[SOLVED] identifying a directional variogram

If you have problems using RGeostats package or if you simply need some help on an unlisted subject.

[SOLVED] identifying a directional variogram

Postby Jeffrey Yarus » Thu Jul 25, 2019 11:04 pm

I have figured out how to create directional variograms using autofitting by asking for the number of directions I would like to see. The question I have is, how do you specify the principle direction to use when creating a kriged or conditionally simulated map? In other words, my assumption was, if I ask for two directions, for example (0,90) I get the directional variograms for North 90 and East, 0. When I specify that variogram model, my result looks exactly the same as the omni-directional variogram. Oddly enough, if I specify (0, 45), I get what looks like a N-S, directional model!

Here is the chunk where I create variograms for multiple directions:

Code: Select all
#Additional to the explanation above on model.auto(), if you included many basic structures into the auto-fitting algorithm of RGeostats, model.auto() will try to fit every included structure and #pick out the best fit to plot with the experimental variogram.

data.model.omni = model.auto(vario.omni,
                             struct = c("Spherical", "Gaussian", "Exponential"),
                             title = paste(property, "Model Omnidirectional"), pos.legend=1,
                             xlab = "Log distance",
                             ylab = expression(paste("Variance (", gamma, "(h))", sep="")))

Here is the chunk where I specify an omnidirectional variogram to use in Kriging:
Code: Select all
dbgrid.omni <- dbgrid
dbgrid.omni = kriging(db, dbgrid, data.model.omni, neigh.unique, radix="NewOmni")
plot(dbgrid.omni, pos.legend=1, cex=0.7,
     title = paste(property, "Kriging with omni-directional Model"),xlab = "X (UTM)", ylab = "Y (UTM)")

Here is a chunk where I specify a directional variogram:
Code: Select all
#Using vario.calc() again, this time to calculate specific directions such as 0, 90, degrees

data.2dir.vario = vario.calc(db, nlag=10, dir=c(0,90))

Here is where I specify to plot the variogram:
Code: Select all
plot(data.2dir.vario,type = "p", pch = 19, npairpt=T, npairdw=F, pos.legend = 1, cex = 0.7,
     title = paste(property, "Experimental 3 Directional Variogram"),
     xlab = "Log distance", ylab = expression(paste("Variance (", gamma, "(h))", sep="")))

Here is where I specify the directional variogram to use in kriging...
Code: Select all
dbgrid.2dir.vario <- dbgrid
dbgrid.2dir.vario = kriging(db, dbgrid.2dir, data.model.2dir, neigh.unique, radix="New2Dir")
plot(dbgrid.2dir, pos.legend=1, cex=0.7,
     title = paste(property, "Kriging with directional-Model"),xlab = "X (UTM)", ylab = "Y (UTM)")

It appears the direction '0' is East and '90' is north (mathematical directions)
I can't past the images here, but it does seem that the omni directional kriging results and the directional results are virtually the same...

Any suggestions would be appreciated.

Jeffrey Yarus
Jeffrey Yarus
 
Posts: 48
Joined: Wed Jun 26, 2019 9:39 pm

Re: identifying a directional variogram

Postby Fabien Ors » Fri Jul 26, 2019 1:14 pm

Dear Jeffrey,
Thank you for your message !
Didier is currently preparing a reply to your question.
He will put it in this thread as soon as possible.
Best regards,
Fabien
Fabien Ors
Administrateur du site
 
Posts: 226
Joined: Thu Sep 20, 2012 1:07 pm

Re: identifying a directional variogram

Postby Fabien Ors » Sat Jul 27, 2019 8:25 am

Dear Jeffrey,
Didier has cooked a Rmd file for answering your question.
Please find it here: http://rgeostats.free.fr/doc/Examples/V ... .Model.Rmd
Hope this helps.
Best regards.
Fabien
Fabien Ors
Administrateur du site
 
Posts: 226
Joined: Thu Sep 20, 2012 1:07 pm


Return to Other Troubleshooting

Who is online

Users browsing this forum: No registered users and 10 guests