Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Natural Hazard Science. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 26 September 2022

The Role of Road Transportation in the Flood Evacuation Processfree

The Role of Road Transportation in the Flood Evacuation Processfree

  • Marta Borowska-StefańskaMarta Borowska-StefańskaUniversity of Lodz
  •  and Szymon WiśniewskiSzymon WiśniewskiUniversity of Lodz


Floods, which are among the most dangerous and frequent disasters in the world, are expected to occur more frequently due to climate change. Floods, and flash floods in particular, generate economic, environmental, and social effects. Economic effects include damage to infrastructure, the negative influence upon transportation and communications networks, and an increase in fuel costs, as well as time loss due to traffic delays (congestion) and the necessity of taking alternative routes. It is therefore important to take action both to prevent and to mitigate these effects. In the 21st century there has been a radical change in the approach to the issue of flood protection (as seen in the approach formulated within the 2007 European Floods Directive)—it is no longer believed that there is such a thing as complete protection against floods, but that the damage and loss it inflicts can only be mitigated, and since floods cannot be completely eradicated, societies must learn how to live with them. In the event of a flood, preprepared procedures to counteract and mitigate the effects of the disaster are followed, including the evacuation of people and movable property from affected areas.

Evacuation planning is meant to reduce the number of disaster-related (including flood) fatalities and material losses. Crucially, this type of planning requires a well-defined, optimum evacuation policy for people and households within flood hazard areas. In addition, evacuation modelling is particularly important for authorities, planners, and other experts managing the process of evacuation, as it allows for more effective relocation of evacuees to safety. Modelling can also facilitate the identification of bottlenecks within the transportation system prior to the occurrence of a disaster; that is, it enables us to determine the impact of flood-related road closures, and to comprehend—among other things—the effects a phased evacuation has on traffic load. Furthermore, not only may the ability to model alternative evacuation scenarios lead us to establish appropriate policies, evacuation strategies, and contingency plans, but it might also facilitate better communication and information flow.

Evacuation from flood-hazard areas is a major challenge for the field of flood risk management as well as the fields of traffic engineering and transport planning. This is particularly true when the research has to include not only those journeys directly related to escape from hazardous places, but also their reflexive relationship with the total number of journeys made in the “background” of the evacuation itself.


  • Floods


Floods and Flood Risk Management

A flood is a surge of water in rivers, reservoirs, canals, or at sea, during which water overflows river valleys or depressions and threatens people and property (Borowska-Stefańska & Wiśniewski, 2018; Jonkman, 2005; Jonkman et al., 2008; Radosavljevic et al., 2017). Flash flooding is defined as flooding that occurs as a result of intense bursts of rainfall over a relatively small area. It is characterized by short warning times and high velocity flows that rise and fall rapidly. This type of flooding is particularly dangerous in urban areas, where there is sealing of catchment areas, overloading of drainage systems, and increased surface runoff (Haynes et al., 2009; Modrick & Georgakakos, 2015; Musolino et al., 2020).

Floods killed at least eight million people in the 20th century alone. Between 1998 and 2017, they accounted for 43.4% of all global disasters, causing 11% of disaster-related deaths and affecting 45% of the victims of all hazardous occurrences (Borowska-Stefańska et al., 2021). In the 20th century, the largest floods in Europe included those that occurred in Italy (1951, 1954, 1963, 1985, and 1998), Romania (1970 and 1991), Spain (1962 and 1973), and in Austria, Poland, Germany, Czech Republic, and Slovakia (1997). In the 21st century, the greatest losses were caused by the floods of 2002, 2005, 2007, 2010, and 2021. In 2002, floods occurred in the area of six European countries (Czech Republic, Germany, Austria, Hungary, Romania, and France), leading to 78 deaths, and causing losses estimated at approximately USD 21.2 billion (Genovese, 2006; Kundzewicz, 2004; Lugeri et al., 2010). In turn, in 2005, floods affected, inter alia, Romania, leading to dozens of fatalities and billions in damage. Another flood in Great Britain in 2007 resulted in material losses estimated at USD 8 billion (Lugeri et al., 2010). The flood of 2010 mainly affected Central Europe (the Czech Republic, Slovakia, Poland, Hungary, Ukraine, Austria, Germany, and Serbia). In Poland, 18,500 buildings, 825 kilometers of road infrastructure, and 587 engineering structures were damaged, and 25 people died (Biedroń et al., 2011). Flooding in 2021 affected the countries of western, central, and southern Europe, including Germany, Belgium, Italy, Croatia, the Czech Republic, Great Britain, Austria, and Romania. More than 180 people died in Germany (Fekete & Sandholz, 2021). It is therefore important to take action both to prevent and to mitigate flooding effects.

At the turn of the 21st century, attitudes towards strategies of flood protection evolved from the assumption that fully-effective methods for flood protection were possible to the conviction that total protection was impossible and that flood-related damage and losses could only be mitigated. This involved a transition from purely structural measures to a more holistic and adaptive approach to flood risk management (Abebe et al., 2019; Hall & Penning-Rowsell, 2011; Lagadec, 2002; Lumbroso & Vinet, 2012). This approach was articulated for European countries in the Flood Directive (Flood Directive, 2007) which obligated all member states to create: (a) a preliminary flood risk assessment (by the end of 2011); (b) flood hazard maps and flood risk maps (by the end of 2013); and (c) flood risk management plans (by the end of 2015) (Borowska-Stefańska, 2015; De Moel et al., 2009; Hartmann & Spit, 2016; Nakamura & Llasat, 2017).

Flood Risk Management Aims

The implementation of the Flood Directive by E.U. member states has resulted in a common framework for flood risk management (Adamson, 2018), the main aim of which is to reduce the risk and consequences of disaster. In order to reduce the magnitude of floods and their effects, measures (flood hazard and flood risk management) that cover different stages of the disaster should be integrated. These include:

mitigation: refraining now and in the future from building homes and industrial buildings in floodplains. It is also extremely important to adapt the buildings that exist or will be constructed to the degree of potential flood hazard, and to promote more appropriate land use and agricultural and forestry practices. Furthermore, it is important to raise public awareness, including the perception of risk.

preparedness: informing and warning the public about flood hazards and how to proceed in the event of a flood (evacuation planning).

response: emergency response based on flood emergency management plans.

recovery: restoring normal conditions and learning the lessons (i.e., returning to “normal” conditions as early as possible and mitigating the social and economic impact on those affected by the flood; Drużyńska & Nachlik 2006, ; Hall & Penning-Rowsell, 2011; Lumbroso & Vinet 2012; Nachlik, 2007, 2008).

