Hi,

After recently updating to SEADAS 5.0 I've noticed that the cloud masking on my nightime L2 SST products has been quite poor.  I had been using the SSTFAIL flag as a mask, using the following code:

flags = rd_sds(ifile, 'l2_flags')
mask = nm_2_msk('SSTFAIL')
mask = mask[0]
clouds = where((flags and mask) ne 0, countclouds)
IF (countclouds GT 0) THEN data(clouds) = -163.8
  
Has something changed with this flag.  I use CLDICE flag for my daytime passes, and it still seems to do a good job.

Thanks,
Ryan

Hi Ryan,

Please see the following for a description of flags and quality
levels in our SST data.

http://oceancolor.gsfc.nasa.gov/DOCS/modis_sst/

Regards,

Norman

Hi Norman.

Thanks for the information - I was already aware of that document.  I can't find anything in it that explains my problem however.  It seems like the SSTFAIL flag is not being set.  I've looked at several (obviously cloud contaminated) nighttime Aqua and Terra images using the SEADAS GUI, and there are no pixels in my study region that have the SSTFAIL flag set.  There are however flag pixels in SSTWARN (way more than I could use as a cloud mask though).

My problem started right after installing SeaDAS 5.0 ...

Ryan

Ryan,

The quality-level and SST flagging algorithm was changed to include a 5th
quality level corresponding to actual SST failure (failure to retrieve a
value at Level-2).  The SSTFAIL flags is set for that case.

Previously, the highest quality level (lowest quality) was set for the above
cases, as well as cases where SST was much different from Reynolds. 

So, SSTFAIL now really means SSTFAIL, but that means the the old logic
of assuming SSTFAIL kicked out the pixels with a large deviation from the
Reynolds OISST (e.g. clouds tops) will now allow cloud contamination through.

You need to use the qual_sst (or qual_sst4 if using sst4) SDS to identify the
appropriate quality level for your purposes.

Sean

Ryan,

The quality-level logic was changed.  There are now 5 levels (0-4).  The old level-3 was divided, such that QL=4 now indicates true SST failure (e.g., band saturation) such that sst could not be computed, and QL=3 now indicates very questionable retreivals (e.g., very different from Reynolds).  The SSTFAIL flag is set for worst quality, so it is now set for QL=4.  SSTWARN is set for QL=1.  Our SST processing does not use these flags; it looks at the quality levels directly.  You will need to do the same.  To get the old results, ignore pixels with qual_sst > 2.

-- bryan

OK - Thanks.

Where/how do I access qual_sst?  It doesn't seem to be a L2 product generated by msl12.  Is it a l2_flag that doesn't show up in the SeaDAS L2 flag viewer?

Ryan

Ryan,

You are the first to notice this.  You will receive a free seadas update.

Apparently the sst4, qual_sst, and qual_sst4 products never made it into the seadas GUI.  Mark will put that on the list.  In the interim, if you know how to run msl12 from the unix commandline, you can just add qual_sst to the l2prod list in the msl12 parameter file.

-- bryan

Hey Ryan,

I have a fix for this for you.  It's located on our ftp site under seadas/pub/ryan.  If you're using your own version of IDL, put the msl12_defval.pro file in your $SEADAS/idl_lib and $SEADAS/src/idl_procs/msl12 directories.  If you're using embedded IDL, put the seadas_idl_lib.sav file in your $SEADAS/idl_rt directory.  Try it and let me know if it doesn't work for you.  This only allows you to select the sst4, qual_sst, and qual_sst4 products with msl12.  Displaying quality levels for level 2 files will take more time.

Mark

OK - thanks,

What is the address of your FTP site?

Ryan

samoa.gsfc.nasa.gov

i just have to say that i really love the way this forun is working...ryan, thanks for bringing this to our attention and sean, bryan and mark, thanks for being so quick and helpful with the response.
gene

The new code seems to be working perfectly ... I need to reprocess some of the poorly masked images through msl12 to be certain, but my initial tests have worked as advirtised.  I've installed the patch on machines with IDL as well as a runtime SeaDAS - both are working.

Thanks again,
Ryan

Ryan,

I've put some code on our ftp site for you that should allow you to display level 2 sst by quality levels.  It's located under our update02 directory in the same location as the update01.  It's not an official update yet, so for right now, you have to download the files and install them manually in the right place.  The INSTRUCTIONS file tells you where to put everything.  Let me know if you have any problems or questions.

Mark