10. Mapping with Style

Turning up the Style

Author

Affiliation

Published

January 1, 2021

A reminder that maps can mislead without context.

“The world is complex, dynamic, multidimensional; the paper is static, flat. How are we to represent the rich visual world of experience and measurement on mere flatland?”

— Edward Tufte

“Science flies you to the moon. Religion flies you into buildings.”

— Victor Stenger

Now that you have learned the basics of creating a beautiful map in ggplot2 it is time to look at some of the more particular things you will need to make your maps extra stylish. There are also a few more things you need to learn how to do before your maps can be truly publication quality.

At this point, it is important to remember that style is not decoration; it is constraint. Maps are arguments, not illustrations. Every stylistic choice should help the reader orient themselves, interpret the data honestly, and see the intended message without confusion.

If we have not yet loaded the tidyverse let us do so.

# Load libraries
library(tidyverse)
library(scales)
library(ggspatial)

# Load Africa map
load(here::here("data", "BCB744", "africa_map.RData"))

1 Default Maps

In order to access the default maps included with the tidyverse we will use the function borders().

Borders vs Shapefiles

borders() is convenient and quick, but it is coarse and imprecise. Use it for rough context or exploratory work. For publishable maps or analysis, load proper spatial data (e.g., shapefiles or sf objects) with correct boundaries and metadata.

ggplot() +
  borders(col = "black", fill = "cornsilk", size = 0.2) + # The global shape file
  coord_equal() # Equal sizing for lon/lat

Figure 1: The built in global shape file.

Jikes! It is as simple as that to load a map of the whole planet. Usually you are not going to want to make a map of the entire planet, so let us see how to focus on just the area around South Africa.

Aspect Ratio Matters

coord_equal() preserves a 1:1 relationship between x and y units. Without it, the map can look fine but be geometrically misleading.

Here is the next iteration. What is different about the code that causes the map to look very different?

sa_1 <- ggplot() +
  borders(size = 0.2, fill = "cornsilk", colour = "black") +
  coord_equal(xlim = c(12, 36), ylim = c(-38, -22), expand = 0) # Force lon/lat extent
sa_1

Figure 2: A better way to get the map of South Africa.

That is a very tidy looking map of South(ern) Africa without needing to load any files. The extent values (xlim, ylim) are a design decision: too tight and you clip context; too loose and you lose focus. Use the scientific question to guide the bounds.

2 Specific Labels

A map is almost always going to need some labels and other visual cues. You saw in the previous section how to add site labels. The following code chunk shows how this differs if you want to add just one label at a time. This can be useful if each label needs to be different from all other labels for whatever reason. You may also see that the text labels we are creating have \n in them. When R sees these two characters together like this it reads this as an instruction to return down a line. Let us run the code to make sure you see what this means.

Label placement is one of the hardest problems in cartography. A few rules you should break only deliberately:

Avoid overlaps and do not obscure the data.
Respect visual hierarchy: water bodies should not visually overpower land, and context labels should not dominate the data.
Use style (size, colour, angle) only to encode meaning, not decoration.

Also note the distinction between manual labels (a small number of contextual labels) and data-driven labels (labels generated from your data). Manual labels are appropriate for oceans, regions, and context; data-driven labels should be automated and consistent.

sa_2 <- sa_1 +
  annotate(
    "text",
    label = "Atlantic\nOcean",
    x = 15.1,
    y = -32.0,
    size = 5.0,
    angle = 30,
    colour = "navy"
  ) +
  annotate(
    "text",
    label = "Indian\nOcean",
    x = 33.2,
    y = -34.2,
    size = 5.0,
    angle = 330,
    colour = "red4"
  )
sa_2

Figure 3: Map of southern Africa with specific labels.

3 Scale Bars

With your fancy labels added, let us insert a scale bar next. There is no default scale bar function in the tidyverse, which is why you have loaded ggspatial. This package is devoted to adding scale bars and North arrows to ggplot2 figures. There are heaps of options so you will just focus on one of them for now. It is a bit finicky so to get it looking exactly how you want it requires some guessing and checking. Please feel free to play around with the coordinates below.

