ArticlePDF Available

Human interaction with animated maps: The portrayal of the passage of time


Abstract and Figures

Methods for interactive map animations are still in an early stage compared to more traditional cartography, and the potentials for improvements are significant. This paper focuses on better design for map animations, particularly for the portrayal of the information regarding the temporal dimension. A series of animations are presented and discussed and subsequently introduced to a wider group of people. In the responses from this group it was not possible to show any significant difference in how well people respond to the different animations, nevertheless people do have preferences for certain animation types depending on which time-scale they represent.
Content may be subject to copyright.
Human Interaction with Animated Maps:
The portrayal of the passage of time
Terje Midtbø1, Keith C. Clarke2 and Sara I. Fabrikant3
1 Division of Geomatics, Norwegian University of Science and Technology
Høgskoleringen 7, 7491 Trondheim, NORWAY
Phone: +47 73 59 45 81
2Department of Geography, University of California at Santa Barbara
3611 Ellison Hall, Santa Barbara, CA 93106-4060, USA
Phone: +1 805 893 3663
3 Department of Geography, University of Zürich
Winterthurerstr. 190 CH-8057 Zurich, SWITZERLAND
Phone: +41 44 635 51 50
Abstract. Methods for interactive map animations are still in an early stage
compared to more traditional cartography, and the potentials for improvements
are significant. This paper focuses on better design for map animations,
particularly for the portrayal of the information regarding the temporal
dimension. A series of animations are presented and discussed and
subsequently introduced to a wider group of people. In the responses from this
group it was not possible to show any significant difference in how well people
respond to the different animations, nevertheless people do have preferences for
certain animation types depending on which time-scale they represent.
Key words: Dynamic maps, interactive maps, visualization, map animations,
spatio-temporal information.
1. Introduction
People have used maps as tools for description of the geographical environment for
ages, while the media for transmission of geographic information have changed over
the years. Materials like stone, mammoth teeth, clay tablets, silk, papyrus, wood,
copper, paper etc have been involved. Today most maps are stored electronic, and
many of them are passed on to the user trough a computer screen. While cartography
has been concentrated on static illustrations during thousands of years, computer
based methods open for a new category of maps: the map animations and dynamic
maps. These maps offer something that never could be a part of the traditional maps;
movement and change in the map. However, the quantity of map animations has not
really increased that much until the last 10 years. And it is quite obvious that the
Internet and the World Wide Web (WWW) has been the most important catalyst for
this increase. Still most of the maps on the web remain static, but several high level
programming languages and applications for preparation of animations are now
making the production of animations easier. However, even though the technical
framework is becoming more accessible we still need more knowledge on how these
animations should appear.
Research has begun to examine how effective animations are for the
communication of information. Reviewing empirical animation research Tversky
(2002) found that many animations are too complex and too fast to be accurately
perceived. The necessity of making animation without too much detail was pointed
out by Harrower (2003). The information has to be well arranged and include as few
details as possible to be effective. Tversky (2002) also pointed out that animations
might generally be more effective when they are interactive, and she makes a sharp
distinction between non-interactive and interactive animations. Although
Tversky(2002) reviewed experimental work dealing with animations in general and
not map animations in particular, she found no advantage in using animations
compared to using a series of static pictures (Tversky, 2002). In the reviewed
experiments, the length of the time period for studying the figures/animation did not
seem to be a controlled variable.
Geographic space is central to all geographic information, yet almost all
geographical data are also highly structured in time. Like space, time has granularity
and extent. In geology and geomorphology, changes occur on a very long-term time-
scale, while in meteorology and traffic analysis, the dynamics are hour to hour and
minute to minute. Cartographic animation has emerged as an important tool for the
interpretation of time-space information (Peterson, 1995). Before the computer age
such animations were rare. The most famous early cartographic animation may be
Disney’s visualization of the German attack on Warsaw in 1939 (Peterson 1999).
Using movie technologies was rather expensive and time consuming, so there were
few examples in cartography until Tobler’s animation of the population growth of
Detroit (Tobler, 1969).
While most map animations focus on depicting spatial change using visual
variables (Bertin, 1981) as extended for dynamics (DiBiase et al., 1992), the design
aspects of temporal change have often been overlooked. Koussoulakou and Kraak
(1992) pointed out that the mapping of spatial data’s multi-dimensional and temporal
component would be an important challenge in the nineteen-nineties. During the last
decade we saw many examples of animations for this kind of data, but there remain
many unsolved research issues connected to the problem. Other than animations
involving changing the map viewing geometry for an unchanging map, most
phenomena that are visualized in an animation do have a strong temporal message. In
such presentations, time itself is an important variable that has to be included, and
clearly needs to be an easy perceivable component of the map. According to Bertin
(1981) only one variable, in addition to the position, should be presented on a
thematic map when the aim is to make a "seeing map", i.e. to keep the perception of
the contents on a global level. In many applications, on the web, it is indeed the only
variable shown on the map. However, on these maps time is symbolized as text, often
46 Proceedings, ScanGIS’2007
“hidden” on the fringe of the application. Slocum et al. (2000) state that it is difficult
to keep one’s focus on the main phenomenon in the animation and to read the text
outside the map simultaneously. One solution might be to use some kind of sound
(Midtbø, 2000), but this is restricted to sequential information, and is consequently
less suited for interactive animations. To keep the information visual we can use an
animation for the temporal component as well as for the main phenomenon. Kraak et
al. (1997) proposes the use of dynamic legends describing the passage of time. In an
experiment Edsall et al. (1997) depicted the passage of time as separate animated
legends, placed below the animated map. However, their results were inconclusive,
showing no significant differences in performance or speed among any of the legends.
