Acolite results in Open oceanic waters

[Asim12]

Hi,

I have applied Acolite on open water and attached are the Rrs plots. It looks like there is a high reflection in the NIR band which is confusing. Please have a look at my Rrs plots and settings file. Any suggestions would be appreciated. I have tried the Python version (May 2021).

aerosol_correction=dark_spectrum
dsf_path_reflectance=tiled
## output resolution (S2 only 10, 20, or 60 m)
s2_target_res=60
ancillary_data=True
glint_correction=False
merge_tiles=False



Regards
Asim

[quinten]

Hi Asim

As ACOLITE was intended for turbid coastal and inland water processing, I am not surprised you have some rather poor results for open ocean waters. Perhaps it is worth looking at some other processors (l2gen or POLYMER?) that are more suited for offshore waters. There is some further discussion here.

This vertical shift looks like glint contamination. Perhaps you can enable the glint correction (dsf_residual_glint_correction=True) to get better results?

It is also possible to change the reference band for the glint correction to the 865 nm band

Code: Select all

dsf_residual_glint_correction=True
glint_force_band=865
glint_mask_rhos_wave=865
glint_mask_rhos_threshold=0.05

Or apply the alternative glint correction:

Code: Select all

dsf_residual_glint_correction=True
dsf_residual_glint_correction_method=alternative
dsf_residual_glint_wave_range=860,870
glint_mask_rhos_threshold=0.05

Let me know how it goes!

Quinten

[Asim12]

Hi,

Results are not poor. Better than C2RCC and Polymer. I have used a threshold RRs at 865 < 0.0009. Polymer results in the NIR band are weird.

Regards
Asim

[quinten]

Hi Asim

It may help if you could share which scenes these spectra come from. Can you separate the spectra with high NIR by season or relative orbit, i.e. indicating this is a glint problem?

Quinten

[Asim12]

Hi Quinteen,

These are some of the scenes where Rrs at 865 is >= 0.002. I have also used glint correction as you had suggested earlier.
S2A_MSIL1C_20180521T105031_N0206_R051_T34WDD_20180521T111613.SAFE (Lat/Long =71.501/19.802 )
S2B_MSIL1C_20180502T110619_N0206_R137_T32WNS_20180502T113342.SAFE (Lat/Long = 64.521/ 9.303)
S2A_MSIL1C_20180918T122751_N0206_R052_T33XVE_20180918T143928.SAFE (Lat/Long = 76.386/12.654 )

After using a threshold Rrs at 865 < 0.0009, I got the attached Rrs plots. The Rrs always peak at blue bands, whereas, with C2RCC sometimes Rrs peaks at the green bands as well.

Regards
Asim

[Asim12]

I was thinking can I change rhos to rhow for glint correction?

dsf_residual_glint_correction=True
dsf_residual_glint_correction_method=alternative
dsf_residual_glint_wave_range=860,870
glint_mask_rhos_threshold=0.05

to

dsf_residual_glint_correction=True
dsf_residual_glint_correction_method=alternative
dsf_residual_glint_wave_range=860,870
glint_mask_rhow_threshold=0.05

[Asim12]

Hi,

I am wondering why the Rrs always peak at blue bands, whereas, with C2RCC sometimes Rrs peaks at the green bands as well.

Regards
Asim

[quinten]

Hi Asim

This is a very generic question and may depend on your study area and scenes...

The peak in the blue may be because your target waters have very low concentration of scattering/absorbing particles or substances. There can also be a residual signal from the atmosphere (which is stronger towards the blue) in the surface level reflectance after the atmospheric correction for a variety of reasons. (Underestimation of aerosol concentration, top-of-atmosphere calibration,...)

The fact that ACOLITE/DSF and C2RCC give different answers is probably mainly a result of their different design. C2RCC uses a neural network to fit modeled atmosphere and water spectra, while ACOLITE estimates and subtracts only a modeled atmospheric signal. C2RCC provided the best fitting result for a given spectrum and geometry according to its neural network training dataset.

Quinten