Our World in Emissions | TidyTutorial

Welcome! If you saw my post for this week’s TidyTuesday, I’m glad you liked it enough to learn from it! If not, you can either scroll to the bottom to see the final product or click here. For this plot, we will use an area plot to visualize the global emissions by type going back to 1900. To start, we will use a bare-bones ggplot2 area chart with no bells or whistles to see what we are working with.

library(tidyverse)

# read data and rename an ugly column ------------------------------------------
emis<- read_csv(paste0(
  "https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/",
  "data/2024/2024-05-21/emissions.csv"
  )
)

emis <- emis |>
  rename(emissions = "total_emissions_MtCO2e")

emis |>
  group_by(year, commodity) |>
  summarise(emissions = sum(emissions), .groups = "drop") |>
  
  # start of plot --------------------------------------------------------------
  ggplot(aes(x = year, y = emissions, fill = commodity)) +
  geom_area(alpha = 0.9)

Okay, we’ve learned a lot. First, there are a lot of categories. A good rule of thumb is that once you get to about seven colors, even non-colorblind humans struggle to differentiate. But there is hope! Notice that there are several types of coal production. Let’s aggregate them. Second, there is a long tail on the left because of near-zero data. Let’s bring our limit to the right to get a better look.

emis |>
  filter(year >= 1900) |> # get rid of that tail
  mutate(
    # aggregate coal
    commodity = if_else(str_detect(commodity, "Coal"), "Coal", commodity),
  ) |>
  group_by(year, commodity) |>
  summarise(emissions = sum(emissions), .groups = "drop") |>
  
  # start of plot --------------------------------------------------------------
  ggplot(aes(x = year, y = emissions, fill = commodity)) +
  geom_area(alpha = 0.9)

Much better! But to me, having the smallest category (cement) on top feels awkward. Let’s reorder the categories! I’ll do so in descending order of emissions in the last year.

LEVS <- c("Coal", "Oil & NGL", "Natural Gas", "Cement") # our desired order

emis |>
  filter(year >= 1900) |>
  mutate(
    commodity = if_else(str_detect(commodity, "Coal"), "Coal", commodity),
    commodity = factor(commodity, levels = LEVS) # re-order 
  ) |>
  group_by(year, commodity) |>
  summarise(emissions = sum(emissions), .groups = "drop") |>
  
  # start of plot --------------------------------------------------------------
  ggplot(aes(x = year, y = emissions, fill = commodity)) +
  geom_area(alpha = 0.9)

Now we’re cooking! It’s time for some style points. I’ll use my favorite aesthetic cheat code: ggthemes. Let’s add a theme and color scheme. I’m going with the FiveThirtyEight theme and a colorblind-friendly palette. I’ll also take this opportunity to adjust the opacity down just a touch. This is a personal choice, but I find it nice to be able to see the grid behind such ink-heavy plots as area plots.

Important

Remember: unless you are making plots for a very small number of people and you know for certain that none are colorblind, making inaccessible plots is inexcusable. Of course, we all make mistakes, so if you ever notice an accessibility issue on my site, reach out and let me know on Discord or via a GitHub issue so I can improve for next time!

LEVS <- c("Coal", "Oil & NGL", "Natural Gas", "Cement") # our desired order

emis |>
  filter(year >= 1900) |>
  mutate(
    commodity = if_else(str_detect(commodity, "Coal"), "Coal", commodity),
    commodity = factor(commodity, levels = LEVS) # re-order 
  ) |>
  group_by(year, commodity) |>
  summarise(emissions = sum(emissions), .groups = "drop") |>
  
  # start of plot --------------------------------------------------------------
  ggplot(aes(x = year, y = emissions, fill = commodity)) +
  geom_area(alpha = 0.9) + # drop the opacity just a touch

  # add theme and colors (love you, ggthemes) 
  ggthemes::scale_fill_colorblind() +
  ggthemes::theme_fivethirtyeight()

And just like that, it feels like we are almost there! Let’s change a few things at once. We will change the background color, add the title/subtitle/axis labels/caption, and format the \(y\)-axis to read 30k instead of 30000. That will give us a feel for the final color scheme and how the fonts feel on the page. Because of the subscript “2” in \(CO_2\), I will use the latex2exp package use \(\LaTeX\) typesetting in the plot.