In this paper we introduce some examples on animation where we attempt to
explore how visual variables can be applied to depict the passage of time jointly with
spatial change. According to Slocum et al. (2000), the temporal legend which has the
width of the map itself might be easier to interpret than a legend occupying a small
portion of the display window. In some of our examples the temporal legend has both
the width and the height of the animation and is, in one way, incorporated into the
map animation itself. The ideas for some of the presented animations were originally
introduced in Midtbø (2000).
2. Animation of time
This paper looks closer into how the presentation of “time” can be included in an
animation. This involves several problems to be addressed like such as; differences in
time scale (one hour, one decade etc.) and different techniques for the presentation of
time within these scales. Cross testing display combinations makes for a wide variety
of animations when the various methods are compared.
For the visualization of the temporal component it might be appropriate to
consider how people perceive time and how natural phenomena influence our
perception of time. Block (1990) outlined how humans perceive time. Three
important factors are succession, duration and temporal perspective. The length of
perceived episodes and the order of incidents are of importance. People also use
“tags” from various experiences as “temporal indexes” much as they use landmarks
during navigation. Natural phenomena, like the earth moving around its own axis and
the earth moving around the sun, pave the way for perception of time as a repeating
pattern. For such repeating temporal patterns it is reasonable to consider a circular
model for the temporal component rather than the more common linear model.
Multiple ways of representing the passage of time in map animations were
designed. Most were based on incorporation of the temporal component into the
animation itself. The “time legend” used in these methods encircles the map and is no
longer situated outside the animation. The intent was to use the peripheral vision of
the map interpreter to visualize the temporal component at the same time as the main
phenomenon is studied in the map animation. Care must be taken when deciding
which visual or dynamic variables to use. It is important that the variable used for
visualizing time differs from the variable used for the main phenomenon. In most of
our examples, dynamics in the thematic map animations are represented by changing
Midtbø, Clarke and Fabrikant 47
the size of circles on a representative proportional circle map. This means that the
visual variable “size” is used (Bertin, 1981), and also to some degree the dynamic
variable “rate of change” (DiBiase et al. 1992). Hence, these variables should be
avoided when visualizing time in the same animation. The list below shows an
overview of the animation types that were evaluated. The various animation types can
be studied at:
xBackground shadow (Figure 1). In these animations the background of the map
changes as a shadow is moving from one end of the map to the other. Alongside
the map there is a time scale, which the shadow passes over. The darker area
(shadow) may move from left to right, right to left, from above to below or from
below to above the map. In our case the shadow moved from left to right behind
the map animation. It is also possible to use a circular movement of the shadow for
a cyclic visualization of the temporal component. Another type in the same
category is when the background of the animation initially is dark, and a lighter
background “slides” in to denote change in time. We also tried a highly transparent
shadow that slides in on top of the map animation. However, this type had less
positive effects on the visualization. By using a moving background shadow, it is
the visual variable position that represents time and the front of the moving shadow
that indicates the present point of time.
xAnalog clock hand (Figure 2). For temporal phenomena that have a circular
character it is possible to use a clock hand as time indicator. The aim is still to
incorporate the visualization of time in the animation, so the clock hand rotates
behind the map and points at a time scale that encircles the map. While this figure
illustrates a circular animation, it is also possible to use a pointer that moves along
a linear scale. The pointer indicates time in the same way as the front of the
shadow in our previous example. It seems appropriate to consider such an
animation for phenomena that are cyclical in a 24-hour period. However, an analog
Figure 1: Background shadow
48 Proceedings, ScanGIS’2007
clock animation will, by most people, be identified as a 12-hour cycle. The
background intensity can be used to distinguish between night and day in a 24-hour
period. We also considered some animations where the hand was on top of the
map, but this seemed less suitable.
xSatellite (Figure 3). Instead of the “clock hand” a satellite object is moved along
the time scale indicating the point of time. Since this is different from an analog
clock with hands, the method might be more appropriate when showing
phenomena with other cyclical time durations than 24-hours. Figure 3 shows
temperature changing over one year. Like the other animations this is interactive. It
is possible to start and stop the animation with the control panel, and the animation
can be run forward and backward continuously or step by step. In addition it is
possible to jump to a chosen time in the animation by clicking on the small
stationary and transparent satellites along the time scale. This “time jump function-
ality” seems particular useful. Finally it is possible to move the legend circles to
compare them with the growing and shrinking circles in the map on close range.
Figure 3: Satellite
Figure 2: Analog clock hand
Midtbø, Clarke and Fabrikant 49
xSun/moon icon (Figure 4). For a 24-hours period the satellite can be substituted
for a sun/moon icon. To point out the difference between night and day, the icon
changes from a moon to a sun at 06:00 in the morning, and back to a moon at
06:00 in the evening. An alternative type of this animation is shown in figure 4.
Here the animation is based on the fact that the sun rises and sets behind a horizon.
