smigen VS l3mapgen?

[blesht]

I've been using smigen for years (and years and years).  I have a very large collection of mapped l3 files created with smigen that I use for various different analyses.  I understand that smigen has been (or is in the process of being) replaced by l3mapgen.  I just did a quick test in which I used both programs to map a binned chlorophyll file that was created using l2bin.  The two relevant lines in the script I used are below.

smigen ifile=${INFILE} prod=${2} ofile=${ofile1} resolution=${resolution} latnorth=${latnorth} latsouth=${latsouth} lonwest=${lonwest} loneast=${loneast} meas=${stat} stype=${stype} precision=${precision}

l3mapgen ifile=${INFILE} product=${2} ofile=${ofile2} resolution=${resolution} north=${latnorth} south=${latsouth} west=${lonwest} east=${loneast} oformat="hdf4"

I'm attaching a scatter plot of the chlorophyll values in ${ofile2} (the l3mapgen output) versus those in ${ofile1} (the smigen output).  Although many of the points fall on the 1:1 line (as one would hope), many do not.

Why???  Does this mean that if I intend to use l3mapgen going forward, I need to reprocess all my binned files that were created with smigen with l3mapgen if I want them consistent?

Thanks.

[OB.DAAC - SeanBailey]

> Why???

Floating point math and output vs input driven mapping.

If you want absolute consistency, then yes, you'd want to process all with l3mapgen.  Is it a problem if you don't?  I wouldn't say so. 

Sean

[blesht]

Thanks, Sean.  Appreciate your help as always.   Best, Barry