By Kenneth Field (Research Cartographer), Damien Demaj (ArcGIS Online Cartography) and Linda Beale (Geoprocessing)
At the 2012 Esri Education GIS Conference and the 2012 Esri International User Conference, we demonstrated how you can build informative thematic maps using the ArcGIS System. The purpose of the sessions was to take relatively simple datasets and create a range of alternative thematic map types that told a story in different ways. This demonstrated the techniques for creating the maps using ArcGIS for Desktop as our authoring environment and ArcGIS Online as our publishing mechanism. As the XXX Olympiad is currently taking place in London, UK we illustrated how alternative maps can be made to tell different stories of the relative success of nations over the period since the first Olympic games in 1896.
This blog entry explores how the message of the map communicates a very different story depending on how it is designed. The gallery of web maps referred to in this blog entry is available here. Future blog entries will focus on some of the details of how the maps were constructed. Each map made use of the Light Gray Canvas basemap to provide a neutral background for the thematic content and popups were designed to reveal further information and encourage exploration of the map.
Figure 1 is a proportional symbol map of the total medal count since 1896. The size of the symbols are scaled to the total number of medals awarded and they are symbolised to reflect the map theme using an Olympic medal graphic. At small scales the symbols overlap which we countered by placing a small white border around each symbol.
Figure 1. Proportional symbols: Olympic medal count
The multiscale characteristic of web maps means that as you zoom in, the symbols maintain their relative size and begin to separate to reveal each distinct symbol. This map reveals the dominant nations as measured by the most medals. The USA total of 2,302 medals is the largest symbol and suggests an overall dominance. Popups reveal the breakdown of medal success by Gold, Silver and Bronze. At smaller scales, however, the symbols are often hard to disentangle so an alternative approach is to construct a Dorling cartogram (Figure 2).
Figure 2. Dorling cartogram: Olympic medal count
This sort of map redistributes the proportional symbols to give each its own unique position in space. Now, all symbols are visible at all scales. We added labels to the map so that they appear at different zoom scales and the popups are retained to give the medal breakdown. Using scale to determine label visibility helps promote those countries with the highest medal totals at the smallest scales with detail being revealed as you zoom in. This map still shows the USA as having a dominant total but the message is altered to one that suggests a European regional dominance.
We also mapped the same data as a choropleth map. Firstly, using raw totals (Figure 3) and then by normalising in relation to the country’s population (Figure 4).
Figure 3. Choropleth map: Olympic medal count as totals
Figure 4. Choropleth map: Olympic medal count normalised by population
The map of total medal counts shows how a choropleth can be extremely misleading when no account of the different size of areas or population is considered. In fact one of the take-aways from our presentation was to avoid mapping totals at all costs as it presents a false picture. When the medal totals are normalised (and we map using a similar quantile classification scheme for visual comparison) we see that the story changes dramatically. Now, the USA’s medal haul is nowhere near as impressive compared to, say, Sweden, Hungary or Australia who have far lower populations which means their relative success could be considered to be more impressive.
And what if we build in the number of competitors that each nation sends to the Olympics because that too ought to show some relationship with medal success? We created a bivariate choropleth to explore the story of the relationship between medal count normalised by population against medal count normalised by the number of competitors (Figure 5).
Figure 5. Bivariate choropleth map: Olympic medal count by population and number of competitors
Because this sort of map is sometimes difficult to interpret as it combines two choropleths into one, we used the popup to explain the interpretation for each country as well as provide a legend for interpretation. Countries like Sweden, Australia and Canada send a relatively high number of athletes per 100,000 people but this investment is matched by a relatively high medal haul. Turkmenistan and Botswana also send a relatively high proportion of athletes but have a far poorer relative medal haul indicating a lower measure of relative success. At the other end of the spectrum, Kenya sends relatively few athletes compared to their high medal haul so their relative success is greater. The USA sends an average number of athletes compared to their total population in relation to other nations but they achieve a high relative medal haul.
Since a choropleth map always requires you to normalise by some meaningful denominator, an alternative approach is to create an area cartogram. An area cartogram uses the values themselves to modify the shape and size of the area in which they are mapped. Thus, the shape of the country is warped to accommodate the inherent differences between areas. This creates a sometimes strange, yet powerfully visual way to display area based thematic data.
We created an area cartogram using the Gastner-Newman method which preserves the general appearance and adjacency of countries while modifying the shape so that the area is proportional to the variable being mapped (Figure 6). The resulting map illustrates the variable without the baggage of having to normalize the data. It’s certainly an eye-catching alternative to our normal view of the world map and shows the bloated shapes of countries with large medal hauls compared to their slimmed down counterparts.
Figure 6. Area cartogram: Olympic medal count
Turning away from the medal count data, we created one further area based map that we used to present reference information about each individual country (Figure 7).
Figure 7. Olympic reference map
For this qualitative map, each country should be given equal visual treatment so we used a four-colour basemap of countries. This is a good way of using a minimum number of colours to create a visually uniform backdrop. Since we wanted to design a popup with reference detail we created symbols using the flags of nations that aid us in identifying the location of each country based on a familiar visual cue. The flags invite users to interact and the popup reveals detail including a link to an audio file of each country’s national anthem. This illustrates how you can embed multimedia inside a popup in an engaging way.
Having considered point and area maps we then turned our attention to linear thematic maps and, using data for the number of athletes, by nation, who competed in the London Olympics in 1908, 1948 and 2012 we created a side-by-side web map app of competitor flow (Figure 8).
Figure 8. Flow map: Competitors for the London Olympics of 1908, 1948 and 2012
This map illustrated the construction of quantitative geodesic flow lines between two points, the nation of origin and the destination…London, where width is proportional to the number of athletes and colour is used to accentuate the appearance of countries with larger numbers of athletes. Presented on Web Mercator the lines would be highly curved and distorted but we used an azimuthal equidistant basemap, centred on London as our basemap in ArcGIS Online. This creates a radial flow map with origins flowing to a central point. Altering the projection and centring the map on the destination point helps to make the lines become more discernable. Using the side-by-side template gives us a great way of viewing three periods of data to show how the number of athletes and competing nations has altered across the three Olympics hosted by London.
Our final map made use of the time aware functionality in ArcGIS for Desktop to create an interactive web map of the Olympic torch route during the days leading up to the opening ceremony (Figure 9). The route of the torch was built as a time enabled layer and we added points for each of the main destinations with popups that revealed detail for that location as well as a picture, again demonstrating the important role that multimedia plays in creating an engaging popup that provides a window to additional information. Published using a time aware web map app template, the map user can animate through the time aware data, pausing and moving to key places as they wish. The authoring and publishing of this map was described in the previous blog entry Creating a time aware web map for the torch relay event.
Figure 9. Time aware map: route of the torch relay prior to the XXX Olympiad
The maps we’ve outlined here form a small collection and the new ArcGIS Online Public Gallery template provides a great way to show them as a unified collection (Figure 10).
Figure 10. ArcGIS Online Public Gallery template for the collection of Olympic thematic maps
In this blog entry we’ve shown how you can design a thematic map in a range of styles and how that choice alone might have a profound impact on the way the map is interpreted. It simply isn’t possible to present all of these different views on the same map effectively but using different maps that focus on a common theme can illuminate different aspects and can be used collectively to build a more rounded story. The new Public Gallery templates also helps organise related maps to group them by theme. Future blog posts will look at some of the detail of the methods used to construct the maps that can be implemented using your own data to create informative thematic maps published on ArcGIS Online.