The sun icon is moving along a semicircle over the “map horizon” at daytime. At
“sunset” a moon icon starts in the same orbit. In addition to the alternating icons it
is possible to change the background intensity to emphasize the day/night
transition. The drawback of the method is that the time scale is no longer directly
compatible with an analog clock. Here a combination of the visual variables
position,form and intensity represents the temporal dimension.
xEmbedded hands (Figure 5). This time illustration depends on the visual variable
that is used to show the primary phenomenon. In our example we use proportional
Figure 4: Sun/moon icon
Figure 5: Embedded hands
50 Proceedings, ScanGIS’2007
circles. As we can see from Figure 5, a hand is embedded in each circle on the
map. All the hands are moving in parallel and their positions/angles denote the
point of time. This analog clock type animation depends on a circular
representation of the primary phenomenon. It is however possible to imagine hands
embedded in other symbol types. One problem with the method is that it might be
more difficult to comprehend the time information when the circles become very
xGrowing bar (Figure 6). This is may be one of the few of our examples that can be
found in other applications. In Figure 6 the bar is growing along a time scale. The
bar is usually horizontal, but vertical variants could be considered. As Slocum et al.
(2000) suggests, our growing bar has the width of the map itself, and is not situated
in a peripheral position in the window. In the growing bar it is mainly the variable
Figure 6: Growing bar
Figure 7: Alternating pictures/icons
Midtbø, Clarke and Fabrikant 51
position that shows time, partly assisted by the variable size. Because of the
qualities of our peripheral vision it is important to give the bar an intensity that is
in contrast with the background.
xAlternating pictures/icons (Figure 7). In human perception of time information is
“stored” in the brain and “linked” through a kind of indexing. Much of this
indexing is individual, but to some degree common experience leads to a common
understanding. This fact is utilized in the moon/sun example. Most people connect
the sun icon with daytime and the moon icon with nighttime. Other icons can be
employed for groups that have common interests. Figure 7 shows an animation in
which different pictures of cars are used for time indexing. Each car represents a
certain decade. In this animation the car models and styles change as the cars move
along the time scale. Consequently, a combination of the variables position and
form are used to visualize the temporal dimension. In one of our animations the
changing cars only represented time (no change in position). Compared to the
animation from Figure 6 this animation, which was based on the variable form
only, seemed less interpretable.
xAlternating text (Figure 8). This represents the most commonly used method for
the representation of time in map animations on the Web today. The time is
indicated by dates, years and clock hours or similar, written as text and numbers.
These are often represented by too small a font and are situated too far outside the
animation. Their distinctness can be improved by increasing the font size, and by
moving the text closer to the animation.
xSliding map (Figure 9). In all the animations the presentation of time has a central
position, but in this type the temporal dimension is even more central. The time
scale is dominant in the display, and the map animation slides along the time scale.
The position of the map itself denotes the point of time. If the map reader is
studying a certain site on the map, this site can be connected to the time scale.
Unfortunately, the map has to be rather small in this animation to avoid inaccuracy
Figure 8: Alternating text
52 Proceedings, ScanGIS’2007
in the perception.
xTransparent/emphasized map (Figure 10). This animation permits more
Figure 9: Sliding Map
Figure 10: Transparent/emphasized map
Midtbø, Clarke and Fabrikant 53
interactivity than animation. As in the sliding map, the time scale is central. A
series of small mutliples are located along the time scale. Initially all the maps are
transparent. By moving the mouse pointer over a map it is highlighted by reducing
the transparency. The position of the map in the time scale supplies the temporal
xMouse over legend (Figure 11). Here we use an active legend (Peterson 1999)
where the temporal change in the animation is controlled by moving the mouse
pointer along a time scale. In Figure 11 the time scale consists of years written as
text. Both the text itself and the position of the text represent the temporal variable.
3. Testing the animations
The animations presented in Section 2 is based on fundamental cartographic
theory, some triaal and errors and the autors subjective ideas for design. However,
would these animations be attractive and easy understandable for a wider group of
people? And which kind of animation would be best adapted to different time-spans?
Fore answering these questions an experimental web-interface was designed. More
details about this application can be studied in Midtbø et al. (200x).
Initially, many different animations were considered for exploration and testing.
For the final trial it was decided to concentrate on a subset of the characteristics of
time animation. A principal conceptual difference among the methods outlined above
is their treatment of time as a circular variable, a linear variable, or a simple text
string. The various tests in the application was designed and organized to explore
different aspects of the animation of time. First, they were intended to investigate
whether this basic division of the animations was reflected in peoples’ perceptions.
Secondly, the tests focused on whether any of the animation types were better suited
for particular time periods. Accordingly, for each animation type (circular, linear and
Figure 11: Mouse over legend
54 Proceedings, ScanGIS’2007
text) we introduced four different time spans: a 24-hour period, 12 days, 1 year, and 4
decades. The conterminous United States (excluding Alaska and Hawaii) was used as
the geographical extent for the experiment. For the animations’ primary phenomenon,
various kinds of weather information were used for time spans up to 1 year, and
population growth was used for variations over decades. Table 1 shows the data sets
together with their corresponding time periods. All the variables were represented
similarly, by proportional symbols, using growing and shrinking circles. For each
phenomenon three different data sets were prepared, each based on information about
weather and population in seven cities situated around the country (Figure 12).
Table 1: Time spans and corresponding phenomena
Figure 12: The Three Distributions of Cities
Each of the animations presented above were implemented in Macromedia
Director to contain an interactive control panel. This included “move forward”,
“move backwards”, “stop”, “step forward” and “step backwards” buttons. It was also
possible to jump in the animations by clicking on the markers along the time-scale.
