Software choice

For the mapping tool, I settled for QGIS, the leading open-source desktop GIS. Unlike many Desktop GIS software, QGIS is free, yet rich with analytical and cartographic capabilities. You just need to download it from the QGIS website. It is cross-platform, available on Windows, Mac, Linux. For this project, I used the most recent QGIS version, QGIS 4.0.

Getting the data

Most of the time, the dataset is not always in a form ready for GIS. We need to prepare data so that we can use it to create maps. All I had was knowledge of winners and participants in text format. How was I going to bring this into GIS?

I had QGIS software installed already, so I launched the software. I closed the start pop-up window that shows up:

I realised I needed the world countries shapefile. I could download the shapefile from places like Natural Earth but I remembered that by default, QGIS includes a sample, basic world map dataset in its installation files. Entering the word “world” into the Coordinate box tells QGIS to fetch and display this file instantly.

Explore the world shapefile

With the world map layer available, I explored the dataset by opening the layer attribute table. I needed to check if perhaps my layer had names of countries which I could use for creating the map. I opened the attribute table by right-clicking on the World Map layer in the Layers panel and choosing Open Attribute Table in the context menu:

The attribute table had a column named NAME which had names of countries. I used the information available in the Wikipedia article to select my countries of interest from the attribute table based on this column.

Saving the QGIS project

At this point, I had started doing something in my project. I had added data to the project. I did not want to lose my work should anything happen (loss of power, or accidentally closing my project without saving). To save the project, I could use the keyboard shortcut Ctrl+S (on a Windows machine). Alternatively, I could click on the Project menu item and select Save Project As…

I saved the project as world-cup.

World cup participating nations in addition to winners

I did not just want to show winners of the previous tournaments, but other countries that had ever participated in the tournament as well. To do that, I needed to select the 83 countries that had ever participated in the tournament from the World Cup layer.

These were Algeria, Angola, Argentina, Australia, Austria, Belgium, Bolivia, Bosnia and Herzegovina, Brazil, Bulgaria, Cameroon, Canada, Chile, China, Colombia, Costa Rica, Croatia, Cuba, Czechoslovakia, Czech Republic, Denmark, DR Congo, East Germany, Ecuador, Egypt, El Salvador, England, France, Germany, West Germany, Ghana, Greece, Haiti, Honduras, Hungary, Iceland, Indonesia, Iran, Iraq, Israel, Italy, Ivory Coast, Jamaica, Japan, Kuwait, Mexico, Morocco, Netherlands, New Zealand, Nigeria, North Korea, Northern Ireland, Norway, Panama, Paraguay, Peru, Poland, Portugal, Qatar, Republic of Ireland, Romania, Russia, Saudi Arabia, Scotland, Senegal, Serbia, Slovakia, Slovenia, South Africa, South Korea, Soviet Union, Spain, Sweden, Switzerland, Togo, Trinidad and Tobago, Tunisia, Turkey, Ukraine, United Arab Emirates, United States, Uruguay, Wales, Yugoslavia.

Missing nations

Checking the list, I realised that some of these names represent teams that no longer exist in that form, such as Czechoslovakia, the Soviet Union, Yugoslavia, and East Germany, whose results are now attributed by FIFA to their recognized successor nations like the Czech Republic and Slovakia, Russia, Serbia, and unified Germany respectively.

Another thing was that England was not available in my World Map layer; rather, my layer had United Kingdom. I decided to use United Kingdom to represent England. By the way, only England has ever won the tournament within the United Kingdom.

Select the nations of interest from the World Map layer

There were a few issues I identified with my World Map layer. Some of the names of countries did not match. They had been coded differently. To solve these issues, I decided to use AI assistance to automatically identify all names of countries that did not match, and to align the names accordingly.

But before I could use AI, I needed to extract all country names, which I would compare with the list above. I couldn’t quickly find a way of copying all the names in the attribute table, so I resorted to exporting the layer to CSV.

I did this by right-clicking on the World Map layer in the Layers panel and then selecting Export > Save Feature As…

In the Save Vector Layer as… dialog, I chose the Comma Separated Value (CSV). Since the World Map layer contained more than the field of interest, specifying the field to save was necessary.