Decision rules:

Include a scale bar when distances matter to interpretation.
Omit it on very small-scale maps or when it adds clutter without value.
Use a north arrow only if orientation is ambiguous or non-standard.

sa_3 <- sa_2 +
  annotation_scale(
    location = "bl", # bottom left; you can also specify "tl", "tr", etc.
    width_hint = 0.2, # relative width of the scale bar
    height = unit(0.3, "cm"), # scale bar height
    text_cex = 0.8, # text size
    pad_x = unit(0.5, "cm"), # horizontal offset
    pad_y = unit(0.5, "cm"), # vertical offset
    bar_cols = c("black", "white")
  ) +
  annotation_north_arrow(
    location = "bl",
    which_north = "true", # true north, not grid north
    pad_x = unit(3, "cm"), # adjust to approximate your original x.min/x.max
    pad_y = unit(2, "cm"), # adjust to approximate your original y.min/y.max
    style = north_arrow_fancy_orienteering
  )

sa_3

Figure 4: Map of southern Africa with labels and a scale bar.

4 Advanced Styling: Insets and Coordinate Labels

Inset maps, coordinate labels, and export are not optional niceties — they are often what distinguishes a competent map from a careless one. If your main map needs geographic context, use an inset. Here we build a simple Africa inset using borders() so it stays lightweight and reliable.

africa_inset <- ggplot() +
  borders(colour = "black", fill = "cornsilk", size = 0.2) +
  coord_equal(xlim = c(-20, 55), ylim = c(-36, 38), expand = 0)
africa_inset

We can add the inset using a grob. This is a controlled way to place an entire plot inside another plot.

africa_grob <- ggplotGrob(africa_inset)

sa_4 <- sa_3 +
  annotation_custom(
    grob = africa_grob,
    xmin = 20.9, xmax = 26.9,
    ymin = -30, ymax = -24
  )
sa_4

Map of southern Africa with labels, scale bar, and an inset map of Africa.

Now tweak coordinate labels for cartographic conventions:

sa_final <- sa_4 +
  scale_x_continuous(
    breaks = seq(16, 32, 4),
    labels = c("16°E", "20°E", "24°E", "28°E", "32°E"),
    position = "bottom"
  ) +
  scale_y_continuous(
    breaks = seq(-36, -24, 4),
    labels = c("36.0°S", "32.0°S", "28.0°S", "24.0°S"),
    position = "right"
  ) +
  labs(x = "", y = "")
sa_final

Export with intention — size, aspect ratio, and resolution must match the medium.

ggsave(
  plot = sa_final,
  filename = "figures/southern_africa_final.pdf",
  height = 6,
  width = 8,
  dpi = 300
)

Map Sanity Checklist

Clear title and legend?
Coordinate system and aspect ratio appropriate?
Labels readable and not obscuring data?
Scale bar or north arrow included only when needed?
Export size and resolution appropriate for the audience?

Common Student Errors

Over‑labelling and visual clutter
Decorative colours that imply meaning
Cropping boundaries in ways that mislead
Using default borders when precision is required

Analytic vs Illustrative Maps

Analytic maps emphasise data integrity and scale; illustrative maps prioritise narrative and clarity. Decide which you are making before you style.

Reuse

CC BY-NC-SA 4.0

Citation

BibTeX citation:

@online{smit,_a._j.2021,
  author = {Smit, A. J.,},
  title = {10. {Mapping} with {Style}},
  date = {2021-01-01},
  url = {http://tangledbank.netlify.app/BCB744/intro_r/10-mapping_style.html},
  langid = {en}
}

For attribution, please cite this work as:

Smit, A. J. (2021) 10. Mapping with Style. http://tangledbank.netlify.app/BCB744/intro_r/10-mapping_style.html.

