In paleolimbot/geoarrow: Extension Types for Spatial Data for Use with 'Arrow'

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

geoarrow

The goal of geoarrow is to leverage the features of the arrow package and larger Apache Arrow ecosystem for geospatial data. The geoarrow package provides an R implementation of the GeoParquet file format of and the draft geoarrow data specification, defining extension array types for vector geospatial data.

Installation

You can install the released version of geoarrow from CRAN with:

install.packages("geoarrow")

You can install the development version of geoarrow from GitHub with:

# install.packages("pak")
pak::pak("geoarrow/geoarrow-r")

Example

The geoarrow package implements conversions to/from various geospatial types (e.g., sf, sfc, s2, wk) with various Arrow representations (e.g., arrow, nanoarrow). The most useful conversions are between the arrow and sf packages, which in most cases allow sf objects to be passed to arrow functions directly after library(geoarrow) or requireNamespace("geoarrow") has been called.

library(geoarrow)
library(arrow, warn.conflicts = FALSE)
library(sf)

nc <- read_sf(system.file("gpkg/nc.gpkg", package = "sf"))
tf <- tempfile(fileext = ".parquet")

nc |> 
  tibble::as_tibble() |> 
  write_parquet(tf)

open_dataset(tf) |> 
  dplyr::filter(startsWith(NAME, "A")) |>
  dplyr::select(NAME, geom) |> 
  st_as_sf()

By default, arrow objects are converted to a neutral wrapper around chunked Arrow memory, which in turn implements conversions to most spatial types:

df <- read_parquet(tf)
df$geom
st_as_sfc(df$geom)

The entry point to creating arrays is as_geoarrow_vctr():

as_geoarrow_vctr(c("POINT (0 1)", "POINT (2 3)"))

By default these do not attempt to create a new storage type; however, you can request a storage type or infer one from the data:

as_geoarrow_vctr(c("POINT (0 1)", "POINT (2 3)"), schema = geoarrow_native("POINT"))

vctr <- as_geoarrow_vctr(c("POINT (0 1)", "POINT (2 3)"))
as_geoarrow_vctr(vctr, schema = infer_geoarrow_schema(vctr))

There are a number of files to use as examples at https://geoarrow.org/data that can be read with arrow::read_ipc_file():

url <- "https://github.com/geoarrow/geoarrow-data/releases/download/v0.1.0/ns-water-basin_point.arrow"
tab <- read_ipc_file(url, as_data_frame = FALSE)
tab$geometry$type