I saved the file as world countries.csv.

I was then ready to visit Claude.ai and enter a prompt that would help me sort the issues. Since I had my CSV file already, I attached it and then copied the names of countries that participated in the World Cup up to 2022, highlighted above, and framed my prompt as below:

Claude identified names that required remapping to match the names in the World Map layer data:

Again, the following names did not have any matching names in the layer:

• Czechoslovakia, East Germany, West Germany, Soviet Union, Yugoslavia — these are historical/dissolved states, not present in modern admin-0 layers
• England, Northern Ireland, Scotland, Wales: these are constituent countries of the United Kingdom, not separate features (only “United Kingdom” exists)

I asked Claude to create an SQL query with corrected and matched names, which I would use to select all countries in QGIS that participated in the World Cup by 2022:

SELECT * FROM world_countries WHERE NAME IN ( 'Algeria', 'Angola', 'Argentina', 'Australia', 'Austria', 'Belgium', 'Bolivia', 'Bosnia and Herz.', 'Brazil', 'Bulgaria', 'Cameroon', 'Canada', 'Chile', 'China', 'Colombia', 'Costa Rica', 'Croatia', 'Cuba', 'Czechia', 'Denmark', 'Dem. Rep. Congo', 'Ecuador', 'Egypt', 'El Salvador', 'France', 'Germany', 'Ghana', 'Greece', 'Haiti', 'Honduras', 'Hungary', 'Iceland', 'Indonesia', 'Iran', 'Iraq', 'Israel', 'Italy', "Côte d'Ivoire", 'Jamaica', 'Japan', 'Kuwait', 'Mexico', 'Morocco', 'Netherlands', 'New Zealand', 'Nigeria', 'North Korea', 'Norway', 'Panama', 'Paraguay', 'Peru', 'Poland', 'Portugal', 'Qatar', 'Ireland', 'Romania', 'Russia', 'Saudi Arabia', 'Senegal', 'Serbia', 'Slovakia', 'Slovenia', 'South Africa', 'South Korea', 'Spain', 'Sweden', 'Switzerland', 'Togo', 'Trinidad and Tobago', 'Tunisia', 'Turkey', 'Ukraine', 'United Arab Emirates', 'United States of America', 'Uruguay', 'United Kingdom' );

In QGIS, I then opened the Query Builder by right-clicking on the World Map layer in the Layers panel and selecting Properties… > Source tab, then expanding Provide Feature Filter and clicking the Query Builder button:

I then pasted the SQL expression from Claude into the query builder:

When I clicked the OK button, I received an error that the table World didn’t exist. You may get a different reference for the table name, such as world_countries. It appears the word Map is a reserved word in SQL (Map is part of SQL vocabulary).

I then renamed the World Map layer to world_map by right-clicking on the layer and opening the Source tab in the Layer Properties window, then editing the layer name.

I further exported the layer as a shapefile named world_map.

I needed to run the SQL query again, but this time working with the new shapefile, ensuring that my query used the correct layer name, world_map.

SELECT * FROM world_map WHERE NAME IN ( 'Algeria', 'Angola', 'Argentina', 'Australia', 'Austria', 'Belgium', 'Bolivia', 'Bosnia and Herz.', 'Brazil', 'Bulgaria', 'Cameroon', 'Canada', 'Chile', 'China', 'Colombia', 'Costa Rica', 'Croatia', 'Cuba', 'Czechia', 'Denmark', 'Dem. Rep. Congo', 'Ecuador', 'Egypt', 'El Salvador', 'France', 'Germany', 'Ghana', 'Greece', 'Haiti', 'Honduras', 'Hungary', 'Iceland', 'Indonesia', 'Iran', 'Iraq', 'Israel', 'Italy', 'Côte d''Ivoire', 'Jamaica', 'Japan', 'Kuwait', 'Mexico', 'Morocco', 'Netherlands', 'New Zealand', 'Nigeria', 'North Korea', 'Norway', 'Panama', 'Paraguay', 'Peru', 'Poland', 'Portugal', 'Qatar', 'Ireland', 'Romania', 'Russia', 'Saudi Arabia', 'Senegal', 'Serbia', 'Slovakia', 'Slovenia', 'South Africa', 'South Korea', 'Spain', 'Sweden', 'Switzerland', 'Togo', 'Trinidad and Tobago', 'Tunisia', 'Turkey', 'Ukraine', 'United Arab Emirates', 'United States of America', 'Uruguay', 'United Kingdom' )

76 rows/countries were selected. These represent the nations that participated in the World Cup tournament between 1930 and 2022.

