Navigation – Plan du site

AccueilRubriquesSystèmes, Modélisation, Géostatis...2007Spatio-temporal exploration of SA...


Spatio-temporal exploration of SARS epidemic

Une stratégie d’exploration spatio-temporelle de l’épidémie de SRAS
Arnaud Banos et Javier Lacasa


Apparu en Novembre 2002 dans la province de Guangdong, en Chine, le Syndrome Respiratoire Aigu Sévère (SRAS) a largement occupé le devant de la scène médiatique pendant l’année 2003, atteignant même rapidement le statut de « première pandémie du XXIe siècle ». Six mois après sa ré-émergence à Hong-Kong en mars 2003, 30 pays étaient touchés, 8500 cas identifiés et 800 personnes décédées des suites de ce coronavirus. Dans cet article, nous proposons une approche de géo-visualisation, dynamique et interactive, permettant une exploration spatio-temporelle de l’épidémie.

Haut de page

Texte intégral

Main lessons from SARS

1While the SARS epidemic was retaining great attention from the media, thus becoming the major world event of the moment with the Iraq war, one couldn’t help questioning the epidemiologic numbers updated daily by WHO (figure 1): what could these numbers represent, compared to the score reached by the lethal trio “AIDS-malaria-tuberculosis” (5 to 6 million deaths each year)?

2First, as Bary R. Bloom quoted, “in the absence of effective vaccines or drugs, infectious diseases that are spread by the respiratory route – whether the influenza epidemic of 1918 that killed 20 to 40 millions people worldwide or measles, which spreads rapidly among children – must be taken very seriously” (Bloom, 2003). Indeed, two epidemiological studies published at the end of the epidemic draw the conclusion that the SARS coronavirus was sufficiently transmissible to infect a majority of people, and from that perspective was able to cause - if uncontrolled - a very large epidemic (Lipsitch and al., 2003 ; Rilez and al., 2003).

3Second, the fatality rate of that new virus proved to be higher than expected: it was estimated to 13.2% for patients younger than 60 years and 43.3% for patients aged 60 years or older (Donnely and al., 2003). Two points that highlight how lucky we were the virus spared Africa…

4Third, beyond this fortunately limited - but real - human burden, economic consequences proved to be dramatic. Thus, Singapore experienced a significant decrease of its GNP (around 4.3%) during the epidemic period, due to a spectacular fall of its tourist activity (minus 63%). The cost for China was evaluated around 20 billions dollars, while for Canada it raised 1.2 billion dollars, medical costs not being taken into account (Zylberman, 2004).

5More generally, the SARS episode revealed a great weakness of our global and networked society: the virus spread dramatically in a few days from the source in Hong-Kong through intercontinental flights, managing to establish itself in a large city of a well developed country, Toronto.

Figure 1: A time-line view of the SARS outbreak

(Source: WHO,​en/​)

6Therefore, one of the great lessons from SARS is that we need an efficient global health network to cope with such unanticipated outbreaks as the security of even rich countries, depends from now on our capacity to identify potential health threats around the world and to address them locally. From that perspective, the SARS epidemic is very remarkable, as Bary Bloom underlined: “in the 1980s, it took 2 years to identify HIV as the cause of AIDS. In 2003, WHO created and extraordinary network of 13 laboratories in 10 countries, which identified a virus associated with SARS in 2 weeks and had its entire genome sequenced in two more. Those labs shared their knowledge in an unprecedented fashion, to the benefit of everyone” (Bloom, 2003).

Geographical spread of the SARS epidemic

7Rapidly identified as a major world event, SARS inspired a large number of epidemiologic studies, dealing with various crucial aspects of the phenomenon. But, from our point of view, a basic perspective of the geographical spread of the SARS epidemic was still missing. Yet, appropriate data existed that may help reveal the global dynamic of the virus propagation. Indeed, as soon as the World Health Organization (WHO) declared a global alert - the first of its history -, daily statistics on the epidemic became available on its website [​en/​]. Table 1 reproduces this daily report for a selected day (27 march 2003). As it can be seen, complementary comments (notes) come with data, underlying specific features. Indeed, assessing number of SARS cases at this global level implied very generic and strict definitions, independent from the countries reporting cases. However, even if WHO provided very early a complete methodology to detect and assess SARS cases in such a generic way, the United States accepted to follow this common procedure only one month after the outbreak’s beginning, leading to very different numbers (Figure 5).

