从R的嵌套循环内的列表中提取3D netcdf变量

Question

Say I have 10 model configurations of n timesteps for 3 different sites, producing a total 30 netcdf files I want to open and manipulate. I can open the 30 files such as

require(ncdf4)

allfiles= list() 
nmod=10     
nsites=3    

for (i in 1:nmod) {
allfiles[[i]] = list(nc_open(paste0('Model',i,'siteA.nc')),
                     nc_open(paste0('Model',i,'siteB.nc')),
                     nc_open(paste0('Model',i,'siteC.nc')))
}

When querying the class of what was opened, I have

class(allfiles)

[1] "list"

class(allfiles[[1]][[1]])

[1] "ncdf4"

as expected. Now what I would like to do is extract the values from a variable in the files such that

var=list()
for (i in 1:nmod) {
  for (j in 1:nsites) {
    var[[i]][[j]] <- ncvar_get(allfiles[[i]][[j]],"var1") 
    nc_close(allfiles[[i]][[j]])
   }}

but I get this error message:

 `Error in *tmp*[[i]] : subscript out of bounds`

If I try

 var[[i]] <- ncvar_get(allfiles[[i]][[j]],"var1")

it (understandbly) only produces a list of 10 model configurations at one site, ie var[[1]][[1]][1] prints out the value of the variable at model configuration 1, site A, timestep 1 but var[[1]][[2]] doesn't exist. How can I declare var in the above loop so that it contains all the values for all models, all sites and all timesteps (eg for var[[1]][[2]][1] to exist)?

Answer 1

In your original version where the error occurs, in the first inner loop, you try to do something: var[[1]][[1]] <- something , but var[[1]] doesn't exist, and R doesn't know what to do, so I guess the following thing should work, you set var[i] <- list() before you do var[[i]][[j]] <- something :

var=list()
for (i in 1:nmod) {
var[i] <- list()
  for (j in 1:nsites) {
    var[[i]][[j]] <- ncvar_get(allfiles[[i]][[j]],"var1") 
    nc_close(allfiles[[i]][[j]])
  }
}

For example, if you do:

var <- list()
for (i in 1:10) {
    for (j in 1:10) {
        var[[i]][[j]] <- 1
    }
}

Then the same error happens. But if you set var[[i]] <- list() before carrying out the inner loop like this:

var <- list()
for (i in 1:10) {
    var[[i]] <- list()
    for (j in 1:10) {
        var[[i]][[j]] <- 1
    }
}

Then the problem will be solved.