Bilinear interpolation for 3d array

spInterp_bilinear(
  z,
  dims = NULL,
  range = c(70, 140, 15, 55),
  res = 1,
  na.rm = TRUE,
  convertTo2d = FALSE
)

Arguments

z

3d array (lon, lat, time), with the dimension of [nlon, nlat, ntime]. image(z[,,1]) should look normal.

dims

A list object with lon and lat.

lon: numeric, [nlon]
lat: numeric, [nlat]

range

[xmin, xmax, ymin, ymax]

res

the grid resolution (degree)

na.rm

If true, Na value in margin will be fixed.

convertTo2d

boolean (default false),

FALSE: a 3d array will be returned, [nlon, nlat, ntime]
TRUE: a 2d array will be returned, [nlon*nlat, ntime]

Note

High resolution interpolating to low resolution will lead to unreliable result, if cellsize.new/cellsize.origin > 2.

Examples

set.seed(1)
## first example: simulated data
range = c(70, 140, 15, 55)
dims = make_dims(range, res = 2)

nlat  <- length(dims$lat)
nlon <- length(dims$lon)
ntime <- 10

z <- array(rnorm(nlon*nlat*ntime), c(nlon, nlat, ntime))
z_bl  <- spInterp_bilinear(z, dims, range, res = 1)
par(mfrow = c(1, 2), mar = c(1, 1, 1, 1))
image(z[,,1])
image(z_bl[,,1])


## second example: real data
# f = system.file("inst/extdata/PMLV2_veg-dynamic_GPP_2014.tif", package = "spInterp")
# PMLV2_2014_gpp = terra::rast(f) %>% sf2::rast_array()
arr = PMLV2_2014_gpp
interp_PML <- function(arr, res = 2, plot = FALSE) {
  range = c(-180, 180, -60, 90)
  dims = make_dims(range, res = 1)
  arr_bl  <- spInterp_bilinear(arr, dims, range, res = res)
    
  if (plot) {
    par(mfrow = c(2, 1), mar = c(0, 0, 0, 0))
    image(arr[,,1])
    image(arr_bl[,,1])
  }
  arr_bl
}

arr_deg2_v0 = interp_PML(arr, res = 2, plot = TRUE)

arr_deg2_v1 = interp_PML(arr[,,1], res = 2)
all.equal(arr_deg2_v0, arr_deg2_v1)
#> [1] TRUE

arr_deg05 = interp_PML(arr, res = 0.5, plot = TRUE)

# Ipaper::write_fig({ tmp = interp_PML(arr, res = 0.5, TRUE)}, "bilinear_gpp.png", 8, 6)
# Ipaper::write_fig({ tmp = interp_PML(arr, res = 2, TRUE)}, "bilinear_gpp.png", 8, 6)