Using R: Working with Geospatial Data (and ggplot2)
Join the DZone community and get the full member experience.
Join For FreeThis post was originally written by Bethany Yollin at the Working With Data blog.
This is a follow-up blog-post to an earlier introductory post by Steven Brey: Using R: Working with Geospatial Data. In this post, we’ll learn how to plot geospatial data in ggplot2
. Why might we want to do this? Well, it’s really about your personal taste. Some people are willing to forfeit the fine-grained control of base
graphics in exchange for the elegance of a ggplot. The choice is entirely yours.
To get started, we’ll need the ggplot2
package and some data! The dataset we’ll look at are shapefiles defining watersheds in Washington state.
NOTE: Check the Department of Ecology GIS data page if any of the links are unavailable.
Loading libraries and data
# load libraries library(ggplot2) library(sp) library(rgdal) library(rgeos) # create a local directory for the data localDir <- "R_GIS_data" if (!file.exists(localDir)) { dir.create(localDir) } # download and unzip the data url <- "ftp://www.ecy.wa.gov/gis_a/inlandWaters/wria.zip" file <- paste(localDir, basename(url), sep='/') if (!file.exists(file)) { download.file(url, file) unzip(file,exdir=localDir) } # create a layer name for the shapefiles (text before file extension) layerName <- "WRIA_poly" # read data into a SpatialPolygonsDataFrame object dataProjected <- readOGR(dsn=localDir, layer=layerName)
Transforming the data
Thus far, we haven’t done anything radically different than before, but in order to prepare the data for plotting in a ggplot, we’ll have to do a couple manipulations to the structure of the data. ggplot2
will only work with a data.frame
object, so our object of class of SpatialPolygonsDataFrame
will not be appropriate for plotting. Let’s write some code and discuss why this kind of transformation is necessary.
# add to data a new column termed "id" composed of the rownames of data dataProjected@data$id <- rownames(dataProjected@data) # create a data.frame from our spatial object watershedPoints <- fortify(dataProjected, region = "id") # merge the "fortified" data with the data from our spatial object watershedDF <- merge(watershedPoints, dataProjected@data, by = "id") # NOTE : If we so choose, we could have loaded the plyr library to use the # : join() function. For those familiar with SQL, this may be a more # : intuitive way to understand the merging of two data.frames. An # : equivalent SQL statement might look something like this: # : SELECT * # : FROM dataProjected@data # : INNER JOIN watershedPoints # : ON dataProjected@data$id = watershedPoints$id # library(plyr) # watershedDF <- join(watershedPoints, dataProjected@data, by = "id")
What does all this code mean and why do we need it? Let’s go through this line by line.
dataProjected@data$id <- rownames(dataProjected@data)
Here we are appending to the data an extra column called “id”. This column will contain the rownames so that we define an explicit relationship between the data and the polygons associated with that data.
watershedPoints <- fortify(dataProjected, region = "id")
Fortify? What does that even mean? A quick search on the internet will yield some helpful documentation. (See fortify.sp documentation). Basically, fortify
take two arguments: model
, which will consist of the SpatialPolygonsDataFrame
object we wish to convert and region
, the name of the variable by which to split regions. If all goes according to plan, some magic happens and we get a data.frame
, just like we wanted… well, not quite. If you inspect this data.frame
, you’ll notice it appears to be missing some critical information. Fret not! Using the relationship we created earlier, we can merge these two datasets with the following command.
watershedDF <- merge(watershedPoints, dataProjected@data, by = "id")
And viola! Now that we’ve created a data.frame
that ggplot2
likes, we can begin plotting. Before we get to plotting, let’s take a quick look at this new data.frame
we’ve created.
head(watershedDF)
## id long lat order hole piece group WRIA_ID WRIA_NR WRIA_AREA_ ## 1 0 2377934 1352106 1 FALSE 1 0.1 1 62 789790 ## 2 0 2378018 1352109 2 FALSE 1 0.1 1 62 789790 ## 3 0 2382417 1352265 3 FALSE 1 0.1 1 62 789790 ## 4 0 2387199 1352434 4 FALSE 1 0.1 1 62 789790 ## 5 0 2387693 1352452 5 FALSE 1 0.1 1 62 789790 ## 6 0 2392524 1352623 6 FALSE 1 0.1 1 62 789790 ## WRIA_NM Shape_Leng Shape_Area ## 1 Pend Oreille 983140 3.44e+10 ## 2 Pend Oreille 983140 3.44e+10 ## 3 Pend Oreille 983140 3.44e+10 ## 4 Pend Oreille 983140 3.44e+10 ## 5 Pend Oreille 983140 3.44e+10 ## 6 Pend Oreille 983140 3.44e+10
Your first ggplot
If you’re coming from base
graphics, some of the syntax may appear intimidating, but’s it’s all part of the “grammar of graphics” after which ggplot2
is modeled. You’ll notice a graph is built layer by layer, beginning with the data and the mapping of data to “aesthetic attributes”. We’ll add “geoms” or geometric objects and perhaps we’ll compute some statistics. We may also want to adjust the scale or coordinate system. All this can be added in a very modular fashion; this is one of the key advantages to using ggplot2
. So, enough talk, let’s make a plot!
Published at DZone with permission of Jonathan Callahan, DZone MVB. See the original article here.
Opinions expressed by DZone contributors are their own.
Comments