One more thing before I could start making the final map: I needed another layer for World Cup winners. Since I already had my World Cup participants layer, all I did was duplicate this same layer in the Layers panel, clear the previous SQL expression, and build another one that would select the names of nations that have ever won the World Cup tournament.

I renamed the filtered world_map layer to world_cup_participants. Then I created another layer from it by right-clicking on the layer and selecting Duplicate Layer:

The duplicate layer draws at the bottom of the original layer in the Layers panel. I moved the duplicate to the top of all the layers in the Layers panel by right-clicking on it and selecting Move to Top:

I then renamed the duplicate layer to world_cup_winners and made it active on the map by clicking the checkbox of the layer in the Layers panel:

To select countries that have ever won the World Cup, I needed to revise the SQL query as follows:

SELECT * FROM world_map WHERE NAME IN ( 'Argentina', 'Brazil', 'France', 'Germany', 'Italy', 'Spain', 'United Kingdom', 'Uruguay' )

Note that I kept the old name world_map in the SQL query, because although the layer is displayed as the duplicate in the panel, the actual file on disk is still world_map:

Again, here I used United Kingdom to represent England, and Germany to represent the former East Germany.

Unchecking the participants layer helped me clearly see the nations that had ever won the tournament. This was necessary because the world_cup_participants layer and the world_cup_winners layer had similar colors on the map:

With all the map layers available in my QGIS project, I could then move on to make the final map.

Styling map layers for the final map

I decided to style my world_map layer with a gray fill color, #e0e0e0, and a white stroke color with a stroke width of 0.26. This was achieved by right-clicking on the world_map layer and opening the layer’s Properties. In the Layer Properties window, I selected the Symbology tab:

This layer was only useful for giving context, hence the choice of a non-dominant color.

I followed the same process to give a green color, #55793f, to World Cup participants, and a golden color, #e6ba32, to countries that have ever won the tournament:

Convinced with my layer styles, I moved on to create the map layout.

Choosing a compromise projection

When creating a map of the world, I like choosing a map projection that minimises the distortion of landmasses. For this map, I chose the Gall Stereographic map projection, ESRI:54016, which is a cylindrical map projection developed by Scottish clergyman James Gall in 1855. It’s a compromise projection which is neither conformal (doesn’t preserve shape/angles) nor equal-area (doesn’t preserve area). It’s meant to balance distortion reasonably across the whole map, similar in spirit to the Robinson or Miller projections.

By default my map was not projected — it had the geographic coordinate system, EPSG:4326. To fix this, I opened the project’s properties:

and then searched for and selected ESRI:54016 in the CRS tab:

giving me the desired look:

At this point, I was ready to move on to the Print Layout to make the final map.

Map layout

The Print Layout is opened via the Project menu item:

I named the new Print Layout World Cup:

I remembered that I had not labelled my map polygons. I went back to my map and, after right-clicking on the world_cup_participants layer in the Layers panel, I selected Show Labels in the context menu:

Nothing showed on the map even after I had selected Show Labels. I needed to tell QGIS which field values would be used to label the polygons. To do that, I opened the Layer Properties of the world_cup_participants layer. In the Layer Properties dialog, I selected the Labels tab. In the Value field, I selected the NAME field. For the Text, I selected the Roboto font, Regular style, and reduced the default size to 6.

