Skip to contents

obs_foot_flip

Source: R/12_obs_nc.R
obs_foot_flip.Rd

Reorders the axes of a 3-D footprint array produced by obs_traj_foot so that its dimension order matches the CF-convention expected by obs_nc / terra::rast(), or converts back in the opposite direction.

Usage

obs_foot_flip(arr, from = c("traj", "cf"), flip_lat = TRUE, flip_lon = TRUE)

Arguments

arr

A 3-D numeric array. Two layouts are accepted:

  • "traj" (default) — [lat, lon, time] as produced by obs_traj_foot. Converted to [lon, lat, time].

  • "cf" — [lon, lat, time] (CF / terra layout). Converted back to [lat, lon, time].

from

Character string, either "traj" or "cf". Describes the current axis order of arr. Default "traj".

flip_lat

Logical. If TRUE (default) the latitude axis is also reversed so that index 1 corresponds to the southern-most row, matching the south-to-north order expected by CF tools and terra. Set to FALSE if the array is already in south-to-north order.

flip_lon

Logical. If TRUE (default) the longitude axis is reversed so that index 1 corresponds to the western-most column. Set to FALSE if the array is already in west-to-east order.

Value

A 3-D array with reordered (and optionally flipped) axes.

Details

Why this is needed. obs_traj_foot returns an array with dimensions [lat, lon, time] (R row-major: row = latitude index). NetCDF files written with obs_nc preserve that order, but terra::rast() follows the CF / GIS convention where the first spatial dimension is longitude (x-axis). Reading the file back with terra therefore transposes the spatial plane, which is why a t() call was needed on the raster.

Calling obs_foot_flip(arr) before passing the array to obs_nc (or after reading it back) resolves the mismatch without manual transposing.

See also

obs_traj_foot, obs_nc

Examples

if (FALSE) { # \dontrun{
# foot.arr comes from obs_traj_foot: dims [lat, lon, time]
foot_cf <- obs_foot_flip(foot.arr)       # now [lon, lat, time]

obs_nc(
  lat      = lats,
  lon      = lons,
  time_nc  = time_vec,
  vars_out = "foot1",
  units_out = "(ppb/nanomol m-2 s-1)",
  nc_out   = "output.nc",
  larrays  = list(foot1 = foot_cf)
)
# terra::rast("output.nc") now reads correctly without t()
} # }