In Chrisjb/rgeoportal: Fetch UK Boundary Files

knitr::opts_chunk$set(echo = TRUE,
                      fig.path = "man/figures/README-")
library(dplyr)
library(sf)

Installing rgeoportal

Install from github using devtools:

library(devtools)
install_github('Chrisjb/rgeoportal')

get_boundary

get_boundary(boundary_type = c("administrative", "census", "electoral",
                               "eurostat", "health", "other", "postcodes"),
             boundary_name = NA,
             bbox = NA,
             redius = NA,
             point = NA,
             names_like = NA,
             names_equal = NA,
             custom_polygon = NA)

boundary_type

We can get the boundary names available within each boundary type by specifying boundary_type with no boundary_name option:

library(rgeoportal)
census_boundaries <- get_boundary(boundary_type = 'census')

head(census_boundaries)

administrative <- get_boundary(boundary_type = 'administrative')

head(administrative)

Some of the boundary names come with a range of suffixes:

BFC: Full resolution - clipped to the coastline (Mean High Water mark).
BFE: Full resolution - extent of the realm (usually this is the Mean Low Water mark but in some cases boundaries extend beyond this to include off shore islands).
BGC: Generalised (20m) - clipped to the coastline (Mean High Water mark).
BUC: Ultra Generalised (500m) - clipped to the coastline (Mean High Water mark). Contains both Ordnance Survey and ONS Intellectual Property Rights.

Note that most boundaries have a maximim record count of 200. Large requests will typically be limited to 200 observations.

boundary_name

We can select one of the boundary names returned from the above request and set it as the boundary_name:

merged_auth <- get_boundary(boundary_type = 'census',
                                  boundary_name='Census_Merged_Local_Authority_Districts_December_2011_Boundaries')

head(merged_auth)

library(ggplot2)
ggplot() + 
  geom_sf(data=merged_auth)

Specifying the request

In general we don't want to request more data than we need. Many boundaries also limit the amount of polygons any given request will return to 200. There are several ways to request only the subset of the total boundary data we need:

bbox: return only results in the specified bbox
radius: return only results within a given distance of a specified point
names_like: return only results where the polygon name contains the given string(s)
names_equal: return only results where the polygon name matches the given string(s)
custom_polygon: return only results in the specified polygon

bbox

One way to request a subset of boundary data is to add a bbox to the request. We can set a bounding box in the form 'xmin,ymin,xmax,ymax' or as an sf bbox as generated by sf::st_bbox().

One way to get a bounding box is to use st_bbox on another set of polygons:

my_auths <- merged_auth %>% filter(cmlad11nm %in% c('Wandsworth'))
my_bbox <- st_bbox(my_auths)

msoa_in_bbox <- get_boundary(boundary_type = 'census',
                             boundary_name='Middle_Super_Output_Areas_December_2011_Boundaries',
                             bbox = my_bbox)

head(msoa_in_bbox)

ggplot() + 
  geom_sf(data=msoa_in_bbox) +
  geom_sf(data = st_as_sfc(my_bbox), colour = 'black', fill=NA)

We can also specify a bbox from a user specified string:

msoa_in_bbox_2 <- get_boundary(boundary_type = 'census',
                             boundary_name='Middle_Super_Output_Areas_December_2011_Boundaries',
                             bbox = '-2.65,51.41,-2.52,51.50')

head(msoa_in_bbox)

ggplot() + 
  geom_sf(data=msoa_in_bbox_2) +
  coord_sf(xlim = c(-2.65,-2.52),  ylim =c(51.41, 51.50))

radius

Another way to filter the request is to get boundaries within a set radius of a given point. If radius is set, point must be set too.

point should be a longitude latitude point in the form 'lng,lat' radius is specified in meters

ward_in_radius <- get_boundary(boundary_type = 'admin',
                             boundary_name='Wards_May_2019_Boundaries_UK_BFC',
                             point = '-2.58791,51.45451',
                             radius = 2000)

names_like

If we don't want to list the exact names of the wards, we can do partial matching by specifying names_like. This is especially useful for boundaries where the area name forms a predictable pattern such as msoa names which all start with their local authority name.

Specifying names_like will return polygons that (partially) match the name of the polygon

msoa_dartford <- get_boundary(boundary_type = 'census',
                             boundary_name='Middle_Super_Output_Areas_December_2011_Boundaries',
                             names_like = c('Dartford'))

head(msoa_dartford)

names_equal

If we know the exact name(s) of the polygon we want to return, we should specify names_equal:

dg <- get_boundary(boundary_type = 'admin',
                             boundary_name='Local_Authority_Districts_December_2019_Boundaries_UK_BFC',
                             names_equal = c('Dartford', 'Gravesham'))

head(dg)

custom_polygon

There is also the option to use a custom polygon and return boundaries that intersect the given polygon.

We may need to simplify the polygon before sending the request:

custom_poly <- ward_in_radius %>% filter(wd19nm == 'Clifton')
custom_poly_simp <- rmapshaper::ms_simplify(custom_poly)

lsoa_in_poly <- get_boundary(boundary_type = 'census',
                             boundary_name='Lower_Super_Output_Areas_December_2011_Boundaries',
                             custom_polygon = custom_poly_simp)


ggplot() +
  geom_sf(data = st_intersection(lsoa_in_poly, custom_poly)) +
  geom_sf(data =custom_poly, color = 'red', fill=NA )