To make sure the labels stood out against the background, I gave the text a white buffer of size 0.5 with 50% opacity:

As a final touch to my labels, I enabled Draw drop shadow:

I was happy with my map labels, so I continued working in the Print Layout:

Adding items to the Map Layout

To add the map to the layout, I used the Add Item menu and then selected Add Map:

Just doing that didn’t add the map to the page. I needed to draw a rectangle on the blank page — after my mouse cursor had turned into a cross-hair, ensuring that the rectangle filled the whole page:

Before I could add more map elements to the layout, I set the map scale to 100000000:

A good map should be balanced, meaning the map body should be centered. I used the Move Item Content tool to move the map body to the center of the page:

I disregarded Antarctica while repositioning the map, since it only provided context to the map.

To add the map title, I used the Add Item menu and selected the Add Label tool:

Again, drawing the rectangle at the top of the page:

I changed the default text to FIFA Men's World Cup Winners and changed the font to Roboto, giving it the Light style and a font size of 36, with horizontal center alignment:

A map title should tell us what, when, and where. I added a subtitle to the map, maintaining a similar font to the main title but reducing the font size to 20:

Lessons learnt

A few things worth remembering for next time:

• Name matching is the real work. Country name mismatches between data sources and reference layers are near-guaranteed, build in time for this step.
• Reserved words can break queries silently. A layer named World Map failed the SQL query builder simply because Map is SQL vocabulary. Rename layers to avoid ambiguity.
• Duplicate layers keep referencing the original source. When duplicating a layer, remember the underlying file name, not the panel display name, is what your queries must reference.
• Historical and constituent nations need a documented decision. Dissolved states (USSR, Yugoslavia, Czechoslovakia, East/West Germany) and UK constituents don’t map cleanly onto modern admin-0 layers. Decide how to handle them early, and disclose the choice on the map itself.

Behold, the final map!

The map was ready for exporting, as an image, PDF, and more:

I exported the map as an image:

Happy mapping, everyone!
– Dickson

Step 1: Load the spatial data

Every map starts with data. Download the dataset from here.. Here I bring in four spatial layers:

1. colombia — the national administrative boundary (admin level 0), sourced from a standard admbnda shapefile.
2. cali — a single-feature layer just for the city of Cali, which gets the special COP16 logo treatment.
3. colombia_cities — a broader set of major Colombian cities, shown as secondary reference points.
4. world — a global boundary layer, loaded here for context but not directly plotted in this version of the map.

I also point to two logo image files that will be layered onto the final map later: the COP16 event logo (used as a custom point symbol for Cali) and the COP16 wordmark (used as a corner logo).

colombia        <- st_read("data/col_admbnda_adm0_mgn_itos_20200416.shp", quiet = TRUE)
cali            <- st_read("data/cali.shp", quiet = TRUE)
colombia_cities <- st_read("data/colombiancities.shp", quiet = TRUE)

logo <- "data/COP16_logo.png"
logoWords <- "data/cop-16-logo_120.png"

Step 2: Draw the country outline

The base layer is Colombia’s national boundary, styled in forest green (#228B22) to evoke the “nature” half of “Peace with Nature.” A lwd of 2 keeps the border visible without overpowering the city symbols that follow.

tm_shape(colombia) +
  tm_borders(col = "#228B22", lwd = 2)

Step 3: Give Cali the spotlight

This is the layer that makes the map distinctly about COP16. Instead of a generic dot, Cali is marked with the COP16 event logo itself via tm_symbols(shape = logo, ...). Underneath it, tm_labels() writes out the city name in a bold rust-red (#bc3110), offset slightly up and to the right (xmod/ymod) so the text doesn’t collide with the logo icon.

tm_shape(cali) +
  tm_symbols(shape = tmap_icons(logo), size = 0.5) +
  tm_labels(
    "city",
    size = 2,
    col = "#bc3110",
    fontface = "bold",
    just = c("left", "bottom"),
    xmod = 0.6,
    ymod = 0.1
  )

Step 4: Add context with other major cities

To help readers orient Cali within the country, I plot a set of other major Colombian cities as smaller reference symbols — orange fill (#ffa101) with a rust outline (#bc3110) — each labeled in a muted olive tone (#b3ac8b) at a much smaller size so they support the map without competing with Cali.