The first two stages focus on predisaster issues, while the latter two address those after the disaster has already occurred (Esposito Amideo et al., 2019). There is a fundamental difference between the different stages of flood protection and emergency management which comprise flood risk management. In the case of the former, the primary objective is to provide effective protection against the effects of the surge so that water does not detrimentally affect people and property. If, on the other hand, the focus is on emergency management, the main aim is not protection in the sense of modernization or the construction of a protective system, but the most effective possible flood control, which consists of:

protection of life and health and the reduction of the tangible effects of the flood during the rescue operation;

transport efficiency with regard to the evacuation of people and property, and the provision of resources and equipment for damage mitigation;

protecting vulnerable structures and equipment that may be a threat to the environment during a flood, such as sewage systems, petrol stations, chemical plants, and so on;

protecting public structures and buildings frequented by large numbers of people; and

other activities specific to the region (Buczek & Nachlik, 2011).

Although each element of flood risk management is extremely important for minimizing damage, evacuation is the most significant in reducing the loss of life when disasters occur (Zhai & Ikeda, 2006).

Evacuation—Types, Stages, Planning

Evacuation is the stage of disaster management (Cova & Johnson, 2003) which involves moving people from flood hazard areas to safety (Na et al., 2012). Transport plays a crucial role in the process, and planning is important for its successful implementation (Lim et al., 2013) as its main objective is to reduce the number of fatalities or the scope of tangible damage. Even if every disaster is different and thus difficult to predict, and it is not possible to fully plan for evacuation, good planning still remains the basis of disaster management (J. Kendra & Wachtendorf, 2003).

Evacuation (in Latin: evacuatio—draining, emptying, weakening) is a common strategy for managing emergency situations and one of the fundamental measures taken to protect the health and lives of people and animals and to save property in the event of any hazard. Evacuation is the process in which people are moved from danger zones to safe places where they can remain until it is appropriate for them to return (Lumbroso et al., 2010; Saadatseresht et al., 2009).

Types of Evacuation

Evacuation can be conducted at different stages of a disaster—either before (e.g., in the case of a flood) or after it begins (e.g., during a fire); it may be planned (preemptive, anticipatory) or an emergency (immediate, urgent, ad hoc). Planned (organized) evacuation refers to the preorganized movement of people from floodplains such as those adjacent to hydrotechnical facilities which pose a potential threat to the population, property, and the environment. Once the probable hazards in a given area have been identified, it is vital to determine the transport options available to transit the evacuees from the affected location to safe areas. This planning should take into account the evacuees’ own means of transport, mass transit (buses) and all other available options, and evacuation on foot (Esposito Amideo et al., 2019; Rabemalanto et al., 2020; Rawłuszko, 2016; Shaw et al., 2014). Furthermore, a planned evacuation can be total, partial, or staged. A total evacuation is conducted when, due to the nature of the hazard, everyone in the affected communities is encouraged or directed to evacuate. A partial evacuation applies only to those who, for instance, live in the lowest areas and will be directly affected by flood waters. A staged evacuation may be conducted under either of the two strategies, when—due to the time available before the arrival of the hazard (e.g., a flood wave) and in order to avoid road congestion—the evacuees are directed to safety in stages (Borowska-Stefańska et al., 2021). As for emergency evacuation, this involves the urgent relocation of people from areas where there is a sudden and unforeseen threat to health or life (e.g., flash floods).

Evacuation may also be classified as voluntary, recommended, and mandatory. No special traffic control or transportation measures are normally taken during voluntary evacuations. People in such cases can remain in their places of residence if they wish. Recommended evacuation refers to disasters with a higher probability of endangering the population. In this case, decisions of whether or not to leave are left to individuals and limited special transportation arrangements are made. Mandatory evacuations represent the most serious type of evacuation, but they are extremely difficult to carry out and, in many countries, are even impossible. This is because the relevant provisions do not exist everywhere and because people often resist orders to leave their homes (Urbina & Wolshon, 2003). However, while the definition of a mandatory evacuation is often unclear from a legal point of view (Baker, 1991), and emergency management and law enforcement agencies realize that it is not realistic to enforce an order compelling people to evacuate, the term itself is of great significance to potential evacuees. It indicates to them that the situation is so serious that it is necessary to evacuate (Murray-Tuite & Wolshon, 2013).

Moreover, it is important whether an evacuation is conducted (supervised and controlled) by an external entity (emergency services in particular) or by the evacuees themselves, in which case it is called self-evacuation (Kolmann, 2020) and may be performed in an organized manner or spontaneously (Gromek & Kozioł, 2015).

A planned (organized) self-evacuation can be implemented when the inhabited area has not yet been flooded but the probability of disaster is high. In such circumstances, residents can self-evacuate: either after receiving a warning there is a flood wave approaching or following their own observations of the watercourses. In most cases, their own means of transport will be used, but if a household does not possess a car or other mode of transport, any planned evacuation should assess whether a neighbor’s means of transport would suffice for this household also. Spontaneous (ad hoc) self-evacuation occurs when there is a direct threat of a flood wave. In such cases, the population should immediately self-evacuate to a safe place using their own means of transport. However, it must be stressed that any self-evacuation is dependent primarily upon people’s individual capabilities, including access to transport and accommodation. What differentiates self-evacuation from coordinated evacuation is primarily that self-evacuation lacks management, supervision, and control over its course by the authorities, leading to the spontaneous nature of the actions taken, and where the lack of accurate information may even result in chaos (Kolmann, 2020).

Evacuation may be conducted by all means of transport, even on foot. Those supervising the evacuation may choose to give priority to essential workers and people with reduced mobility. If the evacuation requires the movement of large numbers of people (or equipment) over long distances, then it is important to use the most efficient means of transport (especially road and rail). However, these means are often not available in sufficient numbers to meet the full range of needs in question. It would then be an acceptable practice to apply a combined approach and use the available organized means of transport together with the evacuees’ own resources or to evacuate people on foot. Since each option has its limitations depending on the situation, the effectiveness and efficiency of the application and the correct selection are considered major issues to be resolved by the decision-makers responsible for the management of an evacuation (Kolmann, 2020).

Drabek and Stephenson (1971) proposed a division into four types of evacuation: by invitation, choice, default, or compromise. Evacuation by invitation occurs when someone outside the at-risk area provides the means or impetus to leave. Evacuation by decision or choice includes those who decide to evacuate on the basis of a warning regarding the disaster. Evacuation by default involves behavior dictated by actions other than seeking safety from the hazard (such as not being allowed by officials to enter an evacuated zone or structure upon one’s return). Evacuation by compromise occurs when people follow evacuation orders even though they do not want or feel the need to evacuate (Sorensen et al., 1987).

Obviously, people should be properly informed about the evacuation process, and the exponential cultural and (primarily) technological changes have influenced the way in which disaster warnings are communicated to the public. These changes include:

new warning technologies (using cell phones or messaging through the Internet)

private warning subscription providers

nationalization of news coverage

increased availability of visual images and information

increased use of a Global Positioning System (GPS) for alerts and notifications

Stages of Evacuation

The process of evacuation can be divided into five stages:

decision to evacuate




return (Lim et al., 2013) (Figure 1).

Figure 1. Evacuation process—From hazard to return.

