I need to substitute or interpolate the NA in a vector of each cell from a rasterstack data. I have two functions fun_sub for substituting NA and fun_interp for interpolating NA. I found fun_sub works very well. But fun_interp does not work but I cannot find a reason. Thank you very much
Tianyi Zhang
Below is a simple example:
#------EXAMPLE
library(timeSeries)
library(raster)
fun_sub <- function(x) {
# substitute the NA to the mean of vector for each cell
v=as.vector(x)
z<-substituteNA(v,type="mean")
return (z)
}
fun_interp <- function(x) {
# interpolate the NA to the the linear regression of vector for each cell
v=as.vector(x)
z=interpNA(v, method="linear")
return (z)
}
# create data
r<-raster(ncols=2,nrows=2)
r1<-r; r2<-r; r3<-r; r4<-r
r1[]<-c(1,1,1,2)
r2[]<-c(5,5,NA,5)
r3[]<-c(3,3,4,2)
r4[]<-c(6,5,5,2)
s<-stack(r1,r2,r3,r4)
# try the two functions; the task is change the NA in r2 to a number;
res_sub<-calc(s,fun_sub) # works great! substitute the NA to the mean of c(1,NA,4,5); I got c(1,3.333,4,5)
res_inter<-calc(s,fun_interp) # cannot interpolate; have an error, don't know the reason; I expected it is c(1, 2.5 ,4, 5). But it returns an error
# try whether interpNA() can work or not
interpNA(c(1,NA,4,5),method="linear") # but this function is OK.
I would go this way using standard functions:
# create data
library(raster)
r1 <- r2 <- r3 <- r4 <-raster(ncol=2,nrow=2)
r1[] <- c(1,1,1,2)
r2[] <- c(5,5,NA,5)
r3[] <- c(3,3,4,2)
r4[] <- c(6,5,5,2)
s < -stack(r1,r2,r3,r4)
m <- mean(s, na.rm=TRUE)
r_sub <- cover(s, m)
r_int <- approxNA(s)
values(s)
## layer.1 layer.2 layer.3 layer.4
##[1,] 1 5 3 6
##[2,] 1 5 3 5
##[3,] 1 NA 4 5
##[4,] 2 5 2 2
values(r_sub)
## layer.1 layer.2 layer.3 layer.4
##[1,] 1 5.000000 3 6
##[2,] 1 5.000000 3 5
##[3,] 1 3.333333 4 5
##[4,] 2 5.000000 2 2
values(r_int)
## layer.1 layer.2 layer.3 layer.4
##[1,] 1 5.0 3 6
##[2,] 1 5.0 3 5
##[3,] 1 2.5 4 5
##[4,] 2 5.0 2 2
To use the functions from the timeSeries package
library(timeSeries)
library(raster)
New data .At least 6 cells needed for calc to test the function, that is an issue.
r1 <- r2 <- r3 <- r4 <-raster(ncol=3,nrow=2)
r1[] <- c(1,1,1,2,1,1)
r2[] <- c(0,5,NA,5,1,1)
r3[] <- c(3,3,4,2,1,1)
r4[] <- c(6,5,5,2,1,1)
s <- stack(r1,r2,r3,r4)
values(s)
values(calc(s, function(v) substituteNA(v,type="mean")))
values(calc(s, function(v) interpNA(v, method="linear")))
values(s)
## layer.1 layer.2 layer.3 layer.4
##[1,] 1 0 3 6
##[2,] 1 5 3 5
##[3,] 1 NA 4 5
##[4,] 2 5 2 2
##[5,] 1 1 1 1
##[6,] 1 1 1 1
values(calc(s, function(v) substituteNA(v,type="mean")))
## layer.1 layer.2 layer.3 layer.4
##[1,] 1 0.000000 3 6
##[2,] 1 5.000000 3 5
##[3,] 1 3.333333 4 5
##[4,] 2 5.000000 2 2
##[5,] 1 1.000000 1 1
##[6,] 1 1.000000 1 1
values(calc(s, function(v) interpNA(v, method="linear")))
## layer.1 layer.2 layer.3 layer.4
##[1,] 1 0.0 3 6
##[2,] 1 5.0 3 5
##[3,] 1 2.5 4 5
##[4,] 2 5.0 2 2
##[5,] 1 1.0 1 1
##[6,] 1 1.0 1 1