local elevation

[Mario Palladino]

I thank you for your quick answer. Maybe either my
answer wasn't clear or I did not explain in detail the
problem. I apologise also for my bad english.

>   I'm curious: how accurate are your elevation
> measurements and how accurate
> are horizontal positions?

Of course there are errors. I cannot quantify 'cuase I
didn't develop that measurements by myself.
 
>   The reason I ask is because most requirements for
> "precisely localized"
> models are based on a lack of understanding of
> natural systems. If your
> sampling points are 50 m apart, what do you mean by
> "precisely localized"?
> 

What I mean with the term "precisely" is that I wish
calculate the elevations of the eight neibourgh point
of each sample point measurement, just to calculate
terrain attributes with different grid size, even
greater than 50 m. I believe that in this way I could
avoid introducing another source of error. If I used
directly the output maps (slope, aspect, etc.) of
s.surf.tps, for example, the results may be non
representative of the grid cell centered "with a
certain casual error" on each of my 100 points not
being just in the middle of the grid cell.

>   The other consideration is the interpolation
> algorithm used to build the 3D
> surface from the point data. Several interpolators
> have been developed. The
> one most commonly used with elevation data assumes
> unevenly-spaced
> measurements and (by design) overshoots the high
> points and undershoots the
> low points. The reason for this is that almost never
> do you have
> measurements at the actual local maxima and minima.
> 
Yes, I knew about that, errors are unavoidable. In
this moment I'm only a user of this kind of programs.
Besides, my point are along a trasect situated along
an hillslope, so there aren't peacke or valleys.

>   What I'm trying to say (in a round-about,
> long-winded way) is that any DEM
> you build is an estimate of the real world, and it
> contains varying amounts
> of inaccuracies. The only way you'll get "precisely
> localized" models is to
> have a survey crew place a number of markers at
> known locations on the
> ground, have the site flown by a skilled aerial
> photography company, then
> pay a photogrammetist big money to produce an
> accurate DEM for you.

I thank you for your words, you've been very kind to
answer to my simple question, but I just want to know
if there are some grass programs that could manipulate
site list files in the same way of r.mapcalc. In this
way i should be able to solve my problem.



=====
Mario Palladino
DIAAT - University of Neaples
Via Università 100 - Portici - Napoli - Italy
tel/fax +39-081-7755341

__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com