ggmap quickstart
For more functionality, see ggmap documentation and
https://dl.dropboxusercontent.com/u/24648660/ggmap%20u
seR%202012.pdf
1. Define location: 3 ways
- location/address
myLocation <- "University of Washington”
- lat/long
myLocation <- c(lon = -95.3632715, lat = 29.7632836)
- bounding box lowerleftlon, lowerleftlat, upperrightlon, upperrightlat
(a little glitchy for google maps)
myLocation <- c(-130, 30, -105, 50)
Part 1: Downloading the map raster
Convert location/address its lat/long coordinates:
geocode(“University of Washington”)
stamen: terrain
stamen: toner
stamen: watercolor
google: terrain google: satellite google: roadmap google: hybrid
osm*
All maps can be displayed
in black and white
color = “bw”
• If you use Rstudio:
Sometimes a plot will not
display. Increasing the size
of the plot window may
help. dev.off() prior to
plotting may also help.
• The urlonly = TRUE will
return a url that will
display your map in a web
browser. Which is pretty
cool and may be handy!
• legend=“topleft” will inset
the legend on the top left
of the map is data is
overlayed (page 2).
*Open street maps may return
a '503 Service Unavailable‘
error. This means their server
is unavailable, and you must
wait for it to become available
again.
3. Fine tune the scale of the map using zoom
The get_map function takes a guess at the zoom level, but
you can alter it:
zoom = integer from 3-21
3 = continent, 10=city, 21=building
(openstreetmap limit of 18)
If you can’t get the
map you want by
adjusting the
location/zoom
variables, the
functions designed
for the different map
sources provide more
options:
get_googlemap,
get_openstreetmap,
get_stamenmap,
get_cloudmademap
myMap <-
get_map(location=myLocation,
source=“osm", color=“bw”))
There are 2 basic steps to making a map using ggmap:
Part 1: Download
map raster
Part 2: Plot raster
and overlay data
Start by loading the package: library(ggmap)
2. Define map source, type, and color
The get_map function provides a general approach for quickly
obtaining maps from multiple sources. I like this option for exploring
different styles of maps.
myMap <- get_map(location=myLocation,
source="stamen", maptype=“watercolor", crop=FALSE)
ggmap(myMap)
This will produce a map that looks something like this
NOTE: crop = FALSE because otherwise, with stamen plots, the map is
slightly shifted when I overlay data.
There are 4 map “sources” to obtain a map raster, and each of these
sources has multiple “map types” (displayed on right).
- stamen: maptype = c(“terrain”, “toner”, “watercolor”)
- google: maptype = c(“roadmap”, “terrain”, “satellite”, “hybrid”)
- osm: open street map
- cloudmade: 1000s of maps, but an api key must be obtained
from http://cloudmade.com
The following maps show different map source/type options (except cloudmade)
The appearance of these maps may be very different depending on zoom/scale
Page 1 ggmap,
Melanie Frazier
1 . Plot the raster:
ggmap(myMap)
3. Add polygons from shp file
The shp file is imported into R using the rgdal package, and must be transformed to geographic
coordinates (latitude/longitude) on the World Geodetic System of 1984 (WGS84) datum using
the rgdal package:
library(rgdal)
shpData <- readOGR(dsn="C:\\Documents and Settings\\Watershed", layer="WS")
proj4string(shpData) # describes data’s current coordinate reference system
# to change to correct projection:
shpData <- spTransform(shpData,
CRS("+proj=longlat +datum=WGS84"))
To plot the data:
geom_polygon(aes(x = long, y = lat, group=id),
data = shpData, color ="white", fill ="orangered4",
alpha = .4, size = .2)
Part 2: Plotting the maps and data
2 . Add points with latitude/longitude coordinates:
ggmap(myMap)+
geom_point(aes(x = Longitude, y = Latitude), data = data,
alpha = .5, color="darkred", size = 3)
alpha = transparency
color = color
size = size of points
The size, color, alpha, and shape of the points can be scaled relative
to another variable (in this case estArea) within the aes function:
ggmap(myMap)+
geom_point(aes(x = Longitude, y = Latitude, size=sqrt(estArea)),
data = data, alpha = .5, color="darkred")
Additional functions can be added to control scaling, e.g.:
ggmap(myMap)+
geom_point(aes(x = Longitude, y = Latitude, size=sqrt(estArea)),
data = data, alpha = .5, color="darkred")+
scale_size(range=c(3,20))
4. Annotate figure
baylor <- get_map('baylor university', zoom = 15, maptype = 'satellite')
ggmap(baylor) +
annotate('rect', xmin=-97.11, ymin=31.54, xmax=-97.12, ymax=31.55, col="red“, fill=“white”)+
annotate('text', x=-97.12, y=31.54, label = 'Statistical Sciences', colour = I(‘red'), size = 8)+
annotate('segment', x=-97.12, xend=-97.12, y=31.55, yend=31.55,
colour=I('red'), arrow = arrow(length=unit(0.3,"cm")), size = 1.5) +
labs(x = 'Longitude', y = 'Latitude') + ggtitle('Baylor University')
size
color
Controlling size and color
scale_size(range = c(3, 20))
But, the following is better for display because
it is based on area (rather than radius)
scale_area(range=c(3,20))
Continuous variables: color gradient between
n colors, e.g.:
scale_colour_gradientn(colours =
rainbow_hcl(7))
Discrete variables, e.g.:
scale_colour_manual(values=rainbow_hcl(7))
scale_colour_manual(values=c("8" = "red",
"4" = "blue","6" = "green“)
*Use colorspace and RColorBrewer to choose
color combinations
color= outline color
fill = polygon color
alpha = transparency
size = outline size
ggmap quickstart
Page 2 ggmap,
Melanie Frazier
