Page 1 of 1

how to remove bowtie for VIIRS imagery in SeaDAS

Posted: Fri Jul 16, 2021 2:46 am America/New_York
by zhigang
Greetings,

I am processing some VIIRS data in SeaDAS with V2021.1 processor. The input data is L1A and GEO data downloaded from the ocean color portal. However, the result processed by l2gen missed some data, which should related to the bowtie effect. I have tried to map it using reproject and mosaic tool following the instructions in https://oceancolor.gsfc.nasa.gov/forum/ ... l?tid=5906, but the data still missed.

My examined data is V2020133054200.L1A_SNPP.nc, and command is like:

Code: Select all

l2gen ifile=V2020133054200.L1A_SNPP.nc geofile=V2020133054200.GEO-M_SNPP.nc ofile=V20201330542_NPP_l2.nc l2prod='rhos_vvv Rrs_vvv' south=28 north=36 west=110 east=122
Could someone provide some suggestions?



Best Regards,
Zhigang

Re: how to remove bowtie for VIIRS imagery in SeaDAS

Posted: Sat Jul 17, 2021 7:36 pm America/New_York
by gnwiii
Can you explain what is meant by "remove bowtie"? Do you want to replace the pixels deleted in the scanline overlap areas at the edges of the scans for "improved" appearance? Do you have access to OCSSW Processing Software. e.g., linux (including WSL) or macOS?

Re: how to remove bowtie for VIIRS imagery in SeaDAS

Posted: Mon Jul 19, 2021 9:24 pm America/New_York
by zhigang
Yes, I would like to improve the appearance thourough filling those deleted pixels at the edges of scans. I processed the data using the ocssw in Ubuntu 18.04.

I found that the data generated by l2gen without subset options could get beautiful image after reprojection. But the results with subset option (south=...,north=...,west=...,east=...) did not work.

Re: how to remove bowtie for VIIRS imagery in SeaDAS

Posted: Tue Jul 20, 2021 7:47 am America/New_York
by gnwiii
Supporting subset would complicate the area-weighting and interpolation at edges of the region, so maybe it should not be a big surprise that you get "unfilled" deletions in mapped images. Given that
mapping a single granual doesn't take long with modern hardware, it makes sense to produce mapped netcdf files of granuals and then apply a subset operation to the mapped image.

Are you using SeaDAS 8.1.0 and OCSSW tag=2021.2 software? There are several ways to make mapped images from a single level-2 granule: export the screen image in the GUI, l2mapgen, l2bin+l3mapgen+"some netcdf to image process". l2mapgen has changed from a python script that used l2bin+l3mapgen to a stand-alone program.

The SeaDAS "benchmark" script (which should be called "check that a fresh install is working properly" and is a download+install option in the OCSSW section of the GUI) produces an SMI image using l2bin+l3mapgen from a single MODIS Aqua granual at 1km resolution:

l2bin ifile=[...] ofile=[...] resolution=1 area_weighting=1
and
"l3mapgen ifile=[...] ofile=[...] resolution=1km interp=area"

For older versions the SMI image has a Moire pattern of missing pixels near the pass edge, but the current version has filled in the missing pixels using the new area_weighting and interp=area options. The
previous benchmark script did use the subset option.

Re: how to remove bowtie for VIIRS imagery in SeaDAS

Posted: Wed Jul 21, 2021 1:05 am America/New_York
by zhigang
Thanks George.

Re: how to remove bowtie for VIIRS imagery in SeaDAS

Posted: Wed Jul 21, 2021 9:15 am America/New_York
by gnwiii
You are very welcome. Hope you find a workflow that fits your use case. I should mention that the deletion artifacts (and Moire effects) are less of a problem at lower resolution. Also, ESA has their own binning tool in the Graph Processing Facility (GPF). It is mentioned in the SeaDAS 8 GUI Help GPF Operator Index, but not all ESA SNAP operators are available in SeaDAS 8.1.0 so you may have to use ESA SNAP.

Re: how to remove bowtie for VIIRS imagery in SeaDAS

Posted: Wed Jul 21, 2021 9:40 am America/New_York
by zhigang
Fortunately, it is easy to use the GPF when SNAP was installed in the same system. I will try it later :).