This detail made the repeated study of a certain time-period efficient. Finally, to get a
better comprehension of the size of the proportional circles in the animation, it was
possible to move the transparent circles from the legend into the animation itself. In
order to make the recording process of the interaction simpler and easier to investigate
user reaction to the animations, this control panel was simplified for the versions
presented on the web. All buttons were made inactive except the start button. In
addition a button to terminate the animation was introduced. This made it possible to
Time span Primary
24-hours Wind speed (I.e.) 1 hour Size Display date,
rate of change
12 days Temperature above
and below normal
temperature (F°)
24 hours Size, color
Display date,
rate of change
1 year Temperature above
and below normal
temperature (F°)
1 month Size, color
Display date,
rate of change
4 decades Population growth (%) 10 years Size, color
Display date,
rate of change
Midtbø, Clarke and Fabrikant 55
measure how long the animation was viewed by the users. On the other hand, this
excluded most kinds of interaction when the animation was running.
For each animation in the experiment a question was asked before the user got
access to the animation. The user was asked to study two particular cities on the map
animation. The question was accompanied by a figure that pointed out the two cities.
Figure 13 shows an example how the questions were asked. Next, the animation was
started and the viewer was asked to terminate the animation as soon as he/she knew
the answer to the question. Finally, the audience had to answer the question by
choosing between alternatives. Table 2 shows the alternatives for the different time
Table 2: Alteratives in the questionnaire
Time span No. of alt. Alternatives
24 hours 8 6pm-9pm, 9pm-12am, 12am-3am, 3am-6am,
6am- 9am, 9am-12pm,12pm-3pm, 3pm-6pm
12 days 4 5-7, 8-10, 11-13, 14-16 (dates)
1 year 4 Jan.-Mar., Apr.-Jun., Jul.-Sep., Oct.-Dec.
4 decades 4 1960-1970, 1970-1980, 1980-1990, 1990-2000
Figure 12: Questions asked before the animation
56 Proceedings, ScanGIS’2007
4. Accomplishments and results
The application for testing the animation on a larger group of people was
announced on some major cartography interests discussion lists in USA. After some
weeks we had received 133 answers, most of them during the first week after the
announcement. Forty-five of these were women, and 101 out of 133 claimed to have a
formal education within Cartography (not a surprise considering the mailing lists for
announcement of the experiment).
In the questionnaires the participants were asked to check off the time periods that
they believed reflected the data shown in the animations. The results from the
questionnaires were quite varied, so initially we studied the time each participant
spent watching the animation before he/she terminated it to answer the questions.
Some answers were rejected due to slow computers on the client side, some
evident misunderstandings etc., leaving 91 subjects. The mean value for the measured
time the participants used before answering the questions showed a marginal
difference between the three animation types. Most variation appeared for the 4
decades time span, which was presumed to be the animation with the easiest time
representation. The results showed a mean value of 12 seconds for the linear
animation, 10.3 seconds for the circular animation and 13.1 second for the text
animation. However, since the standard deviations are respectively 9.4, 6.8 and 8.1
the analysis of variance showed no significant difference between the animation
After the participants had completed the four animations, examples of the two
remaining animation types were introduced. They were then asked some questions
about which type of animation they preferred for the various time spans. The results
from these questions are shown in Table 3. There are clear preferences for some of
the animation types in combination with particular time spans. The majority of the
participants preferred the circular animations for the visualization of a 24-hour period,
while the linear animations seem to be most attractive for time spans of a non-cyclic
character. About 60% of all the participants prefer the linear animation for the 12 days
and the 4 decades periods. One year is a common period related to a natural
phenomenon. However, from the table we can see that the group that prefer a linear
animation is about the same size as the “circular group.”
Table 3: Preferred animation types
24 hours Year 12 days 4 decades
Prefer linear animation 23.3% 45.1% 57.2% 61.7%
Prefer circular animation 68.4% 45.9% 29.3% 24.8%
Prefer textual animation 8.3% 9.0% 13.5% 13.5%
Midtbø, Clarke and Fabrikant 57
Table 4: Preferred animationtypes dependent on employed animation in the experiment
24 hours Year 12 days 4 decades
Linear animation in experiment
Prefer linear animation 29.8% 53.2% 68.1% 74.5%
Prefer circular animation 66% 44.7% 21.3% 17.0%
Prefer textual animation 4.2% 2.1% 10.6% 8.5%
Circular animation in experiment
Prefer linear animation 32.6% 45.7% 56.5% 60.9%
Prefer circular animation 63.0% 50% 34.8% 32.6%
Prefer textual animation 4.3% 4.3% 8.7% 6.5%
Textual animation in experiment
Prefer linear animation 8.7% 34.7% 43.5% 47.8%
Prefer circular animation 76.1% 45.7% 34.8% 28.3%
Prefer textual animation 15.2% 19.6% 21.7% 23.9%
In Table 3 the preferences for all the participants are included, independent of
which type of animation they were assigned during the experiment. However, we are
actually looking at three different groups of participants; a group that used the circular
animations, another that used the linear animations and a third that used the textual
description during the experiment. Were the preferences of the participants in the
different groups influenced by the animation used in the experiment? Table 4 shows
the preference for animation types based on which type of animation that was
employed during the experiment.
If we compare the average results in Table 3 with the numbers in Table 4 there is a
tendency for participants to favour the animation type to which they were exposed
during the experiment. There is only one exception from this tendency; for the
participants in the “linear group” the preference for a circular animation for a 24-hour
period is greater than average. Correspondingly their preference is lower than average
in the “circular group”.
5. Conclusions
It is our belief that map animations of today have significant potentials for
improvements. In this paper we have presented some methods for the portrayal of the
passage of time, and tested some of these on an audience. The practical case in the
experiment is focused on people’s reaction to various animations representing time.