tm_shape(colombia_cities) +
  tm_symbols(col = "#bc3110", fill = "#ffa101", lwd = 3, size = 0.6) +
  tm_labels(
    "city",
    size = 0.7,
    col = "#b3ac8b",
    fontface = "bold",
    just = c("left", "bottom"),
    xmod = 0.5,
    ymod = 0.5
  )

Step 5: Branding — logos and credits

A few finishing touches give the map its editorial identity:

• The COP16 wordmark logo sits in the top-left corner.
• Event hashtags (#PeaceWithNature, #CountryOfBeauty, #CaliColombia, #COP16) appear as bottom-left credits in a deep blue (#1a5783), doubling as both attribution and a nod to the conference’s social media campaign.
• A cartographer credit line (name and date) sits just below the hashtags, in a neutral grey so it doesn’t distract from the campaign messaging.
• An R logo in the top-right corner signals the tool used to build the map — a nice touch for a technical or portfolio audience.

tm_logo(logoWords, position = c("left", "top"), height = 8) +
  tm_credits(
    "#PeaceWithNature\n#CountryOfBeauty\n#CaliColombia\n#COP16\n\n",
    position = c("left", "bottom"),
    size = 1,
    color = "#1a5783"
  ) +
  tm_credits(
    "Cartographer: Dickson Mbeya\n05 July 2026",
    position = c("left", "bottom"),
    size = 0.8,
    color = "grey20"
  ) +
  tm_logo("data/r.png", position = c("right", "top"), height = 1.5)

Step 6: Frame it

Finally, a double frame effect: an outer navy-blue frame (#1a5783, lwd = 2) and an inner gold frame (#efaf48, lwd = 1) applied via two stacked tm_layout() calls. Layering tm_layout() calls like this lets you build up frame effects that would be hard to express in a single call, since tmap merges layout options progressively.

tm_layout(
  frame = TRUE,
  frame.lwd = 2,
  frame.col = c("#1a5783")
) +
tm_layout(
  frame = TRUE,
  frame.lwd = 1,
  frame.col = c("#efaf48")
)

Putting it all together

Here’s the full map, assembled from every layer above into a single tmap object and rendered in static plot mode.

colombia_map <- tm_shape(colombia) +
  tm_borders(col = "#228B22", lwd = 2) +
  tm_shape(cali) +
  tm_symbols(shape = tmap_icons(logo), size = 0.5) +
  tm_labels(
    "city",
    size = 2,
    col = "#bc3110",
    fontface = "bold",
    just = c("left", "bottom"),
    xmod = 0.6,
    ymod = 0.1
  ) +
  tm_shape(colombia_cities) +
  tm_symbols(col = "#bc3110", fill = "#ffa101", lwd = 3, size = 0.6) +
  tm_labels(
    "city",
    size = 0.7,
    col = "#b3ac8b",
    fontface = "bold",
    just = c("left", "bottom"),
    xmod = 0.5,
    ymod = 0.5
  ) +
  tm_logo(logoWords, position = c("left", "top"), height = 8) +
  tm_credits(
    "#PeaceWithNature\n#CountryOfBeauty\n#CaliColombia\n#COP16\n\n",
    position = c("left", "bottom"),
    size = 1,
    color = "#1a5783"
  ) +
  tm_credits(
    "Cartographer: Dickson Mbeya\n05 July 2026",
    position = c("left", "bottom"),
    size = 0.8,
    color = "grey20"
  ) +
  tm_logo("data/r.png", position = c("right", "top"), height = 1.5) +
  tm_layout(
    frame = TRUE,
    frame.lwd = 2,
    frame.col = c("#1a5783")
  ) +
  tm_layout(
    frame = TRUE,
    frame.lwd = 1,
    frame.col = c("#efaf48")
  )

tmap_mode("plot")
colombia_map

Takeaways

A few reusable lessons from this build, useful beyond just COP16 maps:

• Custom logos as point symbols (tm_symbols(shape = tmap_icons("path/to/logo.png"))) are an easy way to turn a plain location marker into something that carries event or brand identity. Remember to wrap the path in tmap_icons(), since a raw file path isn’t a valid shape value on its own.
• Stacked tm_layout() calls let you compose layered visual effects (like a double frame) that would otherwise need custom grid graphics.

