Re: GEOSTATS: interpolation
- Tha is a non-sequitur, given only a finite amount of data and no state
equations you have a non-unique problem. First of all one has to ask what
does "best" mean? Again this is non-unique.
There are quite a number of papers that have compared different
methods/algorithms/implementations of algorithms/applications of algorithms
for rainfall and temperature data but in general one can only conclude that
for a particular data set one method or algorithm seemed to work best.
You mentioned using arcview spatial software and inverse distance weighting
with a specific choice of the search neighborhood. The results can be
sensitive to not only the choice of the search neighborhood but also the
choice of the exponent. See for example, Kane et al, Computers and
If one uses kriging, i.e., a variant of kriging (simple, ordinary,
universal, indicator, etc) then one still has to estimate and model the
variogram as well as to choose the search neighborhood parameters. Unless
the elevation is approximately constant over your region of interest you
may need to incorporate that effect either by cokriging or by using an
Differences in results from using different methods/algorithms may be
considerably affected (or not) by the pattern of data locations.
Do you want point interpolations or do you want average rainfall over
specific subregions? Kriging will allow you to directly do the latter but
IWD does not.
Note that the thin plate spline and smoothing spline are both special cases
My suggestion is to search the literature more completely for papers that
have applied different methods and try to match the specifics of one of
those instances to your problem.
Department of Mathematics
University of Arizona
Tucson, AZ 85721
At 12:24 PM 4/1/98 +0000, you wrote:
>i would like to find out the best method for interpolating rainfall
>and temperature data. the weather stations are randomly spread over
>an area of approx. 8 million hectares. there are approx. 200 temp
>stations and approx 1000 rainfall stations. i have tried inverse
>distance weighting with 4 nearest neighbours for temp and 8 for
>rainfall, both using a 300m grid size. i am using arcview spatial
>any help is appreciated.
>craig email: c_vhagen@...
>*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!
*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!