[GRASS-user] bug? r.watershed strange outputs

[Ken Mankoff]

On 2018-11-19 at 08:26 +0100, Ken Mankoff <mankoff at gmail.com> wrote:
> Attached is a graphic showing the elevation input and direction
> output.
>
> If you look near the burgundy square in the elevation at the 4.0 value
> to the NW, *everything* near that cell should flow into it (all
> neighbors are >4.0). However, if you look at the direction output, the
> 4.8 value (south of 4.0 and west of the burgundy 4.4) flows to the NE
> out of the domain (direction = -1). This seems like a bug.
>
> Viewing the output direction raster, the burgundy "3" flows into the
> "-2" to the NW. That means that is ONE catchment. But it is bisected
> by a different catchment (the -1 flowing out of the domain).

Well I can't re-create this with an MWE so for now please disregard this issue for now. I've lost the location in the large raster where I observed this, so I can't look into what caused the bug until it is re-created.

  -k.