GEOSTATS: guided interpolation
- This is an inherent problem in spatial, and temporal, data analysis
where we don't have enough information from the "out" side of the
boundary to avoid a biased estimate. From my experience, this problem
will always have an effect on your solution, no matter what you do.
Unfortunately, Kriging does not treat this problem thoroughly. In
non-parametric statistics, for example Kernel Estimators, several
techniques were adopted. For example the reflection method may be used
to generate fictitious data points in the "out" side of the boundary as
mirror image of the real data points which exist in the "in" side of the
boundary. This technique is reasonable only if the spatial, or the
temporal, derivative of the function is very small near this boundary.
Other techniques used with Kernel Estimators are "cut and normalize",
and "boundary Kernels". I will be more than happy to provide references
for those who are interested.
Alaa I. Ali, Ph.D., P.E.
New Mexico State Engineer Office
P.O.Box 25102, Santa Fe, NM 87504
e-mails: aali@...; or aali@...
Phone: 505-827-6125 Fax: 505-827-6188
> contours at the edges (edge effects). Recently, I talked to someonewho
> gets around this problem by constraining the analysis as averages overwhat
> was expected to be homogeneous areas (polygons). Of course, thehomogeneous
> areas were subject to modeling and mapping errors.polygons
> I'm interested in hearing how people deal with edges during surface
> modeling (short of cropping out the edges, designating homogeneous
> or making assumptions about outliers). [i.e. assume you need tointerpolate
> a surface over a given area, but you only have data within that area]and
> Also, I've heard of constraining/tying rainfall to elevation during
> interpolation (Hutchinson's work). Have others used ancillary data to
> constrain their surfaces?
> *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
> "unsubscribe ai-geostats" in the message body.--
> DO NOT SEND Subscribe/Unsubscribe requests to the list!
*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!