The first three stages of evacuation are critical. Delays in any of these may result in insufficient time to carry out the planned evacuation, or in a situation when not all evacuees reach the desired destination in good time, or even fail to leave hazard areas (Jonkman, 2007; Kolen & Helsloot, 2012; Kolen et al., 2013; Urbina & Wolshon, 2003). A key objective of planning an evacuation in emergency situations is to ensure that evacuees leave the affected area as quickly as possible and reach safe locations. Therefore, planning an evacuation should be about minimizing the total evacuation time (Dulebenets, 2021).

Planning the Evacuation Process

In the literature, two conflicting premises can be found in relation to disaster management. The first states that evacuation plans and their implementation are pointless, while the second argues that better schemes and better planning can improve disaster management (Wilson, 2004; Cook & Lourdes Melo Zurita, 2016). Regarding the first premise, Clarke (1999) claimed that those institutions whose area of operation may be exposed to large-scale disasters must engage in planning emergency procedures (including evacuation) even if it is obvious that such occurrences cannot be predicted (Boin & McConnell, 2007); however, Clarke argued that this type of planning is merely rhetorical and is designed to reassure external recipients that issues of safety are under control. According to this argument, decision-makers—guided by technical analyses regarding responses to emergency situations—operate in a state of great uncertainty, which transforms the nature of their planning to such an extent that their ability to cope with a disaster is, in effect, a fantasy. Clarke’s analysis (1999) does not allow us to determine unambiguously whether documentation regarding emergency management in the event of a disaster has no value due to its unrealistic assumptions, or if it actually increases safety. Rather, the position is that relying on emergency plans leads units responsible for emergency management to lower their guard and become complacent. Moreover, the implementation of preventive measures can also suffer if one assumes that there actually is a truly effective scheme to deal with emergency situations (Borowska-Stefańska et al., 2019a).

In contrast, J. Kendra and Wachtendorf (2003) have posited that planning with regard to disaster management is so vital that schemes must be created even though no disaster is the same and each one always differs from what was predicted. While it might not be possible to make accurate predictions or effective responses (Clarke, 1999; McConnell & Drennan, 2006), efforts to improve disaster management remain primarily limited to better predictions and better responses, usually through improved communications (Borowska-Stefańska et al., 2019a; Cook & Lourdes Melo Zurita, 2016).

In order to achieve the goal of reducing the loss of human life and tangible damage caused by disasters (including floods; Jafari, 2005), it is important to define the optimum evacuation policy for people/households from hazard areas (Stepanov & Smith, 2009). Evacuation modelling is particularly important for authorities, planners, and those managing actual evacuations where evacuees must be moved efficiently to safety, as it can help identify bottlenecks and other weaknesses in the transport system before a disaster strikes, determine the effects of road closures due to flooding, and understand the impact of a staged evacuation on traffic load. In addition, modelling a range of evacuation scenarios may help establish appropriate evacuation policies, strategies, and contingency plans, and facilitate communication and information flow (Lim et al., 2013).

It should be stressed, however, that the evacuation and its planning are heavily dependent on, inter alia, the speed with which a disaster occurs. This is because disasters can have a slow or fast onset and the more time people have to react, the better the evacuation can be organized—in which case existing plans are useful. In the event of a tsunami (particularly a near-field tsunami) or flash floods, the possibility of rapidly withdrawing to safety is paramount to the survival of the populations at risk, but in these types of disasters there is not always enough time to implement existing plans. People’s responses to a catastrophe are therefore closely dependent on the nature of the hazard and their individual perception and personalization of risk (Lindell, 2013; Mostafizi et al., 2019). Near-field tsunamis and flash floods are more difficult in terms of managing the evacuation process because the population sometimes has only minutes to decide whether to flee. In the case of tsunamis, the difficulty additionally lies in the fact that they can form as a result of local earthquakes—which can destroy roads, bridges, and other infrastructure—which also affects the evacuation process (Mostafizi et al., 2019; Wood et al., 2018).

The idea of vertical evacuation emerged in the first decade of the 21st century. Vertical evacuation is an alternative protective measure (to horizontal evacuation shelters) aimed at keeping people safe by “going up” in multistory buildings in the event of disasters with rapid onset (i.e., tsunamis or flash floods). Vertical evacuation shelters (VES) are therefore essential if evacuation from the flood zone is not possible. Typical vertical evacuation shelters include existing or designed elevated sites, garages, public utility facilities, and other designated buildings. However, as it is in the case of disasters with a slow onset, in the event of tsunamis or flash floods, the effectiveness of vertical evacuation shelters in saving lives depends largely on the population knowing the location of these facilities and their willingness to make the decision to evacuate. In such a situation, “planning” should refer primarily to the appropriate designation of shelters within the boundaries of rapid onset disasters (Mostafizi et al., 2019; Wolshon et al., 2005).

Evacuation planning is a very complex issue involving many behavioral and management aspects. Mass evacuation tools used for a range of hazards worldwide can be grouped as: (a) behavioral based simulation models; (b) traffic simulation models; and (c) time-line/critical path management diagrams. An evacuation behavioral analysis needs to address the following questions:

How many people will evacuate? (What is the evacuation participation rate?)

When will evacuees leave in relation to an evacuation order?

What will be the rate of public shelter usage?

How many evacuees will leave the local area?

How many of the available vehicles will be used? (Alsnih & Stopher, 2004)

For an evacuation plan to be successful, it is essential to understand the evacuees’ likely reactions since human behavior plays a key role in the effective process of flood-related evacuation (Alonso Vicario et al., 2020; Hamilton et al., 2020). Unfortunately, behavior is difficult to predict and control in emergency situations, which makes it vital to understand how people perceive evacuations in terms of threat and risk and how this perception impacts their decision-making.

Decisions made at the individual level are:

whether to evacuate

when to evacuate

what to take

how to travel

route of travel

where to go

when to return (Alsnih & Stopher, 2004)

As stressed by Bubeck et al. (2020), the incorporation of human behavior into flood risk modelling is absolutely critical for the development of effective strategies of flood risk management. However, one must emphasize that such sociodemographic traits as age, gender, health, and related issues may significantly affect the ability of individuals to evacuate (Dulebenets et al., 2018/2019). Boyce (2017)argued, for instance, that age-related “disability” has a significant impact on the timing and manner of evacuation. Timely emergency evacuation is especially crucial for vulnerable citizens who are disproportionately affected by such situations, primarily due to age. Importantly, the decisions made by and for evacuees also depend on a number of factors, including an awareness of the evacuation plan and its management (voluntary or mandatory evacuation) and confidence in the evacuation instructions, among others (Murray-Tuite & Wolshon, 2013; Southworth, 1991; Sun et al., 2020). Education on disaster situations and evacuation practices can reduce people’s vulnerability and increase their ability to respond appropriately (Mondino et al., 2020).

Traffic simulation models predict evacuation-related traffic flows from a departure point, which is usually a residential area, to a destination (destinations are the “exits” of the affected threatened area) (Lumbroso et al., 2010). Evacuation modelling can be divided into three classes, depending on the scale of the model:

microscopic simulations

mesoscopic simulations

macroscopic simulations (Mens et al., 2009)

Macroscopic models are typically used for large-scale evacuations (Bayram, 2016) and are based on traffic flow theory (Barceló, 2010). Microsimulation models are most commonly used by transport engineers and take into account traffic studies at a more detailed level, focusing on individual vehicles (Bayram, 2016), while mesoscopic models combine features of both macroscopic and microscopic models (Boxill & Yu, 2000).

