Differences in raster extents in gdalinfo, SAGA & QGIS after resampling [closed]

Question

I recently resampled a raster file using SAGA's Resampling module. The aim was to widen the extent of the new raster to two decimal places, where the original raster has three decimal places. The resolution of the raster stays the same with 0.02. I take the extent of the original raster and calculate the extent of the additionally needed input-raster from it.

For example - Orginal raster upper left corner:

• This is what gdalinfo tells me:

    589262.561, 5955004.195
  

• What I see with SAGA hard zoom to displayed raster:

    589262.561, 5955004.195
  

• What I see in SAGA-Metadata instead:

    589262.571, 5955004.185
  

• What I see with QGIS hard zoom to displayed raster:

    589262.561, 5955004.195
  

• What I see in QGIS Metadata:

    589262.5609, 5955004.1949
  

The given numbers of lines and columns are all the same from gdalinfo to SAGA and QGIS.

Why is there such a difference in displaying the raster in SAGA?

After resampling with the calculated input grid I get the following:

gdalinfo:         589262.550, 5955004.210
SAGA-zoom:        589262.570., 5955004.190
SAGA-metadata:    589262.56, 5955004.2
QGIS-zoom:        589262.570., 5955004.190
QGIS-metadata:    589262.550, ,5955004.209

Again such differences between displaying the raster and its actual metadata. However, the amounts of lines and columns are identical again, though the other corners also differ slightly.

Is there generally such inaccuracy in these programs?

Is it somehow caused by the resampling module?

Which information should I believe for further use of the raster?

A lot of information and questions.

Answer 1

I found the following by coincidence in SAGA's "Create Grid System" Module-Description:

...the module follows the philosophy of SAGA in that the values describing the extent refer to the cell centers...

With a raster, using a 0.02m²-Resolution, it perfectly explains the "phenomena".

So SAGA metadata shows the cell's center, and SAGA-viewer the border - as well as gdalinfo does. Same counts for QGIS i guess. If I had paid more attention on the numbers themselves i could have figured out on my own earlier.

Volker Wichmann wrote the module description.