Hello Hao
Regarding your question about how to apply a correction for the 0.5
pixel offset, we need to estimate the size of a pixel. For this purpose,
we need the number of rows and columns in an image. I'm not sure which
Landsat attribute give that information. It may be the REFLECTIVE_LINES
and REFLECTIVE_SAMPLES properties, but this needs to be verified:
* Verify documentation provided in the LSDS-809.pdf file.
* Be aware that there is also PANCHROMATIC_LINES and THERMAL_LINES
properties. Which property to use depends on the band - the
LSDS-809.pdf documentation should tell which bands.
* Compare with the size of one of the TIFF images that come with the
Landsat metadata. For example open the image with Gimp and verify if
its size match the numbers in the Landsat file.
After we verified that the image size information that we got is
correct, we can use that information (together with the latitude and
longitude values) for getting a rough estimation of pixel size.
Martin