Note

One note that is unique to this plot. When we use theme_fivethirtyeight, it removes the \(y\)-axis title. So, although we normally wouldn’t have to explicitly set the axis title to element_text in the theme function, we will here.

LEVS <- c("Coal", "Oil & NGL", "Natural Gas", "Cement")
BG_COLOR <- "#F0F0F0" # this will be our background color

emis |>
  filter(year >= 1900) |>
  mutate(
    commodity = if_else(str_detect(commodity, "Coal"), "Coal", commodity),
    commodity = factor(commodity, levels = LEVS) 
  ) |>
  group_by(year, commodity) |>
  summarise(emissions = sum(emissions), .groups = "drop") |>
  
  # start of plot --------------------------------------------------------------
  ggplot(aes(x = year, y = emissions, fill = commodity)) +
  geom_area(alpha = 0.9) +

  ggthemes::scale_fill_colorblind() +
  ggthemes::theme_fivethirtyeight() +

  # abbreviate the y axis labels using the scales package
  scale_y_continuous(label = scales::label_number(scale = 1e-3, suffix = "k")) +

  # add labels to the plot -----------------------------------------------------
  labs(
    x = element_blank(),
    y = latex2exp::TeX("Emissions ($MtCO_2e$)"), # LaTeX typesetting with TeX()
    title = "Our World in Emissions",
    subtitle = latex2exp::TeX(
      paste(
        "Emissions are measured in Millions of Tons of $CO_2$ equivalent",
        "($MtCO_2e$)"
      )
    ),
    caption = paste(
      "Made with love by Mitch Harrison",
      # long blank line to "hack" a an annotation in the bottom-left corner
      "                                                                       ",
      "Source: Carbon Majors database and TidyTuesday"
    )
  ) +
  theme(
    axis.title.y = element_text(size = 10),
    plot.background = element_rect(fill = BG_COLOR) # change background color
  ) 

You could submit this plot for public consumption without shame, but we can do better! For example, I think we could safely remove the legend by annotating the colors directly on the plot. Let’s use a geom_text to do just that. While this entire process has been creative, we are getting into highly subjective territory here. So if you don’t like these changes, do something else! I would love to see your ideas.

To make the annotations, I want the text to be right-justified and directly atop one another. To accomplish that, I will give geom_text a single \(x\) value but several \(y\) values (one for each category).

LEVS <- c("Coal", "Oil & NGL", "Natural Gas", "Cement")
BG_COLOR <- "#F0F0F0"

emis |>
  filter(year >= 1900) |>
  mutate(
    commodity = if_else(str_detect(commodity, "Coal"), "Coal", commodity),
    commodity = factor(commodity, levels = LEVS) 
  ) |>
  group_by(year, commodity) |>
  summarise(emissions = sum(emissions), .groups = "drop") |>
  
  # start of plot --------------------------------------------------------------
  ggplot(aes(x = year, y = emissions, fill = commodity)) +
  geom_area(alpha = 0.9) +

  ggthemes::scale_fill_colorblind() +
  ggthemes::theme_fivethirtyeight() +
  scale_y_continuous(label = scales::label_number(scale = 1e-3, suffix = "k")) +

  # add annotation text to replace the legend ----------------------------------
  annotate(
    geom = "text",
    color = "white",
    x = 2020,
    y = c(1000, 4700, 13000, 26000),
    label = c("Cement", "Natural Gas", "Oil & NGL", "Coal"),
    hjust = "right",
    fontface = "bold"
  ) +

  labs(
    x = element_blank(),
    y = latex2exp::TeX("Emissions ($MtCO_2e$)"),
    title = "Our World in Emissions",
    subtitle = latex2exp::TeX(
      paste(
        "Emissions are measured in Millions of Tons of $CO_2$ equivalent",
        "($MtCO_2e$)"
      )
    ),
    caption = paste(
      "Made with love by Mitch Harrison",
      "                                                                       ",
      "Source: Carbon Majors database and TidyTuesday"
    )
  ) +

  theme(
    legend.position = "none", # hide the legend
    axis.title.y = element_text(size = 10),
    plot.background = element_rect(fill = BG_COLOR)
  ) 

Nailed it. Now, I will happily take criticism here. I don’t love that the “Cement” label isn’t entirely encompassed by its data. But I think it’s much cleaner than having a legend drawing our eye away from the plot, so I’ll keep it.

