[GRASS-dev] Watershed delineation result problem

GRASS GRASS developer
1

Hi Everyone

Hope this email find you safe and healthy

I tried watershed delineation with both ArcGIS and GRASS. but I got different results.

I was using D8 flow direction method in both ArcGIS and GRASS. My study area is very flat. Is there any tool in GRASS will generate the same result with ArcGIS

Thanks
Ming

2

On Wed, 7 Oct 2020, ming han wrote:

  Hope this email find you safe and healthy

  I tried watershed delineation with both ArcGIS and GRASS. but I got
different results.

  I was using D8 flow direction method in both ArcGIS and GRASS. My study
area is very flat. Is there any tool in GRASS will generate the same result
with ArcGIS

Ming,

I did wetland determinations and delineations for almost a decade but
dropped it when it became a commodity. I offer two points for your
consideration:

1) The ArcGIS may not be any more 'accurate' than the GRASS results. Never
having used the former I cannot comment on how they do this.

2) Of greater importance is that you can pick either one; it doesn't matter
in the real world. In 1994 I was the first environmental consultant
authorized by Oregon's Department of State Lands to use GPS receivers to
delineate wetland boundaries. They had insisted that only professional land
surveyors could do this and they set a 2cm accuracy standard. Really?
Wetland boundaries are transistion zones that can be several meters wide,
depending on topography, soils, and antecedent precipitation conditions when
the boundary is flagged. A stream bank is an exception to this broad
transition area. When I made the case that there is no sharp line of
demarkation between wetland and upland they accepted my delineations.

Of similar disconnect between engineering and natural ecosystems, I worked
for a brief time for a water management district in the 1980s. They decided
to digitize the 7.5 min (1:24000) topographic maps covering the District's
area and contracted with a company in India to do the work. The contract
specified that the digitized lines had to aline withine 1/2 the width of
roads and other boundaries on the maps. When I pointed out to my Division
Director that the maps themselves said "this map is accurate to +/- 24 feet"
so they were trying to be more accurate than the maps themselves it was not
well received. :slight_smile: (That's one reason I left a government position.)

Anyway, draw your boundary and in most cases you'll be within that
transition zone.

HTH,

Rich

3

Hi Rich

Thanks for sharing, I see, I will continue with GRASS version I guess.

Cheers
Ming

Rich Shepard <rshepard@appl-ecosys.com> 于2020年10月7日周三 下午1:42写道：

On Wed, 7 Oct 2020, ming han wrote:

Hope this email find you safe and healthy

I tried watershed delineation with both ArcGIS and GRASS. but I got
different results.

I was using D8 flow direction method in both ArcGIS and GRASS. My study
area is very flat. Is there any tool in GRASS will generate the same result
with ArcGIS

Ming,

I did wetland determinations and delineations for almost a decade but
dropped it when it became a commodity. I offer two points for your
consideration:

  1. The ArcGIS may not be any more ‘accurate’ than the GRASS results. Never
    having used the former I cannot comment on how they do this.

  2. Of greater importance is that you can pick either one; it doesn’t matter
    in the real world. In 1994 I was the first environmental consultant
    authorized by Oregon’s Department of State Lands to use GPS receivers to
    delineate wetland boundaries. They had insisted that only professional land
    surveyors could do this and they set a 2cm accuracy standard. Really?
    Wetland boundaries are transistion zones that can be several meters wide,
    depending on topography, soils, and antecedent precipitation conditions when
    the boundary is flagged. A stream bank is an exception to this broad
    transition area. When I made the case that there is no sharp line of
    demarkation between wetland and upland they accepted my delineations.

Of similar disconnect between engineering and natural ecosystems, I worked
for a brief time for a water management district in the 1980s. They decided
to digitize the 7.5 min (1:24000) topographic maps covering the District’s
area and contracted with a company in India to do the work. The contract
specified that the digitized lines had to aline withine 1/2 the width of
roads and other boundaries on the maps. When I pointed out to my Division
Director that the maps themselves said “this map is accurate to +/- 24 feet”
so they were trying to be more accurate than the maps themselves it was not
well received. :slight_smile: (That’s one reason I left a government position.)

Anyway, draw your boundary and in most cases you’ll be within that
transition zone.

HTH,

Rich

grass-user mailing list
grass-user@lists.osgeo.org
https://lists.osgeo.org/mailman/listinfo/grass-user

4

On Wed, 7 Oct 2020, ming han wrote:

   Thanks for sharing, I see, I will continue with GRASS version I guess.

Ming,

While I mis-read your 'watershed' as 'wetlands' the same argument holds
true. I don't know the size of each cell in your DEM raster or whether
arcGIS uses the cell center or a corner for calculations. There are so many
factors involved that unless there are major differences pick one and use
it.

As I run only linux and have used GRASS since the mid-1990s I can't speak
about arcGIS.

Stay well,

Rich

5

Hi all,

I would like to briefly weigh in in this discussion, especially about
the following:

2) Of greater importance is that you can pick either one; it doesn't matter
in the real world. In 1994 I was the first environmental consultant
authorized by Oregon's Department of State Lands to use GPS receivers to
delineate wetland boundaries. They had insisted that only professional land
surveyors could do this and they set a 2cm accuracy standard. Really?
Wetland boundaries are transistion zones that can be several meters wide,
depending on topography, soils, and antecedent precipitation conditions when
the boundary is flagged. A stream bank is an exception to this broad
transition area. When I made the case that there is no sharp line of
demarkation between wetland and upland they accepted my delineations.

I couldn't agree more on this point - at times people really have fancy
interpretations of maps' resolution and what they can actually tell about
the real world. I also would like to add that watershed delineated from
a digital topography may depend on the scale at which one looks at the
landscape. I mean, not the DEM resolution by the very definition of a
watershed - if we agree that this definition depends on the number of
cells draining on the outlet (parameter "threshold" on r.watershed), we
get very different results depending on this parameter.

And now for a little advertising: we discuss these topics and their use
to produce slope unit tools (in GRASS GIS) and maps in these papers:

"Automatic delineation of geomorphological slope units with r.slopeunits v1.0
and their optimization for landslide susceptibility modeling "
https://doi.org/10.5194/gmd-9-3975-2016

"Parameter-free delineation of slope units and terrain subdivision of Italy"
https://doi.org/10.1016/j.geomorph.2020.107124

M