Crossing the international date line

[yvanrichard]

Hi. I can't manage to center a map with tiles when data are crossing the international date line.

In the example below, the points are split and placed on opposite sides of the world. When the layer_spatial line is above annotation_map_tile, the centering of the map around the data points is right, but the tiles above 180 degrees of longitude are missing (and it also takes much longer).

Is there a way to do it right?

library(sf)
library(ggspatial)
library(ggplot2)


## Create some data
pts <- st_as_sf(data.frame(lat = seq(-52, -32, length.out = 50),
                           lon = seq(165, 200, length.out = 50)),
                coords = c('lon', 'lat'), crs = 4326)
## Map
g <- ggplot() +
    annotation_map_tile(type = "https://server.arcgisonline.com/ArcGIS/rest/services/Ocean_Basemap/MapServer/tile/${z}/${y}/${x}.jpg", zoomin = 0) +
    layer_spatial(data = pts)
g

Cheers

[paleolimbot]

The answer is...not yet! There's a chance that the "ggmap" package has this figured out...it's a hard problem!

[yvanrichard]

Thanks. That's unfortunate. I would have thought with the caching and indexing of individual tiles it would be easier for ggspatial to shift the western tiles to the right of the eastern ones. I haven't had much luck with ggmap either, but maybe I should persevere.

[paleolimbot]

Transferring to rosm because this package has the implementation for tile wrapping. Currently tiles are resolved using a min/max approach, but should really be resolved using something more like a intersection approach so that osm.raster() can properly detect situations in which tiles should be wrapped (osm.plot() does this already, so it shouldn't be too bad!)

[mdsumner]

a GDAL api will help solve this, but there's some issues with auto-overview detection just fyi

OSGeo/gdal#7019 (comment)