## Re: GEOSTATS: weighted interpolation method

Expand Messages
• Gregoire, Defining alternative metrics seems like a good idea on the face of it although a great deal of subjective input is needed and a sensible metric for
Message 1 of 2 , Feb 9, 1997
Gregoire,

Defining alternative metrics seems like a good idea on the face of it
although a great deal of subjective input is needed and a sensible metric
for one variable may not be sensible for other measured or mapped variables
available. This poses an interesting problem in which different metrics are
integrated into a single cost function for interpolation. An alternative I
try to employ if possible is to investigate linear/nonlinear regression /
ancova relationships between the variable of interest and any other
available covariates such as elevation, slope aspect, satelite image data,
and on and on depending on the particular application. After partialling
out large scale trends with categorical and continuous variables, the
residual process is often independent or only weakly spatially correlated
and predictions are made in a universal kriging type analysis where the
design matrix contains more than just geographic coordinates or a single
variable external drift. I do this in a MATLAB toolbox which I have been
developing over the last year or so. However, this could be conducted in
ARC by fitting regression models in S+, SAS, SPSS or your favorite
statistical software package followed by ordinary kriging the residuals and
adding. This can probably be done in GSLIB also without a great deal of
trouble. Especially in the case of pollutants, I have found this approach
especially useful for pollutants such as PCB or heavy metals.

>Greetings to all Arc/Users and or AI-geostatisticians,
>
>The IDW is an inverse distance weighted interpolation method.
>The problem here is the definition of the distance.
>2 measurements of a variable X close together can be separated
>geographically by a hill which won't be taken into account during
>the interpolation,
>
>Therefore, if I use a DEM (for exemple) to add another weight to
>the interpolation function ( with cost distance functions like
>those proposed by ESRI for ARc/Info) I wouldn't be limited
>anymore to euclidian distances.
>
>
>The use of a "cost distance weight", a measure of the "effort" to reach
>2 points, could also be usefull for the variogram calculation.
>
>Well, these are things I'm trying out for the moment with a pollutant.
>
>
>Any suggestions on how to do it with/without Grid (Arc/Info) ?
>
>
>Thanks for any help,
>
>
>Best regards,
>
>
>Gregoire
>--
>Gregoire Dubois (PhD student) Tel. 39-332-78.99.44
>Joint Research Centre Fax. 39-332-78.54.66
>Environment Inst. TP 321 Email: gregoire.dubois@...
>I-21020 Ispra (Va), ITALY URL: http://java.ei.jrc.it/rem/gregoire/
>--
>*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 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.
Your message has been successfully submitted and would be delivered to recipients shortly.