Basic config l2w_mask

[andressa.muraro]

Hello Quinten,

I came across a doubt about these two parameters:
l2w_mask=True
l2w_mask_water_parameters=True
I tried to use both with "true and false" and in both processing all the products returned the same, without any difference.
What's the difference in actually using l2w_mask true or false? And for the l2w_mask_water_parameters?
I'm extracting as parameters l2w = rhos _ *, spm_nechad2016, t_dogliotti, fai, kd490_qaasw

Example using in the first picture the l2w_mask = True and the second picture l2w_mask = False. The same, without differences...

2.jpg (31.05 KiB) Viewed 107155 times

1.jpg (31.05 KiB) Viewed 107155 times

Best Regards,

[quinten]

Hi Andressa

The effect of these keywords is rather subtle: l2w_mask determines whether a mask is computed. If it is set to True you will have a l2_flags dataset in your output file. l2w_mask_water_parameters determines whether this l2_flags dataset is used to hard mask the output data, i.e. the data values will be replaced by NaN for pixels where the mask != 0.

So there are 3 options:
1) If l2w_mask = False, nothing happens. Changing l2w_mask_water_parameters has no effect. Parameters will be provided for all pixels.
2) If l2w_mask = True and l2w_mask_water_parameters = True, a flags dataset will be computed and the output data will be removed (set to NaN) for flagged pixels.
3) If l2w_mask = True and l2w_mask_water_parameters = False, a flags dataset will be computed but no output data will be removed.

I would recommend setting both to True and only changing the masking procedure if you understand why the flags were triggered and are using/developing an alternative masking procedure. If you disable the l2w_mask or l2w_mask_water_parameters, the retrieval algorithms will be applied to non-water pixels as well, which will lead to nonsensical results ("land turbidity" and "cloud chlorophyll" ).

Hope this helps!

Quinten

[andressa.muraro]

Thanks for the answer Quinten!

I just had one more question: If I set both to "true" will the islands or aquatic plants on the surface of the water be mask? I say this for the "rhos" product.
l2w_mask_water_parameters and l2w_parameters = True affect the rhos?

This is a part of config txt that I use:
l2w_parameters=rhos_*,spm_nechad2016,t_dogliotti,fai,kd490_qaasw
l2w_mask=True
l2w_mask_wave=1600
l2w_mask_threshold=0.0215
l2w_mask_water_parameters=True
l2w_mask_negative_rhow=True
l2w_mask_cirrus=True
l2w_mask_cirrus_threshold=0.005

[quinten]

Hi Andressa

rhos_* should not be affected by l2w_mask or l2w_mask_water_parameters. Let me know if you find otherwise!

Quinten

[andressa.muraro]

Nice Quinten, I'll try.

About the file ".nc". Is it possible to not create this product in processing?

[quinten]

No, the .nc NetCDF files are in fact the main output. The PNG and GeoTIFF files are created from the NetCDF files.

[stcarp]

Quinten,

I am currently getting an odd l2_flag output in the L2W file using the settings attached on Linux using the binary version of the code.

mask file

settings file

This has been the case on all of the tiles I have looked at when processing many.

Do you have any idea why this would be the case?

[quinten]

We had a database crash of the forum on 27 November and further answers to this thread were lost. For completeness, I believe the issue you had was related to the auto scaling of the datasets in SNAP to a NetCDF fill value (some really big number). I hope to improve masking of the NetCDF data in the future.

Quinten

[guoqiangqiu]

Hi Quinten，
I am processing S2 data, and the readings of l2_flags are 0,1,2, 3. What do these mumbers mean?
Thanks!
Guoqinag

[quinten]

Hi Guoqiang

The flags are represented in binary, so each bit can represent a different flag. By default the l2_flags are computed according to these exponents:

Code: Select all

flag_exponent_swir=0
flag_exponent_cirrus=1
flag_exponent_toa=2
flag_exponent_negative=3
flag_exponent_outofscene=4

A 0 occurs when no flags are set, a 1 means the SWIR threshold is reached (2**0), a 2 means that the cirrus threshold is reached (2**1), and a three occurs when both SWIR and cirrus thresholds are reached (2**0+2**1).

Quinten