The last thing we have to do before we can worry about the big annotation in the middle of the plot is change where the axes break. That is, set the years and emission amount displayed on the x and y axes, respectively. And while I’m at it, I will use a geom_hline to make the \(x\)-axis a bit bolder since it melts into the background a little bit too much for my liking.

LEVS <- c("Coal", "Oil & NGL", "Natural Gas", "Cement")
BG_COLOR <- "#F0F0F0"
GRAY <- "gray35"

emis |>
  filter(year >= 1900) |>
  mutate(
    commodity = if_else(str_detect(commodity, "Coal"), "Coal", commodity),
    commodity = factor(commodity, levels = LEVS) 
  ) |>
  group_by(year, commodity) |>
  summarise(emissions = sum(emissions), .groups = "drop") |>
  
  # start of plot --------------------------------------------------------------
  ggplot(aes(x = year, y = emissions, fill = commodity)) +
  geom_area(alpha = 0.9) +

  ggthemes::scale_fill_colorblind() +
  ggthemes::theme_fivethirtyeight() +

  # change where the axis breaks occur -----------------------------------------
  scale_x_continuous(breaks = seq(1900, 2020, 20)) +
  scale_y_continuous(
    breaks = seq(0, 40000, 5000), 
    label = scales::label_number(scale = 1e-3, suffix = "k")
  ) +

  annotate(
    geom = "text",
    color = "white",
    x = 2020,
    y = c(1000, 4700, 13000, 26000),
    label = c("Cement", "Natural Gas", "Oil & NGL", "Coal"),
    hjust = "right",
    fontface = "bold"
  ) +

  labs(
    x = element_blank(),
    y = latex2exp::TeX("Emissions ($MtCO_2e$)"),
    title = "Our World in Emissions",
    subtitle = latex2exp::TeX(
      paste(
        "Emissions are measured in Millions of Tons of $CO_2$ equivalent",
        "($MtCO_2e$)"
      )
    ),
    caption = paste(
      "Made with love by Mitch Harrison",
      "                                                                       ",
      "Source: Carbon Majors database and TidyTuesday"
    )
  ) +

  geom_hline(yintercept = 0, linewidth = 0.7, color = GRAY) + # bold axis
  theme(
    legend.position = "none", # hide the legend
    axis.title.y = element_text(size = 10),
    plot.background = element_rect(fill = BG_COLOR)
  ) 

Once I write-in the line breaks, I’ll use the annotate function as before. But that’s not all. By default, there is no background with text annotations, so the grid overlaps the text and decreases legibility. To fix this, I’ll use annotate to put a rectangle the same color as the plot background behind the text, which “removes” the grid lines behind the text.

Finally, to accomplish the arrow, we will use our final annotate to draw a line segment and put an arrowhead at the end.

Note

Normally, the order that we put things in a ggplot2 pipeline doesn’t matter. But here, if you put the background rectangle after the text annotation, it will cover the text, rendering it invisible.

Because this is our last edit, I will take this opportunity to make one very oft-forgotten change: write my alt text. Since you’re here, I know you respect the power of data communication. Alt text lets us communicate with those who sometimes miss out on learning from plots online. As our color palette did for colorblind viewers, we owe it to our non-sighted friends to let them participate.

And finally, I’ll change the aspect ratio of the plot. You may have heard of the golden ratio, which is a ratio that many humans find inherently satisfying to look at. That ratio is approximately 1.618:1. The inverse of that number is 0.618, which will be our horizontal aspect ratio (1.618 is vertical). Because the quarto headers won’t render with the document, my final header is below:

#| label: plt-final
#| fig-width: 8
#| fig-align: "center"
#| fig-asp: 0.618
#| fig-alt: |
#|   This plot is titled Our World in Emissions. It is an area plot that shows
#|   global emissions over time by type. The types are coal, natural gas,
#|   cement, and oil and NGL. The plot notes that in 1995, the UN first met to
#|   discuss the climate threat. The plot shows near-zero emissions from 1900 to
#|   1920, when a slow increase begins. From there, emission growth seems to be
#|   exponentially increasing, with no decline since the UN first met. Coal is
#|   the largest emitter, then oil and NGL, then natural gas, and finally,
#|   cement.

