write_ncdf() is an S3 generic for writing R objects to a netCDF file.
Methods are provided for writing a single array-like object
(write_ncdf.default()) and a list of arrays (write_ncdf.list()).
Usage
write_ncdf(x, filename, ...)
# Default S3 method
write_ncdf(
x,
filename,
varid,
dim,
longname,
units,
prec = "float",
missval = NA,
compression = 9,
chunksizes = NA,
verbose = FALSE,
dim.names,
dim.units,
dim.longname,
unlim = FALSE,
global = list(),
force_v4 = FALSE,
...
)
# S3 method for class 'list'
write_ncdf(
x,
filename,
varid,
dim,
longname,
units,
prec = "float",
missval = NA,
compression = 9,
chunksizes = NA,
verbose = FALSE,
dim.names,
dim.units,
dim.longname,
unlim = FALSE,
global = list(),
force_v4 = FALSE,
...
)Arguments
- x
Object to write. Supported methods currently accept either a single array-like object or a list of array-like objects.
- filename
Character string giving the path to the netCDF file to create.
- ...
Additional arguments passed to methods.
- varid
Character string giving the name of the variable to create in the output file.
- dim
A named or unnamed list defining the coordinates of each dimension. Each element contains the coordinate values for one dimension. If omitted, dimensions are generated as
seq_len()for each dimension ofx.- longname
Character string giving the long name of the output variable. Defaults to
"".- units
Character string giving the units of the output variable. Defaults to
"".- prec
Character string giving the storage precision passed to
ncdf4::ncvar_def(), for example"short","integer","float"or"double".- missval
Missing value used in the netCDF variable definition.
- compression
Numeric compression level passed to
ncdf4::ncvar_def(). Compression usually requires netCDF4 output.- chunksizes
Optional chunk sizes for compressed variables.
- verbose
Logical. Should
ncdf4report progress while creating and filling the file?- dim.names
Character vector with names for the dimensions when
dimis unnamed. Ignored whendimalready has names.- dim.units
Character vector giving the units of the dimensions. If omitted, empty strings are used.
- dim.longname
Character vector giving the long names of the dimensions. If omitted, empty strings are used.
- unlim
Character string naming the unlimited dimension. Use
FALSEfor no unlimited dimension.- global
Named list of global attributes to add to the output file. A
historyattribute documenting the function call is added automatically.- force_v4
Logical. Should the output file be forced to netCDF4 format?
Details
write_ncdf.default() writes a single object x as one variable in a new
netCDF file. The dimensions of the variable are defined by dim.
If dim is unnamed, dimension names are taken from dim.names, or generated
automatically as "dim1", "dim2", and so on. Dimension units and long
names can be supplied through dim.units and dim.longname.
write_ncdf.list() writes each element of x as a separate variable in the
same netCDF file.
If varid is omitted, variable names are taken from names(x). Arguments
such as longname, units, and prec can be supplied either as length-one
values, to be recycled to all variables, or as one value per variable.
The object dim defines the full set of dimensions available in the file.
Individual variables may use all or a subset of those dimensions, provided
their sizes are consistent with the declared dimension lengths.
Examples
local({
# Synthetic daily temperature for January 2001, as in the bundled data
lon = seq(-80, -65, length.out = 15)
lat = seq(-25, -6, length.out = 20)
days = 0:30
spatial = outer(lon, lat, function(x, y)
16 + 6 * exp(-((x + 72)^2 / 16 + (y + 15)^2 / 36)) - 0.15 * (y + 15))
temperature = outer(spatial, days, function(field, day)
field + 2 * sin(2 * pi * day / 365) + 0.2 * sin(2 * pi * day / 7))
temperature[1:2, 1:2, ] = NA_real_
files = vapply(seq_len(3), function(i) tempfile(fileext = ".nc"), "")
on.exit(unlink(files))
# Quick defaults: dimensions are named dim1, dim2, dim3, with index values
write_ncdf(temperature, filename = files[1], varid = "temperature")
# Supply geographic coordinates, time coordinates, and variable metadata
coordinates = list(lon = lon, lat = lat, time = days)
coordinate_units = c("degrees_east", "degrees_north", "days since 2001-01-01")
write_ncdf(temperature, filename = files[2], varid = "temperature",
dim = coordinates, dim.units = coordinate_units,
longname = "Synthetic daily temperature", units = "degree_C")
# Add descriptive attributes, unlimited time, and explicit storage settings
write_ncdf(temperature, filename = files[3], varid = "temperature",
dim = coordinates, dim.units = coordinate_units,
dim.longname = c("Longitude", "Latitude", "Time"),
longname = "Synthetic daily temperature", units = "degree_C",
unlim = "time", prec = "float", missval = -9999,
compression = 4, force_v4 = TRUE,
global = list(
title = "Synthetic temperature for January 2001",
source = "Spatial gradient and Gaussian centre with annual and weekly cycles",
comment = "Synthetic data, not observations; a fixed corner patch is missing."
))
nc = ncdf4::nc_open(files[3])
tryCatch({
print(ncvar_dim(nc, "temperature"))
print(ncdf4::ncatt_get(nc, 0, "title"))
}, finally = ncdf4::nc_close(nc))
})
#> [1] "lon" "lat" "time"
#> $hasatt
#> [1] TRUE
#>
#> $value
#> [1] "Synthetic temperature for January 2001"
#>