unable to find an inherited method for function ‘krige’ for signature ‘"lm", "SpatialPointsDataFrame"’

[Dave Dauphine]

Hi!

Can anyone tell me why I'm getting this "Error in (function (classes, fdef, mtable)"?

I see something about a second "grid" database in the R examples. Do I need to create this? So far my target variables, predictors, and coordinates are all in the same database. 

Or do I need to specify a coordinate system (CRS)?

Thanks!

library(gstat)

library(maptools)

clean3145 = read.csv(file="clean3145.csv")

coordinates(clean3145)=~UTMEM+UTMNM

summary(clean3145)

#Variogram of LM residuals

logAs<-lm(logAs1 ~ Elev_m + Basin, clean3145)

residuals(logAs)

variogram(residuals(logAs)~UTMEM+UTMNM, clean3145)

plot(variogram(residuals(logAs)~UTMEM+UTMNM, clean3145), plot.nu=T, pch="+")

#Fit variogram

As.rev=variogram(residuals(logAs)~UTMEM+UTMNM, clean3145)

as.rvgm=fit.variogram(As.rev, vgm(nugget=0.25, model="Exp", range=15000, sill=0.6))

plot(As.rev, as.rvgm, plot.nu=T)

str(as.rvgm)

#Regression Kriging (rk)

library(sp)

AS.rk=krige(logAs, clean3145, as.rvgm)

[Eric Holmes]

Hi Dave,

You are correct in thinking you need the grd.  This will give you the sampling grid for your kriging interpolation.  I modified your code to include a sample grid.  You may need to experiment with the resolution parameter to get it to work for you.  Good luck!

library(gstat)
library(maptools)

clean3145 = read.csv(file="clean3145.csv")
coordinates(clean3145)=~UTMEM+UTMNM
summary(clean3145)

##Create sampling grid
x.range <- as.integer(range(clean3145@coords[, 1]))
y.range <- as.integer(range(clean3145@coords[, 2]))

resolution <- 4
grd <- expand.grid(x = seq(from = x.range[1], to = x.range[2], by = resolution),
                   y = seq(from = y.range[1],to = y.range[2], by = resolution))

coordinates(grd) <- ~x + y
gridded(grd) <- TRUE

#Variogram of LM residuals
logAs<-lm(logAs1 ~ Elev_m + Basin, clean3145)
residuals(logAs)
variogram(residuals(logAs)~UTMEM+UTMNM, clean3145)
plot(variogram(residuals(logAs)~UTMEM+UTMNM, clean3145), plot.nu=T, pch="+")

#Fit variogram
As.rev=variogram(residuals(logAs)~UTMEM+UTMNM, clean3145)
as.rvgm=fit.variogram(As.rev, vgm(nugget=0.25, model="Exp", range=15000, sill=0.6))
plot(As.rev, as.rvgm, plot.nu=T)
str(as.rvgm)

#Regression Kriging (rk)
library(sp)

AS.rk=krige(logAs, clean3145, as.rvgm, newdata = grd)

--
Check out our R resources at http://www.noamross.net/davis-r-users-group.html
---
You received this message because you are subscribed to the Google Groups "Davis R Users' Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to davis-rug+...@googlegroups.com.
Visit this group at http://groups.google.com/group/davis-rug.
For more options, visit https://groups.google.com/d/optout.

--

Eric Holmes

Research Ecologist

Center for Watershed Sciences

University of California, Davis

ejho...@ucdavis.edu
(530)305-1688

[Dave Dauphine]

woohoo, thanks so much eric, D-RUG to the rescue again!

You received this message because you are subscribed to a topic in the Google Groups "Davis R Users' Group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/davis-rug/taywzPxWvu4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to davis-rug+...@googlegroups.com.

[Dave Dauphine]

Hmmm, still getting the same error for some reason!

 AS.rk=krige(logAs, clean3145, as.rvgm, newdata = grd)

Error in (function (classes, fdef, mtable)  : 

  unable to find an inherited method for function ‘krige’ for signature ‘"lm", "SpatialPointsDataFrame"’ 

[Eric Holmes]

The error looks like it is not handling the linear model in the first argument of the krige function.  It is hard to diagnose without sample data, but you could try something like this:

krige(formula = logAs1 ~ Elev_m + Basin, data = clean3145, model = as.rvgm, newdata = grd)

[Dave Dauphine]

merci encore Eric!

[Dave Dauphine]

still a similar error:   unable to find an inherited method for function ‘krige’ for signature ‘"formula", "missing"’

attached my code and data if your kindness still abounds!

dave 530 416 0418

[Dave Dauphine]

here's the code again, not sure the most recent changes were saved on the previous

[Eric Holmes]

Hey Dave,

So I am not sure if this is what you are looking for, but at least it runs and will give you something to tweak.  I removed some duplicate coordinates at the top of the code as this can cause an error in the krige function (https://stat.ethz.ch/pipermail/r-sig-geo/2012-July/015540.html).

library(gstat)
library(maptools)
library(raster)

clean3145 = read.csv("clean3145.csv")
clean3145 = clean3145[duplicated(paste(clean3145$UTMEM, clean3145$UTMNM, sep="")) != TRUE,]

coordinates(clean3145)=~UTMEM+UTMNM
summary(clean3145)

