[GRASSLIST:7897] Re: Hutchinson's Adaptive Alogrithm for sound DEMs?

[Dylan Beaudette]

Maciek and other GRASS users,

These are great points to consider. My work leads me into the realm of  
terrain analysis, so this topic is near and dear to me. I think that it  
would be great if the pooled efforts of the GRASS community could be  
used to make GRASS the premier DEM creation / modification environment.  
Since the quality of a DEM is of the utmost importance when calculating  
primary and secondary terrain parameters, documentation of and  
confidence in the algorithms used for DEM creation should be a  
priority.

Currently v.surf.rst provides a very flexible means to produce  
elevation surfaces from point and contour data. The recent work by  
Tomas Cebecauer and others (See "Processing digital terrain models by  
regularized spline with tension: tuning interpolation parameters for  
different input datasets" from the proceedings to the 2002 GRASS  
conference.) shows how v.surf.rst can be used in a method similar to  
ANUDEM to enforce proper drainage networks, by adding a "terrain  
skeleton".

In addition, in reference to a recent message from David Finlayson  
regarding cross-validation of v.surf.rst derrived data, it might be a  
good idea for someone with some experience doing so to make a how-to  
document. I for one am still a bit baffled at how to properly use the  
cross-validation tools in v.surf.rst.

Of course all of this is easy for me to say, and much harder to  
implement. As my C skills are not the best, perhaps I  can contribute  
to this aspect of future work in the form of documentation...

Any thoughts, ideas from other users?

--
Dylan Beaudette
Soils and Biogeochemistry Graduate Group
University of California at Davis
530.754.7341


On Aug 14, 2005, at 12:03 PM, Maciek Sieczka wrote:

> Helena, Markus,
>
> Since improving v.surf.rst is discussed, I would like to ad a few  
> notes from
> a simple user, who doesn't understand all the maths of DEM  
> interpolation but
> who would like to obtain decent results. Sorry for any naiveness.  
> Please let
> me know what you think and if any of my wishes you find worthy of  
> adding to
> v.surf.rst.
>
> From: "Markus Neteler" <neteler at itc.it>
>
>> Here a FWD from Helena Mitasova:
>>
>> On Tue, Aug 02, 2005 at 07:01:33PM -0400, Helena Mitasova wrote:
>
>>> We are currently testing different interpolation methods including
>>> topogrid (or whatever its current name is)
>
> I gues we should call it ANUDEM - after Hutchinson. The TOPOGRID is an
> ArcInfo program, which is an implementation of ANUDEM v. 4.5 or 4.6.3,
> depending ArcInfo/ArcGIS version
> http://support.esri.com/index.cfm? 
> fa=knowledgebase.techarticles.articleShow&d=20779.
>
>>> and v.surf.rst along with other methods and we
>>> plan to implement modifications/improvements that we find beneficial.
>
> If I may point 4 features I find very usefull in ANUDEM (and two other
> tools) for creating DEMs from data extracted from topo maps. It would  
> be
> great if v.surf.rst could provide them.
>
> ANUDEM is able to utilize following data besides elevation points and
> contours:
>
> 1. watercourses, including their flow direction
> I used it in TOPOGRID. At little effort - only digitise the  
> watercourses in
> the direction needed - it is possible to obtain a DEM where the water  
> flows exactly the way you want it.
>
> 2. elevation discontinuity lines, called 'cliffs' in ANUDEM
> This was introduced in version 5.1. I haven't had an occasion to use  
> it,
> since it is not available in TOPOGRID and I never had the original  
> ANUDEM at
> hand. I only read it's supported in new version
> http://cres.anu.edu.au/outputs/anudem.php and think it is a great  
> feature.
> Often elevation contour lines and points alone simply can't express  
> all the
> complexity of terrain when gullies, scarps, embakments, walls, other
> breaklines are involved. That's due to these are not parallel to  
> elevation
> isolines, thus cannot be presented as elevation isolines. Trying to
> represent them as points, although in theory doable in *some* cases,  
> would
> require a lot of work to digitise points dense enough and to correctly
> estimate each point elevation manually. Yet utilising elevation
> discontinuity lines, digitised from topo maps, could greatly improve  
> DEM
> accuracy - at very little effort. Especially in areas of land slides,
> erosion drived by river or flooding, land deformation due to mining,  
> cliff
> shoreline - to name those I can think of right now.
>
> Such a functionality is also present in SURGE interpolation software.
> http://www.geocities.com/miroslavdressler/surgemain.htm
> But it's only freeware/shareware for Windows, not free software, is  
> very
> limited as to amount of data it can handle in one turn and the input  
> data
> format is non standard and pretty complex.
>
> 3. waterbodies
> I used it in TOPOGRID. The elevation of waterbodies interpolated  
> agreed very
> well with their actual elevation as seen on topo maps and they are  
> flat like
> they should be. I bet many folks would find it usefull for accurate
> elevation representation in lakelands, visualisation of areas in the
> vicinity of water bodies etc.
>
> The other feature, not supported in ANUDEM but practical I think, are
> ridges. I found it supported in another DEM interpolation software,
> CatchmentSIM (again freeware for Windows, sigh), as "Interpolation  
> Training
> Lines". The user can digitise them from topo maps and inlude during
> interpolation to model the mountain ridges in his DEM as he wishes
> http://www.toolkit.net.au/catchsim/.
>
> Maciek
>
>
>