Welcome to the Earthdata Forum! Here, the scientific user community and subject matter experts from NASA Distributed Active Archive Centers (DAACs), and other contributors, discuss research needs, data, and data applications.
by avmehta » Sat Feb 06, 2021 1:48 pm America/New_York
Hello, I am trying to use gpt MosaicGeneral to merge MODIS and OLCI images for each day. I have the following questions: 1) It looks like this procedure will also reproject images. Or should I use gpt to reproject seperatly? 2) Can I use this procedure to combine images for a week? If yes, how are the overlapping pixels treated? Are the averaged? Thank you for your help. Amita
by gnwiii » Sun Feb 07, 2021 7:52 am America/New_York
See Mosaic Algorithm. Because level-2 pixels have a larger ground footprint towards the edge of the pass, edge pixels (which are generally lower quality due to long atmospheric paths) are over-represented in mapped images (binning gives each level-2 pixel the same weight). You should first compare values from your MODIS and OLCI images where they overlap as there are can be systematic differences between sensors for a given time and geographic region. You may want to rescale one of the products if these differences are large.
by avmehta » Sun Feb 07, 2021 9:30 am America/New_York
Thanks for reply George. Initially I will not be merging images from MODIS and OLCI - but merge them separately. I just looked at the link you sent -- great it explained what I was asking for! Thanks a lot for your help. Amita
by avmehta » Mon Feb 08, 2021 9:51 am America/New_York
Hello, I have been able to mosaic several images using gpt. I already have reprojected files so I use them to mosaic multiple images. I have the following questions I need help with. 1) Is it OK to use reprojected files and l2flags (for criteria to mosaic) for mosaic? 2) I see reproject, mosaic, and L3-binning options - how are they different? Does mosaic take care of reprojection and L3-binning? 3) In my MosaicEx01.par I am using geographical projection. Also, I use qualityExpression=!l2_flags.LAND OR !l2_flags.PRODFAIL (I thought this would just make land pixels invalid -- but that does not seem to be the case!). I'd appreciate your help. Thanks. Amita
by gnwiii » Mon Feb 08, 2021 12:33 pm America/New_York
Both Binning and Mosaic use a valid pixel expression. The `PRODFAIL` flag is set if any product fails, so if you are interested in a few products you will want to use flags specific to those products. In general, flags must be applied to level-2 data because it doesn't make sense to average flag values from multiple level-2 pixels. Mosaic does mapping then averaging so can overweight pass edge pixels, GPT binning avoids that problem, but may produce Moire patterns if too high resolution is requested.
by avmehta » Mon Feb 08, 2021 1:03 pm America/New_York
Thanks George. I am interested in just Rhos for all bands. If I did not specify PRODFAIL, according to you message it would it still exclude invalid pixels - correct? In my averaged image Rhos which are not valid show up as '0'. Is there a way to set them as NaN or other undefined value? Thanks agian. Amita
by avmehta » Tue Feb 09, 2021 8:43 am America/New_York
Hello, I am trying one flag at a time, in addition to LAND, to see the impact. I tried qualityExpression=!l2_flags.LAND and !l2_flags.CLDICE -- but I still see data when clouds are present. Should it be 'or' so that either land or cloudy pixel will be excluded? Sorry about all these questions and I appreciate your patience. Thanks. Amita
by gnwiii » Tue Feb 09, 2021 10:22 am America/New_York
It may be helpful to start by looking at the impact of the flags on one product. You can test valid pixel expressions on individual bands in the GUI using band properties. For some products, the valid range will have already masked out pixels that are flagged as land or cloud.