I cant resist since I looked at the "spiky" data of the first author below
(Pancho, Francois Molle) and responded as below in quotes on 5/19/98. I
might also add that gridding methods should be chosen appropriate to your
dataset not just whether or not you want your contours to be smooth.
Kriging, for instance, can grossly modify the actual data distribution and
is inappropriate for surfaces that need to specifically honor points such
as topography or the water-well example noted by James Harvie. All of
these tools, (MI, Vertical mapper, Surfer and others) allow for making
very attractive things very easily and quickly but not necessarily
appropriately.
Regards
George M Smith
On 5\19\98 I wrote:
"This may not be an answer you want. I am no longer using VM 1.5 and dont
have it on any machines around here to test. I can say frankly that for
creating a grid file to represent topography your number of nodes is far
too great given the distance between your contour lines. Particularly for
any inverse distance method, you will produce a stairstep effect. The
program doesnt know if you are doing topography or concentration of
elements in soils for instance. This will also have an effect on
tin-derived data although less. I did run your curved dataset through using
the default tin parameters in VM 2.0 and got a very good to excellent
result. There are some patches to VM1.5 for which you should contact
Northwoods directly and/or check their website if you dont want to upgrade
to 2.0. However, too high a concentration of nodes for your line spacing is
your biggest problem.
Its probably no consolation, but I just checked your parallel lines file
and the default tin parameters in VM 2.0 produced a perfectly smooth result
Regards
George M Smith"
At 08:33 AM 2/2/99 -0500, you wrote:
>We use both VM and Surfer, and my interpretation of the 'spiky' comment
>relates to the contour objects as opposed to the histogram. We have
>found that for many of our needs kriging is necessary to achieve smooth,
>evenly spaced contouring. This is the only area where VM falls short
>and where we turn to Surfer. Even then we typically bring the results
>back into MapInfo. If VM incorporated kriging as a modeling option it
>would be all we'd need.
>
>Will Mitchell
>Geographic Services Coordinator
>The Environmental Company, Inc.
>Charlottesville, Virginia
>804-295-4446
>[EMAIL PROTECTED]
>
>
>>-----Original Message-----
>>From: James Harvie [SMTP:[EMAIL PROTECTED]]
>>Sent: Monday, February 01, 1999 5:56 PM
>>To: Mapinfo L
>>Subject: MI VM vs. Surfer (or other)
>>
>>>Personnaly I use VM and find the interface (version 2.0) quite OK, but
>>>the interpolation algorithms yield surprising results : when
>>>transforming contour lines (say elevation 1,2,3,... to 10) into points
>>>and creating a grid surface based on these points, the cell values
>>>distribution is incredibly "SPIKY", with huge concentrations of cell
>>>values around integer values (while we could expect a smooth
>>>distribution from regularly spaced contour lines).
>>
>>I assume that the "SPIKY" behaviour you are speaking of is that shown in the
>>histogram displayed in the Colourizer dialogue. If this is the case, then
>>the spikiness is really just an artifact of the default histogram interval
>>used in the graph function. As you know, with histograms, it is possible to
>>get very different looking graphs of the same data simply by adjusting the
>>width of the sampling interval. This could be improved by allowing the
>>user to set the number of intervals as desired. We'll consider this for the
>>future.
>>613-224-2020 << File: ATT00061.html >>
>----------------------------------------------------------------------
>To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put
>"unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED]
>
>
GeoMax Information: GIS for Exploration and Geoscience
Data, Research, Maps
Exploration and Mining Information
VOICE 303.440.4539
FAX 303.727.4679
EMAIL [EMAIL PROTECTED]
WEB www.geomax.com (in development)
----------------------------------------------------------------------
To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put
"unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED]
