I downloaded Landsat 8 Level-2 scenes from https://espa.cr.usgs.gov/ for land cover classification. Following are the product list in the package, cause I ticked Top of Atmosphere Reflectance (TOA), Surface Reflectance (SR) and Pixel QA to be processed.

LC08_L1TP_129026_20140712_20170421_01_T1 = *

*_b1 to b11
*_sr_band1 to band7
*_toa_band1 to band9

I am new to image calibration and correction things and I thought the image quality will be significantly increased after these processing. My understanding was that the calibration and corrections will be done on same raster band of same scene, For example all of the SR and TOA and all other angle corrections can be processed on same raster layer of Band 1 of LC08_L1TP_129026_20140712_20170421_01_T1 scene all together. But what I got separate files from the ESPA web, such as apart from original band data (b1-b11), I got surface reflectance (b1-b7), top of atmosphere reflectance (b1-b9) and several other azimuth and sensor related files.

Now I am confused, are they corrected/calibrated data of a certain band (which one is corrected and calibrated if they are) or are they data to be used for calibration/correction process of the original band data?

  • 1
  • 1
    I don't think they are same. In this question I wanted to know whether SR and TOA are corrected/calibrated data products or data to be used for correction/calibration process. In the another question I wanted to know what kind of correction and calibration is necessary for land cover classification analysis. It is annoying. When I ask them in same post, people were asking to post the questions separately and let the post single question focused, but when I post questions separately, you guys say they are duplicated. Please let it go as different, I did't get my answer yet for this one.
    – Vandka
    Commented Dec 8, 2017 at 3:45
  • 1
    Yes because there many preprocessing tools/steps for correcting/calibrating satellite data (Landsat 8) before doing land use/cover classification. Beside some Landsat data are already corrected/calibrated depending on the product level.
    – ahmadhanb
    Commented Dec 8, 2017 at 4:08
  • 4
    Agreed @ahmadhanb, Our GIS SE model works best with focused questions. However, I think this question is sufficiently different from the one you reference to remain open.
    – Aaron
    Commented Dec 8, 2017 at 4:19
  • 3
    @ahmadhanb I've reclassified the post you reference as a duplicate of a different question. I see this as a unique question such that editing the closed question to resemble this one would be inappropriate. Problem solved?
    – Aaron
    Commented Dec 8, 2017 at 4:37

1 Answer 1


You will want to use the *_sr_band1 to band7 data for your analyses as these are atmospherically corrected surface reflectance products (Figure 1). Only the Surface Reflectance data products and quality assessment bands are included in the deliverables (Figure 2). The ancillary bands (e.g. *_sensor_azimuth_band4) are used in the process of atmospheric correction and/or cloud detection.

Figure 1; Source: USGS enter image description here

Figure 2; Source: USGS, p.16 enter image description here

  • 1
    Okay, now I got it. Thanks. I looked at the guide from Landsat, but it seems that I overlooked it. I was expecting to use modified (corrected) version of the raw band data, instead of newly created data called SR.
    – Vandka
    Commented Dec 8, 2017 at 4:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.