=Paper=
{{Paper
|id=Vol-2323/SKI-Canada-2019-7-1-5
|storemode=property
|title=Web Maps for Global Data Visualization: Does Mercator Matter?
|pdfUrl=https://ceur-ws.org/Vol-2323/SKI-Canada-2019-7-1-5.pdf
|volume=Vol-2323
|authors=Sam Lumley,Renee Sieber
}}
==Web Maps for Global Data Visualization: Does Mercator Matter?==
https://ceur-ws.org/Vol-2323/SKI-Canada-2019-7-1-5.pdf
Spatial Knowledge and Information Canada, 2019, 7(1), 5
Web Maps for Global Data Visualization:
Does Mercator Matter?
SAM LUMLEY R ENEE SIEBER
Department of Geography Department of Geography
School of Environment School of Environment
McGill University McGill University
sam.lumley@mail.mcgill.ca renee.sieber@mcgill.ca
ABSTRACT 1. Introduction
1.1 Web maps for data visualization
The Mercator projection has become a Web map visualization describes the
standard across web mapping platforms, but interactive display of geographic
has long been considered inappropriate for
information on a computer-based map
global data display due to its distortion of high
(Kraak & Brown, 2014). By giving users an
latitude areas. With the ever-rising popularity
of web maps, the Mercator projection has seen
intuitive schema for navigation, web maps
a resurgence in its use for spatial data represent a popular communication tool for
visualizations. In this study we investigated sharing spatial information (Elwood, 2011;
the implications of the area distortion effects Johnson & Sieber, 2012). Further, the
of the Mercator projection for public data development of map tiling services over the
interpretation. We recruited 120 participants past decade has dramatically reduced the
via Amazon’s Mechanical Turk platform to computational demands for data retrieval
complete an online survey assessing their and display (Haklay, Singleton, & Parker,
ability to identify and account for the 2008), enabling user-friendly interaction
distortion effects. Participants were asked to and serving the information seeking
estimate the areas covered by five colored Mantra: “overview first, details on demand”
regions on a global map, having been split into (Shneiderman, 1996).
a control group using an equal-area projection
and a treatment group using a Mercator As a result, there is a growing adoption of
projection. On average, participants did not web mapping applications, such as Google
discount for the projection and their data Maps, OpenLayers and Mapbox APIs, in
interpretation differed between the two public data portals and interactive maps
conditions as a result. Our findings provide an (Batty, Hudson-Smith, Milton, & Crooks,
empirical basis for the distortion effects of the 2010). This resurgence demands further
Mercator projection currently used in web
research into the perceptual implications of
maps, and further implicate its
the Mercator projection’s area distortions.
appropriateness for displaying global data.
More broadly, they introduce experimental Understanding how such representational
methods for research exploring cartographic features influence public data interpretation
biases in non-expert groups. represents a critical issue in GIScience, and
will be key to improving cartographic
communication more generally.
1.2 Mercator in web maps
The Mercator projection has become the
standard across web mapping applications
(Battersby, Finn, Usery, & Yamamoto,
2014). The preservation of angles
(conformality) and universally upward
2 Web Maps for Global Data Visualization: Does Mercator Matter?
pointing north (cylindricality) make it scale cognitive maps. Their results from 194
ideally suited for street mapping services student participants’ area estimations of
(Strebe, 2012). The variant used in web world regions suggested that projection
mapping represents the earth as a square at choice had a lower-than-expected impact on
its lowest zoom level by truncating each pole cognitive maps, a finding further explicated
by 5°. These properties come at the expense in a follow up review by Battersby et al.
of area distortions that increase from the (2014). Aside from a study on map
equator to the poles. projection preferences (Šavrič, Jenny,
White, & Strebe, 2015), most recent
While mapping platforms provide a experimental research on map projections
powerful and convenient tool for data has been focused on academic or expert
visualization, past research has shown that populations. As the number of web mapping
even experienced users can struggle to applications used to display scientific data
compensate for distortions when making rises, it will be increasingly important to
on-the-spot judgements (Downs & Liben, understand the implications of projection
1991; MacEachren, 2004). The “Mercator choices in digital interfaces for non-expert
Effect” predicts that people overemphasize audiences (Nocke, Flechsig, & Bohm, 2007;
the importance of the enlarged high latitude Slocum et al., 2001), a primary objective of
regions (Saarinen, 1988), which can lead to the present study.
an inaccurate interpretation of any global
data being overlaid. Critical geographers 1.3 The present study
further argue that the distortion and This study assesses the influence of the
orientation effects have served to reinforce Mercator projection on area estimation and
European colonialism (Harpold, 1999), and data interpretation in non-expert audiences.
more recently new forms of ‘digital Specifically, we were interested in whether
imperialism’ (Farman, 2010). people identify the distortions, and if they
do, how able they are to account for them.
