X
PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 15 Mar 2007 11:05 AM by  anon
Convert to reflectance and DOS
 1 Replies
Sort:
You are not authorized to post a reply.
Author Messages

anon



New Member


Posts:
New Member


--
15 Mar 2007 11:05 AM
    Hello, I'm looking for some advice on how to process several Landsat ETM scenes. I need to convert them to reflectance values and perform DOS in one step. The ENVI tools to do this are seperated and once you have converted an image to radiance and done DOS you cannot easily convert many bands to reflectance at once. Maybe I am missing something but I have not found an easy solution to process large numbers of scenes in an efficient way. Perhaps someone has a batch routine or IDL script they would be willing to share. My IDL scripting skills are non existent so far. Much appreciated, Martin Gamache UNBC

    Deleted User



    New Member


    Posts:
    New Member


    --
    15 Mar 2007 11:05 AM
    Hi Martin, By DOS, I assume you are referring to dark object subtraction? If so, then you would definitely have to build a batch program in order to process a large number of scenes. Doing so is not trivial, though--especially if you want to end up with bands 1-7 in the same file and you want the background value to be constant across all bands. The program would likely involve the following steps: - Perform DOS on all bands - Retrieve calibration parameters for each band (from the web using ENVI_L7_CPF, if necessary) - Call TMCAL_DOIT to calibrate each band individually - Build a mask for each band that highlights valid pixels and supresses background pixels - Apply the masks to each band, setting the masked values to something constant ('NaN' or maybe 0.0) - Call ENVI_LAYER_STACKING_DOIT to bring all bands back together into a single file - Call ENVI_FILE_MNG to remove and delete all intermediary files (single band calibration and masking results). I've built programs like this before, but it takes time to set them up. Depending on how many scenes you have to process, the manual route might be faster. I usually build in an additional step that multiplies the floating point TOA reflectance output by 255 and then converts the result back to byte prior to storage in a new layer stacked output file, via a custom tiled processing routine. Since the dynamic range of Landsat is 0-255, it makes sense to stuff the reflectance back into that range since the shape of the spectral curve for each pixel is more important than the absolute DN values associated with it. A side benefit is an output file that is one quarter the size of what it would otherwise. I don't have a sample program that I can post here, unfortunately.
    You are not authorized to post a reply.