Cloud flagging adjustments

[alaroy]

Hi again SeaDAS'ers :)

Is there a way (or ways) to adjust the cloud flag determination in l2gen to be "more aggressive" in flagging cloudy pixels? I'm finding that we're getting contaminated data (for the purpose of our algorithm) at / near the edge of cloudy areas.

In case it makes a difference in the answer we're operating on MODISA granules.

-Andrew L.

[OB.DAACx - SeanBailey]

For MODIS, the STRAYLIGHT flag is effectively a dilation of the CLDICE (and HILT) flags.
If you apply the STRAYLIGHT flag as a mask, you may not need to change the cloud threshold. 

If you really want to change the CLDICE flag, you can provide l2gen with a stricter cloud_thresh value.
The default is 0.027 (effectively a 2.7% reflectance at 869nm - the wavelength to use is determined by the cloud_wave parameter).
When doing vicarious calibration work, I usually set it to 0.022.

Sean

[alaroy]

Sean,

Thank you, as always, for your helpful hints. I tried with the straylight flag and while it removed almost all of the cloud contamination (yaay!) it also dropped more pixels than we would like :(.

You said the flag is based on reflectance at 869nm - is that rrs_869, rhos_869, the cloud_albedo product or something else entirely? Since I'm processing (slowly!) at 500m resolution it would be nice to write the relevant value to the L2 file and then decide which pixels to keep during post-processing.

-Andrew L.

[OB.DAACx - SeanBailey]

If you want to assess in a post-l2gen step, output the cloud_albedo product.  This is the one on which the CLDICE threshold is based.

Sean