0

Page 12 from this guide https://www.usgs.gov/media/files/landsat-8-9-collection-2-level-2-science-product-guide has the following row. It says that Band 10, can be multiplied by 0.00341802, and then 149 can be added in order to get the temperature in Kelvin. But my doubt is the following. The valid range is 293-61440, but 61440 Kelvin = 61166.85 Celcius which is not right (I think that, it is too high to exist on earth). Also, why the valid range can't be less than 293 Kelvin = 19.85 Celcius?

enter image description here

Improve this question

1 Answer 1

Reset to default
2

The valid range is for the stored data values before applying the scale and offset.

The valid range in degrees is 150 - 359° Kelvin (-123 - 86° Celsius).

293 * 0.00341802 + 149 = 150
61440 * 0.00341802 + 149 = 359
Improve this answer
6
  • Good to know! So, I must apply scale factor and offset on band 10 in order to get the surface temperature. Is it correct to not apply scale and offset on the other bands, in order to compute vegetation indices (Bands 1 to 7 etc.)? I understand that using scale and offset on the other bands, makes them floating point units from 0 to 1 and that's all. @user2856
    – Chris
    Commented Jan 2, 2023 at 20:31
  • So, scaled NDVI will be the correct one. But NDVI valid range will still be -1 to 1? or because we scaled our bands it will have a new valid range from 0 to 1? @user2856
    – Chris
    Commented Jan 2, 2023 at 22:03
  • And for future reference, please ask new questions as a separate question not in comments.
    – user2856
    Commented Jan 2, 2023 at 22:58
  • I understand, but I need to clarify because you and @GeoMonkey are saying 2 different things. If you read the comments of the following post, "gis.stackexchange.com/questions/446640/…" you will find this, "You cannot apply math operations on scaled values. You should apply the offsets and then do the math on the unscaled data". It is not clear, who is right because there are 2 different opinions. Any suggestion?
    – Chris
    Commented Jan 3, 2023 at 0:28
  • 1
    No, that says the same thing. The data is scaled to fit into an integer datatype for storage. Apply the scale and offset before calculating NDVI to unscale/rescale/scale to 0-1 reflectance. And this question has nothing to do with the question you link to and the questions you posed in comments.
    – user2856
    Commented Jan 3, 2023 at 1:36

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.