Dear Gregoire,
I just recently read your original question. I've not read all the
replies since, so I hope what I say is not redundant or repetitive.
The phenomenon you are seeing is due to the smoothing effect of
kriging. This effect increases with geographic distance of estimates
from observations. If you partition validation estimates by estimation
variance, you will find an increasing correlation between observed and
estimated values with decreasing estimation variance. Estimates for
points beyond the variogram range will be uncorrelated with observed
values, and therefore residuals and original values will be negatively
correlated. This is illustrated and discussed in "Geostaistics and Pet.
Geol." pg. 103 (2nd Ed.) or pg. 128 (1st ed.)
Now, did I understand your question?
Mike
--
____________________________________________________
Michael E. Hohn,
Director and State Geologist
West Virginia Geological and Economic Survey
1 Mont Chateau Rd.
Morgantown, WV 26508-8079
Tel. 304-594-2331
Fax. 304-594-2575
____________________________________________________
+
+ To post a message to the list, send it to ai-geostats@jrc.it
+ 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/
