Hi to all,
This is my scenario followed by several questions since I am new to the
world of geostats, variograms, and kriging. I collected soil samples at 46
sites on a valley floor and wish to interpolate to 130 sites in order to
determine the distribution of each soil property on the landscape. Each
property will be analyzed individually. So far I have worked using S+,
attempting each step of the variogram and kriging processes. These are my
(1) Some of the data needs transformation due to skewness, etc. When I
apply an omnidirectional to the transformed data 7 of the properties appear
to be spatially autocorrelated versus 5 when the data are not transformed.
Is it appropriate for me to be transforming my data based on (a) some
properties do exhibit spat. auto. following transformation and (b)
information posted in ai-geostats-FAQ states that a bias is introduced when
using transformed data in kriging?
(2) My sample size is too small to consider looking for anisotropic affects,
or directional trends - what is recommended in this instance to check for
(a) possibility of stationarity and/or (b) remove the affect of a potential
trend, or at this sample size I should not worry about these?
(3) Is there anything out there that would guide me in the interpretation of
the y-axis of the semivariogram results - that is, the semivariance values
that are computed for each property? Some of property results are low while
others are high, my guess somewhat a function of scale of measurement, but
(a) how much and (b) at what point would they indicate that even if there is
spat. autocorr. the variances are too high in general to consider using the
results for the kriging process - or is this a worthy question?
Thats enough for now - thanks, Cathy
Utah State University
Dept. Fisheries and Wildlife
Logan, UT 84322-5210
*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!