#1. Transform data to log-concentrations and calculate the empirical variogram

loglead<-lead
loglead$data<-log(lead$data)
u<-5000*(1:25)
V.loglead<-variog(loglead,uvec=u)
plot(V.loglead)

#2. Fit  theoretical variogram ``by eye'' assuming (by default) an exponential correlation function

result<-eyefit(V.loglead)
result
#    cov.model sigmasq      phi              tausq kappa kappa2   practicalRange
#1 exponential    0.19 29729.73 0.0213133527731625  <NA>   <NA> 89062.3116429668


#3. Fit  theoretical variogram by maximum likelihood

MLE<-likfit(loglead,ini.cov.pars=c(0.19,29730))