1 avenue Montaigne
93160 Noisy le Grand
Keywords: Cartographic animation, Fourth dimension, Crime Mapping, Visualization, Visual Thinking, Visual Communication
In this paper we want to show a simple using of SVG standard to automatic creation of cartographic animation.
Although we are working for crime mapping, our project may interest several people working on spatio-temporal database.
In the first part of paper we explain the sort of data we are working with and the Visual thinking approach.
The second part talks about time in cartography and presents our theoretical approach of cartographic animation.
The third part presents the workflow, the use of SVG to animate our maps, and our animated maps.
The last and fourth part gives some ideas about doing a better animation.
The RATP exploits the greatest public transport network in France. It serves the "Ile de France" area, i.e. Paris and its suburbs. Within the framework of the quality of the service which it proposes, the company must take care of safety as of the security of its customers.
Via its "Headquarters" of intervention in real time "PC 2000", the RATP security service within its competence field records the delinquent acts which are reported, and the circumstances, more especially space and time.
The statistical office of the Environment and Security department handles all these data for operational purposes. It analyzes them within the framework of its statistical competence but also from a cartographic point of view.
Many maps are elaborated. In fact there are inventory maps and each one represents a state of the facts at a given date or at a given period.
The treated data are space-time data, the place where the facts took place and the time : hour, day, month, year are known precisely.
NB. The data presented in this article were falsified intentionally.
Individually, each map produced automatically by the security service gives a good approach of the phenomenon. Their accumulation in great number however makes it difficult to sense how things evolve in the long run.
However, the security integrates a double dimension: forecast and search of the repetitive space-time behaviors (or cyclic). These static maps can only partially reach this double objective and they imply significant efforts of analysis and memory.
The security service is an operational service and it has very little time for analysis, so it is very important to propose "immediate" tools for understanding of the events.
The developed tool must bring a fast and automated solution. This operational purpose passes by the realization of two phases :
As they appeared GIS entailed an evolution of map statute.Many authors(Tobler, MacEachren...) were interested in this slipping in the map function summarized thus by Dr Kraak M-J (1994)  « [..] GIS as introduced a new perspective in visualisation by shifting the focus from maps produced by cartographers as a final product, to maps created by users as part of their analytical process. »
The map integrates from now on a broader process of "scientific visualization". Under the term "visualization" are gathered the different methods of graphic representation in relation to the widespread use of data processing. « Visualization offers a set of techniques to explore, analyse and present information. » Dr Kraak M-J (1994) 
Visualization is applied to two types of activities (DiBiase and al. 1992 ) : visual thinking and visual communication
As a conclusion " Visual thinking is exploratory ; visual communication is explanatory ". (DiBiase et al. 1992 )
The question of time was already raised by the geographers and the cartographers. It implies specific processing and poses a problem of representation. We will try to show which solutions the cartographers considered.
For RATP safety statistics department time has been a question from the very beginning of map representation. As a matter of fact, in the field of security, a situation is generally observed in comparison with the situation over the previous period. Thus, it is possible to see whether one is within the framework of an emergent, a regressive or a stable "hot spot".
How to represent space-time data was a question raised very early in cartography.One finds thus in the Graphic Album of Statistics (Paris, Imprimerie National, 1889), various maps representing space-time data by means of cartograms (diagrams satellite, helicoid diagrams) and of anamorphoses. (See Palsky, 1996 )
But these "cartograms" have a low visual effectiveness. In general, cartograms, i.e. the diagrams scattered on a map are perfectly useless for a global reading and insufficient for an elementary reading. (Bertin J. 1977 p.155 )
The best representation of an evolution on a static map thus remains the representation of the rates of evolution which can be combined with the numbers of facts per station over the recent period. This is the principle we applied in our first automatisation of static map production.
A map shows where the facts took place and in which proportion (where and how much). The two dimensions of map X and Y are used for localization, the quantification of the facts thus uses the third dimension, Z.
If one piles up maps (with a common Z scale so as to be able to compare the maps) until there is a real collection, the whole chronological set of maps must be saved in order to perceive the time. No new variable is brought in.
Indeed, to perceive the evolutions it is necessary to compare the maps with one another and for each represented place to retain the variations of the associated data.
Perception = (n x Z x XY)
n, amount of maps in the chronological set
Z, amount of recorded facts
The third solution consists in adding a fourth variable to represent time: time itself. Time becomes an intrinsic part of the representation mode.
The first cartographic animations are movies in the first half of the XXth century (to represent displacements or space diffusion processes ). The experiments are more and more numerous with the introduction of digital processing, in the Sixties and Seventies but the animation of maps takes its true rise with personal computers. (Campbell and Egbert, 1990 )
The "animation" solution , such as we propose it, is based on two principles: the compression of the facts on one time unit, and an accelerated accumulation of compressions over a given period.
The choice of the unit of time proceeds directly from the question that the analyst asks. The facts recorded over the selected time unit are then added up. This total corresponds to a compression of time since we summarize in a single figure. a time interval made of long periods without noticeable facts and much shorter moments when the facts take place (a few seconds or a few minutes).
This compressioncan be done with various "time scales " (hourly, weekly, monthly, etc). The smaller the time scale (for example annual), the greater the ratio compression and thus the loss of precision of information . The compression of time is like a "discretization" of the time variable. Each compression generates a map. It is the first principle.
Traditional cartographic animations make the two dimensions of the plan vary. Displacements of an army will be represented by the movement of an arrow (linear symbol). The analysis of the urban growth will be perceptible through the increase in a surface.
