Hello

I'm doing some test using NIR/SWIR aerosol model option versus standard and MUMM options. We would like retrieving remote sensing reflectance on turbid water due to inorganic matter

The implementation on Seadas 6.3 of NIR/SWIR correspond to Wang et al 2009 (forum post), at least at switch parameters (tindx >1.05,  nwL (869) >=  0.08 and chl > 1). As  showed in Werder et al. 2010 (SWIR/NIR evaluation), I loss a significant amount of data. As suggested by these authors I have tried to modify this thresholds.

To accomplish that, I modified the thresholds in atmocor2.c (lines 307 tindx, line  500 tindx and, line 506 chl and  nLw[nir_l]. Then I rebuild l2gen binary and replace the original found in /seadas/run/bin/linux_64. Then, I  reprocess the granule but no changes occur. I am not sure if I'm doing is correct and/or enough.

MUMM option seems to produce interesting results , although we can not verify this due to a lack of in situ data .

Thank's in advance
Josep

Josep,

I would suggest that you comment-out that entire block of code around line 500 (where it checks chl and nLw and reverts back to NIR if needed).  That would ensure that the switching is only done based on the tindx.  You can then add tindx_shi as an output product when you run the code, and you can use it to determine what pixels are processed with SWIR and what pixels are processed with NIR.  In my experience, you WILL lose a lot of data when processing with SWIR bands, because of the very poor signal to noise (high digitization noise).  

-- bryan

Bryan

Thank you very much for your help. I will do some tests using your suggestions. According to our tests and to your work, NIR/SWIR will not probably be an option for us. However MUMM could be an alternative to standard NIR.

Josep