If you’re working on your own event or conference map, the same six-step structure boundary, feature highlight, context layer, branding, framing, render should adapt cleanly to most tmap projects.

Happy mapping, everyone!
– Dickson

1. Load the required packages

We need a handful of specialised packages:

• sf – reads and handles spatial vector data (shapefiles).
• tmap – our main mapping engine; it uses a grammar of graphics similar to ggplot2 but is tailored for maps.
• dplyr – for quick data manipulation (renaming columns).
• extrafont – to use system fonts like Roboto for a modern, clean look. If you haven’t used extrafont before, you can run font_import() and loadfonts() once to make your system fonts available in R.

library(tmap)
library(dplyr)
library(sf)

popn <- st_read("data/mw_popn.shp")

Reading layer `mw_popn' from data source 
  `C:\Users\devmbeya\Documents\blog\posts\make_popn_map\data\mw_popn.shp' 
  using driver `ESRI Shapefile'
Simple feature collection with 32 features and 23 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: 3636988 ymin: -1935618 xmax: 3998420 ymax: -1047139
Projected CRS: WGS 84 / Pseudo-Mercator

bnd <- st_read("data/mw_bnd.shp")

Reading layer `mw_bnd' from data source 
  `C:\Users\devmbeya\Documents\blog\posts\make_popn_map\data\mw_bnd.shp' 
  using driver `ESRI Shapefile'
Simple feature collection with 22 features and 1 field
Geometry type: POLYGON
Dimension:     XY
Bounding box:  xmin: 464479.7 ymin: 8105086 xmax: 814025.4 ymax: 8964834
Projected CRS: WGS 84 / UTM zone 36S

2. Read the spatial data

The data comes from the Malawi National Statistical Office and contains district boundaries and population counts for 2008.

In this example I store the shapefiles inside a data/ folder relative to the project root. This makes the script portable and reproducible.

First, I read the population point/centroid layer which contains the district names and population counts:

popn <- st_read("data/mw_popn.shp")

Reading layer `mw_popn' from data source 
  `C:\Users\devmbeya\Documents\blog\posts\make_popn_map\data\mw_popn.shp' 
  using driver `ESRI Shapefile'
Simple feature collection with 32 features and 23 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: 3636988 ymin: -1935618 xmax: 3998420 ymax: -1047139
Projected CRS: WGS 84 / Pseudo-Mercator

Next, I read the district boundary polygon layer which will serve as the background map:

bnd <- st_read("data/mw_bnd.shp")

Reading layer `mw_bnd' from data source 
  `C:\Users\devmbeya\Documents\blog\posts\make_popn_map\data\mw_bnd.shp' 
  using driver `ESRI Shapefile'
Simple feature collection with 22 features and 1 field
Geometry type: POLYGON
Dimension:     XY
Bounding box:  xmin: 464479.7 ymin: 8105086 xmax: 814025.4 ymax: 8964834
Projected CRS: WGS 84 / UTM zone 36S

Let’s inspect the population data to see what columns we have:

head(popn)

