gdal_contour: min value unexpected

[pmeems]

What is the bug?

I have a Surfer grid which I want to convert into polygons using gdal_contour.
The min value of the lowest polygon is 0 which I didn't expect. Since the max value is 107000 I expected the min value to be 101000.
In my next step, I calculate the average of the min and max values, and in this case that average is wrong.

I've attached the Surfer grid.
These are the values of the grid:

Grid Size (X*Y):       113 x 121
NoData Value:          1.70141E+38
Minimum:               107000
Maximum:               133000

Surfer.zip

Steps to reproduce the issue

gdal_contour -p -f "ESRI Shapefile" -fl -97000 -37000 23000 83000 89000 95000 101000 107000 113650 121000 129000 137000 143000 149000 155000 215000 275000 335000 -amin Min -amax Max "Surfer.grd" "tmp3p4f4i.shp"

Versions and provenance

I tried it with GDAL 3.6.3, released 2023/03/07 on Windows and with GDAL 3.8.4, released 2024/02/08 also on Windows

Additional context

I don't have this problem with just this grid file, I experience the issue with all grid-files I convert using gdal_contour.

We now have to manually update the shapefile to fix this. This is a lot of work for us. I think I can free up some budget if needed to fix this issue.

[pmeems]

I haven't tested it with v3.9.0 because it is not yet available at gisinternals.com.
But I did try with the development version of gisinternals.com: GDAL 3.10.0dev-cccb40e39c, released 2024/05/24

@rouault: Should this fix work for that version as well? I assume it does because it has a higher version number.

But in this version it doesn't seems to work. The lowest value is no longer 0 but it is still not the correct value.
The value is now -4294967295 which looks like the lowest possible int.

This time I use these gdal_contour.exe arguments:
-p -f "ESRI Shapefile" -i 20 -amin Min -amax Max "TC.asc" "tmpefbulv.shp"
The result:

If needed I can provide some sample data.

[rouault]

Should this fix work for that version as well? I

from what I can see, you're not using -fl but -i, so this is a slightly different code path that might require extra adjustment as well. Please re-open a new issue with the gdal_contour command line arguments and the input dataset

[pmeems]

I was working on the new issue: gathering data, describing my steps when I noticed that the lowest value is set to the lowest value when calling first gdalinfo -stats TC.asc. In this case 72.818
Without gathering the stats in v3.8.5 the lowest value is 0 and in v3.10.0 it is -4294967295.

So I will update my workflow and create the statistics before I call gdal_contour, that seems to solve my problem.