In hypertidy/ceramic: Download Online Imagery Tiles

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%", 
  dev = "png"
  #, dev.args = list( jpeg = list(quality =50))
)
options(warn = -1)

ceramic

The goal of ceramic is to obtain web map tiles. Use a spatial object to define the region of interest.

library(ceramic)
roi <- ext(100, 160, -50, 10)
im <- cc_location(roi)

plotRGB(im)

The terra package is always loaded by ceramic, so we can assume the use of its functions, ceramic accepts a wider range of inputs than terra does however.

We can use wk, geos, terra, raster, sp, sf, or stars objects, or an input lon,lat point and a buffer (in metres) to define an extent. This provides a very easy way to obtain imagery or elevation data for any almost any region using our own data.

sql <- "SELECT shapeGroup FROM geoBoundariesCGAZ_ADM0 WHERE shapeGroup IN ('BOL')"
dsn <- "/vsizip//vsicurl/https://github.com/wmgeolab/geoBoundaries/raw/main/releaseData/CGAZ/geoBoundariesCGAZ_ADM0.zip"
bol <- vect(dsn, query = sql)

im <- cc_location(bol)
plotRGB(im)

Even if the data uses a map projection it will be converted into a region to match the Mercator extents used by Mapbox image servers.

options(warn = -1)

data("nz", package = "spData")

library(sf)
im_nz2 <- cc_location(nz)
plotRGB(im_nz2)
plot(st_transform(nz, crs(im_nz2))[0], add = TRUE, col = rainbow(nrow(nz), alpha = 0.5))

There are basic heuristics to decide if data is projected or just in "longitude,latitude" in the usual way.

Raster elevation data is also available.

north <- nz[nz$Island == "North", ]
dem_nz <- cc_elevation(north, type = "elevation-tiles-prod" )


## plot elevation data for NZ north
dem_nz[!dem_nz > 0] <- NA

plot(dem_nz, col = grey.colors(128))
plot(st_transform(st_cast(north, "MULTILINESTRING")["Name"], terra::crs(dem_nz)), add = TRUE, lwd = 5)

I thought you said tiles?

Indeed, here's a function called read_tiles(), it shares the same interface as get_tiles().

read_tiles()

Note that, the cc_location() and cc_elevation() functions no longer use tiles, they read directly from the internet using GDAL and are not related to the tile download facilities.

But, they used to run get_tiles() behind the scenes. The separation is still a little unfinished, but I want ceramic to have separation of loading data from the internet with downloading tiles.

This function and its counterparts get_tiles_zoom(), get_tiles_dim() and get_tiles_buffer() will only download files.

tile_summ <- get_tiles_zoom(north, zoom = 8)
length(tile_summ$files)
str(tile_summ$tiles)

(tile_rect <- tiles_to_polygon(ceramic_tiles(zoom = 8)))

This is really for expert use when you want to control the downloaded tile files yourself directly.

Providers

The default map provider is Mapbox, but ceramic is written for general usage and also provides access to the joerd AWS tiles via the type = "elevation-tiles-prod" argument.

pt <- cbind(175.6082, -37.994)
nz_z12 <- cc_location(pt, buffer  = 100000,  type = "elevation-tiles-prod")

north_carolina <- sf::read_sf(system.file("gpkg/nc.gpkg", package = "sf", mustWork = TRUE))
nc_image <- cc_location(north_carolina)

rowan_dem <- cc_elevation(dplyr::filter(north_carolina, NAME == "Rowan"))
rowan_dem

Installation

Install ceramic from CRAN with:

install.packages("ceramic")

You can install the development version of ceramic from Github.

## install.packages("remotes")
remotes::install_github("hypertidy/ceramic")

Set your mapbox API key with

Sys.setenv(MAPBOX_API_KEY = "<yourkey>")

Example

This complete example gets tiled imagery that we can use as real data.

The code here

• generates a bounding box in longitude-latitude
• reads the raster data using GDAL

then we look at the actual tiles involved,

• uses slippymath to find sensible tiles for the region
• downloads them to a local cache
• summarizes the tiles as a spatial (wk) object

library(ceramic)
## a point in longlat, and a buffer with in metres
pt <- cbind(136, -34)
im <- cc_location(pt, buffer = c(1e6, 5e5), type = "mapbox.satellite")
op <- par(bg = "black")
plotRGB(im)

## get the approximately matching tiles (zoom is magic here, it's all wrapped - needs thought)

tileset <- get_tiles(pt, buffer = c(1e6, 5e5))
tiles <- ceramic_tiles(zoom = tileset$tiles$zoom, type = "mapbox.satellite")
plot(tiles_to_polygon(tiles), add = TRUE, border = "white")



middle <- function(x, y) {
  x + (y - x)/2
}
text(middle(tiles$xmin, tiles$xmax), middle(tiles$ymin, tiles$ymax), lab = sprintf("[%i,%i]", tiles$tile_x, tiles$tile_y), 
     col = "yellow")

par(op)

Tasmap maps

library(ceramic)
library(terra)
template <- rast(ext(527358, 527880, 5252204, 5252704), res = .3, crs = "EPSG:32755")
ortho <- cc_location(template, type = "tasmap_orthophoto")

plot(ortho)

plot(cc_location(template, type = "tasmap_street"))

plot(cc_location(template, type = "tasmap_tasmapraster"))

plot(cc_location(template, type = "tasmap_hillshade"))

plot(cc_location(template, type = "tasmap_hillshadegrey"))
plot(cc_location(template, type = "tasmap_esgismapbookpublic"))  ## nope
plot(cc_location(template, type = "tasmap_topographic"))


plot(cc_location(template, type = "tasmap_tasmap25k"))  ## also 100k, 250k, 500k

quantize_figs <- function(dir = "man/figures") {
  f <- fs::dir_ls(dir, regexp = "png$")
  for (i in seq_along(f)) {
    tf <- tempfile(fileext = ".png")
    im <- magick::image_read(f[i])
    magick::image_write(magick::image_quantize(im), tf)

    ## I don't know how to properly unlink the original pointer, this just a guess (not sure it matters)
    rm(im)
    fs::file_delete(f[i])
    fs::file_move(tf, f[i])
  }
} 
quantize_figs()

---

Please note that the 'ceramic' project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.

hypertidy/ceramic documentation built on Feb. 28, 2024, 11:27 p.m.