GEOSTATS: Kriging Median Polish detrended data?
- Hi folks,
Well, I'm back again, a little wiser (thanks to you all) and a little
farther along in my analysis, but I've hit a new snag.....namely, how do
you create surface maps of data that have been detrended using median polish?
I'm following the advice of several papers published in soil science
literature, and they all suggest: 1) detrend 2) fit a semivariogram model
to the residuals 3) develop kriged estimates of the residuals at unsampled
locations AND 4) Add these estimated values back to the trend to create
surfaces. Funny, but that crucial last step is never demonstrated, and
neither Cressie nor Issaks/Srivastava address the issue directly.
After reading the literature, I'm convinced that the best way to detrend my
data is median polish, although I have the software (SPLUS) to support
polynomial detrending. So, I choose to do median polish, and now I don't
know how to create estimates of the *trend* at the unsampled locations for
which I've estimated the residual values. Can anybody help?
1) can I create a trend surface from the results of a median polish? I
have row, column and all values from the median polish - are these
expressing some kind of trend along the rows and columns that I can use to
estimate trend at unsampled locations?
2) what do you think of the relative value of median polish vs. polynomial
detrending - should I be concerned about the claims that polynomial
detrending is invalid, because spatially correlated data do not meet the
assumptions required for valid regression?
*To post a message to the list, send it to ai-geostats@....
*As a general service to list users, please remember to post a summary
of any useful responses to your questions.
*To unsubscribe, send email to majordomo@... with no subject and
"unsubscribe ai-geostats" in the message body.
DO NOT SEND Subscribe/Unsubscribe requests to the list!