The type of evacuation model that is appropriate for a particular flood risk area will depend on the level of risk and the processes which the evacuation modelling is seeking to inform. For instance, a densely populated urban area—where the scale of potential evacuation is large—requires a more detailed simulation model, while for an area covering the whole country—due to the enormous amount of data—the simulation model should have more general assumptions.

A timeline diagram or critical path tool is the most basic mass evacuation “model” available, showing the critical path of emergency response for flood evacuation. The derived timeline can then be used as instructions for those responsible for managing the evacuation regarding what needs to be done, when it needs to begin, and approximately how long it might take for a given flood scenario (Lumbroso et al., 2010). Ensuring people’s safety during disasters is extremely vital; however, to make this possible, it is necessary to develop simulation tools that can be used in evacuation planning (Dias et al., 2021).

Road Transport Function During a Flood Occurrence

A Flood as an Unusual Traffic Event

In order to approach the subject of the role of road transport in the evacuation process associated with floods as reliably as possible, one cannot but refer—at least synthetically—to the role of the flood itself in the functioning of the road transport system. This is because it is a factor that can dramatically change the equilibrium level of the system, and the costs (e.g., expressed in units of travel time) of moving to this new lower level can burden the group of people being evacuated at the least opportune moment, namely at the time of the evacuation. The equilibrium of the road transport system should be understood as the state in which individual transport needs are met (i.e., transport demand and supply including the road network) and lead to a periodic balance through interacting with each other. It cannot therefore be said that the road transport system has a fixed and single equilibrium level. These levels fluctuate together with changes in the demand (i.e., the need for evacuation) or supply (i.e., flooding of part of the road network). The period in which the system reaches equilibrium at the new level following the events analyzed here involves resistance and costs on both sides of this peculiar mobility game. Within the scope of traffic modelling, the equilibrium state occurs when there is a satisfactory positive correlation of traffic volumes of subsequent iteration steps. In capacity-constrained modelling methods, different iteration techniques and capacity-constrained functions lead to an equilibrium level based on the adopted equilibrium criterion (Krych, 2018).

The occurrence of a flood and the ensuing evacuation puts the road transport system in a state where the average service time depends on the volume of traffic flows on network elements at a level of inconvenience that forces a change in the transport decisions of its users: both evacuees and those from the transport background. The coincidence of flooding and evacuation is associated with the presence of critical points in the road network where the demand reported to be handled by evacuees exceeds their capacity over a given period of time. As the volume of the evacuating population increases, the congestion area expands into adjacent sections of the network. The lowest level of such inconvenience may be manifested by a change in the path of travel. As congestion increases further, a change in transport decisions may involve postponing the start time of an evacuation, changing modes of transport, or even abandoning evacuation.

A large portion of the research conducted to determine the vulnerability of the road network to flood hazards is of a qualitative nature. This is largely due to a lack of precisely defined quantitative measures for this issue (Chen et al., 2007). In the literature, there has been a fairly extensive debate on defining the concept of vulnerability of road transport to flood risks (e.g., Berdica, 2002;Duy et al., 2019; Einarsson & Rausand, 1998; Mattsson & Jenelius, 2015; Singh et al., 2018). A number of authors have defined this phenomenon as vulnerability to events that may result in a serious reduction in the operability of the road network (Berdica, 2002; Bocanegra & Francés, 2021; Ha et al., 2021; Lin et al., 2019), while others have indicated that a section of the network is vulnerable if the loss (or significant reduction in performance) of even a small number of connections severely reduces the accessibility of that particular section when measured by means of the standard methodology for testing transport accessibility (Coles et al., 2017; Forero-Ortiz et al., 2020; He et al., 2021; Taylor & D’Este, 2007). The significance of a given component in the network encompasses both the probability of that component failing due to a flood and the consequences of this failure for the system as a whole. The more significant the component is, the more severe the failure will be for the system when that component is missing. If the probability of flooding is high, such a component is at risk. In turn, if the consequences are significant, the component is essential. Therefore, a component can be considered critical if it is both at risk and significant (de Bruijn et al., 2019; Kim et al., 2018; Murdock et al., 2018; Nicholson & Du, 1994; Figure 2).

Figure 2. Main components of a critical element within the road infrastructure.

Performance-related changes generated on the road network as a result of flooding typically cover an area significantly larger than just that where the hazard occurred, and may last considerably longer than the duration of the hazard itself (Do & Jung, 2018; Jenelius & Mattsson, 2012). Disruptions to the transport network involve a wide range of factors in terms of both the likelihood of flooding and the associated negative effects: from minor traffic collisions and accidents, which occur relatively frequently, to unforeseen failures which can have severe repercussions such as with bottlenecked, flooded, or damaged bridges (Berdica, 2002; Diakakis et al., 2016; Pregnolato et al., 2017, 2020).

Road transport disturbances can generally be divided into typical and nontypical (Chen et al., 2016). The former is caused by the imbalance of transport demand in time and space, and are cyclical in nature over a specific time horizon. However, not all disturbances are caused by the insufficient capacity of network components. In addition to typical disturbances, nontypical ones may be caused by random occurrences—such as engineering works being conducted on the road strip or in its immediate vicinity or public gathering events—which require appropriate traffic organization measures to be implemented. Additional causes can also include various types of road incidents, unfavorable weather conditions, the temporary deterioration of the road surface, the occurrence of floods, and so on. Typical and nontypical disturbances may lead to congestion (Chung, 2012; Grant-Muller & Laird, 2006; Jacyna, 2009; Kuo & Tang, 2010; Muller & Laird, 2006; Skabardonis et al., 2003; Żochowska & Karoń, 2012), the probability and magnitude of which vary depending on the type of network, its redundancy, and its susceptibility to disruption, which is related, inter alia, to the effectiveness of risk management measures and the proper planning of road works or the efficiency in removing obstacles hindering the flow of traffic.

Types of Impact on the Road Transport System

Taking into consideration the reduced capacity of the road network due to floods is necessary in the modelling of traffic distribution related to the evacuation process. It allows for taking into account the impact of traffic volume and the congestion level on the choice of transport routes and travel time in the form of resistance resulting from road resistance function modified by flood factors. The problem of evacuation path selection in the face of flood impacts on the road network is related to the information (or lack thereof) about path selection by other traffic participants, especially those involved in the evacuation. In the process of modelling these elements, therefore, great importance is attached to the function of the dependence of the handling time of the transport demand reported by evacuees on the traffic volume coming from the transport background and the capacity (often reduced by the impact of flood water).

