[R-sig-Geo] anisotropy - intamap

[seba]

Hi Jon

Thank you for the quick reply.
For my job I did my task (this morning) exactly following the 
procedure you mentioned (i.e. I calculated first the residuals)
and now things work well, very well!
I think that the proposed approximation is the only one possible, 
i.e. first remove the trend and then check for anisotropy of residuals.
As soon as possible I'll try the package you changed.
Again, thank you for the support!
Sebas


At 15.19 14/01/2010, Jon Olav Skoien wrote:
>Hi Seba,
>
>Your impression was unfortunately right, in the current version of 
>the package on CRAN it is not possible for the anisotropy detection 
>method to take trends into account. Although the trend and the 
>anisotropy parameters should ideally be estimated jointly, I think 
>it would be a decent approximation to estimate the trend first, and 
>the anisotropy from these residuals. This has been included in a 
>test version that you can download from:
>
>http://www.intamap.org/downloads/intamap_1.3-3.zip
>http://www.intamap.org/downloads/intamap_1.3-3.tar.gz
>
>I will check the results of the change a bit more, then upload to CRAN.
>
>Cheers,
>Jon
>
>
>seba wrote:
>>Dear list members
>>
>>I have a tecnical question in relation to the calculation of 
>>variogram anisotropy in presence of a trend with estimateParameters function.
>>I have a data set "w" with a clear trend (degree 1) so I would like 
>>to calculate geometric anisotropy of residuals.
>>I do the following:
>> 
>>bingo=createIntamapObject(observations=w,formulaString=as.formula(z~x+y),params=getIntamapParams(),class="automap") 
>>
>>  estimateParameters(bingo,model=c("Pow"))->p
>>
>>But when I go to check for anisotropy I have the impression that 
>>the anisotropy calculated is relative to the raw variogram not to 
>>the residual one:
>>Is it possible?
>>Is there something wrong in my expression?
>>Thank you in advance for your help.
>>
>>Bye
>>Sebastiano
>>
>>_______________________________________________
>>R-sig-Geo mailing list
>>R-sig-Geo at stat.math.ethz.ch
>>https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>