In our case, the facts occur at precise places on the map, (for example the platform of a subway station), these places are fixed and constant and animation is done on the third dimension, the quantitative variable.
By analogy, one can compare the map with a photograph, a P state of the observed phenomenon, and the animated map with a movie. The movement enhances the passage of the P state of the phenomenon to the P+1 state.
In an animated map one "encapsulates" the map set into the unit map. It is then as if one single map contained all the others. It is no longer the eye which moves from one map to the other , but the symbol which "moves" from one value to the other. This movement of the represented variable, the third dimension, is nothing else than the time which elapses.
Animation of course cannot be done in real time. The objective being to perceive in a few tenths of seconds what occurred in one month, or even one year, it is necessary to accelerate time.
The animation process creates connections from one compression to the following one. Since one does not perceive any visual rupture during the animation, there is continuity. Time is a linear, continuus, phenomenon but the discretization induced by compressions inevitably creates discontinuities which it is then advisable to fill. This is the second principle.
This fluidity objective entails a true constraint in the choice of the adopted technical solution. As a matter of fact there are many ways of displaying maps successively, but they do not offer the possibility of an interpolation between the successive situations (animated GIF or diaporama in MS Power point).
Once our goals had been defined, we searched a technical solution available for our knowledge. We are geographers and not developers. Our algorithm's knowledge, let us consider anyway some small and simple developments in order to meet our needs.
For the Security Service, we wrote scripts for Arcview 3.2 (Esri) in Avenue Language. This application draws automatic maps, through request to an Oracle database from an MS Access Interface.
Decision makers of the Safety department can thus in an autonomous way draw all the maps they want to consult. This automatic realization, with a choice of the desired parameters (Choice of period, geographical space, victim's type…) let the user have an exploratory approach.
At the same time, the good graphic quality of produced maps, permit us to publish them in officials reports.
The maps animation, developed in the same way as the previous solution for static maps, wanted to keep the self-realization context for users. This wish of automatic creation leed us to choose a standard like SVG.
Our position being more one of user, than one of developer, we tried to find existing scripts in order to fit them to our needs. The Esri User Community let us collect an Avenue Language Script « shp2svg » by Nedjo Rogers () allowing export of Arcview Views into SVG. We adjusted this script to our animation needs.
The operator makes a request to an Oracle Data Base, choosing his parameters through a MS Access interface. This request entails the export of two tables, one containing the request result and the other one containing the parameters of the request.
Application begins with importing those two tables into Arcview 3.2. The creation of the background map precedes the calculation of the circle radius and the export of the map and the radius values calculated for each step of time in standard SVG into an HTML page. The type of map produced are maps for stations of the choosing network on the studied period.
Operating System: Windows NT 4.00
Software Platform : Access 2.00, Arcview 3.2, Browser
DOS Routine to convert bmp to png : bmp2png.exe
User Communities let all users get free some developments. The average GIS user can thus adapt the scripts to his needs.
The facility and simplicity of parameter setting in SVG allows to develop quickly.
In case of an animation application, the asset of SVG relates to the fluidity of the graphic aspect, due more especially to the light weight of the produced files.
So the aim of our animation: " to see the time " has been reached and the analyst can concentrate on searching constants or on the contrary irregularities.
Following examples illustrate the type of evolution we are looking for.
The results we get are full of promise but insufficient. Beyond the cartographic animation, it will be necessary to make dynamic and interactive cartography.
In order to facilitate analysis and « Visual Thinking », we envisage several improvements.
 Kraak Meno-Jean in Craglia M. 1994 Report of the specialist meeting on GIS and Multimedia.http://www.shef.ac.uk/uni/academic/D-H/gis/multimed.html.
 DiBiase D., MacEachren A. Krygier B., Reeves C. 1992. " Animation and the Role of Map Design in Scientific Visualization ". Cartography and Geographic Information Systems ; Vol. 19, n°4, pp 201-214, 265-266.
 BERTIN J. 1967. Sémiologie Graphique. Les diagrammes, les réseaux, les cartes. Paris-La Haye : Mouton Gauthier-Villars, 431 p.
 Palsky G. 1996. Des chiffres et des cartes - La cartographie quantitative au XIXème siècle - Paris - Ministère de l'Enseignement Supérieur et de la Recherche - Comité des travaux historiques et scientifiques
 BERTIN J.1977. La graphique et le traitement graphique de l'information. Paris : Flammarion, 277 p. (coll Nouvelle Bibliothèque Scientifique).
 Campbell C.S et Egbert S. L. 1990. « Animated Cartography : Thirty years of scratching the surface. Cartographica », Vol 27, n° 2 – Summer 1990. Pp 24 –46
 Nedjo Rogers, Environmental Mining Council of BC (firstname.lastname@example.org). This is an open source project (see definition at http://www.opensource.org/docs/definition_plain.html) and is distributed under the GNU General Public License (see http://www.gnu.org/copyleft/gpl.html). This script is available for download at http://www.carto.net/projects/shp2svg/
 bmp2png.exe available for download on http://hp.vector.co.jp/authors/VA010446/b2p-home (NB : download the DOS version)
 Slocum T.A., Egbrt, S. L., Prante M.C., Robeson S.H., 1988. Developping an information system for coropleth maps. Proceedings of the Third International Symposium on Spatial Data Handling. Sydney, Australia, pp ; 293-305