X
PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 15 Aug 2011 09:08 AM by  anon
Problem with NDVI after FLAASH
 4 Replies
Sort:
You are not authorized to post a reply.
Author Messages

anon



New Member


Posts:
New Member


--
15 Aug 2011 09:08 AM
    Hello, Excuse me about my precarial English. I'm a new ENVI user from Ecuador and I have some questions about an ENVI process that I'm developing. I have 3 ASTER 1B images from different ages of the Ecuadorian Amazon Rainforest and I need to analyze them for a multitemporal study. In first place I transformed them into a BIL image and then, I ran the FLAASH algorithm from ENVI 4.7, In the requirement of "Radiance Scale Factors" I typed 10 as it is said in many forums because of the radiance units of my files. After this process, In "model athmosphere", I selected Tropical for obvious reasons and that's all. After this I ran the "band math" process with the equation "(float(b3)-float(b2))/(float(b3)+float(b2))" for a NDVI, but the result of this process is an image with values between -0.04 (Instead of -1) and 1 aproximately. So the image tends to a brightly white. I suspect I'm making a mistake in the radiance scale factors because I've read that this images are recognized as reflectance automatically but I'm not sure about what the problem is. I beg for your help. Greetings

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    15 Aug 2011 09:56 AM
    Your process sounds correct. You should be aware that the output from FLAASH is scaled reflectance with a multiplier of 10,000. Perhaps you may have better results if you divided the data from FLAASH by 10.000 to get to percent reflectance which should range between 0-1.0. However, there are cases where some pixels do not model well and you can get values less than 0 or more than 1. These pixels should be masked or ignored. Your NDVI values sound reasonable. NDVI does not *scale* the data between -1 and 1. The results should simply fall within this range. You will typically not get very low values near -1 unless you have ice and snow in the image. You will also not get values of 1 very often and if you see many of these, I would think that they are suspect.

    Deleted User



    New Member


    Posts:
    New Member


    --
    15 Aug 2011 10:19 AM
    Thank you for your answer MariM, I will check the results after applying your tips. Also, I would like to know if its possible that ENVI does not automatically read the header of mi file. May that causes problems in the radiance scale? or also I should apply values diferent from 10 in any case? Greetings

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    15 Aug 2011 10:22 AM
    The header of which file? If you are reading in your ASTER 1B data natively (as EOS ->ASTER), then ENVI is opening the file and reading it in as floating point radiance data. You should only need to convert the interleave and use a scale factor of 10 in FLAASH.

    Deleted User



    New Member


    Posts:
    New Member


    --
    15 Aug 2011 10:39 AM
    [QUOTE]MariM wrote The header of which file? If you are reading in your ASTER 1B data natively (as EOS ->ASTER), then ENVI is opening the file and reading it in as floating point radiance data. You should only need to convert the interleave and use a scale factor of 10 in FLAASH.[/QUOTE] I'm in fact doing that, so I think I should have no problem... Thank you very much Greetings
    You are not authorized to post a reply.