[SeaBASS - MODISA] SST retrieval at night

[madjidh]

Hello,

I am using SeaBASS dataset for SST, MODIS Aqua, R2018.
I have 3 questions:
1) How are the BT in the MWIR/LWIR corrected? Can you point me to the appropriate atmospheric correction document for these specific bands please?
2) When using night data (flagged as night), I have an issue at isolating clear sky matchups. The CLDICE flag returns 50 000 cloudy pixels over the 12 million matchups available at night, while it returns 5 million cloudy pixels for daylight matchups. Is there an issue at identifying clouds during the night, does the method differ from daylight? Maybe I missed something.
3) What algorithm is used? Because I use the R2018 reprocess of MODIS Aqua, I downloaded the data in 2021. I am uncertain if the SST4 and SST algorithms come from Kilpatrick et al. (2015) or from the latest formula described in Luo et al. (2019).

Thanks in advance

[OB General Science - guoqingw]

Hi,

The SST product you downloaded should be the most recent R2019 version.
Here is the document about it:
https://oceancolor.gsfc.nasa.gov/reprocessing/r2019/sst/

Details about the algorithms used in SST:
https://oceancolor.gsfc.nasa.gov/atbd/sst/
https://oceancolor.gsfc.nasa.gov/atbd/sst4/
https://oceancolor.gsfc.nasa.gov/atbd/sst_triple/

Please check these documents and see if you can find the answers for your questions. Let us know if you need more help.
Guoqing

[madjidh]

Thank you for the answer, I now understand 1) and 3).
However for my 2nd question, I am still a bit lost.
Any thoughts about why I fail at identifying clear sky pixels at night?

Cheers

[madjidh]

Hello, my question is still up, if someone can please get in touch with me to unlock the situation? I can provide more details if required.

[OB.DAACx - SeanBailey]

The CLDICE flag is an ocean color flag, not a SST flag. Cloud masking is entirely different for SST (and differs a bit between day and night). For SST, unless you have a reason to do otherwise, you should use the quality flags (qual_sst) for selecting good quality SST retrievals. There is an flags_sst array with bits set for various conditions. These are combined into the qual_sst flag, for which 0=Best; 1=Good; 2=Questionable; 3=Bad

For more details, see:

https://oceancolor.gsfc.nasa.gov/atbd/sst/flag/

Sean

[madjidh]

I understand now. I was using l2flags because I included visible bands along with IR/MWIR/LWIR to try to improve retrievals over noisy matchups. But I guess I will switch to qual_sst.

Thanks for the details!