[GRASS5] Re: r.drain -- new version

[Rich Shepard]

On Thu, 19 Jul 2001, Jaro Hofierka wrote:

> I think some of these modules (r.watershed, r.fill.dir, r.drain) use older
> methods of flowtracing based on so-called D8 algorithm, i.e. use only
> limited 8 directions of flow with zig-zag flow pattern. Also, the
> flowaccumulation is computed using a single flow direction (flow is
> assigned only to 1 downhill cell).
>
> The second group of flowrouting modules contains r.flow and r.flowmd. They
> use a D-infinity algorithm, i.e. consider all possible flow directions and
> therefore produce smooth flowlines. The flowaccumulation maps are produced
> using a single flow direction model in r.flow and multiple flow direction
> algorithm in r.flowmd.
>
> Helena sent me a very interesting link to TerraFlow project at Duke
> University (http://www.cs.duke.edu/geo*/terraflow/). Those interested in
> this topic might find a lot of useful information here including
> comparisons with r.watershed and ArcInfo.

Jaro,

  Thank you for the insight. I suspected something like this was the case.
What I've learned so far from the terrain analysis book is that the D8
(discrete, 8 adjacent cells) algorithm is not nearly as representative as
others -- particularly DEMON. The book cite's Helena's efforts at
implementing this method in one of the GRASS modules (I forget which one it
is right now).

  I greatly appreciate this information, and I need to look into the modules
more closely so that I can understand and document just how GRASS is doing
its thing.

Rich

Dr. Richard B. Shepard, President

                       Applied Ecosystem Services, Inc. (TM)
            2404 SW 22nd Street | Troutdale, OR 97060-1247 | U.S.A.
 + 1 503-667-4517 (voice) | + 1 503-667-8863 (fax) | rshepard at appl-ecosys.com
                 Making environmentally-responsible mining happen.