We measured the time each participant studied the animations before he/she was
prepared to answer a series of questions. The results gave no significant indications
that one of the animation types in the experiment was better suited for temporal
representation. This could be a negative result, or it could indicate that different
people have different personal preferences. However, the answers to the questions
were quite scattered, and made it difficult to make any conclusions. By studying the
data we can see that the questions related to the animations were too ambitious for
58 Proceedings, ScanGIS’2007
this type of experiment. To discover the usefulness of the various temporal
representations in web-based experiments it is necessary to prepare simpler and more
unambiguous tasks.
On the other hand there was a distinct difference in which type of animation the
participants preferred. For the representation of a period of non-cyclic nature, most of
the participants preferred a linear animation for the representation of time. For the
natural cyclic period of a year, the preference was about even between the circular
animation and the linear animation, while most participants preferred a circular
animation for a 24-hour period. It was also evident that most of the participants
preferred a figurative animation above representing time by changing text.
1. Bertin, J. (1981). Graphics and Graphic Information Processing. Translated by
William J. Berg and Paul Scott. Walter de Gruyter, Berlin*New York.
2. Block, R.A. (1990). “Models of Psychological Time”. In Block, R.A. (ed.)
Cognitive models of psychological time. Lawrence Erlbaum Associated, New
Jersey, 1-35.
3. DiBiase, D. (1992) “Stretching Space and Splicing Time: from cartographic
animation to interactive visualization.”, Cartography and Geographic Information
Systems. 19(4):215-227;
4. Edsall, R.M., Kraak, M.-J., MacEachren, A.M. and Peuquet, D.J. (1997).
“Assessing the effectiveness of temporal legends in environmental visualization.”
In: Proceedings of GIS/LIS `97, Cincinnati, Ohio, pp 677-685.
5. Harrower, M. (2003). “Tips for Designing Effective Animated Maps”,
Cartographic Perspectives, Number 44 (Winter 2003):63-65.
6. Koussoulakou, A. and Kraak, M-J. (1992). “Spatiotemporal maps and cartographic
communication”, The Cartographic Journal, 29(2):101-108.
7. Kraak, M-J., Edsall R. and MacEachren, A.M. (1997). “Cartographic animation
and legends for temporal maps: exploration and or interaction”, Proceedings 18th
ICA/ACI International Cartographic Conference, 23-27 June 1997, Stockholm,
Sweden, 253-260.
8. Midtbø,T (2000) ”Visualization of the temporal dimension in multimedia
presentations of spatial phenomena”, Proceedings 9th International Symposium on
Spatial Data Handling, 10-12 August, Beijing, People’s Republic of China.
9. Midtbø, T,, Clarke, K.C. and Fabrikant, S.I. (200x). “Web-based experiments in
the study of human interaction with animated maps”. To be published.
10.Peterson, M.P. (1995). Interactive and Animated Cartography. Prentice Hall, New
11.Peterson, M.P. (1999). “Active legends for interactive cartographic animation”,
International Journal of Geographical Information System, 13(4):375-383.
12.Slocum, Terry A., Yoder, Stephen C., Kessler, Fritz C., Sluter, Robert S. (2000).
“MapTime: Software for Exploring Spatiotemporal data Associated with Point
Locations”, Cartographica, vol 27, no 1, spring 2000, pp 15-31
Midtbø, Clarke and Fabrikant 59
13.Tobler, W. R. (1969) “A computer movie simulating urban growth in the Detroit
region.” Paper prepared for the meeting of the International Geographical Union,
Commission on Quantitative Methods, Ann Arbor, MI. August.
14.Tversky, B., Morrison, J.B. and Betrancourt, M. (2002). “Animation: can it
facilitate?”, International Journal of Human-Computer Studies (2002) 57, 247-262.
60 Proceedings, ScanGIS’2007
... It also depends on the flow of the animation in regard to order and synchronization [45] . For example , Midtbø et al. [50] found out that the usage of animation types depends on the used time span. The participants in their study preferred circular animations for time spans with repeating patterns and linear animations for time spans which had non-cyclical character. ...
... Interaction possibilities allow users to control the animation which can improve the visual exploration of the data than for the case if users only watch animation passively [11, 74]. Most approaches (see, e.g., [6, 10, 20, 26, 50, 52, 59, 61, 71] ) consider wellknown and widely understood standard Video Cassette Recording (VCR)-style control elements like play, pause, (fast) forwarding, and (fast) rewind in combination with a slider in order to control the speed (see Figure 6for an example). User interactions are analyzed in several studies (see, e.g., [44, 52, 74]) with the insights that a) interactivity can be good solution to overcome the difficulties of perception and b) the comprehension in animations helps users to get a feeling of the data. ...
... Additional to the interaction possibilities in order to influence the flow of an animation , a combination of animation with further indicators like colors, traces or sound is recommended in the evaluation studies (see, e.g., [10, 15, 49, 50, 59, 66, 71, 82]). For example, indicators like colors or sound can be used to lead users' attention to specific parts in the animation or to specific animated objects. ...
Full-text available
Representing time-dependent data plays an important role in information visualization. Time presents specific challenges for the representation of data because time is a complex and highly abstract concept. Basically, there are two ways to support reasoning about time: time can be represented by space, and time can also be represented by time (animation). From the point of view of the users, both forms of representation have their strengths and weaknesses which we will illustrate in this chapter. In recent years, a large number of visualizations has been developed to solve the problem of representing time-dependent data. Nevertheless, it is still not clear which types of visualizations support the cognitive processes of the users. It is necessary to investigate the interactions of real users with visualizations to clarify this issue. The following chapter will give an overview of empirical evaluations and recommendations for the design of visualizations for time-dependent data.