Simple feature collection with 6 features and 23 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: 3667131 ymin: -1884697 xmax: 3932595 ymax: -1047139
Projected CRS: WGS 84 / Pseudo-Mercator
      DIST_NAME      DISTRICT  REGION Shape__Are Shape__Len
1        Balaka        Balaka Central 2309288545  227264.32
2      Blantyre      Blantyre   South 1935900658  314395.31
3 Blantyre City Blantyre City   South  261111996   92609.81
4      Chikwawa      Chikwawa   South 5318053799  413383.01
5    Chiradzulu    Chiradzulu   South  825563720  184760.60
6       Chitipa       Chitipa   North 4499298409  627147.82
                                GlobalID CreationDa Creator   EditDate Editor
1 {7732b160-5d5f-4f3c-a381-d8e5bbe4f9c2} 2022-02-06  mbeyad 2022-02-07 mbeyad
2 {1f059ed3-7289-4a55-a6d5-9b45872048fe} 2022-02-06  mbeyad 2022-02-07 mbeyad
3 {ead7d5d2-408f-4cce-bcba-827bb9a71dab} 2022-02-06  mbeyad 2022-02-07 mbeyad
4 {3ea276d9-a990-4a77-8e3b-d2807b223417} 2022-02-06  mbeyad 2022-02-07 mbeyad
5 {fa51c606-08ea-4b41-8340-ba68833bad02} 2022-02-06  mbeyad 2022-02-07 mbeyad
6 {20595550-2501-44ae-8f91-5786859edc50} 2022-02-06  mbeyad 2022-02-07 mbeyad
  MALE1998 FEM1998 TOTAL1998 MALE2008 FEM2008 TOTAL2008 FEM2018 TOTAL2018
1   120706  132397    253103   151637  165111    316748  229105    438379
2   413429  395968    809397   164546  173501    338047  232756    451220
3        0       0    502053   337655  323789    661444  399092    800264
4   178217  178465    356682   217981  220914    438895  287794    564684
5   111376  124674    236050   137194  153752    290946  187196    356875
6    60682   66117    126799    86152   92920    179072  120535    234927
  MALE2018 POPDEN2008 POPDEN1998  AREA_KM2 DEN2018
1   209274        148        119 2133.8513     205
2   218464        189        453 1788.3027     252
3   401172       2805       2129  235.8099    3394
4   276890         90         73 4891.6565     115
5   169679        382        310  762.5165     468
6   114392         42         30 4247.7344      55
                        geometry
1 MULTIPOLYGON (((3916682 -16...
2 MULTIPOLYGON (((3908694 -17...
3 MULTIPOLYGON (((3905701 -17...
4 MULTIPOLYGON (((3822326 -17...
5 MULTIPOLYGON (((3913071 -17...
6 MULTIPOLYGON (((3673613 -10...

Looking at the output, I can see the column TOTAL2008 holds the total population for each district. I’ll rename it to something more intuitive for our mapping:

popn <- popn %>%
  rename(Population = TOTAL2008)

Now we have a clean attribute called Population that we can use in our map.

3. Design the main typographic map

This is the heart of the visual. Instead of colouring polygons, I use tm_text() to place the district names directly on the map. The size argument is mapped to the Population column—so larger populations appear in bigger text.

Let’s break down the styling choices:

Background – a very light grey ("grey99") for the districts, with thin grey borders ("grey70"). This keeps the focus on the text.

Text colour – bright orange ("orange") stands out beautifully against the light grey.

Layout – I added generous inner margins so the text doesn’t touch the frame. The frame itself is rounded (frame.r = 10) and coloured orange to tie the design together.

Legend – I place it at the top right (position c(0.7, 0.8)) with a light grey frame.

Title – a multi‑line, attention‑grabbing question. The title is placed on the right side, with a white background to improve readability.

Credits – three separate credits cover the challenge tag, my name, and the data source. I use Roboto throughout for consistency.

mypop <- tm_shape(bnd) + 
  tm_polygons(fill = "grey99", col = "grey70") +
  
  tm_shape(popn) + 
  tm_text(text = "DISTRICT", 
          size = "Population", 
          col = "orange", 
          fontface = "bold") + 
  
  tm_layout(inner.margins = c(0.02, 0.3, 0.02, 0.3),
            outer.margins = c(0.03, 0.02, 0.03, 0.02),
            frame.r = 10,
            frame.lwd = 1,
            frame.color = "orange",
            legend.frame.lwd = 1,
            legend.frame.color = "grey80",
            legend.title.fontfamily = "Roboto Light",
            legend.position = c(0.7, 0.8)) + 
  
  tm_title("Which Districts\nHad the Largest\nPopulation in Malawi\nin 2008?", 
           color = "orange",
           size = 1.2,
           bg = TRUE,
           bg.alpha = 1,
           position = c("right", "top"),
           frame = FALSE) +
  
  tm_credits("Date: 05 July 2026",
             position = c("left", "bottom"),
             size = 0.7, color = "orange",
             fontfamily = "Roboto") + 
  
  tm_credits("© Dickson Mbeya",
             position = c("center", "bottom"),
             size = 0.7, color = "orange",
             fontfamily = "Roboto") +   
  tm_credits("Data Source: Malawi\nNational Statistical Office\n\nwww.dicksonmbeya.com",
             fontfamily = "Roboto",
             position = c("right", "bottom"),
             size = 0.6, color = "orange")

Why text‑sized names instead of a classic choropleth?
Because readers can immediately identify the largest districts without needing to consult a legend. The text itself becomes the visual variable, which is both intuitive and visually striking.

