Population movement under extreme events

The last century has witnessed the fastest population growth in human history, with the result that, for the first time, more than 50% of the world population lives in cities. This process is still advancing. According to United Nations predictions, urban population will exceed 75% of the world’s population by 2070. Furthermore, estimates are that if the current trends continue, by the end of the 21st century the entire world will be urbanized (1). Beyond this statistical–demographic fact, this prediction implies that the effects of extreme events (ExEvs), such as blackout, flood, tsunami, or terrorist attack, will have increasingly dramatic and intimate affects on the specific structure and dynamics of cities and urbanism. Thus, there is a critical need to understand the ways in which ExEvs will affect urban life and population dynamics. This new understanding can be achieved through the exponentially growing field of computational social sciences, as illustrated by a study in PNAS (2). In this work, the authors use mobile phone use data in Haiti to investigate the population dynamics and displacement as a result of the tragic 2010 earthquake. This unique work allows a study of how population concentrations, which are established in cities, respond to and are affected by ExEvs.

ExEvs have concerned human societies since their early history (e.g., the biblical flood). However, it is only recently that the issue entered the core of public discourse. This new attention reflects the fact that catastrophes in modern society tend to be less localized, with a significant impact on a larger hazard zone. The mere movement of people inside cities depends on the integration of the electricity grid, the railway network, the roads, the control and communication of traffic lights, and many more infrastructures. Furthermore, a significant amount of people uses the same infrastructure extensively. These two elements (interdependencies between different infrastructures and massive use of infrastructures) yield new vulnerabilities that did not exist a few decades ago (see for example ref. 3). A scenario of cascading failures that rapidly diffuse throughout the entire urban structure is thus a relatively new threat. As such, new methods are necessary to mitigate such threats, with insights such as those presented by Lu et al. (2).

Cities are not merely collections of people. They are interdependent multilevel network systems shaped by interacting and multidimensional driving forces, such as the various urban agents (e.g., individuals, households, firms, planning agencies, and other entities that are acting in the city), infrastructures (e.g., power grids, transportation), economy (e.g., trade, innovation), dependency on resources, sustainability, demography, culture, and values. The distinct role of urbanism in creating new vulnerabilities is strikingly demonstrated in the recent tsunami in Southeast Asia, the September 11, 2001 attack on New York City, Hurricane Katrina in New Orleans, and the 2011 earthquake followed by tsunami leading to nuclear reactor meltdowns and radioactive air pollution at the Fukushima Daiichi Nuclear Power Station in Japan. These events testify that even the most technologically advanced societies are in general ill prepared for ExEvs, but particularly in cities.

The last 30 years have witnessed the emergence of complexity theories of cities (CTC)—a domain of research that applies the various complexity theories to the study of cities. CTC portray cities as complex, self-organizing systemic networks. They suggest that cities have originally emerged and are still developing out of the space–time interactions between the many urban agents, that is to say, the individuals, families, households, firms, and other entities that act and interact in the city. The activities and interactions between these urban agents give rise to the global urban multilevel network and structure that in turn affects the agent’s cognition, behavior, movement, and action in the city in a circular causality. CTC have demonstrated a whole set of resemblances between cities on the one hand, and natural, material, and organic networks on the other (4–7). However, these studies have also exposed important differences that are a consequence of, first, the fact that cities are large-scale collective artifacts, and second, that cities are dual-complex systems; that is, the city as a whole is a complex system/network, and each of its components (the urban agents) is a complex system/network by itself. Their behavior depends on many factors, and when facing a danger, people react in different ways. The new science of big data is now allowing researchers such as Lu et al. to quantify the interplay between micro and macro hierarchical levels in these multilevel complex systems (see for example ref. 8).