... A particular area that provides relevant information regarding spatio-temporal visualizations, like the one used in paper I, is the use of Geographical Information Systems (GIS). Research in GIS has indicated that the influence of various time representations (a timeline, a time wheel, and a text-based legend) is of minor importance to users (Edsall et al., 1997;Midtbø, Clarke, & Fabrikant, 2007). Relevant for this thesis is also the research performed by Vít and Bláha (2012), which showed that time represented as a temporal axis, with time units of consistent length and a movable cursor indicating the rate of time, proved to be most user-friendly in comparison with a representation with units of varying length and an animated cursor moving at a constant speed. ...
... Thus, informed decisions have to be made when choosing an appropriate visualization for specific educational purposes. In contrast to some prior research (Midtbø, Clarke, & Fabrikant, 2007), this study indicates that the design of the time representation does matter, and can affect learners' abilities to comprehend at least some temporal aspects. ...
... It has shown how different representations of time affect users. In a study carried out by Edsall et al. (1997), and later confirmed by Midtbø et al. (2007), three different representations of time were investigated: a timeline, a time wheel and a text-based legend. No significant differences in the performance among groups working with the different representations could be shown. ...
... Through an analysis of four representations of time in a visualization of hominin evolution, we argue that informed and conscious choices of animations with appropriate depictions of time can facilitate the teaching and learning of subject matter regarding evolutionary scenarios involving deep time. In contrast to earlier research (Midtbø et al. 2007), this study indicates that different representations affect learners' abilities to perceive and comprehend at least some temporal aspects. In line with results from geographic information systems studies, time units of consistent length were proven to be most helpful (Vít and Bláha 2012) in giving students a sense of the relative time spans of evolution, which is one of the most difficult aspects to master in understanding evolution. ...
Full-text available
Background Macroevolutionary time is a difficult idea to grasp and is considered to be a threshold concept in teaching and learning evolution. One way of addressing this subject is to use animations that represent evolutionary time. The aim of this descriptive and exploratory study was to investigate how various representations of time in an animation affect the way undergraduate students comprehend different temporal aspects of hominin evolution. Two factors, namely differences in timelines (the number of timelines with different scales) and the mode of the default animated time rate (either constant throughout the animation or decreasing as the animation progressed) were combined to give four different time representations. The temporal aspects were investigated using undergraduate students’ ability to find events at specific times, to comprehend relative order, to comprehend concurrent events, to estimate the duration of time intervals and their ability to compare the lengths of time intervals. Results The results revealed that “finding events at specific times” near to the end of the animation (closer to present time), where the sequence of events appeared very quickly, was more difficult for groups working with animations with only one timeline. We also found that the ability to comprehend concurrent events can be impaired if several timelines are displayed and the animation speed is relatively high. The ability to estimate the duration of a time interval was more difficult for groups working with animations with only one timeline, especially at the end of the animation where the sequence of events occurred quickly. Making correct comparisons of time intervals was relatively independent of which animation was used with one notable exception: groups working with an animation featuring several timelines and a decreasing default animated time rate performed worst at comparing events with intervals that spanned parts of the timeline with different scales. Conclusions Our results indicate that the choice of animation should depend on the teaching intention. However, a visualization with several timelines, and an animated time which slowed down toward present time, generated the best results for the majority of items tested. Temporal scale shift may interfere with the perception of time in cases where durations are compared.
... Some of these interactive solutions were part of the research agenda during desktop studies. Different levels of interactivity for specific tasks helped reduce cognitive load and address some perceptual problems (Midtbø et al. 2007). ...
Full-text available
This study investigated the impact of graphical user interface (GUI) design on the efficiency and effectiveness of map-based tasks on mobile devices, using time-based weather data as a case study. Three different GUI designs (button-type, circle-type, and slidebar) were tested in a between-subjects design, with 50 participants completing a set of map-based tasks on each GUI design. The results showed that GUI design significantly affected the effectiveness of map-based tasks. Participants performed better at tasks involving the search for the highest and lowest temperature amplitudes on the button-type GUI whereas the circle-type GUI showed lower effectiveness for tasks involving the search for day temperatures. Analysis of the visual attention distribution based on fixation count revealed that different GUI designs led to different patterns of visual attention. The study highlights the importance of considering GUI design in the development of mobile map applications, particularly for map-based tasks involving time-based data. The study shows that separating the date from the time navigation panel reduces necessary visual focus on the GUI itself and is a valuable insight for future GUI design.
... The experiment was concluded after eight days, as earlier web-based experiments show that most participants respond a short time after an experiment is published online (Midtbø, Clarke, & Fabrikant, 2007). 164 participants had registered and completed at least one task-set. ...
Crowd-sourced geospatial data can often be enriched by importing open governmental datasets as long as they are up-to date and of good quality. Unfortunately, merging datasets is not straight forward. In the context of geospatial data, spatial overlaps pose a particular problem, as existing data may be overwritten when a naïve, automated import strategy is employed. For example: OpenStreetMap has imported over 100 open geospatial datasets, but the requirement for human assessment makes this a time-consuming process which requires experienced volunteers or training. In this paper, we propose a hybrid import workflow that combines algorithmic filtering with human assessment using the micro-tasking method. This enables human assessment without the need for complex tools or prior experience. Using an online experiment, we investigated how import speed and accuracy is affected by volunteer experience and partitioning of the micro-task. We conclude that micro-tasking is a viable method for massive quality assessment that does not require volunteers to have prior experience working with geospatial data.