The occurrence of nontypical events may lead to damage to the physical road network. This includes failures of road components and equipment and may result in impediments for traffic, or the partial or complete closure of a given section of the network. Such damage may stem from factors which are unrelated to the use of the network, or may result from the nature of this use, or may be the effect of intentional damage. The unrelated factors include weather and geological conditions as well as human activity which does not stem from people using the road network. This group also includes a wide spectrum of flood-related effects (Antoniou et al., 2013; Brijs et al., 2008; Gil & Steinbach, 2008; Kilgarriff et al., 2019; Klipper et al., 2021; Mitsakis et al., 2014).

Flooding may affect the operability of road transport in both a direct and indirect way. The degree of impact depends not only on the nature of the phenomenon itself, but also on the surroundings and the time period during which road transport is affected. Therefore, it would be extremely difficult to provide a complete list of the direct and indirect flood-related factors that may impede the operation of road transport, which also means that it would be virtually impossible to make a comprehensible list of the properties and effects of these individual factors. Direct factors occur as a result of the direct impact of flooding on the network and its users. Indirect factors, on the other hand, may occur over an area far larger than that directly affected by the flood. Moreover, the duration of its impact may be considerably longer than the duration of the flood itself (Borowska-Stefańska & Wiśniewski, 2018; Merz & Thieken, 2004). Examples of indirect factors (indirect damage) are traffic disturbances (changes to traffic conditions) or losses resulting from changes to traffic (occurrence of induced or suppressed traffic) due to destroyed or preventively deactivated infrastructure (Merz & Thieken, 2004). The two types of factors can also be divided more specifically into tangible and intangible, depending on whether they can be expressed in monetary terms. Tangible losses are those that can be easily measured monetarily (e.g., the destruction of a stretch of road), while intangible losses (e.g., transport delays or drivers feeling unsafe)—cannot be commercially valued (Jonkman et al., 2008; Merz et al., 2010). Direct damage is definitely easier to estimate than indirect damage (Merz et al., 2010), which is perhaps why the largest volume of literature relates to the assessment of direct tangible damage (Merz et al., 2004).

Following a flood, the road transport network may be exposed to factors that reduce its performance. A disaster of sufficient intensity or duration may cause the capacity of individual elements of the network to be compromised, thus affecting their operability. These factors may include, inter alia, exposure to high pressure exerted (e.g., by water, or an abnormal traffic load), or excessive structural load (e.g., due to scouring; Jamrussri & Toda, 2018; Marcinkowski, 2009; Mulyono et al., 2009; Winter et al., 2016). Flooding may lead to structural deformation, which in turn might reduce the load-bearing capacity of the structure, rendering it unserviceable. The destructive effect of pressure can be caused by, for instance, the passage of flood waves, ice floes, and other objects carried by water (Lindenschmidt et al., 2018; Vuojala-Magga & Turunen, 2013). Exceeding the permissible load on network elements may stem from the movement of overweight vehicles or an excessively large flow of vehicles caused by flood-related changes in traffic organization (e.g., designation of evacuation routes). Reduction in the load-bearing capacity of road structures or a loss in their structural stability may also stem from changes in the foundation of structures due to the impact of flood waters (Ismail et al., 2020; Martinez-Gomariz et al., 2017; Smith et al., 2019).

Forms of Interference With the Resilience of the Road Transport System

The functioning of the road transport system in the face of floods redefines the concept of optimal transport task. In such a complex combination of transport tasks that occur when an evacuation has to be initiated, the destination function can be decomposed by introducing constraints according to the formula of the transport task, or the evacuation. Under these specific circumstances, the concept of optimality should be based on the selection of the most effective solution in the conditions where only a heuristic approach is possible due to the limitations of having access to complete data on the circumstances of the evacuation process. However, a distinction must be made between the notion of optimality and the notion of effectiveness in the evaluation of the result of the adopted solutions.

Determining a precise and exhaustive list of flood-related effects on road transport is an extremely complex task. This is due to the multitude of elements that influence the design of transport networks and systems, their subsequent use, or the co-occurrence of other nontypical events. Therefore, it is necessary to make certain generalizations and categorizations in this matter. A holistic view of the factors determining the effectiveness of transport systems impacted by a flood should also concern spatial development (the spatial component), the transport network (the transport component), and the socioeconomic habits of users (the individual component) for a specific moment in time (the time component). The diversity of variables in each group of factors gives great leeway to develop different types of models and simulations. Invariably, the occurrence of a flood is accompanied by a drop in the attractiveness of journeys and their destinations. This nontypical event may increase distance decay on the route to a chosen destination, and sometimes also be so strong a deterrent that a trip may be abandoned (Li et al., 2021; Sohn, 2006).

When the operation of road transport is disrupted by external occurrences (here: flooding), any negative impact will primarily affect its users. Some locations will become less accessible to them, which may be an inconvenience or even an added danger. However, not all sections or engineering structures of the road network are equally important for its operation at times when the entire network is exposed to flooding (Balijepalli & Oppong, 2014). It is therefore crucial to assess the vulnerability of transport networks and its individual sections so that emergency response plans can be developed and the impairment of the system’s capacity after a flood can be minimized (Chen et al., 2007; Kim & Yeo, 2016).

A vital element for restoring a road transport system to an acceptable state of balance is to take action as soon as flooding has occurred. The debris carried by water (remains of trees, structures, etc.) must be prevented from blocking the openings of bridges or culverts, otherwise it will cause water flow to dam up. In extreme cases, when the level of accumulated water is already high enough that destruction to a bridge may occur, and all other measures have failed to produce the desired effect, the embankment on the access to the bridge may be broken to allow the pent-up water to escape. For such reasons, it is extremely important for infrastructure managers to remain alert when there is a risk of flooding, and for the local communities to be involved (Abana et al., 2019; Madaj & Wołowicki, 2013; Wang et al., 2018).

Flooding (particularly flash floods that occur rapidly as a result of heavy rainfall) is a major cause of transport disruption (Department for Transport [DfT], 2014; Pregnolato et al., 2017) and it is expected that this problem will escalate in the future (Dawson et al., 2016). It has become particularly severe on urban road networks due to the high proportion of impervious surfaces that prevent water infiltration into the ground (Pregnolato et al., 2017).

Research indicates that there are a number of possible responses to a reduction in the performance of a transport system affected by, or at risk of, destructive flood waters. In some cases, this response may involve altering the operational or technical parameters of key elements of the transport infrastructure, for instance by raising its height above the assumed maximum flood levels (vulnerability reduction). Frequently, road conditions are also improved during such disruptive improvements. Sometimes this renders the network more dependent on these key elements (their importance increases) and, thus, become more vulnerable to their failure (Fekete et al., 2017; Li et al., 2020; Figure 3). An alternative approach is to add new sections to the network (especially alternatives to possibly-affected segments), thereby increasing evacuation route options (Taylor & D’Este, 2007).

Figure 3. The cycle of increasing flood resistance of the road network.