Now, let’s see the plot!

# constants for ease of code legibility ----------------------------------------
LEVS <- c("Coal", "Oil & NGL", "Natural Gas", "Cement")
BG_COLOR <- "#F0F0F0"
UN_TEXT <- paste(
  "In 1995, the United Nations\nConference of the Parties met for\nthe first", 
  "time to discuss the looming\nthreat of climate change. The COP\nhas",
  "met twenty-eight times since."
)

# data cleanup -----------------------------------------------------------------
emis |>
  filter(year >= 1900) |> # lots of near-zero space without this filter
  mutate(
    commodity = if_else(str_detect(commodity, "Coal"), "Coal", commodity),
    commodity = factor(commodity, levels = LEVS) # re-order areas
  ) |>
  group_by(year, commodity) |>
  summarise(emissions = sum(emissions), .groups = "drop") |>
  
  # start of plot --------------------------------------------------------------
  ggplot(aes(x = year, y = emissions, fill = commodity)) +
  geom_area(alpha = 0.9) +
  
  # UN COP annotation text box -------------------------------------------------

  # the arrow
  annotate(
    geom = "segment",
    x = 1995,
    xend = 1995,
    y = 35500,
    yend = 20500,
    linetype = "solid",
    linejoin = "round",
    linewidth = 1,
    color = "grey35",
    arrow = arrow(type = "closed", length = unit(0.2, "cm"))
  ) +

  # the background rectangle (must be before the text)
  annotate(
    geom = "rect",
    xmin = 1945.5,
    xmax = 1993.5,
    ymin = 23500,
    ymax = 35800,
    fill = BG_COLOR
  ) +

  # annotation text
  annotate(
    geom = "text",
    x = 1992,
    y = 30000,
    label = UN_TEXT,
    color = GRAY,
    fontface = "italic",
    hjust = "right"
  ) +
  
  # replace legend with annotation text ----------------------------------------
  annotate(
    geom = "text",
    color = "white",
    x = 2020,
    y = c(1000, 4700, 13000, 26000),
    label = c("Cement", "Natural Gas", "Oil & NGL", "Coal"),
    hjust = "right",
    fontface = "bold"
  ) +
  
  # visual style elements (love you, ggthemes) ---------------------------------
  ggthemes::scale_fill_colorblind() +
  ggthemes::theme_fivethirtyeight() +
  
  # customize axis breaks and labels -------------------------------------------
  scale_x_continuous(breaks = seq(1900, 2020, 20)) +
  scale_y_continuous(
    breaks = seq(0, 40000, 5000), 
    label = scales::label_number(scale = 1e-3, suffix = "k")
  ) +
  labs(
    x = element_blank(),
    y = latex2exp::TeX("Emissions ($MtCO_2e$)"),
    title = "Our World in Emissions",
    subtitle = latex2exp::TeX(
      paste(
        "Emissions are measured in Millions of Tons of $CO_2$ equivalent",
        "($MtCO_2e$)"
      )
    ),
    caption = paste(
      "Made with love by Mitch Harrison",
      "                                                                       ",
      "Source: Carbon Majors database and TidyTuesday"
    )
  ) + 
  
  # theme cleanup --------------------------------------------------------------
  geom_hline(yintercept = 0, linewidth = 0.7, color = GRAY) + # bold axis
  theme(
    legend.position = "none", 
    axis.title.y = element_text(size = 10),
    plot.background = element_rect(fill = BG_COLOR)
  ) 

No plot is perfect, but I am happy with what we have accomplished, and I hope you are too! If you have any questions or corrections, feel free to reach out on Discord, and I’ll be happy to help. And, of course, if you want to contribute to this effort financially, you are more than welcome to buy me a coffee.

Thanks for sticking around, and good luck with your TidyTuesday adventures!