##Create sampling grid
x.range <- as.integer(range(clean3145@coords[, 1]))
y.range <- as.integer(range(clean3145@coords[, 2]))

resolution <- 4000
grd <- expand.grid(UTMEM = seq(from = x.range[1], to = x.range[2], by = resolution),
                   UTMNM = seq(from = y.range[1],to = y.range[2], by = resolution))

coordinates(grd) <- ~UTMEM + UTMNM

gridded(grd) <- TRUE

#Variogram of LM residuals
logAs<-lm(logAs1 ~ Elev_m + Basin, clean3145)
residuals(logAs)
variogram(residuals(logAs)~UTMEM+UTMNM, clean3145)
plot(variogram(residuals(logAs)~UTMEM+UTMNM, clean3145), plot.nu=T, pch="+")

#Fit variogram
As.rev=variogram(residuals(logAs)~UTMEM+UTMNM, clean3145)
as.rvgm=fit.variogram(As.rev, vgm(nugget=0.25, model="Exp", range=15000, sill=0.6))
plot(As.rev, as.rvgm, plot.nu=T)
str(as.rvgm)

class(as.rvgm)

#Regression Kriging (rk)
library(sp)

AS.rk = krige(logAs1 ~ UTMEM + UTMNM, clean3145, grd, model = as.rvgm)

plot(raster(AS.rk))

[Dave Dauphine]

Thanks so much for sticking with me Eric!

Could you get it to work? It's getting past the krige function for me now, but with a new error message after that:

AS.rk = krige(logAs1 ~ UTMEM + UTMNM, clean3145, grd, model = as.rvgm)

[using universal kriging]

> 

> plot(raster(AS.rk))

Error in plot(raster(AS.rk)) : 

  error in evaluating the argument 'x' in selecting a method for function 'plot': Error: could not find function "raster"

[Erica Rettig]

I assume raster package isn't loaded?

library(raster)

Thanks,

Erica

Erica Rettig

PhD Candidate

Graduate Group in Ecology, UC Davis

[Dave Dauphine]

silly me! 

Eric had the code to load it in there but I never installed it, and the warning message got buried in the rest of the output. Thanks Erica!

Now I'll see if it still works with a GAM instead of a LM.

[Dave Dauphine]

Now could someone help me flip the color scheme? It has high arsenic values in green and low in red. The opposite would be more intuitive, with bad=red. 

I tried to figure this out from the help files, but I'm such a rookie. 

[Erica Rettig]

Take a look at package RColorBrewer. It makes it easy to set up any color scheme you'd like.

[Dave Dauphine]

nevermind, found it, thanks again Eric and Erica!

plot(raster(AS.rk), col= colorRampPalette(c("blue", "green", "yellow", "red"))(100)) 

http://r-sig-geo.2731867.n2.nabble.com/Inverting-the-scale-in-heat-colors-td5523375.html

[Dave Dauphine]

Hi again!

I'm really happy it's working with the log linear model, but when I try to run it with my GAM I get:

> AS.rk = krige(gam3145, clean3145, grd, as.rvgm)

Error in (function (classes, fdef, mtable)  : 

  unable to find an inherited method for function ‘krige’ for signature ‘"gam", "SpatialPointsDataFrame"’

Here's the entire code:

library(mgcv)

library(gstat)

library(maptools)

library(raster)

library(sp)

#Load database and tell R which variables are coordinates

clean3145 = read.csv("clean3145.csv")

clean3145 = clean3145[duplicated(paste(clean3145$UTMEM, clean3145$UTMNM, sep="")) != TRUE,]

coordinates(clean3145)=~UTMEM+UTMNM

summary(clean3145)

#Create sampling grid. 

#Higher resolution parameter seems to REDUCE resolution

x.range <- as.integer(range(clean3145@coords[, 1]))

y.range <- as.integer(range(clean3145@coords[, 2]))

resolution <- 1000

grd <- expand.grid(UTMEM = seq(from = x.range[1], to = x.range[2], by = resolution), 

                   UTMNM = seq(from = y.range[1],to = y.range[2], by = resolution))

coordinates(grd) <- ~UTMEM + UTMNM

gridded(grd) <- TRUE

#Plot experimental variogram for GAM residuals 

gam3145<-gam(As1 ~ s(UTMEM, UTMNM) + Elev_m + Basin, family=Gamma(link=log), clean3145)

residuals(gam3145)

variogram(residuals(gam3145)~UTMEM+UTMNM, clean3145)

#Fit exponential variogram model

As.rev=variogram(residuals(gam3145)~UTMEM+UTMNM, clean3145)

as.rvgm=fit.variogram(As.rev, vgm(nugget=0.5, model="Exp", range=10000, sill=1.5))

plot(As.rev, as.rvgm, plot.nu=T)

str(as.rvgm)

class(as.rvgm)

#Regression Kriging (rk)

#This shows a map of the residuals, not arsenic itself. 

AS.rk = krige(residuals(gam3145) ~ UTMEM + UTMNM, clean3145, grd, model = as.rvgm)

plot(raster(AS.rk), col= colorRampPalette(c("blue", "green", "yellow", "red"))(100)) 

#These codes didn't work. 

AS.rk = krige(gam3145 ~ UTMEM + UTMNM, clean3145, grd, model = as.rvgm)

AS.rk = krige(gam3145 ~ UTMEM + UTMNM, clean3145, grd, model = as.rvgm)

AS.rk = krige(residuals(gam3145), clean3145, grd, model = as.rvgm)

AS.rk = krige(gam3145, clean3145, grd, as.rvgm)

[Dave Dauphine]

now I see it's not working with the linear model (logArsenic=river basin +elevation] either, just the log Arsenic values themselves (logAs1). 

That's a great start, but I'm still trying to do regression kriging. 

Happy Halloween and thanks again for all the help so far!