The Role of Road Transport in the Evacuation Process

The Role of the Time Factor

The occurrence of a flood or the appearance of its symptoms may result in the implementation of the evacuation process, which usually affects the functioning of the transport system (Suzuki et al., 2019; Tanaka & Shimomura, 2021; Teichmann et al., 2021). The road network may become overloaded with traffic due to the movement of people (or the transportation of animals or property) from areas at risk to safety. However, some journeys with typical motivations (including those by private car) may not be made as a consequence of the flood, which means that people may alter their usual travel behavior (e.g., commuting to work or going to the shops) in order not to endanger their health and lives. The consequences for road transport should therefore be analyzed both in terms of mobility related to evacuation and its impact on the balance of the system as a whole.

Guidelines on transport-related aspects of the evacuation process typical for the emergency procedures adopted in a given area (with a considerable diversity of detail in the provisions) can be found in numerous sources. These may take the shape of emergency management plans, flood protection operational plans, or evacuation plans. Provisions for the organization of transport can also be found in the directives concerning the responsibilities of the particular services (e.g., fire brigades, police, or civil defense) which participate in organized evacuations—that is, the principles governing the formation and conduct of vehicle columns during self-evacuation or an organized evacuation.

The course of an evacuation and the related management of road transport may take various forms, depending on the dynamics of the threat. An evacuation may take place immediately after a sudden, unexpected, and direct danger to the public. Flash floods or tsunami floods are of particular importance in this context. The dynamics of these phenomena force evacuees to make almost automatic decisions. In this case, self-evacuating drivers are exposed to high stress, which can significantly condition their transport behavior (Ahmed et al., 2018; Balakrishna et al., 2008; Drobot et al., 2007). Changes to the organization of traffic which are planned in advance and the establishment of information channels which can be used to communicate these changes to road users at an early stage make it possible to bring the transport system to a new state of balance in a more time-efficient manner. This process should be conducted with the active participation of Intelligent Transport System (ITS) operators (Giovanna et al., 2016; Yusof & Raffiee, 2018) as they usually have a variety of tools at their disposal which, being tailored to the scale of a given system, provide real-time information to users about difficulties on the road network and how to avoid them. In addition, the rerouting of transit traffic gives both emergency services and evacuees the opportunity to access either the area affected or safety. What may be problematic here, however, is the conflicting messages issued by personal navigation devices (Haghpanah et al., 2021; Knyazkov et al., 2014).

On the local scale where communities are directly and immediately affected by flooding, the key is a preplanned response regarding the selection of the transport mode and the evacuation route and timing, which should have already been made at the level of individual households (family evacuation plans). Communities potentially susceptible to highly dynamic elements (flash floods, tsunamis) are particularly predisposed to developing this automaticity. In this respect, an obvious difficulty is to determine the vulnerability of individual evacuation “actors” by developing trusted and reliable sources of data, and incorporating risk and vulnerability mapping (Pel et al., 2010). From a methodological approach, one should note agent-based modelling (ABM) (Wang et al., 2016), which is based on creating agents with attributes defining their current state who make spatial decisions by exchanging information with each other. This allows an element of randomness to be introduced into the model (Priest et al., 2011; Samuels et al., 2008; Tagg et al., 2016).

An evacuation may also be a preplanned relocation of people (and possibly animals and property) from the vicinity of potentially at-risk areas or structures that are damaged or malfunctioning. The decision to do so is often taken when the signs of a threat occur. If properly prepared, this provides a longer time for both the emergency services and the public to react. The moment the evacuation is instigated, its timely rollout, and the status of the flood (i.e., it is forecast or already occurring) are all variables that impact the organization of transport during the evacuation and the traffic on the road network where the evacuation is being carried out (Durst et al., 2014; Esm et al., 2010; Liu et al., 2010; Tagg et al., 2013).

The Ability to Self-Evacuate

Regardless of the timeline of the evacuation decisions, a key issue for the organization of transport is to consider it with respect to the people’s ability to self-evacuate (Czerechowski, 2015; Mondal, 2021; Parker et al., 2009). The people affected will exhibit slightly different features during the evacuation depending on whether the arrival of the flood wave is imminent or the evacuation occurs during the flood itself, or whether it is a self-evacuation (scheduled and spontaneous) or a planned evacuation (Rawłuszko, 2016).

The flow of self-evacuating vehicles affects the balance of the transport system, especially in view of the simultaneously planned or implemented closures of some sections of the road network (e.g., due to the threat of flooding, damage to the structural stability, or the “cutting off” of a road segment once a neighboring section has been flooded). Therefore, it is crucial to make an assessment of the possible degree of self-evacuation and to identify its potential directions, destinations, and evacuation routes. The provision of a fuel supply and technical assistance along evacuation routes may also be worthwhile (Figure 4).

Figure 4. Groups of factors which determine the effectiveness of evacuation conducted by means of road vehicle transport.

However, self-evacuees will not be the only “new” traffic from the danger zone to safer places (Gromek et al., 2014). After all, people who, for various reasons, are unable to ensure their own safety will also be additional and nontypical users of the road network who must be evacuated as well. This includes people who are alone, the elderly, those in poor health, and those who have reduced mobility or restricted freedom (e.g., prisoners). This extra load needs to be taken into consideration in evacuation plans, and the means of transport dedicated to the evacuation of people with such special needs must be prepared well in advance, including the specificity of their transport options and movement in the transport scenarios of the whole evacuation process. For the proper execution of transport organized by the authorities, the traffic created by self-evacuating vehicles may constitute a significant obstacle, including reaching local capacity limitations (Lindell et al., 2018; Lumbroso et al., 2008; Perera et al., 2020; Zhang et al., 2020). Apart from a decrease in travel speed or flow, the evacuation may also be disrupted by collisions and accidents.

Therefore, the discussion should take into account the needs and challenges of vulnerable groups of the population during an emergency evacuation. Assessing social vulnerability to disasters is seen as a key factor in reducing its risk (McEntire, 2005). Vulnerability in this regard is defined as the ability of people to anticipate catastrophes, deal with their consequences, resist them, and return to conditions prior to their occurrence (Garatwa & Bollin, 2002; Schneiderbauer & Ehrlich, 2004). Therefore, it depends on the social, cultural, economic, and political conditions that constitute the basis of social differentiation. Some of these conditions are reflected in the evacuation process. The effective communication of information related to the organization of transport during evacuation to vulnerable populations is an essential need in crisis situations. Various factors (e.g., the diversity of vulnerable populations, their specific needs, existing barriers to contact and communication, and a lack of resources and mechanisms to locate and assist these populations) make it difficult to develop comprehensive and effective plans for communicating such information to people with reduced mobility during evacuation (Turner et al., 2010). It is also problematic to collect sufficiently detailed (both spatially and substantially) data on the characteristics of the population (Chakraborty et al., 2005). Social vulnerability can create diverse spatial patterns, which makes crisis management difficult. Emergency evacuation in connection with the impending disaster requires the local and state administration to make important decisions regarding, inter alia, escape routes, shelters, and evacuation dates (Abioye et al., 2020; Dulebenets et al., 2020). This task is further hampered by a highly diverse society. Therefore, it is necessary to develop models (e.g., based on driving simulators) for optimizing emergency evacuation planning that take into account the mental and physical characteristics and frustration of evacuees, because they affect the driving performance of individual people, including vulnerable groups of the population (Dulebenets et al., 2020; Rufat et al., 2019; Tate et al., 2021).

