), (f) Rrs(561), (g) Rrs(655), and (h) Rrs(865) nm, and the SD-SWIR algorithm, for which (i) Rrs(483), (j) Rrs(561), (k) Rrs(655), and (l) Rrs(865) nm.
Fig 4.
L8/OLI Rrs derived using the (a) M6SV, (b) VR-SWIR-F, and (c) SD-SWIR models plotted against in situ measurements from Taihu Lake at 13 synchropoints. The solid line is a 1:1 line; the dashed lines represent ± 15% error around the 1:1 line. The mean ratio (MR) represents the ratio between the model-derived Rrs and in situ-measured Rrs for each band. (d) A comparison of the mean Rrs spectra from the three models and the in situ measurements at the 13 synchropoints.
Table 2.
Accuracy evaluation indices for the L8/OLI Rrs derived using the M6SV, VR-SWIR-F, and SD-SWIR models at each of the four spectral bands.
Fig 5.
(a-c) Examples of validation comparisons between model-derived Rrs and in situ measurements from Taihu Lake matched sampling points. The latitudes and longitudes of the matched sampling points, times of image and in situ spectra acquisition, and models used to derive Rrs are listed. (d) Values of M6SV-derived Rrs at 4 bands changing with the value of AOD550.
Fig 6.
Rrs (λ) extracted from pixels along the transect ‘T’ in Fig 3(E).
Rrs (λ) were derived using the (green) M6SV, (yellow) VR-SWIR-F and (blue) SD-SWIR models at wavelengths of (a) 483, (b) 561, (c) 655, and (d) 865 nm.
Fig 7.
Rrs (λ) density-sliced scatterplots at wavelengths of (a) 483, (b) 561, (c) 655, and (d) 865 nm. The plots show the relationship between the M6SV and VR-SWIR-F models for a L8/OLI image taken over Taihu Lake, China, on October 26, 2014.