Crucial questions that entail ExEvs in cities have not as yet been given sufficient answers: (i) How are the many urban networks of which a city is composed related to each other in space and time, and how might the complex interdependencies between them influence the effects of different kinds of ExEv? (ii) How can the vulnerability and resilience of geographic locations be evaluated with respect to the interdependencies between spatial networks, activity of the residents, and users of cities and their possible reactions to ExEv? (iii) How does a cascade of ExEvs emerge (i.e., the earthquake, tsunami, and nuclear crisis in Japan in 2011), and how can ExEvs be monitored before and after they erupt? (iv) How can ExEvs be planned and prepared for to mitigate the catastrophic? Answers to each of these questions will require the effective communication of analytical findings to human decision makers for the purpose of assessing the status of critical infrastructures and taking appropriate measures in real time. Visual–interactive methods are very promising in integrating analytical methods with human experts and real-world workflows to achieve this goal. Although early examples for such systems exist (9, 10), much research is needed to more closely connect analytical, visual, and interactive methods.

The article by Lu et al. (2) presents the first answers to the third question, relating how the population reacts to an ExEv. The authors quantitatively demonstrate that the widely held hypothesis that a large-scale ExEv causes people to move irregularly while fleeing unrest and searching for material support is wrong. An alternative hypothesis is that, at times of ExEvs, people act on instinct and follow patterns they are highly familiar with; as such, consistent with the results presented by Lu et al., people tend to act in a predictable fashion. People’s movements are highly influenced by their historic behavior and their social bonds.

Animals’ behavior, movement in space included, is highly routinized and as a consequence predictable (11–13). Humans are no exception. Each has his/her own daily routine for the working days as well as for weekends—a fact that shows up in commuting patterns, for instance. Each has his/her favorite coffee shops, theaters, friends, relatives, etc., to which he/she goes regularly or routinely. Such behavioral routines play a central role in the making of human cultures, with the consequence that human cultures differ, among other things, in their spatial routines. For example, both sedentary and nomadic societies exhibit routinized and predictable movement in space; they differ in the

The article by Lu et al. presents empirical investigation of the concept of human spatial regularities and routines.

patterns of their movement. As another example, in some countries (Spain being the prominent example) siesta is a must, whereas in others (United States, England) it is not—a difference that entails significant effects on the very structure of cities.

Hagerstrand’s time geography was an attempt to theorize about humans’ spatial daily routines (14). He suggested that humans’ routinized daily movement in space is determined by three sets of constraints: capability constraints, referring to humans’ basic need to regularly sleep and eat, etc., as well as to their movement capabilities (walking, biking, driving); coupling constraints resulting from the fact that a lot of humans’ action (e.g., work) requires collaboration with others in specific places (e.g., workplaces); and authority constraints, referring to cultural (e.g., siesta) or formal constraints on moving or using space (e.g., law on opening/closing time of shops). The outcome became known as the “web model of time geography” (Fig. 1).

Fig. 1.

Hägerstrand’s web model of time geography. The path represents the daily movement in space–time of an individual; the bundle is the space–time place/cylinder where the individual congregates with other individuals (e.g., home, work, school, etc.), whereas the domain describes the space–time area/cylinder within which the individual’s daily movement takes place. Reprinted with kind permission from Springer Science + Business Media: Cognition, Complexity and the City, 2011, Portugali, J., Figure 3.3. (7).

Routinized action, including movement in space, is thus a basic property of human behavior; in fact, humans have no choice but to act routinely. ExEvs break the routines of daily life and movement in space, but not this basic human need. After an ExEv, the natural tendency of people is thus to go back to the previous routines and if this is not possible to develop alternative (temporary or permanent) routines.

In this respect, the article by Lu et al. presents empirical investigation of the concept of human spatial regularities and routines, which are shown to be predictable. Using the newly available sources of big data, the authors provide empirical evidence to these long-standing theories. Moreover, the more interesting findings of the study refer to the patterns of the routines developed after the earthquake. This information is critical for the management of future urban ExEvs. In fact, many times during such ExEvs, the main infrastructure breaks down immediately after the event (mobile networks, roads, electricity, etc.), and substantial informed information on the predicted movement of the population will dramatically improve relief efforts. Despite its few limitations, as discussed by Lu et al., their work presents an important step in the direction of using big data to predict and manage human movement under ExEvs.