Model Approaches

A solution to model this type of issue can be, for instance, an approach based on the Critical Cluster Model (CCM). Church and Cova (2000) presented this methodology to identify the particular areas where, during an evacuation, it may be necessary to evacuate large numbers of people using a road network which exhibits low-capacity values. To determine the time required to evacuate a given residential area, they used information on: (a) the evacuated population (estimated on the basis of the number of houses and the average ratio of residents per home), (b) the number of vehicles at the disposal of those evacuating, and (c) the capacity of the roads leading away from the affected area.

Researchers representing a spectrum of disciplines have devoted much analysis to optimizing the transport component of evacuation (Murray-Tuite & Wolshon, 2013). Some have focused exclusively on evacuation in terms of the evacuees’ accessibility to temporary accommodation until the danger has ceased (Borowska-Stefańska et al., 2017), while others have researched the potential evacuation itself, taking into account the departure of evacuees from the hazard zones (Church & Cova, 2000). As for the transport component, an issue commonly addressed is the determination of optimum evacuation routes. For instance, Chen et al. (2012), while taking into consideration the Critical Cluster Model (CCM), studied the process of evacuation, applying Dijkstra’s algorithm for this purpose. In contrast, Borowska-Stefańska et al. (in press) introduced the A* algorithm, while Shahabi and Wilson (2014) proposed modelling evacuation in urban systems with the CASPER algorithm. Richter et al. (2013) noted that evacuation management can yield good results, and Yuan et al. (2017) addressed the issue of traffic simulation during evacuation based on a multilevel decision model.

Borowska-Stefańska et al. (2019b) proposed the Dinic, Edmonds-Karp, and Ford-Fulkerson algorithms to determine evacuation routes. Broadly speaking, these tools are used to determine the largest flow that can be obtained in a noncyclic directed graph and can also be sometimes applied in studies of congestion (Abdullah & Hua, 2017) and road capacity, including research analyzing optimum vehicle speeds in terms of capacity (Moore et al., 2013) for which purpose Liu et al. (2006) applied the Adaptive Evacuation Route Algorithm.

Understanding the circumstances under which individuals at risk make their evacuation decisions is also crucial for the successful modelling of evacuation (Dash & Gladwin, 2007). According to Southworth (1991) and Murray-Tuite and Wolshon (2013), transport behavior in this regard depends on, inter alia, confidence in evacuation instructions, risk perception, and the stage at which the decision (in psychological terms) is made to leave the affected area. Laska (1990) lists four stages here: anxiety, danger recognition, acceptance, and deciding to evacuate.

The variety of methodological approaches, the diversity of the transport network, the distribution of people at risk, the safe places, and most importantly, the considerable unpredictability of their behavior (for instance, during a chaotic evacuation) (Masłowska-Szczerba, 2015), all lead one to believe that it is justified to conduct research on the organization of evacuation-related transport only when local conditions and spatial scale are taken into account, as the adoption of universal assumptions in this respect could overly affect the returned results (Wiśniewski, 2021).

The procedures for modeling the evacuation process presented in the section Model Approaches suggest a differentiated approach to the study of transport aspects. It is difficult to clearly indicate which is the most effective. As a flood is a phenomenon difficult to accurately predict, the related evacuation should be modeled taking into account local factors and by specialists with interdisciplinary knowledge in the field of disaster management—the natural and socioeconomic context (with particular emphasis on transport).

Awareness of the wide range of measures taken to ensure people’s safety, including during their transit to safe places, may lead to complacency, but one should remember not to treat emergency plans as the perfect remedy to crisis situations. By maintaining a balance between confidence in the effectiveness of forecasts (usually planned as optimization of the transport component) and vigilance, the authorities responsible for preparing emergency strategies can increase the feeling of safety among those residing in hazard areas without losing sight of their responsibilities for crisis management itself (Boin & McConnell, 2007; Cook & Lourdes Melo Zurita, 2016; J. M. Kendra & Wachtendorf, 2003; McConnell & Drennan, 2006).

Influence of Evacuation on Transport Balance

Local Determinants

An important research challenge in the field of traffic engineering and transport planning (for cognitive and pragmatic reasons) is the identification of the local determinants of an evacuation—for both self-evacuation and an organized evacuation. Knowledge of the transport behavior that may potentially occur during an evacuation allows conclusions to be drawn about its effects on the transport balance at different spatial levels of transport systems. However, the difficulty of such research lies, inter alia, in the variability of the analysis and the indication of how the individual stages of the evacuation affect the entire road transport system.

The strength of the impact that the evacuation has on the entirety of transport relationships depends, among other things, on the social traits of given evacuees, their experience with this type of emergency, their knowledge of current evacuation guidelines, the location of the evacuated area within the road network, and the size and course of traffic flows and its traffic-generating potential (Du et al., 2017; Figure 5). The assessment of this impact can be performed through the inclusion of empirical data when calibrating the traffic model. Depending on its detail and researchers’ needs, micro-, meso- and macrosimulation modelling can be applied. These models provide a picture of the traffic distribution on the network in the event of a flood and enable the optimal evacuation routes to be determined based on data regarding the type of evacuation and the intended means of transport. This type of modelling is particularly relevant for transport authorities directing it at different stages of evacuation, including the emergency services. Identification of bottlenecks in the transport system prior to any flooding and assessment of the impact of flood-related road closures or different traffic organization all offer the opportunity to model alternative evacuation scenarios (Lim et al., 2013) which can take into account the magnitude of disruptions to the “typical” road traffic in a given area (Naghawi & Wolshon, 2010; Zhao & Wong, 2021).

Figure 5. Groups of factors determining the impact of the evacuation process on the balance of the road transport system.

Generation and Organization of Traffic

In order to determine as accurately as possible the impact of the evacuation on the balance of the transport system in a broader perspective, several key variables must be taken into consideration, the first being the traffic-generating component. One may assume that the “normal” traffic-generating potential of areas under evacuation is reduced by the number of people at risk of flooding, because, as they are preparing to leave their homes, they are not making journeys for any other motivation. What matters greatly is that the modal division typical for a given research area be taken into account in the simulations, which requires a comprehensive study or a similar analysis to be conducted beforehand (Abad & Fillone, 2020; Du et al., 2017; Luathep et al., 2013). The second element determining the dynamics of the road transport system is the traffic-attracting component. Its variability in relation to evacuation simulations stems from the supposition that the attraction potential of the areas under evacuation decreases by the size of the population that undergoes this process. This is because initiating preparations for the planned evacuation from flood-hazard areas in the face of the first signs of the disaster will also block those journeys that would otherwise end in these areas. The high dynamics of the evacuation process prevents those who would make trips to flood-hazard areas from doing so due to a lack of time to find an alternative destination that would satisfy the same needs. The indicated changes in the volume of generated traffic to and from the area under evacuation (without the evacuation itself) should not pose a problem for the transport system. On the contrary, in a short-term perspective they may actually improve the vehicle flows that are unrelated to the affected area (Alam & Habib, 2020; Alam et al., 2018).

