Improve check_antimeridian

[jflowernet]

#check_antimeridian incorrectly determines that bounding box crossing the antimeridian does not

kiribati <- spatialgridr::get_boundary(name = "Kiribati", country_type = "sovereign")

kir_crs <- '+proj=laea +lon_0=-159.609375 +lat_0=0 +datum=WGS84 +units=m +no_defs'

kir_local_crs <- sf::st_transform(kiribati, kir_crs)

kir_local_bbox <- sf::st_bbox(kir_local_crs) |>
  sf::st_bbox() |>
  sf::st_as_sfc() |>
  sf::st_as_sf()

#returns FALSE, i.e. does not cross antimeridian (but it does!)
spatialgridr:::check_antimeridian(kir_local_bbox, 4326)
#> [1] FALSE

#check_antimeridian projects the bounding box of the polygon into epsg:4326, but results in a bounding box that is 'inverse' - this is the code it uses:
sf::sf_use_s2(TRUE)
sf::st_transform(kir_local_bbox, 4326) |> 
  sf::st_bbox() |>
  sf::st_as_sfc() |>
  plot(axes = T)

sf::st_transform(kir_local_bbox, 4326) |> 
  sf::st_bbox()
#>        xmin        ymin        xmax        ymax 
#> -146.868970  -13.789183  167.699940    7.830588

#this behaviour is only for the bbox, the polygon boundaries work ok

spatialgridr:::check_antimeridian(kir_local_crs, 4326)
#> [1] TRUE

sf::st_transform(kir_local_crs, 4326) |>
  sf::st_bbox()
#>        xmin        ymin        xmax        ymax 
#> -180.000000  -13.838333  179.998831    7.879056

#the problem is likely due to lack of nodes - when polygon is split at antimeridian, it is broken into two sets of points either side of the antimeridian which are then re-joined after projecting, but the join is the 'wrong' way since it cannot be joined across the antimeridian in 4326

#try densifying before transforming - this add nodes and fixes the problem
kir_local_bbox_dense <- terra::densify(terra::vect(kir_local_bbox), interval = 1e4) |>
  sf::st_as_sf()

spatialgridr:::check_antimeridian(kir_local_bbox_dense, 4326)
#> [1] TRUE

sf::st_transform(kir_local_bbox_dense, 4326) |> 
  sf::st_bbox()
#>        xmin        ymin        xmax        ymax 
#> -179.962616  -13.876564  179.969959    7.879078

sf::st_transform(kir_local_bbox_dense, 4326) |> 
  plot(axes = T)

#could improve check_antimeridian code using this idea: https://gis.stackexchange.com/questions/256329/determining-whether-polygon-crosses-the-antimeridian

^{Created on 2024-12-19 with reprex v2.1.1}