Generate unconditional simulations for power variogram model

[Martin20494]

I have a question about how to generate unconditional simulations for power variogram model. I have tried as below:

# Calculate original variogram
var <- gstat::variogram(object=z~1, locations=<some_dataset>)
fit_var <- gstat::fit.variogram(object=var, model=gstat::vgm(range=1.8, model="Pow"))

# Generate simulations
z.pred <- gstat::gstat(formula = z ~ 1, locations = ~x + y, dummy = T, beta = 0,  model = gstat::vgm(range=1.8, model='Pow'), nmax = 10)
set.seed(1)
z.usim <- gstat::predict(z.pred, newdata = <new_some_dataset>, nsim = 50)

And I got this error

Error in predict.gstat(z.pred, newdata = newbathy_sdf, nsim = 50) : 
value not allowed for: covariance from non-transitive variogram not allowed

As I understand from the error and from here, covariance modelling and simple kriging cannot produce simulations of power variogram because there is no sill. If this is the case, do you know any other way that I can generate unconditional simulations for power variogram model with gstat?

[edzer]

I don't. @BenGraeler ?