--- date: "2021-01-01" title: "10. Mapping with Style" subtitle: "Turning up the Style" --- ```{r code-brewing-opts, echo=FALSE} knitr::opts_chunk$set( comment = "R>", warning = FALSE, message = FALSE, fig.width = 4.5, fig.height = 2.625, out.width = "75%", fig.asp = NULL, # control via width/height dpi = 300 ) ggplot2::theme_set( ggplot2::theme_minimal(base_size = 8) ) ggplot2::theme_set( ggplot2::theme_bw(base_size = 8) ) ``` ![A reminder that maps can mislead without context.](../../images/internet_truth.jpg){width=75%} > “*The world is complex, dynamic, multidimensional; the paper is static, flat. How are we to represent the rich visual world of experience and measurement on mere flatland?*” > > --- Edward Tufte > "*Science flies you to the moon. Religion flies you into buildings.*" > > --- Victor Stenger Now that you have learned the basics of creating a beautiful map in **`ggplot2`** it is time to look at some of the more particular things you will need to make your maps extra stylish. There are also a few more things you need to learn how to do before your maps can be truly publication quality. At this point, it is important to remember that **style is not decoration; it is constraint**. Maps are arguments, not illustrations. Every stylistic choice should help the reader orient themselves, interpret the data honestly, and see the intended message without confusion. If we have not yet loaded the **`tidyverse`** let us do so. ```{r code-maps-load} #| eval: true # Load libraries library(tidyverse) library(scales) library(ggspatial) # Load Africa map load(here::here("data", "BCB744", "africa_map.RData")) ``` # Default Maps In order to access the default maps included with the **`tidyverse`** we will use the function `borders()`. ::: {.callout-note appearance="simple"} ## Borders vs Shapefiles `borders()` is convenient and quick, but it is coarse and imprecise. Use it for rough context or exploratory work. For publishable maps or analysis, load proper spatial data (*e.g.*, shapefiles or **sf** objects) with correct boundaries and metadata. ::: ```{r fig-maps-world, fig.cap="The built in global shape file."} #| eval: true ggplot() + borders(col = "black", fill = "cornsilk", size = 0.2) + # The global shape file coord_equal() # Equal sizing for lon/lat ``` Jikes! It is as simple as that to load a map of the whole planet. Usually you are not going to want to make a map of the entire planet, so let us see how to focus on just the area around South Africa. ::: {.callout-warning appearance="simple"} ## Aspect Ratio Matters `coord_equal()` preserves a 1:1 relationship between x and y units. Without it, the map can look fine but be **geometrically misleading**. ::: Here is the next iteration. What is different about the code that causes the map to look very different? ```{r fig-maps-sa-1, fig.cap="A better way to get the map of South Africa."} #| eval: true sa_1 <- ggplot() + borders(size = 0.2, fill = "cornsilk", colour = "black") + coord_equal(xlim = c(12, 36), ylim = c(-38, -22), expand = 0) # Force lon/lat extent sa_1 ``` That is a very tidy looking map of South(ern) Africa without needing to load any files. The extent values (`xlim`, `ylim`) are a design decision: too tight and you clip context; too loose and you lose focus. Use the scientific question to guide the bounds. # Specific Labels A map is almost always going to need some labels and other visual cues. You saw in the previous section how to add site labels. The following code chunk shows how this differs if you want to add just one label at a time. This can be useful if each label needs to be different from all other labels for whatever reason. You may also see that the text labels we are creating have `\n` in them. When R sees these two characters together like this it reads this as an instruction to return down a line. Let us run the code to make sure you see what this means. Label placement is one of the hardest problems in cartography. A few rules you should break only deliberately: - Avoid overlaps and do not obscure the data. - Respect visual hierarchy: water bodies should not visually overpower land, and context labels should not dominate the data. - Use style (size, colour, angle) **only** to encode meaning, not decoration. Also note the distinction between **manual labels** (a small number of contextual labels) and **data-driven labels** (labels generated from your data). Manual labels are appropriate for oceans, regions, and context; data-driven labels should be automated and consistent. ```{r fig-maps-labels, fig.cap="Map of southern Africa with specific labels."