For this reason, the Mercator projection has This question was addressed through an
long been renounced for use in scientific online experimental survey exploring
visualization on the grounds that its area impacts on area-based judgements about
distortions mislead map readers (Robinson, global geospatial data. To this end, we
1966). Despite this turbulent history and advanced two hypotheses: (1) individuals
recent resurgence, there are still relatively making on-the-fly judgements about spatial
few empirical studies investigating the data presented on a map are unlikely to
cognitive implications of map projections identify or correct for projection distortions
for data display (Battersby et al., 2014). and; (2) even if individuals are aware of the
Further, it is unclear whether past results distortions, they will struggle to accurately
remain relevant (Montello, Waller, Hegarty, convert back to the corresponding areas.
& Richardson, 2004), particularly in light of
recent mapping technologies (Lapon, Ooms, 2. Methods and Data:
& Maeyer, 2017). Digital interfaces offer
new opportunities and new modalities 2.1 Participants
through which people can engage with Participants (N = 120) were recruited using
spatial data (Haklay et al., 2008). The Amazon’s Mechanical Turk online hiring
resulting shifts in use warrant further platform (Amazon, 2014). Mechanical Turk
investigation. is a well-established recruitment tool used
widely in social science research (Berinsky
A recent body of research has begun to et al., 2012; Litman et al., 2017), and has
explore these implications. Notably, been implemented successfully in
(Battersby & Montello, 2009) investigated cartographic research more recently (e.g.
the influence of map projections on global- Retchless & Brewer, 2015; Šavrič et al.,
2015). All of our respondents were adults
Web Maps for Global Data Visualization: Does Mercator Matter? 3
living in the United States and participated earth’s surface. This construction
through a Qualtrics online survey. Our corresponded closely enough to a relatable
sample had a mean self-reported age of 35 real-world example, but was abstract
years (SD = 13.0), with 29% female and 42% enough for participants to engage without
with a bachelor’s degree as their highest strong prior perceptions influencing their
attained level of education. Participants responses (a common problem encountered
were offered $1.00 for completing the during our pilot surveys which used a
survey, plus a $0.50 performance-based temperature labelling scheme). Participants
bonus. After eliminating responses with were asked to estimate the total area
incomplete or unusable answers, we covered by each of the five pollutants.
retained 113 valid responses. Further interpretation of the data was
evaluated by asking respondents to choose
2.2 Design which of two particular colored pollutants
Participants were randomly assigned to one they perceived to be a greater threat to the
of two conditions: a treatment condition earth.
using a Mercator version of the map (N =
60) and a control condition using an equal- Participants were next given a short
area (Lambert cylindrical) version (N = 53). explanation of how different projections
The control map projection was chosen unavoidably distort areas and/or shapes
because areas could be compared at face- displayed on maps. After this briefing, it was
value across the image, while also being a hoped that some participants would decide
commonly used projection (Šavrič et al., that their previous area judgements had
2015). been be influenced by the projection they
had been given. They were then shown a
The data used in the map visualizations was blank version of both projections and asked
derived from a global temperature dataset which one they thought was more suitable
downloaded from the University of East for an area estimation task, and given the
Anglia Climatic Research Unit’s website option to alter their original estimates in
(Jones, New, Parker, Martin, & Rigor, light of the briefing. Participants in the
1999). The data was interpolated and color- treatment condition changing their
quantized to produce five lateral regions estimations would provide evidence that
that emphasized the Mercator Effect, and they had identified and attempted to
then overlaid on a country outline map. account for the Mercator Effect.
Figure 1.0 shows a greyscale version the two
map projections given to participants. We
used the Image Color Summarizer tool
(Krzywinski, 2016) to calculate the face-
value areas for each shaded region,
measured as a percentage of the entire
image, such that the face-value areas for the
control map represented the undistorted
area values.
