X
PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 24 Jul 2015 09:29 AM by  anon
Abnormal NDVI value after Flaash
 1 Replies
Sort:
You are not authorized to post a reply.
Author Messages

anon



New Member


Posts:
New Member


--
24 Jul 2015 09:29 AM
    Hi, I am using ENVI 5.1, I want to conduct image pre-processing before calculating the NDVI. However, I encounter some anomalies as below: If calculate the NDVI before pre-processing, the values of the forest are around 0.4 - 0.5. After radiometric calibration, I choose reflectance, the values are around 0.5- 0.7, I think this result is more accurate as it matches the findings from other literature. I continue to use the radiometric calibrated image to conduct atmospheric correction using Flaash, Abnormal result occurred. The values of NDVI for all pixels are negative values. I have tried many different settings, such as calibration type, output interleave, or apply Flaash settings in the Radiometric Calibration; for the settings of Flaash, most of them are just default values, the KT Upper Channel is defined as SWIR2 (Landsat 8), Maximum Upper Channel Reflectance as 0.08, Reflectance Ratio as 0.5 and Index to first band: 0. Different parameters lead to different strange results and none of them are normal. I have also tried QUAC and other dataset such as images from Landsat 5, and the results are also strange. Settings of radiometric calibration and Flaash can be found as below: http://www.fotop.net/windyfung/windyfung Anyone have idea about this problem? I want to conduct study about vegetation recovery after wildfire, is it acceptable if I only use radiometric calibrated images without atmospheric correction?

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    29 Jul 2015 12:52 PM
    Can you post your template.txt that is produced in the output directory from FLAASH? I cannot access the files you link to below. Did you use 'Apply FLAASH Settings' during Radiometric Calibration? If so, then you will use an input scale factor of 1.0 for FLAASH. The input scale factor is often the cause of abnormal output reflectance. Also note that FLAASH scales the output by 10,000 to integers. So to get reflectance values, you would divide your FLAASH result by 10000 in band math. Some negative values can be expected in areas of very low radiance (water and shadow) because those areas do not model well but your image should not be mostly negative.
    You are not authorized to post a reply.