[GRASS-user] r.watershed and big files

[Markus Metz]

On Thu, Sep 1, 2011 at 4:07 PM, Arno Gerretsen <arno at agerrius.nl> wrote:
>
> Quoting Arno Gerretsen <arno at agerrius.nl>:
>
>>> I guess so too. The first message indicates that the on-disk option
>>> (-m flag) is used. One reason could be not enough free disk space to
>>> create temporary files. BTW, there seems to be a cut'n paste error,
>>> the first message should read "SECTION 1 beginning: ..." and not
>>> "SECTION 3 beginning: ..."
>>>
>>
>> I probably forgot to copy the first part then. Let me try again with the
>> new knowledge about the maximum size. Hopefully that will prevent these
>> errors.
>
> I did another attempt to use r.watershed on my big file. This time I
> compiled GRASS 6.4.1 from source code and made sure it was configured for
> largefiles (--enable-largefile). However I still get errors from r.watershed
> and it seems related to the segment file:
>
> GRASS 6.4.1 (ML):~/grass/bin > r.watershed elevation=ML_elev
> stream=stream_total accumulation=accumulation_total threshold=10000 --o -m
> memory=2000
> SECTION 1 beginning: Initiating Variables. 5 sections total.
> WARNING: No such file or directory
> WARNING: dseg_open(): could not write segment file
> WARNING: Subprocess failed with exit code 136
>
> Can anybody tell me what is causing this (and how I could solve it)? In this
> case I was using a region of 19800x16200 pixels.
>
I think the reason is that GRASS could not initialize a temporary
segment file, maybe because of insufficient free disk space?
dseg_open() s called relative late in the initialization process when
a number of other segment files have already been successfully
created. You could try to create a new GRASS database on a different
partition with more free disk space and try r.watershed again.

If possible, you could also try GRASS7 and run r.watershed --verbose,
this will print out estimated disk and memory requirements (and is
much faster than r.watershed -m in GRASS 6).

Markus M