2.3 Procedure
To investigate the effects of projection
choice on data interpretation, we designed
an area estimation and threat perception
task. Participants were shown a global-scale
map with categorical data displayed (Figure
1.0), which they were told represented the
presence of five different pollutants over the
4 Web Maps for Global Data Visualization: Does Mercator Matter?
B 10.2 7.2 17.7 15.0
C 13.3 15.5 20.1 23
D 26.7 27.5 18.2 19.
E 42.3 45.9 20.7 26.0
3.2 Data interpretation and map
suitability
A chi-square test of independence was used
to examine the relationship between data
interpretation and projection choice. The
difference between conditions was
significant, 𝜒2 (1, N = 113) = 13.58, p < 0.01.
In particular, 17% of participants in the
equal-area condition (N = 53) perceived
pollutant A to be a greater threat than
pollutant E, compared to 50% in the
Mercator condition (N = 60). A chi-square
test was used to test for differences in the
answers to the map suitability questions. No
Figure 1.0: The equal-area (top) and Mercator significant difference was found between
(bottom) data visualizations given to conditions; participants did not judge one
participants via an online survey.
projection to be better than the other for
making area judgements.
3. Results
3.1 Area estimation 4. Conclusion
We tested for the effects of projection type The results from the area estimation and
using independent-samples t-tests to data interpretation tasks indicated that
compare the equal-area and Mercator participants’ judgements were significantly
conditions across the five area estimations affected by the choice of projection.
made by participants. We found a Specifically, participants took the maps at
significant difference across all the regions. face-value and interpreted the data
Specifically, participants overestimated the accordingly. This result was corroborated by
areas which had been enlarged by the responses to follow-up questions, which
Mercator projection, in line with face-value suggested that participants identified the
area judgements, as shown in Table 1.0. Mercator projection as being equally
Similarly, the control condition estimates appropriate to the control projection for
corresponded closely with the face-value area estimation tasks, as well as the fact that
measurements for the equal-area projection. they chose not to adjust their answers to the
Surprisingly, the answers to the second area second part of the survey.
estimation question did not differ
significantly from the original answers; Further work would be necessary to refine
while some participants in both categories the methods used in this study. It is possible
chose to alter their answers, most stuck with that some of the documented effects could
their original estimates. have been observed if participants had not
fully understood the wording of the
Table 1.0: Comparison of the mean estimate and questions. Additionally, there were several
face-value proportions (%) across conditions. unaddressed confounds between the two
Equal-area Mercator conditions which could have contributed
Mean Face- Mean Face-
towards the observed differences, such as
Region estimate value estimate value
A 7.5 4.0 23.3 16.2
the image dimensions and differences in
Web Maps for Global Data Visualization: Does Mercator Matter? 5
granularity between the maps which arose Mapping. Cartographica: The
due to scaling deformations. Despite these International Journal for Geographic
limitations, the central result, that the Information and Geovisualization, 49(2),
Mercator projection biases global data 85–101.
https://doi.org/10.3138/carto.49.2.2313
interpretation, has concrete implications for
Battersby, S. E., & Montello, D. R. (2009). Area
geovisualization and GIScience research. estimation of world regions and the
projection of the global-scale cognitive
This study has provided empirical evidence map. Annals of the Association of
for the Mercator Effect in web maps. We American Geographers, 99(2), 273–291.
found that individuals were unlikely or Batty, M., Hudson-Smith, A., Milton, R., &
unable to identify and re-project area data Crooks, A. (2010). Map mashups, Web 2.0
displayed on a Mercator projection to and the GIS revolution. Annals of GIS,
corresponding areas on the earth’s surface, 16(1), 1–13.
corroborating past research (Monmonier, https://doi.org/10.1080/19475681003700
831
1996; Robinson, 1966). Our framing of the
Downs, R. M., & Liben, L. S. (1991). The
tasks deliberately pointed towards the Development of Expertise in Geography: A
potential for misinterpretation of data in Cognitive-Developmental Approach to
real-world decision-making scenarios. More Geographic Education. Annals of the
broadly, the results emphasize the strong Association of American Geographers,
influence of cartographic design on public 81(2), 304–327.
interpretation of geographic information. https://doi.org/10.1111/j.1467-
Further work should critically assess efforts 8306.1991.tb01692.x
to address the Mercator effect in web maps, Elwood, S. (2011). Geographic information
such as the inclusion of gridlines, alternative science: Visualization, visual methods, and
the geoweb. Progress in Human
web mapping projections and adaptive
Geography, 35(3), 401–408.
maps (Jenny, 2012). Further GIScience Farman, J. (2010). Mapping the digital empire:
research can continue to broaden our Google Earth and the process of
understanding of the complex relationships postmodern cartography. New Media &
between visual representation and Society, 12(6), 869–888.
perception of geospatial information. https://doi.org/10.1177/146144480935090
0
Acknowledgements Haklay, M., Singleton, A., & Parker, C. (2008).
