[R-sig-Geo] Differences of variogram parameters fitted by gstat and automap

[Kerry Ritter]

 I don't know much about "autofitVariogram" fuction in the "automap"
package but I am pretty familiar with the "fit.variogram" function in
the "gstat". One possibility for getting different parameter estimates
is that the two methods are using two different objective functions for
fitting the variogram. For example, I know that gstat has options for
various weighted least squares solutions. I have played around with a
few of the weighted least squares options and I get highly variable
results for the same data, depending on which weights I use. Also, check
that your exponential models are expressed the same in both programs.

Hope this helps,
Kerry

On 9/9/2010 7:44 AM, K Z wrote:
> Dear all,
>
> I am currently fitting variograms using both the "autofitVariogram" fuction in the "automap" package and the "fit.variogram" function in the "gstat" package. I used "autofitVariogram" to fit variograms at first, and then I used estimated parameters as the input of the "fit.variogram" function. However, I found the estimated parameters are largely different.  For example, the estimated parameters by the "autofitVariogram" function are
>
> $var_model
>   model psill range kappa
> 1   Nug  2.47     0     0
> 2   Ste  1.41  2506    10
>
> the estimated parameters by the "fit.variogram" function are:
>   model psill range kappa
> 1   Nug  0.74     0     0
> 2   Ste  3.03  1269    10
>
> I tried the "exponential" model structure, and got similar results. I understand I probably cannot exactly same results, but I expect I can get similar results. Any suggestion?
>
> I appreciate your response and best regards,
> Kai
>
>
> __________________________________________________
> 赶快注册雅虎超大容量免费邮箱?
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo


-- 
**********************
Kerry Ritter, Ph.D.
statistician
Southern California Coastal Water Research Project
3535 Harbor Blvd., Suite 110

work: 714-755-3210
cell: 714-420-3346
fax:  714-755-3299

email: kerryr at sccwrp.org