... Časová legenda je ekvivalentem "klasické" legendy a znázorňuje vztah reálného času (tj. čas, ve kterém se znázorněná tematika opravdu odehrála) a jeho obrazu v animované mapě (Mitbø, Clarke, Fabrikant 2007;Vít 2010). ...
Thearticleprovides an analysis of the possibilities for the graphic presentation of time in animated cartographic works. After a brief introduction presenting the problem of capturing time and action within the framework of cartographic production, the article mentions methods commonly used for the expression of time in cartographic animations and indicates their weakness: as a rule, these methods don't reflect the temporo-spatial progression of events that a cartographic work strives to capture. The article then proposes a general approach recommended for the design and creation of temporal legends for animated maps in order to eliminate this flaw. This general approach is then applied to a sample group of cartographic works that depict historical battles.
... Tversky et al. (2002) reports that animations are not necessarily better than static maps. Several issues associated with map animations are of concern, such as split attention often related to the temporal legend (Midtbø et al. 2007) and disappearance (Harrower 2009, Tversky et al. 2002. The static counterpart is often used as a means of avoiding these issues. ...
Full-text available
Visualisation of spatio-temporal information has been a challenge for cartography for decades. The earliest approaches focused on static visualisations, several of them with great success (Monmonier, 1990; Friendly, 2002; Tufte, 2006). With the digital age, dynamic and animated maps have become possible. Much work has gone in to the design and study of perception of both of these map types (Harrower et al., 2000; Midtbo, 2001a; Midtbo et al., 2007; Fabrikant et al., 2008) Several very effective animated maps have been developed and rigorously evaluated. Both static and animated visualisations have their distinct qualities. Static maps are able to provide the user with unlimited viewing time of the information, allowing the level of detail to be high. On the other hand, static maps do not have the intuitiveness of visualizing temporal information due to their inherent static nature. Animated maps are often more intuitive in their representation of temporal information. Displaying frames of information in rapid succession is perceived as more intuitive representation of the temporal information. However, the fact that the frames have a very limited viewing time often hinders the user from perceiving and gaining knowledge from the information (Harrower, 2009). There has been a sharp distinction between animated and static maps in cartography. Evaluation of either two has been following this distinction and often compared animated against static maps. In this article, we propose moving beyond the sharp distinction of animated and static maps by combining the qualities of both in one new representation method. To emphasize the concept's qualities from both static and animated maps, we call this concept, semistatic animations. The core idea of the concept is to make all information visually available to the user at any given time of the animation. This allows the user to look both back and forth in the animation, while it is playing without interaction. The design draws inspiration from both static multiple map displays and diagram maps - qualities from both of these are integrated in an animated representation. A proof-of-concept implementation has been made using weather map animations. In order to assess the perceptual performance of the concept, a web experiment has been conducted. Realizing that the maps do not necessarily support all usage scenarios, a task-oriented approach was applied on the experiment. In total, the experiment included 240 participants. Results from the experiment revealed that the semistatic concept significantly increases the performance on several of the tasks compared to the equivalent traditional animated map. However, for some of the tasks, the semistatic approach did not improve the performance, or worsened it. This illustrates the importance of evaluating not only on a general level but delve further in different user tasks. The results obtained additionally motivates for further investigation of the semistatic concept as well as user behaviour.
Full-text available
Animated maps provide an intuitive method for representing univariate time-series data, but often fail in presenting additional relevant information saliently, making recognition of certain patterns difficult. Using a second visual variable in animations to represent the magnitude of change between time states has been suggested as an effective method for enabling users to more easily recognize patterns of change in a geographic time-series. This work seeks to answer the question: Does explicitly representing geographic change in animated maps enable users to answer questions about patterns of change easily? To address this research question, bivariate symbols (with both the value of the data and the magnitude of change between time frames represented) were created and tested. Selective attention theory (SAT) was used in selecting bivariate symbol types (separable and integral). Domain analysis with experts from the Avian Knowledge Network (AKN) was performed to determine appropriate map reading tasks for use in task-based experiments using AKN data. Combined with existing task typologies, material from the domain analysis helped form a new task typology of movement patterns found in aggregated spatiotemporal point data. Formal task-based experiments followed, where participants were placed into one of five experiment groups (each using a different symbol) and asked to perform the same series of statement agreement and certainty ratings while studying map animations. Results show that aside from questions explicitly about change, univariate non-change symbolization may be most appropriate. Future studies should focus on testing different data relationships (independent, interdependent, or unrelated) with symbol variations that may have different attention behaviors as predicted by SAT. The results presented here improve the understanding of whether explicit change symbolization helps elucidate geographic time-series patterns or hinders the overall effectiveness of map animation.
Conference Paper
The usage of visualizations to aid the analysis of time oriented data plays an important role in various fields of applications. The need to visualize such data was decisive for the development of different visualization techniques over the last years. One of the frequently applied techniques is animation in order to illustrate the movements in such a way to make changes in the data transparent. However, evaluation studies of such animated interfaces for time-oriented data with potential users are still difficult to find. In this paper, we present our observations based on a systematic literature review with the motivation to support researchers and designers to identify future directions for their research. The literature review is split in two parts: (1) research on animation from the field of psychology, and (2) evaluation studies with the focus on animation of time-oriented data.