Table 1: WHO daily report (27 march 2003)

(Source: WHO,​en/​)

  • 1 Several other websites have proposed maps of the SARS outbreak, but given the life expectancy of su (...)

8Believing, as a geographer, that a map is worth a thousand numbers, we tried to shed new light on that fascinating world phenomenon, from the public database available online. Of course, several maps have been produced during or after that epidemic, as reviewed by Boulos (2004) 1. Our approach is slightly different, in the sense that we wanted to focus on the dynamic component of the phenomenon. Having such a perspective in mind, we therefore created two “lively” maps and made them available on the Internet, allowing anyone to explore the SARS epidemic from a geographic perspective [​SARS/​sars_epidemic.html].

Figure 2: The generic map designed

[Full animation:​SARS/​maps.html]

9The first animation [​SARS/​maps.html] provides therefore a global view of the world diffusion of SARS (Figure 2). It is based on day-by-day maps of the phenomenon, such as the one below.

10Note that China’symbol is localised on the Guangdong province, where most of the cases occurred. Symbols’ size follows a logarithmic scale, in order to allow countries concerned with only few cases of SARS to be visible. The graph at the bottom shows the global cumulative curve of the SARS cases, from the beginning to the end of the alert. Finally, upper buttons allow interactive and automated navigation through the animation. Figure 3 shows the kind of pattern revealed by such an animation.

11The global cumulative curve of the SARS cases grows gradually, revealing the day-by-day progression of the virus. Note also the dramatic geographic spread of the virus, which had reached most of its “target” countries by 1st of April. Note finally the cumulative self-enforcement of the Asian original source, the rapid growth of the Canadian one (Toronto) and the remarkable quietness of Africa and South-America.

12One may wonder if the nearly absence of reported cases there is plausible, or if it highlights an intrinsic limitation of data collection, as each country was asked to self report its cases. After all, as we will see in the next developments, this methodology led to huge over-estimation in the United States. However, given the extreme diffusion and fatality rate of the disease, the extreme vulnerability of these two continents to such and epidemic outbreak and the unprecedented attention paid by the public health community2, a systematic under-estimation seems unrealistic.

Figure 3: Overview of the map animation

[Full animation:​SARS/​sars_epidemic.html]

13Such a dynamic animation proves useful to reveal the spreading process of the virus at a global scale and to communicate about it. Since it is based on raw data produced by the World Health Organization, it could be generated and updated daily, providing the scientific community and the public with a more eloquent view of the phenomenon running. However, it can hardly be considered as a self-sufficient document, based as it is on a single scale, the macro-geographic one.

Back to local

14It is our belief that modern science must - and more and more often does - provide multiple views on a single problem (Banos and al., 2003). Therefore, we provided a second animation, allowing for more localised explorations to be led (Figure 4).

Figure 4: Interactive local exploration of the SARS epidemic

[Full animation:​SARS/​series.html]

15Dynamic linked graphics are thus tools as simple as powerful to investigate one's data set, particularly in a space-time perspective, by simply linking the spatial and temporal dimensions through appropriate graphs. We exploit here the Flash technology, allowing the user to click on a SARS case symbol on the map, so as to display different cumulative curves drawn from WHO data: cumulative number of cases, recoveries and deaths.

16For each selected country, the desktop is composed as follows: the world cumulative curves appear below the map, while the specific curves for different countries appear on the right side, first jointly then separately. The sigmoidally (or S-shaped) curved displayed here is quite typical of epidemic outbreaks that tend to occur as well-marked epidemic waves rather than in temporally uniform or random forms (Cliff and al., 1986). Between the starting point (WHO alert, here) and the saturation level (upper limit for each curve), the shape of the curve shows the day-to-day rate of growth. The key point here concerns our ability to compare different curves for different countries, identifying at a glance various local dynamics of the epidemic (figure 5).

