[gdal-dev] Resampling causes change of max value

Denis Rykov rykovd at gmail.com
Thu May 20 13:13:04 PDT 2021


Thank you Even. You are totally right as always. The reason is the broken
overview.

$ gdal_translate -outsize 1% 1% -oo OVERVIEW_LEVEL=0
input_raster_to_resize_with_max_val_92.tiff output-0ovr.tif

$ gdal_translate -outsize 1% 1% input_raster_to_resize_with_max_val_92.tiff
output.tif

$ gdalinfo -stats -json output.tif | jq
'.bands[0].metadata[""].STATISTICS_MAXIMUM'
"3149.9255371094"

$ gdalinfo -stats -json output-0ovr.tif | jq
'.bands[0].metadata[""].STATISTICS_MAXIMUM'
"70.651634216309"

On Thu, May 20, 2021 at 9:51 PM Even Rouault <even.rouault at spatialys.com>
wrote:

> Denis,
> The maximum of nearest downsampling should be <= maximum of full
> resolution.
>
> Potential explanations:
>
> - your source file has overviews that contain completely different values
> than the full resolution image.
>
> - or perhaps your input_raster_to_resize_with_max_val_92.tiff file has a
> .aux.xml with outdated stats (gdalinfo -stats doesn't regenerate stats if
> already found)
>
> Even
> Le 20/05/2021 à 21:42, Denis Rykov a écrit :
>
> $ gdalinfo -stats -json input_raster_to_resize_with_max_val_92.tiff | jq '.bands[0].metadata[""].STATISTICS_MAXIMUM'
> "92"
> $ gdal_translate -outsize 1% 1% input_raster_to_resize_with_max_val_92.tiff output.tif
> Input file size is 6845, 10302
> 0...10...20...30...40...50...60...70...80...90...100 - done.
> $ gdalinfo -stats -json output.tif | jq '.bands[0].metadata[""].STATISTICS_MAXIMUM'
> "3149.9255371094"I expected to get the same or at least close value as in the original raster. What might be the reason for that behavior?
>
>
> _______________________________________________
> gdal-dev mailing listgdal-dev at lists.osgeo.orghttps://lists.osgeo.org/mailman/listinfo/gdal-dev
>
> -- http://www.spatialys.com
> My software is free, but my time generally not.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20210520/91259e29/attachment.html>


More information about the gdal-dev mailing list