Despite dramatic increases in the availability and cartographic processing capabilities of Geographic technologies, state delimited red and blue maps remain the dominant form of electoral result visualization disseminated by the mainstream media. While scaling the human voting process to physical geographic space and spatial aggregation are inherent weaknesses of red and blue state mapping, the power of such maps in affecting political perceptions is undeniable. Cartograms are familiar to most geographers but largely unknown to the voting public, and can produce intuitive electoral maps that overcome the limitations and biases of this traditional format. In addition, animated time series cartogram electoral visualizations can further understanding of the population shifts and demographic changes that underlie the continual realignment of national electoral constituencies. Application of these techniques provides the basis for a new paradigm in presidential election visualization that may reverse the popular misconceptions associated with traditional election maps. Animated cartograms can overcome the spatial mismatch inherent in projecting political space onto geographic place, and may prove invaluable to researchers and the voting public alike.
Full-text available
The increasing prevalence of interactive visualization applications for the representation of spatiotemporal environmental data has created a need for research into the cartographic design of these dynamic applications. The understanding of temporal animations requires that users not only perceive the spatial information presented, but also, just as importantly, be able to locate that information in time and understand changes through time. Temporal legends can support these needs by prompting appropriate mental views (schemata) for interpreting the temporal component of dynamic maps and by providing a method through which users can control what they see. Little empirical research has been directed toward discovering the relative advantages and disadvantages of the various different legend styles that have been designed to indicate the temporal location of an animation. This paper reports o n an assessment of the relative effectiveness of three types of temporal legends to communicate information and facilitate visual th inking. These three types have been selected from a much larger set of legend styles currently in use for spatiotemporal data presentat ion and exploration. Our general hypothesis is that the effectiveness of a temporal legend depends to a significant extent on both the task (what questions are being answered) and the application (what phenomenon is being investigated) for which it is implemented. The resu lts of this assessment will be important for the future design of well-reasoned cartographic animations intended to facilitate study o f any type of spatiotemporal process.
Full-text available
Active legends in cartographic animation serve the dual purpose of controlling the display of a map sequence and providing necessary information about the individual frames. Legends for cartographic animations have typically been implemented in a non-interactive form as individual legends that are synchronized with the display of the map. An active legend creates a tactual/visual association between the individual maps in the animation and what they depict. The development of more interactive cartographic animations is a function of both the available technology and the operational metaphors that make the technology accessible. The Web is providing a means to distribute interactive animations. A particular interactive cartographic animation developed with JavaScript is evaluated for functionality and content.
We introduce MapTime, a software package for exploring spatiotemporal data associated with point locations. Three basic exploration methods are available in Map-Time: animation, small multiples, and change maps. Animated maps can be presented either automatically (at a specified frame rate) or under user control (by dragging a scroll box along a scroll bar). We found the user-controlled approach most effective, but this and other Map-Time features ultimately need to be evaluated by map users. Potential research issues related to animation include developing a temporal legend that can facilitate understanding animations (a key problem is associating the correct dates with changes in the spatiotemporal pattern) and selecting an appropriate frame rate for the automatic display of various phenomena. Small multiples involve presenting multiple temporal elements simultaneously; they are thus useful for comparing individual temporal elements with one another. We argue that small multiples could be particularly useful as guided discovery tools through which students learn about physical geography principles by comparing temporal map elements with one another. Change maps are single, static maps that display the change over time in one of three forms: raw magnitude, percent, or rate of change. Using change maps as individual elements of a small multiple is particularly interesting, as they permit users to "see" changes that may not be apparent during an animation. A limitation of MapTime is that only proportional circles can be used to symbolize point data. This is problematic because users may have difficulty (1) in interpreting the correct relation between circle areas, (2) in associating these abstract symbols with particular phenomena, and (3) in associating the areas of these circles with point locations of phenomena. Therefore, MapTime should ultimately include a greater variety of point symbols (for example, squares, pictographs, and three-dimensional bars).
Animation and cartography present very different traditions to combine. This paper offers some ideas about the directions such a combination might take and presents a series of cartographic animation and visualization case studies involving several unusual representations. These examples range from the interactive exploration of high-resolution, two-dimensional images, to the use of animation in understanding temporal change and three-dimensional structure. Some of the conventional wisdom about the appropriate software applications and visual representations to use is questioned. Exploratory analysis, presenting facts to an interested audience and creating a dramatic image, are seen as distinct tasks, requiring distinctly different animation methods.
A major obstacle in the development of cartographic animation has been a lack of automated techniques for the interactive creation and viewing of map sequences. A microcomputer program for choropleth mapping makes the interactive creation and viewing of cartographic animations possible. The potential uses of this type of animation extend beyond the depiction of temporal change. The interactive cartographic animation procedure involves the creation of individual maps, at less than one second each; their storage as screen images in the computer's memory; and their subsequent display at speeds up to 60 frames a second. User-interface considerations were an important aspect of the implementation.
An abstract is not available.
During the nineties important challenges to cartography will be related to mapping spatial data's multi-dimensional and temporal component. From a cartographic point of view it is necessary to look at the implications of the use of animated maps. The paper concentrates on communicative aspects of the spatio-temporal map. In order to obtain some concrete evidence of the response of users to spatio-temporal map displays a map user test was carried out. Animated maps and the respective static ones depicting the same subjects were presented to the users and questions were asked. The results indicate that although the correctness of answers is not influenced by the type of map (i.e. static or animated), the users perceive faster from animated map displays.