4. Build a locator inset map

When you share maps online, not everyone knows exactly where Malawi is. A small locator map solves that problem.

I use an orthographic projection centered on Malawi’s coordinates (lat_0 = -13.98, lon_0 = 33.77). This creates a globe‑like effect that makes the country pop out from the rest of Africa.

First, I load the world dataset using the rnaturalearth package:

library(rnaturalearth)

Warning: package 'rnaturalearth' was built under R version 4.5.3

world <- ne_countries(scale = "medium", returnclass = "sf")

Next, I define the orthographic projection and transform both the world and Malawi boundaries:

ortho_crs <- "+proj=ortho +lat_0=-13.98 +lon_0=33.77"
world_ortho <- st_transform(world, crs = ortho_crs)
bnd_ortho <- st_transform(bnd, crs = ortho_crs)

Now I build the locator map. African countries are filled with light grey, while Malawi is outlined in bright orange:

locator_map <- tm_shape(world_ortho) +
  tm_polygons(col = "grey80", border.col = "white", lwd = 0.3) +
  tm_shape(bnd_ortho) +
  tm_borders(col = "orange", lwd = 1) +
  tm_layout(frame = FALSE, bg.color = "white")

── tmap v3 code detected ───────────────────────────────────────────────────────

[v3->v4] `tm_polygons()`: use 'fill' for the fill color of polygons/symbols
(instead of 'col'), and 'col' for the outlines (instead of 'border.col').
This message is displayed once every 8 hours.

The white background and no frame ensure the inset integrates cleanly with the main map.

5. Combine everything with tm_inset()

Now I overlay the locator map onto the main map using tm_inset().

I set the inset size to 9 which works well with the existing layout dimensions. The position c(0, 0.3) places it at the left side, slightly below the top margin.

final_map <- mypop +
  tm_inset(locator_map,
           width = 9,
           height = 9,
           position = c(0, 0.3))

final_map

Warning in graphics::strwidth(comp$title, units = "inch", cex = titleS, : font
family not found in Windows font database

Warning in graphics::strwidth(comp$text, units = "inch", family =
comp$fontfamily, : font family not found in Windows font database

Warning in grid.Call.graphics(C_text, as.graphicsAnnot(x$label), x$x, x$y, :
font family not found in Windows font database
Warning in grid.Call.graphics(C_text, as.graphicsAnnot(x$label), x$x, x$y, :
font family not found in Windows font database
Warning in grid.Call.graphics(C_text, as.graphicsAnnot(x$label), x$x, x$y, :
font family not found in Windows font database

6. Save the final output

You can export the map as a high‑resolution PNG for sharing on social media, in presentations, or for print.

#tmap_save(final_map, "2008_Population.png", width = 10, height = 8, dpi = 300)

What the map reveals

Looking at the scaled district names, Lilongwe (the capital), Blantyre (the commercial hub), and Mzimba (the northern giant) clearly dominate the population distribution. The text‑size approach makes this pattern visible instantly—without needing a separate colour legend.

The map also reveals interesting spatial patterns: - The central region (Lilongwe) has the highest population - The southern region (Blantyre) is the second most populated - The northern region (Mzimba) has large populations but more spread out

Final thoughts

This map was a fun exercise in using tmap’s typographic capabilities. By focusing on text rather than fill colours, I created a poster‑style visual that is both informative and decorative.

Key takeaways for your own maps:

• Don’t be afraid to use tm_text() for quantitative data – size scaling works beautifully
• Always add a locator inset when your audience might be global
• Small typographic touches (fonts, frame colours, margin control) elevate the final product
• Using a light background with bold text creates a clean, professional look

If you have any questions about the code or want to adapt it for your own country, feel free to reach out!

Happy mapping, everyone!
– Dickson