Hello Kelly,
Your project stands to benefit
when you construct sampling variograms as defined in several ISO Standards
including those developed by ISO Technical Committee 69-Applications of
Statistical Methods. It may also benefit from the interleaved sampling
protocol, which is a cost-effective technique to estimate extraneous measurement
variances. Such variances add to the variance of the set and to the first
variance term of the ordered set and may therefore be subtracted before Fishers
F-test is applied to verify associative dependence between temporally
ordered data. In this context, I refer to spatial dependence
rather than "associative dependence". Here are links to a paper on
Sampling in Mineral Processing and to a spreadsheet template with a
temporally ordered set of on-stream data at a mineral processing plant and its
sampling variogram. You'll notice that a sampling variogram shows where
spatial dependence dissipates into randomness.
http://www.geostatscam.com/Adobe/Sampling_Processing.pdf
http://www.geostatscam.com/Excel/Appendix%20D.xls
I couldnt possibly provide
more details in this email
Kind regards, Jan
W Merks
----- Original Message -----
Sent: Thursday, July 20, 2006 2:59
PM
Subject: AI-GEOSTATS: PCA and spatial
designs
Dear Colleagues,
I am in the process of modeling a
semivariogram for a host of chemical
contaminants that are released into
the ocean via a sewage outfall in
the near shore ocean. The
goal is to create a new design or "optimal"
grid for predicting
contamination for future surveys. In particular I
would like to create a
cost efficiency analysis with kriging error on
the y-axis and distance
between grid points on the x-axis to look at the
trade offs (eg. as
suggested by McBratney). I am not looking for
something complicated,
but something rather straightforward. My
questions are as
follows:
1. Is there a problem with first running a PCA on the
chemistry and
using the 1st PC scores as a surrogate that will give me a
useful
semivariogram for this purpose? In other words, rather than
modeling
individual chemicals would it be better to use scores along the
PCA. Am
I violating some assumption here in the presence of spatial
correlation?
2. Should I include covariates like depth and grain size
in the PCA or
should I use the residual fit with grain size/depth as
covariates.
3. Is there a useful function in Splus for creating the
cost-efficiency
curve.
4. Can you suggest any useful
references.
+
+ To post a message to the list, send it to [email protected]
+ To unsubscribe,
send email to majordomo@ jrc.it with no subject and "unsubscribe ai-geostats"
in the message body. DO NOT SEND Subscribe/Unsubscribe requests to the
list
+ As a general service to list users, please remember to post a
summary of any useful responses to your questions.
+ Support to the forum
can be found at http://www.ai-geostats.org/
|