Figure 5: A bouquet of epidemiologic curves

17Each curve indeed tells a story: the China one reveals the late acknowledgement of the local SARS disaster, as well as the lasting will of China authorities to minimize its consequences. Singapore and Taiwan reveal two different timings of the epidemic. The Canadian curve warns about the need to maintain vigilance, as a resurgence of cases is always possible. The French curve shows a strictly controlled and contained epidemic, where each new case has been identified and isolated rapidly. Finally, the US curve reveals a systematic over-estimation of cases at the beginning of the epidemic, underlying the problem of accuracy in such sensitive data and the possible bias introduced by side effects like fear.


18Our objective was to highlight the exploratory power of animated and interactive maps as spatial surveillance tools. Such framework may be very useful to reveal local and global dynamics of a given epidemic, here SARS. Hence, the two documents proposed and made available on the Internet underline the complexity of the epidemic process and the high variability of local situations that a “smooth” global pattern can be composed of. Two crucial parameters -amongst many - we need to control before building any diffusion model. Anyway, a major GIS shift has to be achieved if we want to carry on, from data management to data analysis: as Stan Openshaw recalled, “analysis is no longer an option” [8]. In that spirit, local based methods and models are still to be defined, leading to more realistic and efficient identification of underlying patterns. While being of concern for the entire spatial analysis field, this challenge may sounds particularly appealing to the geographer, given his capacity to assume a major role in the process.

Haut de page


BANOS A., BANOS F., 2003: “Towards integrated spatial surveillance tools of urban risks: space-time analysis of road accidents in Lille”, in Géographie des risques des transports, BANOS Arnaud et al. Editors, Paradigme, Orléans, pp. 9-22.

BLOOM B., 2003: Editorial: “Lessons from SARS”, Science, vol. 300, p. 701.

BOULOS M., 2004: “Descriptive review of geographic mapping of severe acute respiratory syndrome (SARS) on the Internet”, International Journal of Health Geographics, vol. 3, no 2,

CLIFF A., HAGGETT P., ORD J., 1986: Spatial aspects of influenza epidemics, Pion Limited, 280 p.

DONNELY C., GHANI A., LEUNG G. et al., 2003: “Epidemiological determinants of spread of causal agent of severe acute respiratory syndrome in Hong-Kong”, The Lancet, vol. 361, pp. 761-1766.

LIPSITCH M., COHEN T., COOPER B. et al., 2003: “Transmission dynamics and control of severe acute respiratory syndrome”, Science, vol. 300, 1966-1970.

OPENSHAW S., 1995: “Developping automated and smart spatial pattern exploration tools for geographical information systems applications”, The Statistician, vol. 44, no 1, 3-16.

RILEY S., FRASER C., DONNELLY C. et al., 2003: “Transmission dynamics of the etiological agent of SARS in Hong-Kong: impact of public health interventions”, Science, vol. 300, 1961-1966.

ZYLBERMAN P., 2004: « Le SRAS, une épidémie postmoderne ? », La Recherche, no 371, pp. 46-47.

Haut de page


1 Several other websites have proposed maps of the SARS outbreak, but given the life expectancy of such supports, only a few remain accessible. For example:

2 See for example the article published by “Afrol news” in may 2003 (02/05/2003) [].

Haut de page

Pour citer cet article

Référence électronique

Arnaud Banos et Javier Lacasa, « Spatio-temporal exploration of SARS epidemic », Cybergeo: European Journal of Geography [En ligne], Systèmes, Modélisation, Géostatistiques, document 408, mis en ligne le 27 novembre 2007, consulté le 02 février 2023. URL : ; DOI :

Haut de page


Arnaud Banos

Image et Ville (UMR 7011 ULP/CNRS)

Articles du même auteur

Javier Lacasa


Haut de page

Droits d’auteur


Creative Commons - Attribution 4.0 International - CC BY 4.0

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search