} #| eval: true sa_2 <- sa_1 + annotate( "text", label = "Atlantic\nOcean", x = 15.1, y = -32.0, size = 5.0, angle = 30, colour = "navy" ) + annotate( "text", label = "Indian\nOcean", x = 33.2, y = -34.2, size = 5.0, angle = 330, colour = "red4" ) sa_2 ``` # Scale Bars With your fancy labels added, let us insert a scale bar next. There is no default scale bar function in the **`tidyverse`**, which is why you have loaded **`ggspatial`**. This package is devoted to adding scale bars and North arrows to **`ggplot2`** figures. There are heaps of options so you will just focus on one of them for now. It is a bit finicky so to get it looking exactly how you want it requires some guessing and checking. Please feel free to play around with the coordinates below. Decision rules: - **Include a scale bar** when distances matter to interpretation. - **Omit it** on very small-scale maps or when it adds clutter without value. - **Use a north arrow** only if orientation is ambiguous or non-standard. ```{r fig-maps-scale, fig.cap="Map of southern Africa with labels and a scale bar."} #| eval: true sa_3 <- sa_2 + annotation_scale( location = "bl", # bottom left; you can also specify "tl", "tr", etc. width_hint = 0.2, # relative width of the scale bar height = unit(0.3, "cm"), # scale bar height text_cex = 0.8, # text size pad_x = unit(0.5, "cm"), # horizontal offset pad_y = unit(0.5, "cm"), # vertical offset bar_cols = c("black", "white") ) + annotation_north_arrow( location = "bl", which_north = "true", # true north, not grid north pad_x = unit(3, "cm"), # adjust to approximate your original x.min/x.max pad_y = unit(2, "cm"), # adjust to approximate your original y.min/y.max style = north_arrow_fancy_orienteering ) sa_3 ``` # Advanced Styling: Insets and Coordinate Labels Inset maps, coordinate labels, and export are not optional niceties --- they are often what distinguishes a competent map from a careless one. If your main map needs geographic context, use an inset. Here we build a simple Africa inset using `borders()` so it stays lightweight and reliable. ```{r maps-mini} #| fig.cap: "Inset map of Africa (context)." africa_inset <- ggplot() + borders(colour = "black", fill = "cornsilk", size = 0.2) + coord_equal(xlim = c(-20, 55), ylim = c(-36, 38), expand = 0) africa_inset ``` We can add the inset using a grob. This is a controlled way to place an entire plot inside another plot. ```{r maps-inset} #| fig.cap: "Map of southern Africa with labels, scale bar, and an inset map of Africa." africa_grob <- ggplotGrob(africa_inset) sa_4 <- sa_3 + annotation_custom( grob = africa_grob, xmin = 20.9, xmax = 26.9, ymin = -30, ymax = -24 ) sa_4 ``` Now tweak coordinate labels for cartographic conventions: ```{r maps-final} #| fig.cap: "The final map with coordinate labels." sa_final <- sa_4 + scale_x_continuous( breaks = seq(16, 32, 4), labels = c("16°E", "20°E", "24°E", "28°E", "32°E"), position = "bottom" ) + scale_y_continuous( breaks = seq(-36, -24, 4), labels = c("36.0°S", "32.0°S", "28.0°S", "24.0°S"), position = "right" ) + labs(x = "", y = "") sa_final ``` Export with intention --- size, aspect ratio, and resolution must match the medium. ```{r maps-final-save} #| eval: false ggsave( plot = sa_final, filename = "figures/southern_africa_final.pdf", height = 6, width = 8, dpi = 300 ) ``` ::: {.callout-note appearance="simple"} ## Map Sanity Checklist - Clear title and legend? - Coordinate system and aspect ratio appropriate? - Labels readable and not obscuring data? - Scale bar or north arrow included only when needed? - Export size and resolution appropriate for the audience? ::: ::: {.callout-warning appearance="simple"} ## Common Student Errors - Over‑labelling and visual clutter - Decorative colours that imply meaning - Cropping boundaries in ways that mislead - Using default borders when precision is required ::: ::: {.callout-note appearance="simple"} ## Analytic vs Illustrative Maps Analytic maps emphasise data integrity and scale; illustrative maps prioritise narrative and clarity. Decide which you are making before you style. :::