[Gstat-info] Block kriging odd shaped fragmented areas

[Hoogland, Tom]

Edzer,

Thanks for your rapid response. 

I use a local neighbourhood in the data.
What particular value do you mean, and where or how should I try (-d -1)

Tom

-----Original Message-----
From: Edzer Pebesma [mailto:edzer.pebesma at uni-muenster.de] 
Sent: donderdag 31 januari 2008 9:33
To: Hoogland, Tom
Cc: gstat-info at geo.uu.nl
Subject: Re: [Gstat-info] Block kriging odd shaped fragmented areas

Tom, it seems you use data in a local neighbourhood. Have you their
particular value? (try -d -1)
--
Edzer

Hoogland, Tom wrote:
>
> Dear all,
>
> I am trying to calculate an area averaged estimate and kriging 
> variance for an odd shaped fragmented area using ordinary block 
> kriging using the following command file:
>
> #
> data(d):'resgxgtr_sel.txt',x=1,y=2,z=3,v=10, mv=-999, min=8, max=24, 
> radius=10000, average; #
> variogram(d): 650 nug(0) + 1250 exp(17500,0,0,0,1,0.0002); #
> area: 'xyz100-mask.txt', x=1, y=2, z=3; set output = 'blockmean.txt'; 
> #
>
> The file specified in : 'xyz100-mask.txt',  is a list of centre-points
> (x,y,z) for blocks with demensions 100*100*1, together defining an odd

> shaped fragmented area.
>
> This calculation yields the folowing output:
>
> prediction results from file:  variable: col[0]
> 5
> X
> Y
> Jaar1
> pred(d)
> var(d)
>  176510 469736 1990 68.4298 274.002
>
> The calculated area average is about 68, this is strange since the 
> sample mean of the data used for kriging is 0.178.
> Kriging cell values on a mask map using the same data yields an 
> average for all cellvalues of about 12. The mask map is an asciigrid 
> with the same area used in the area-option (using: xyz100-mask.txt)
>
> It seems that I'am not calculating the area average value. Can anyone 
> tell me whats wrong?
>  
> Greetings
>
> Tom Hoogland
> Alterra, Centrum Bodem
> Tel: 0317-484251
> E-mail: Tom.Hoogland at wur.nl
> Internet: _www.grondwaterdynamiek.wur.nl_ 
> <file://www.grondwaterdynamiek.wur.nl>
>
>
> ----------------------------------------------------------------------
> --
>
> _______________________________________________
> Gstat-info mailing list
> Gstat-info at geo.uu.nl
> http://mailman.geo.uu.nl/mailman/listinfo/gstat-info
>