Ordinary kriging using gstat

Question

I am trying to generate variogram and perform ordinary kriging on log10 transformed data in r using gstat package. My issue is on error in fitting. I have posted the code below and bubble plot of transformed rainfall data. If someone help me/show me directions on how to fix this issue? Is it because uneven distribution of sample points?

# loading necessary packages
install.packages("gstat")
library(gstat)  
install.packages("sp")
library(sp)

# give file location

setwd("C:/Users/c/Downloads/Documents")
data <- read.csv("ECtrans.csv",TRUE,",")


# fit variogram using gstat
data.vgm <- variogram(logtrans~1,data)
data.fit = fit.variogram(data.vgm, model = vgm(1, "Sph", 1000000, 1))
Error in fit.variogram(data.vgm, model = vgm(1, "Sph", 1e+06, 1)) : 
  fit.method 7 will not work with zero distance semivariances; use another fit.method value
plot(data.vgm, data.fit)
Error in h(simpleError(msg, call)) : 
  error in evaluating the argument 'y' in selecting a method for function 'plot': object 'data.fit' not found

# to do kriging
data.grid <- read.csv("newgrid.csv",TRUE,",")
coordinates(data.grid) = ~X+Y
> class(data.grid)
[1] "SpatialPoints"
attr(,"package")
[1] "sp"
gridded(data.grid) = TRUE
class(data.grid)
[1] "SpatialPixels"
attr(,"package")
[1] "sp"
data.kriged = krige(logtrans~1, data, data.grid, model = data.fit)
Error in gstat(formula = formula, data = locations, model = model, beta = beta,  : 
  object 'data.fit' not found
spplot(data.kriged["var1.pred"])
Error in h(simpleError(msg, call)) : 
  error in evaluating the argument 'obj' in selecting a method for function 'spplot': object 'data.kriged' not found
plot(data.grid)
data.dir = variogram(logtrans~1, data, alpha = c(0, 45, 90, 135))
datadir.fit = vgm(.59, "Sph", 1200, .05, anis = c(45, .4))
plot(data.dir, datadir.fit, as.table = TRUE)
data.fit
Error: object 'data.fit' not found

Answer 1

The first error is here:

data.fit = fit.variogram(data.vgm, model = vgm(1, "Sph", 1000000, 1))
Error in fit.variogram(data.vgm, model = vgm(1, "Sph", 1e+06, 1)) : 
  fit.method 7 will not work with zero distance semivariances; use another fit.method value

(the rest of your question after that is basically noise because data.fit isn't found - please try and only include relevant code up to the first error).

"Zero distance" means you have coincident points. Check the help for fit.variogram and try another value for fit.method.