I am trying to model the extraction of impervious surfaces in landsat images using tasseled cap transformation (TCT). i chose two year periods and performed radiometric calibration and top of the atmosphere at reflectance on them before doing the tasseled cap transformation. a regression analysis performed on them showed consistency with the landsat7 images both for radiometricaly calibrated and TOA processed images before TCT and images which where directly tasseled cap transformed, although the later had higher adjusted r sqd values. The landsat8 images however performed poorly in both instances. has anybody had a similar experience with Landsat 8? and how do i improve performance? I used PCI geomatica in the TCT. Used both Erdas imagine 2013 and Arcmap 10.1 in the radiometric calibration and top of the atmosphere at reflectance processing.
Dear Mr. Asiedu,
Tasseled Cap Transformation (TCT) uses the green band, orthogonal to green band and Brightness orthogonal to Wetness.
This means that TCT is richer in information than Vegetation indices. However, I would try to compute NDVI and SAVI and check if there are still problem in this case I recommend to redo your atmospheric correction for Landsat 8.
Dear Awad, yes the analysis was based on the greenness component. i will redo the atmospheric correction and see what come up.
You also have the option of obtaining USGS atmospherically corrected Landsat 8 data
I had a similar problem years ago and I purchased atmospherically corrected data from USGS. Try that.
Dear Asiedu,
On the earth explorer web platform you also download surface reflectance data that have undergone some preprocessing.
I appreciate all your contributions. After checking the acquisition date for the images i decided to use same date of image acquisition for both time periods and that changed everything, now there is consistency. I also observed that images which had been radiomatrically calibrated and top of the atmosphere at reflectance processed before tasseled cap transformed had very low correlation compared with those which were tasseled cap transformed without those processes. I may have to try the suggestion to use USGS atmospheric corrected images to see if it makes a difference.
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing