Opened 10 years ago
Closed 10 years ago
#666 closed defect (wontfix)
Wrong GeoTIFF output for WCS subsetting
| Reported by: | abeccati | Owned by: | Vlad Merticariu |
|---|---|---|---|
| Priority: | minor | Milestone: | 8.5.3 |
| Component: | petascope | Version: | 8.5 |
| Keywords: | maintenance | Cc: | |
| Complexity: | Medium |
Description
WCS subsetting does not keep the original pixel size of the underlying coverage.
For the sample query fragment over a rectified coverage with 100x100 meter GeoTIFF input (which is a 1kmx1km subset request):
FORMAT=image/tiff&SUBSET=x(5200000,5201000)&SUBSET=y(1950000,1951000)
The returned GeoTIFF has image size of 11x11 pixels with a pixel size of 90.9091m
Note: For a 1kmx1km subset, we should expect a 10x10 pixels image with a pixel size of 100m however, if the subset box is not aligned with the underlying pixel boundaries then there might be a +1 in each direction where this happens (the returned dataset is not resampled to the bbox, but extended to include all "original" pixels that intersect the bbox, so we might also have valid 10x11, 11x10 or 11x11 image sizes (the case on hand is a non-aligned one on both so 11x11 is the expected image size).
The resolution issue instead is definitely to be resolved since changing the output pixel size with a subset is an invalid result (it implies increased accuracy in the output)
Discovered in 8.5.2
