Importing the Spatial Data in R

To import OSM spatial data into R, you can either make direct queries in the OSM’s API or you can download an OSM file an then open it into R. The first alternative only operates for small bounding boxes (0.25 x 0.25 coordinates range), whereas the second alternative allows bigger bounding boxes. For either alternative, we will need the package osmar.

When importing OSM spatial data, the key is knowing which file format you are dealing with. In this tutorial we will work with .osm, the native OSM file extension and geoJSON, the default file extension in the platform Overpass Turbo. We will open .osm files with the package osmar, whereas we will open .geoJSON files with the package rgdal.

library(osmar)

## Loading required package: XML

## Loading required package: RCurl

## Loading required package: geosphere

##
## Attaching package: 'osmar'

## The following object is masked from 'package:utils':
##
##     find

library(rgdal)

## Loading required package: sp

## rgdal: version: 1.4-8, (SVN revision 845)
##  Geospatial Data Abstraction Library extensions to R successfully loaded
##  Loaded GDAL runtime: GDAL 2.2.3, released 2017/11/20
##  Path to GDAL shared files: C:/Users/rente/OneDrive/Documents/R/win-library/4.0/rgdal/gdal
##  GDAL binary built with GEOS: TRUE
##  Loaded PROJ.4 runtime: Rel. 4.9.3, 15 August 2016, [PJ_VERSION: 493]
##  Path to PROJ.4 shared files: C:/Users/rente/OneDrive/Documents/R/win-library/4.0/rgdal/proj
##  Linking to sp version: 1.4-1

A. Importing directly from an API

We will use the Open Street Map API, which URL is https://api.openstreetmap.org/api/0.6/ (an alternative is this mirror: http://overpass-api.de/api/). As this alternative has a limited bounding box span, first, we will define our bounding box. In this example, I am creating an object with the bounding box for the Washington Park in Albany, NY by using the function corner_bbox().

box.wpark = corner_bbox(-73.7774,42.6530,-73.7652,42.6623)

Next, I will create an object with the API’s URL with the function function osmsource_api().

api.wpark = osmsource_api(url = "https://api.openstreetmap.org/api/0.6/")

The function get_osm will download the map wpark for the bounding box box.wpark from the repository api.wpark.

wpark = get_osm(box.wpark, source = api.wpark)
class(wpark)

## [1] "osmar" "list"

You can now manage the map in R as an OSM object. However, this object is not rendered yet as a map (try, for example, typing plot(wpark)). A great value of OSM and GeoJSON files is that they comprise multiple layers (e.g., points, lines and polygons) in the same file, but this complicates the processing of maps in R. Every layer should be in a separate object. Therefore, to effectively use OSM data, we need to transform the .osm file into a spatial object. To do so, we must use the function as_sp().

area.wpark   = as_sp(wpark,what="polygons") # Extract the polygons
street.wpark = as_sp(wpark,what="lines")    # Extract the lynes
attr.wpark   = as_sp(wpark,what="points")   # Extract the points

Now we have spatial data. As an example, I will plot the three types of spatial data that were residing in the .osm:

class(area.wpark)

## [1] "SpatialPolygonsDataFrame"
## attr(,"package")
## [1] "sp"

plot(area.wpark,   col = "gray")

plot(area.wpark,   col = "gray")
plot(street.wpark, add = TRUE, col = "blue")

plot(area.wpark,   col = "gray")
plot(attr.wpark,   add = TRUE, col="red", pch = 1)

B. Importing from OSM files

To follow this example, you must download into your computer the follwing file https://github.com/crenteriam/crenteriam.github.io/blob/master/files/tutorials/tutorial-osm/_albany.osm.

Once you have downloaded the file, set the working directory and open the file by using the function osmsource_file().

source.albany=osmsource_file("_albany.osm")

Next, we wil ltransform the file into an osmar object. In this case, since the file already contains the bounding box, we replace the information of the bounding box by complete_file() and transform the osmar object into spatial objects.

albany = get_osm(complete_file(), source = source.albany)
area.albany = as_sp(albany, what = "polygons")
street.albany = as_sp(albany,what = "lines")

Plot the results.

plot(area.albany)
plot(street.albany,add=TRUE,col="green")

C. Importing from GeoJSON files¹

To follow this example, you must download into your computer the follwing file https://github.com/crenteriam/crenteriam.github.io/blob/master/files/tutorials/tutorial-osm/_highways.geojson.osm. This file was downloaded from Overpass Turbo (OT).

In this case, you will need the rgdal package. In the firs line below, I inspect the structure of the GeoJSON file with ogrListLayers(). In the second line I inspect the attributes for each layer by using ogrInfo(). You need to specify the file, the extension (“OGRGeoJSON”). The option wkbLineString provides the attributes for the lines layer.

ogrListLayers("highways.geojson")
ogrInfo("highways.geojson", "OGRGeoJSON", require_geomType = "wkbLineString")

# The alternative layers are require_geomType="wkbPoint" OR require_geomType="wkbPolygon"

Then, I will convert this file into a spatial object. To do so, I will use the function readOGR, I will define the object highways.geojson as the source, "OGRGeoJSON" to tell R that the source is a GeoJSON file and require_geomType="wkbLineString" to extract only the lines layer.

highways = readOGR("highways.geojson", "OGRGeoJSON", require_geomType = "wkbLineString")
class(highways)

D. Export to Shapefile.

With the information provided above you would be allowed to do your spatial analysis. However, if you need to export the objects as ESRI’s Shapefiles, use one of the two following lines:²

write a new shapefile with package rgdal (including .prj component)

writeOGR(highways, ".", "highways_prj", driver="ESRI Shapefile")