Error using tmap on large sf object in R

Question

I'm trying to use Resolve ecoregion data to map the biomes of the African continent. The shapefile is global, and there are over 800 polygons, as each polygon represents a unique ecological area (whereas there are only seven terrestrial biomes, biomes are the larger spatial unit and are composed of similar ecological areas). Perhaps because of the data size or number of polygons, I am having a difficult time producing a map using tmap.

I first cropped the shapefile to the African continent:

resolve = st_read("Data/Ecoregions2017/Ecoregions2017.shp")

Africa = st_read("Data/Africa_SHP/Africa/Africa.shp")
st_crs(Africa) = crs(resolve)

resolveAfrica =st_intersection(st_make_valid(resolve), st_make_valid(Africa))

But when I try to then map this cropped shapefile:

tm_shape(Africa) + tm_borders() +
  tm_shape(resolveAfrica) + tm_fill("BIOME_NAME")

R returns an error:

Error in vapply(lst, class, rep(NA_character_, 3)) : 
  values must be length 3,
 but FUN(X[[56]]) result is length 2

Does anyone know what steps I can take to address this error?

Answer 1

This error can arise when trying to use a tmap function that expects polygons (like tm_polygons(), tm_fill(), or tm_borders()) on an object that contains GeometryCollection geometries (i.e., a mixture of polygons, lines, points, etc. in the same geometry).

You can use sf::st_geometry_type(Africa) or sf::st_geometry_type(resolveAfrica) to confirm this is the case.

You can use sf::st_collection_extract(Africa, type = "POLYGON") or sf::st_collection_extract(resolveAfrica, type = "POLYGON") to pull out just the polygon geometries from within each of those sf objects, which should then let your tmap function calls work. Even with some of the rows of the object being LINESTRING geometries, tm_fill() and tm_borders() still seems to know what to do with them.

In case that doesn't just work, perhaps you can pull out the individual types of geometries using multiple sf::st_collection_extract() function calls and then rbind() them back together. As an example with the Africa dataset:

Africa_pts <- sf::st_collection_extract(Africa, type = "POINTS")
Africa_lines <- sf::st_collection_extract(Africa, type = "LINES")
Africa_polys <- sf::st_collection_extract(Africa, type = "POLYGONS")

Africa_combined <- rbind(Africa_pts, Africa_lines, Africa_polys)