Chapter Contents |
Previous |
Next |
The VARIOGRAM Procedure |
Using the values of LAGDISTANCE=7.0 and MAXLAGS=10 computed previously, rerun PROC VARIOGRAM without the NOVARIOGRAM option. Also, request a robust version of the semivariogram; then, plot both results against the pairwise distance of each class.
proc variogram data=thick outv=outv; compute lagd=7 maxlag=10 robust; coordinates xc=east yc=north; var thick; run; title 'OUTVAR= Data Set Showing Sample Variogram Results'; proc print data=outv label; var lag count distance variog rvario; run; data outv2; set outv; vari=variog; type = 'regular'; output; vari=rvario; type = 'robust'; output; run; title 'Standard and Robust Semivariogram for Coal Seam Thickness Data'; proc gplot data=outv2; plot vari*distance=type / frame cframe=ligr vaxis=axis2 haxis=axis1; symbol1 i=join l=1 c=blue /* v=star */; symbol2 i=join l=1 c=yellow /* v=square */; axis1 minor=none label=(c=black 'Lag Distance') /* offset=(3,3) */; axis2 order=(0 to 9 by 1) minor=none label=(angle=90 rotate=0 c=black 'Variogram') /* offset=(3,3) */; run;
Figure 70.8 shows first a slow, then a rapid rise from the origin, suggesting a Gaussian type form:
See the section "Theoretical and Computational Details of the Semivariogram" for graphs of the standard semivariogram forms.
By experimentation, you find that a scale of c0=7.5 and a range of a0=30 fits reasonably well for both the robust and standard semivariogram
The following statements plot the sample and theoretical variograms:
data outv3; set outv; c0=7.5; a0=30; vari = c0*(1-exp(-distance*distance/(a0*a0))); type = 'Gaussian'; output; vari = variog; type = 'regular'; output; vari = rvario; type = 'robust'; output; run; title 'Theoretical and Sample Semivariogram for Coal Seam Thickness Data'; proc gplot data=outv3; plot vari*distance=type / frame cframe=ligr vaxis=axis2 haxis=axis1; symbol1 i=join l=1 c=blue /* v=star */; symbol2 i=join l=1 c=yellow /* v=square */; symbol3 i=join l=1 c=cyan /* v=diamond */; axis1 minor=none label=(c=black 'Lag Distance') /* offset=(3,3) */; axis2 order=(0 to 9 by 1) minor=none label=(angle=90 rotate=0 c=black 'Variogram') /* offset=(3,3) */; run;
Figure 70.9 shows that the choice of a semivariogram model is adequate. You can use this Gaussian form and these particular parameters in PROC KRIGE2D to produce a contour plot of the kriging estimates and the associated standard errors.
Chapter Contents |
Previous |
Next |
Top |
Copyright © 1999 by SAS Institute Inc., Cary, NC, USA. All rights reserved.