X
PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 20 Apr 2010 01:06 AM by  anon
negative value of FLAASH output for Landsat
 1 Replies
Sort:
You are not authorized to post a reply.
Author Messages

anon



New Member


Posts:
New Member


--
20 Apr 2010 01:06 AM
    Hi all, I ran atmospheric correction for Landsat ETM and the result has negative value. The Landsat statistics: Basic Stats    Min    Max       Mean      Stdev      Band 1      0    255  59.255267  48.261881      Band 2      0    255  49.079807  42.286283      Band 3      0    255  45.542444  43.446834      Band 4      0    255  39.367304  35.053712      Band 5      0    255  37.212798  37.839530      Band 6      0    255  22.609134  25.111304 The steps I did following below: 1. Run Landsat calibration to convert Landsat to radiance image . The radiance image statistic: Basic Stats       Min         Max       Mean      Stdev      Band 1  0.000000  190.060699  41.346365  34.799378      Band 2  0.000000  194.921005  34.351350  31.114553      Band 3  0.000000  151.671204  24.546114  25.106662      Band 4  0.000000  239.184052  34.209174  31.825399      Band 5  0.000000   30.810505   3.955160   4.426191      Band 6  0.000000   10.713229   0.740456   0.999171 2. Run FLAASH: scale factor: 10 The result has negative value as the statistic: Basic Stats    Min    Max         Mean        Stdev      Band 1  -3589   7016   130.795096   800.978509      Band 2  -2078   7539   511.932813   848.115801      Band 3  -1174   6335   552.350694   834.562358      Band 4   -987  13984  1625.021885  1644.365808      Band 5   -115   6942   828.752375   967.200516      Band 6    -62   7586   497.561363   696.026196 Please let me know the results are correct? Thanks Thy  

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    20 Apr 2010 07:59 AM
    It is not unusual to have some negative values in some pixels of every band.  It just means that FLAASH was unable to create an appropriate model for those pixels.  They are often areas of very low (or sometimes very high) radiance.  You can try and reduce this effect by increasing the visibility amount to 80-100km if the scene is very clear.  While this may increase your scaled reflectance to 0 or some small amount, it is still likely to be a very insignificant reflectance value when considering the output is scaled by 10,000.  For exaple, if it now returns a value of 10, the percent reflectance is 10/10000 = 0.01 % reflectance.
    You are not authorized to post a reply.