Web Mapping 2.0: The Neogeography of
the GeoWeb. Geography Compass, 2(6),
This study was completed in the 2011–2039. https://doi.org/10.1111/j.1749-
Department of Geography at McGill 8198.2008.00167.x
University under the supervision of Harpold, T. (1999). Dark continents: A critique
Professor Renee Sieber and Dr. Jin Xing, of Internet metageographies. Postmodern
both of whom I thank for their support and Culture, 9(2).
guidance. We also thank the two anonymous Jenny, B. (2012). Adaptive composite map
reviewers for their helpful comments on projections. IEEE Transactions on
Visualization and Computer Graphics,
earlier drafts of the manuscript. 18(12), 2575–2582.
Johnson, P. A., & Sieber, R. E. (2012).
Motivations driving government adoption
References of the Geoweb. GeoJournal, 77(5), 667–
680. https://doi.org/10.1007/s10708-011-
Amazon. (2014). Amazon Mechanical Turk - 9416-8
Welcome. Retrieved October 21, 2017, from Jones, P. D., New, M., Parker, D. E., Martin, S.,
https://www.mturk.com/mturk/welcome & Rigor, I. G. (1999). Surface air
Battersby, S. E., Finn, M. P., Usery, E. L., & temperature and its changes over the past
Yamamoto, K. H. (2014). Implications of 150 years. Reviews of Geophysics, 37(2),
Web Mercator and Its Use in Online 173–199.
https://doi.org/10.1029/1999RG900002
6 Web Maps for Global Data Visualization: Does Mercator Matter?
Kraak, J.-M., & Brown, A. (2014). Web Retchless, D. P., & Brewer, C. A. (2015).
cartography. CRC Press. Guidance for representing uncertainty on
Krzywinski, M. (2016). Image Color Summarizer global temperature change maps.
- RGB and HSV Image Statistics. Retrieved International Journal of Climatology,
December 14, 2017, from 36(3), 1143–1159.
http://mkweb.bcgsc.ca/color- https://doi.org/10.1002/joc.4408
summarizer/?analyze Robinson, A. H. (1966). The look of maps: An
Lapon, L., Ooms, K., & Maeyer, P. D. (2017). The examination of cartography design.
Influence of the Web Mercator Projection University of Wisconsin Press.
on the Global-Scale Cognitive Map of Web Saarinen, T. F. (1988). Centering of mental maps
Map Users. In Proceedings of Workshops of the world. National Geographic
and Posters at the 13th International Research, 4(1), 112–127.
Conference on Spatial Information Theory Šavrič, B., Jenny, B., White, D., & Strebe, D. R.
(COSIT 2017) (pp. 79–82). Springer, Cham. (2015). User preferences for world map
https://doi.org/10.1007/978-3-319-63946- projections. Cartography and Geographic
8_17 Information Science, 42(5), 398–409.
MacEachren, A. M. (2004). How Maps Work: https://doi.org/10.1080/15230406.2015.10
Representation, Visualization, and Design. 14425
Guilford Press. Shneiderman, B. (1996). The eyes have it: A task
Monmonier, M. (1996). How to lie with maps. by data type taxonomy for information
University of Chicago Press. visualizations. In Visual Languages, 1996.
Montello, D. R., Waller, D., Hegarty, M., & Proceedings., IEEE Symposium on (pp.
Richardson, A. E. (2004). Spatial memory 336–343). IEEE.
of real environments, virtual environments, Slocum, T. A., Blok, C., Jiang, B., Koussoulakou,
and maps. Human Spatial Memory: A., Montello, D. R., Fuhrmann, S., &
Remembering Where, 251–285. Hedley, N. R. (2001). Cognitive and
Nocke, T., Flechsig, M., & Bohm, U. (2007). Usability Issues in Geovisualization.
Visual exploration and evaluation of Cartography and Geographic Information
climate-related simulation data. In Science, 28(1), 61–75.
Simulation Conference, 2007 Winter (pp. https://doi.org/10.1559/152304001782173
703–711). IEEE. Retrieved from 998
http://ieeexplore.ieee.org/abstract/docume
nt/4419664/