STRAYLIGHT flag on coastal areas

dem1 · by **dem1** » Mon Sep 02, 2019 9:07 am America/New_York

Hi,

We found that the STRAYLIGHT flag logic on coastal areas is not the same for all sensors:
- SeaWiFS: there is always a band of STRAYLIGHT pixels along the coastline (1 GAC or 3 MLAC pixels), and LAND pixels are always fully flagged STRAYLIGHT
- VIIRSN/J1: the STRAYLIGHT on coastal areas follow the same logic than on other areas: it's a margin around cloud pixels
- MODISA/T: there is always a band of STRAYLIGHT pixels along the coastline but only on the water side (and on both sides if there are cloudy pixels)

Is this normal?

Thanks in advance,
Julien

by **OB.DAAC - SeanBailey** » Tue Sep 03, 2019 12:30 pm America/New_York

Julien,

Yes, it is normal. As I mentioned earlier, for SeaWiFS there is a specific function to correct for straylight - that function sets the STRAYLIGHT flag.
For MODIS and VIIRS (most sensors in fact) the STRAYLIGHT flag is effectively a dilation of the CLDICE (and/or HILT) flags.
The band used in cloud detection for MODIS (869nm) saturates over bright targets. Land is typically bright enough. The coastline, therefore, typically is
of high enough signal to set either the CLDICE and/or the HILT flag, and thus the STRAYLIGHT flag.

Keep in mind, this flag is set just as a warning that the data may be contaminated. It should not be used in isolation from the LAND, CLDICE and HILT flags.

Sean

dem1 · by **dem1** » Wed Sep 04, 2019 10:28 am America/New_York

Thanks Sean.

Now I understand the differences we observed, they come from the HILT flag: it is always raised on lands for MODIST/A and SeaWiFS, but for VIIRSN/J1 this flag seems not used because I did not found any HILT pixels on products, am I right?

Thanks,
Julien

by **OB.DAAC - SeanBailey** » Wed Sep 04, 2019 10:41 am America/New_York

For MODIS the HILT flag is set when one of the bands saturates - the ocean bands tend to do so over land.
For SeaWiFS the HILT flag is set when data are "above the knee"...SeaWiFS has a bi-linear gain to prevent saturation. However, this creates a "knee" in radiance space for the DN-to-radiance conversion.
VIIRS doesn't saturate like MODIS and while some bands have dual gains, it's not the same as SeaWiFS. Yes, HILT is less likely (if ever) to be set in VIIRS.

Sean