[GRASS-user] Selecting starting points for r.flow

[Vincent Godard]

Hi Ken,

Thanks for the advice. The reason I want to use r.flow is because of the 
specific way it handles flow routing (not constrained along a 8 
directions pixel-based framework, as r.drain or other solutions you 
mentioned), which is needed for the hillslope dynamic problem I am 
working on.

Best,

Vincent


Le 16/12/2016 à 12:52, Ken Mankoff a écrit :
> Hi Vincent,
>
> Would r.drain work? It allows multiple pairs of initial locations.
>
> I had a similar issue - I wanted to extract a single flowline from a watershed. I think the correct way to do this is r.drain. But I wanted to compare it to results derived from r.watershed, and the flow algorithm from r.watershed and r.drain are different.
>
> My solution was to use r.stream.extract to extract all the streams, and then build a mask, and then run r.drain on the DEM with the mask. This forced r.drain to follow the same flowpaths as those generated from r.watershed.
>
> Perhaps something similar might help,
>
>    -k.
>
> On 2016-12-13 at 17:55, Vincent Godard <godard at cerege.fr> wrote:
>> I am using r.flow to generate hillslope flow paths over a high
>> resolution DEM. I am actually only interested in a limited number of
>> these paths, originating at some selected hilltop points.
>>
>> r.flow computes flow paths over the whole DEM, and in order to get the
>> vectors I am interested in I would need to run the computation over
>> the whole region (with skip=1) and select afterward some of these flow
>> lines from the output, which is computationally prohibitive on my DEM
>> (potentially millions of lines and I am only interested in a few 1000s
>> of them).
>>
>> Is there way to force r.flow to limit the computation to selected flow
>> starting points?

-- 
Vincent Godard
CEREGE - OSU Pytheas
Aix-Marseille Université
Europôle Méditerranéen de l’Arbois - BP 80
13545 Aix-en-Provence cedex 04, France
godard at cerege.fr