|
|
| GeoCommunity Mailing List |
| |
| Mailing List Archives |
| Subject: | Re: GISList: sampling point location |
| Date: |
12/07/2001 10:34:42 AM |
| From: |
Quantitative Decisions |
|
|
At 11:05 AM 12/3/01 -0500, Martin Chevrier wrote:
>I am trying to find a way to located sampling point. Starting from a
>polygon and some existing sampling point (point shape file) inside that
>polygon, I want to locate new sampling point in an automated way. The only
>critera I need to use are the area. for example, I need to have one samplin
>point for every 100 square meters.
>
>Did someone know a way to achive that ?
There are many, many ways. Perhaps the most natural is to create a regular
array of polygonal tiles (typically rectangles, triangles, or hexagons),
each of 100 square meters, intersect those tiles with your polygon, and
choose one point in some way within each tile--at its center, at a
designated point, or even randomly. This leaves a lot left to be
determined, including the shape and orientation of the basic tile. You can
designate these auxiliary parameters or choose them randomly.
A GIS is a very good platform for creating such sampling plans. Most
(all?) GISes do not do something this complex out of the box, however. You
have to write a script or application. Our web page,
http://www.quantdec.com/sample/, discusses the situation. It has some nice
illustrations, IMHO, especially of the relationship between the tiles, the
sampling points, and their associated Thiessen polygons.
Another approach first determines how many sampling points are needed
(divide the polygon area by 100 square meters), then attempts to distribute
the points among the polygon so that (a) the starting point is sampled and
(b) all points are spaced "far" apart according to some meaningful
criterion. This is a difficult problem to solve. The best known solutions
for arbitrary polygons and sample sizes are computed with simulated
annealing (SA) techniques. Some of this work is documented in the soils
literature: see, for instance, J. W. van Groenigen, W. Siderius, and A.
Stein, "Constrained optimisation of soil sampling for minimisation of the
kriging variance." Geoderma 87 (1999) 239-259. Simulated annealing
requires many iterated solutions (40,000 to over 100,000 in practice) and
so is usually coded in an efficiently compiled language, rather than as a
GIS application. However, in joint work with Alterra (Wagenigen, The
Netherlands) I have demonstrated practical SA solutions coded in Avenue
(ArcView 3.2a).
--Bill Huber
Quantitative Decisions
To unsubscribe, write to gislist-unsubscribe@geocomm.com
________________________________________________________________________
Setup a GeoCommunity Account and have access to FAST DataDownloads
and Premium Career Posting at a discounted rate!
https://www.geocomm.com/cgi-bin/accounts/login
On-line Archives available at
http://spatialnews.geocomm.com/community/lists/
|
|
Sponsored by:
For information
regarding
advertising rates
Click Here!
|
