Lecture 17 - Geospatial Data Part One
Miao Yu
2024-03-18
Vector vs Raster data
We will briefly cover the two main types of geospatial data - vector and raster.
In this module, we will start with vector data.
Creating maps from okcounty
To generate a map of counties using ggplot() with a sf
object, the geom_sf() function is used. This map shows the
borders of all counties in Oklahoma.
ggplot(data = okcounty) +
geom_sf(fill = NA)
Exercise: Try what will happen without
fill=NA in the geom_sf function.
Plot tornoda touchdown places into the county map
Now let’s plot the tornoda touchdown point tpoint into
the county map. They will be overlapped by the geospatial information
(longitude and latitude).
ggplot() +
geom_sf(data = okcounty, fill = NA) +
geom_sf(data = tpoint_16_21) +
theme_bw()
Plot path of tornadoes
Now let’s work on tpath and plot the paths of each
tornado in red lines.
ggplot() +
geom_sf(data = okcounty, fill = NA) +
geom_sf(data = tpath_16_21,
color = "red",
size = 1) +
theme_bw()
Plot tornado map by year using facet_wrap()
ggplot() +
geom_sf(data = okcounty, fill = NA, color = "gray") +
geom_sf(data = tpoint_16_21, size = 0.75) +
facet_wrap(vars(yr), ncol = 2) +
coord_sf(datum = NA) +
theme_void()
Choropleth Maps
A typical way to display geospatial data is as a choropleth map, where summary statistics for each unit region are displayed as different colors. In the tornado example, we can do
ggplot(data = countymap) +
geom_sf(aes(fill = tdens)) +
theme_void()
Graduated symbol map
ggplot() +
geom_sf(data = okcntrd, aes(size = tcnt)) +
geom_sf(data = okcounty, fill = NA) +
theme_void()
Modifying the Appearance of the Map
Additional ggplot2 functions can be added to improve the
appearance of the map. The scale_fill_distiller() function
allows the specification of a different color ramp. This example uses
the “YlOrRd” palette from the RColorBrewer package. The
pretty_breaks() function from the scales
package is used to automatically select a set of breaks for the legend.
The legend is moved to the bottom of the map to better accommodate the
longer legend title.
ggplot(data = countymap) +
geom_sf(aes(fill = tdens)) +
scale_fill_distiller(name = expression("Tornadoes/1000 km"^2),
palette = "YlOrRd",
breaks = pretty_breaks(),
direction = 1) +
theme_void() +
theme(legend.position = "bottom")
Palettes
The RColorBrewer package provides a selection of
palettes designed for choropleth mapping (Harrower and Brewer 2003). The
display_brewer_all() function generates a chart with
examples of all the available palettes.
display.brewer.all()
The top palette group is suitable to map ordered data or numeric data.
The middle palette group is suitable to map categorical data without an order.
The bottom palette group is suitbale to indicate values that are above or below the mean or to highlight values that are higher or lower than zero.
Preview paletts for a given finite number of colors
display.brewer.pal(5, "YlGnBu")
Plot tdens_c1
ggplot(data = countymap) +
geom_sf(aes(fill = tdens_c1)) +
scale_fill_brewer(name = expression("Tornadoes/1000 km"^2),
palette = "YlOrRd") +
theme_void() +
theme(legend.position = "bottom") 
Another example
Similar to choropleth maps, graduated symbol maps are often easier to interpret if they include a limited number of symbol sizes.
maxcnt <- max(okcntrd$tcnt)
brkpts <- c(0, 2, 5, 10, maxcnt)
okcntrd <- okcntrd %>%
mutate(tcnt_c1 = cut(tcnt,
breaks = brkpts,
include.lowest = T))
ggplot(data = okcntrd) +
geom_sf(aes(size = tcnt_c1)) +
scale_size_discrete(name="Tornadoes") +
geom_sf(data = okcounty, fill = NA) +
theme_void() +
theme(legend.position = "bottom")
Save multiple images in a grid
Saved graphs and maps can also be combined into a composite figure
using the cowplot package (Wilke 2020). The
plot_grid() function provides a variety of options for
arranging figures in a regular grid. Below is an example.
plot_grid(choropleth, gradsymbol,
labels = c("A) Choropleth Map",
"B) Graduated Symbol Map"),
label_size = 12,
ncol = 1,
scale = 0.8,
hjust = 0,
vjust = 1.5,
label_x = 0,
align = "v")