X
PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 29 Jan 2020 09:38 AM by  Lari Pilot
Regarding FLAASH correction
 21 Replies
Sort:
You are not authorized to post a reply.
Page 1 of 212 > >>
Author Messages

Bisma Qazi



New Member


Posts:16
New Member


--
14 Mar 2017 06:41 AM
    Hi, I am trying to do atmospheric correction on AWiFS data onboard IRS-P6 using FLAASH module of ENVI 5.1. I have some doubts in the parameters that need to be inputted into the module. In the multispectral settings tab, there is an option for water reference channel, what does it mean? Also an option for index to first band is there, If the value is set from 1 to 4, then it displays the following error message. 'filterfunctionfile'.sli doenot contain enough bands for the given input image. please check sensor type. The sensor type was, however, set to unknown. And if index is set to 0, then it says that 'filterfunctionfile'.sli doenot contain any non-zero responses for band : 0 However, at this step some processing starts and then it stops after a while with the following error message: ACC error: modrd5: Non finite numbers in coefficiebt array coeff. IDL error: OPENR: Error opening file, Unit: 100, File , Filterfunctionfile.sli.hdr The system cannot find the file specified ACC_MODRD5 ACC_HYPER2 ACC_ACC ENVI_ACC_EVENT ENVI_FLAASH_PROCESS_MANAGER IDLRTMAIN $MAIN$ What is exactly the problem, Am I using wrong parameters or else inputting files in wrong format or what? Please help!

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    15 Mar 2017 07:55 AM
    What is the path to the filter function you are using? Is it empty? There is no filter function for AWiFS data in ENVI so you would need to supply your own filter function for this. Alternately, you could use Quac.

    Bisma Qazi



    New Member


    Posts:16
    New Member


    --
    15 Mar 2017 11:09 PM
    Of course, I am inputting the filter function file that has been created using spectral library builder tool available in the software.

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    16 Mar 2017 07:52 AM
    Does the input image and filter function use the same units (microns or nm)?

    Bisma Qazi



    New Member


    Posts:16
    New Member


    --
    17 Mar 2017 12:18 AM
    How do I check the wavelength units of the input image?

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    17 Mar 2017 07:44 AM
    Right click on the image in the Layer Manager and select View Metadata and then view the spectral information. It should show the units as microns or nanometers. Does this also match the filter function units?

    Bisma Qazi



    New Member


    Posts:16
    New Member


    --
    18 Mar 2017 02:25 PM
    Oh, the meta data of input image does not contain this information. What now?

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    20 Mar 2017 06:54 AM
    You will need to add this information using Edit Header (for ENVI 5.1) or Edit Metadata for ENVI 5.3 and higher. You can also do this using ENVI Classic, File->Edit ENVI Header.

    Bisma Qazi



    New Member


    Posts:16
    New Member


    --
    23 Mar 2017 04:15 AM
    This did not help!

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    23 Mar 2017 07:05 AM
    Can you post the template.txt that is created in the output directory and what is the error message shown at the bottom of the journal.txt?

    Bisma Qazi



    New Member


    Posts:16
    New Member


    --
    26 Mar 2017 11:26 PM
    I run the model repeatedly and finally it worked. I don't know what the issue was? However the ouptut values are not appropriate. I wanted to know what scale factor should I use so that I can get percent reflectance i.e. from 0-100%?

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    27 Mar 2017 06:49 AM
    FLAASH and Quac both output to scale integers. The scale factor is 10000. You can use band math to convert the data to reflectance percent using: float(b1) / 10000 where you map the b1 variable to the file.

    Bisma Qazi



    New Member


    Posts:16
    New Member


    --
    27 Mar 2017 10:33 PM
    Input radiance image should be in the units of µW/cm2/sr/µm to apply FLAASH correction. However, I am unable to generate the radiance in ENVI as it says calibration requires gain and offset settings, even after applying the same. Why is this so? Well I generated the radiance image using ERDAS model buider in the units of mW/cm2/sr/µm. To convert this into µW/cm2/sr/µm so as to apply FLAASH correction, what scale factor should I use for input image? 1000 or 0.001. Well I tried with both but I am not getting the appropriate values.

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    28 Mar 2017 06:09 AM
    If the data are in the expected units of uW(cm^2*sr*nm), then the scale factor is 1.0. You can check by looking at the statistics of the data. The bands should range from about 2-35 or so in floating point values.

    Bisma Qazi



    New Member


    Posts:16
    New Member


    --
    28 Mar 2017 08:02 AM
    No, the input is in mW/cm2/sr/µm units. I did convert it into W/m2/sr/µm and now I am using a scale factor of 10. Hope that is right? What do the water absorption and reference channel refer to that needs to be set if opted for water retrieval? What is the basis for selection of the water and aerosol retrieval?

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    28 Mar 2017 08:05 AM
    The parameters are discussed in the help here: https://www.harrisgeospatial.com/docs...

    Bisma Qazi



    New Member


    Posts:16
    New Member


    --
    29 Mar 2017 12:33 AM
    https://www.harrisgeospatial.com/docs/FLAASH.html This link says that for Landsat TM, for example, KT upper channel can be set to band 6 and lower channel to band 3. If we have a look at Landsat TM bandwidths, B1 spans between 450-520 nm, B2 (520-600 nm), B3 (630-690 nm), B4 (760-900 nm), B5(1550-1750nm), B6 (10400-12500 nm) and B7 (2080-2350). Spectral bandwidth of B3(630-690 nm) matches with the recommended one in the above link i.e. (640-680nm) but that of TM band 6 doesnot match with the recommended one for aerosol retrieval. However, the range falls with in B7. so what is the basis of selecting band 6 here as KT upper channel?

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    29 Mar 2017 05:58 AM
    Landsat band 6 would not be stacked with the VNIR-SWIR bands because it has a different pixel size and it is outside of the spectral range that works in FLAASH. So here 'band 6' is the 7th band of Landsat.

    Bisma Qazi



    New Member


    Posts:16
    New Member


    --
    29 Mar 2017 06:18 AM
    True that, I took it for thermal band. Ok, Thanks!

    Bisma Qazi



    New Member


    Posts:16
    New Member


    --
    17 Apr 2017 01:33 AM
    Hi,
    Here I am again with some query on FLAASH correction. I have applied the FLAASH correction on IRS-P6/AWiFS data covering glaciated region.
    I was able to run this module successfully when I set the zenith and azimuth angles to default values i.e. 180 and 0 respectively.
    However, as I changed these to 37 deg and 158 deg as mentioned in the metadata file, it shows zenith angle> calculated horizon angle.
    Why so? Also in the ENVI help, it is mentioned zenith should be between 90 and 180 degrees. What is the reason behind this, when the metadata file shows different values for these? Not only this the atmospherically corrected reflectance values calculated are higher than the TOA reflectance values, which however, should not be the case, I guess.
    You are not authorized to post a reply.
    Page 1 of 212 > >>