2

I got the Landsat 8 image and removed the black area by doing nodata = 0. enter image description here enter image description here

I have a question here. What is the definition of nodata? Is it a pixel with a DN value of 0? Or is it an area with no DN value?

A value with a DN of 0 may also be necessary information. I want to remove only pixels, not satellite images, but if I give a zero value in the nodata option, I'm worried that the data in the satellite images will be erased.

What is the criteria for categorizing nodata areas?

Improve this question
4
  • 1
    No data classically means outside collection area but rasters must be rectangular, therefore irregular capture areas need a value that is nothing to fill that area, thus the concept of NoData is born. As for setting or asserting that's a different problem altogether, in the simplest form asserting a value of 0 means that GIS software won't draw those pixels but also any pixel that has a legitimate value of 0 will also be considered NoData which causes problems when using the data.
    – Michael Stimson
    Commented Mar 24, 2022 at 6:24
  • 1
    NoData value is probably the most common way to handle nodata even it is rather a workaround and with 8-bit data leads to issues that you were worried about. Water bodies are often totally black 0,0,0 and reflections from tin roofs etc. totally white 255,255,255 and nodata will burn holes to valid data area. Mask band or alpha are better in theory but all software do not support them.
    – user30184
    Commented Mar 24, 2022 at 7:51
  • @user30184 Thank you for your detailed reply. I understand the problem. You said that mask bands and alpha bands are theoretically better, but I wonder in what ways they are better for these problems. I still don't know the functions of mask bands and alpha bands. Can you explain it briefly?
    – bsg
    Commented Mar 25, 2022 at 8:31
  • Pixels of the mask band have only one meaning: there is/there is not data in this area. Pixels of the data bands have a primary use for the data with an additional rule "if value is xxx then it means nodata" but that has a side effect that xxx cannot appear in data. Another problem comes when image is compressed with a lossy method like JPEG. JPEG compression may change the pixel values slightly and RGB (0,0,0) can became (0,0,1) or (1,0,0) and nodata turns into data by the same.
    – user30184
    Commented Mar 25, 2022 at 9:03

1 Answer 1

Reset to default
1

Different satellite imagery providers may define NoData areas differently. Landsat and Sentinel-2 use a zero value for this purpose. You can identify NoData value for a particular image yourself using the Identify tool or the Value tool plug-in.

Improve this answer

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.