Another component that can uniquely impact the transport balance is planned evacuation. Under the assumption that flooding has not yet occurred (except for some early signs) it can also be presumed that no closures in the road network have occurred. The road transport system may face a rise in traffic on the network due to the change in destinations and travel routes when compared to “regular” conditions and modifications of the modal division. This is because the part of the traffic originally served by public transport disappears and is partially replaced by private car transport—mostly for self-evacuation (Borowska-Stefańska et al., 2019a; Fahad et al., 2019).

As for the impact that evacuation has on the balance of the transport system, the role of traffic organization is also worth considering (Borowska-Stefańska et al., 2019b; Liu & Hu, 2008). Road network administrators have a wide range of possibilities—from no interference in the “regular” organization to the imposition of absolute priority for vehicles evacuating from hazard areas or a ban on entering certain sections of the network. Depending on the dynamics of the evacuation and the preparedness of transport authorities, it may be necessary for the bodies conducting the evacuation to have provisional control of directing traffic, or it may also be possible to apply any previously prepared changes to traffic organization to reduce the negative external effects of a nontypical event. In the case of an emergency evacuation, it will be difficult to implement a traffic organization adjusted to the needs of the evacuees. Furthermore, due to the high dynamics of the situation, it is also unlikely that the attraction potential of individual hazard areas will be reduced—as, in the absence of prior warnings, journeys destined for these areas are likely to have already begun.

Properties of the Road Network

Flood-related evacuation from hazard areas may mainly have a significant impact on changing the spatial distribution of road network load—especially when the actual evacuation begins and is conducted along roads that have been closed to other users, or when the flooding has already occurred and some roads have already been closed to traffic (Dhingra & Roy, 2015; Kasmalkar et al., 2020; Mioc et al., 2008). Evacuation will only have a limited impact on the travel speed on the road network when the transport system displays a large adaptive capacity owing to a high level of variability. High flexibility of the system, in combination with large capacity from individual network sections, guarantees the absorption of the traffic flow from private car transport as a result of a change in the transport cost ratio between the different modes of public transport that are usual for this section of the road network, without a significant decrease in the system capacity (Deria et al., 2020; Duy et al., 2019).

Figure 6. Balancing the volume of induced and suppressed traffic that accompanies evacuation.

In the face of an evacuation, the part of the traffic flow within the network that is absorbed by alternative transport routes as a result of a change in the transport cost on certain sections due to changes of its shape, traffic organization, or the capacity of its elements brings the system to a new level of balance. Induced and suppressed types of traffic which are the resultant of a shift in the generation and attraction potential of flood-hazard areas, and which accompany the different stages of evacuation, can, to a large extent, cancel each other out in terms of their impact on the efficient performance of the entirety of the transport network (Chiu et al., 2008; Fu et al., 2014; Pel et al., 2010; Figure 6).


Well-prepared evacuation processes are crucial in light of the unpredictability of occurring hazards, including floods, which will continue to occur with an increasing frequency around the world. Evacuation is a complex activity and its course is difficult to predict, which makes prevention measures or rescue operations challenging to implement. Evacuation itself is affected by many different factors, all of which are crucial for human health and life. Evacuation from flood-hazard areas is a major challenge for the field of flood risk management, as well as the fields of traffic engineering and transport planning. This is particularly true when the research has to include not only those journeys directly related to escape from hazardous places, but also their reflexive relationship with the total number of journeys made in the “background” of the evacuation itself. Therefore, in order to enable and facilitate evacuation, a number of planning and preparatory measures need to be taken based on scientific and technical knowledge, taking into account primarily those to be evacuated and well-thought-out intermediate and final destinations to which they should be transported, as well as the authorities and resources necessary to conduct the evacuation effectively and efficiently. For this purpose, not only data from flood hazard and risk maps but also evacuation and traffic models should be used. In the case of flooding, an early warning may allow sufficient time for evacuation to take place before the actual flood even occurs. It is therefore important to have at hand an evacuation plan that defines the optimum evacuation procedures for people in hazard areas and assigns evacuees to specific routes or destinations for before the disaster strikes (Figure 7; Apte, 2009; Borowska-Stefańska et al., 2019a; Campos et al., 2012; Stepanov & Smith, 2009).

Figure 7. The scheme of the optimization pattern for the evacuation process.

The time taken for rescue services to arrive at the scene and the duration of an evacuation from hazard areas have an impact on potential injuries and loss of life, and, therefore, the effectiveness of rescue operations. Previous experience has shown that the main issue in conducting an evacuation is that routes from hazard areas may be limited in number and insufficient in capacity to handle the traffic surge during a large-scale emergency evacuation (Han et al., 2006; Urbina & Wolshon, 2003). Stepanov and Smith (2009) argued that the capacity of transport networks is generally unable to meet the intense transport demand during evacuation. Even in the case of a locally conducted operation, transportation networks impede the fast clearing of the population from an affected area (Church & Sexton, 2002; Cova & Johnson, 2003). In order to manage such emergency situations more effectively, evacuation plans from hazard areas should be drawn up for the scenarios that are most likely to occur, even though they may well have to be revised again after the disaster (Alexander, 2002; Borowska-Stefańska et al., 2019a; Campos et al., 2012).

In turn, in the event of disasters where the response time is extremely short, such as near-field tsunami or flash floods, it is also essential for the success of an evacuation process to know, in the first place, how they should behave and, secondly, to “plan”—yet this planning is of a completely different nature, namely it is aimed primarily at designating shelters adapted to protect the lives of the population within the limits of the areas at risk. Future research could be expanded on:

the individual properties of the population (e.g., age, gender, etc.), which are important in the context of social vulnerability;

the way the population is informed about the hazard, their reaction time to it, and their awareness of the hazard and their readiness to evacuate;

different travel motivations in the case of the transport background;

the inclusion of weather-related factors in the modelling, depending on the location of the analyzed area and the types of floods that occur there;

different evacuation scenarios or the different modes of transport used to evacuate (including evacuation on foot); and

the impact of the dynamics of flooding into the process of evacuation.


The authors would like to thank Professor Edmund Penning-Rowsell for the opportunity to participate in such a prestigious research project and for his invaluable substantive support.

Further Reading

  • Flanagan, B. E., Gregory, E. W., Hallisey, E. J., Heitgerd, J. L., & Lewis, B. (2011). A social vulnerability index for disaster management. Journal of Homeland Security and Emergency Management, 8(1).
  • Khorram-Manesh, A., Goniewicz, K., Hertelendy, A., & Dulebenets, M. (Eds.). (2021). Handbook of disaster and emergency management (2nd ed., pp. 102–108). Gothenburg.
  • Penning-Rowsell, E. C., Sultana, P., & Thompson, P. M. (2013). The “last resort”? Population movement in response to climate-related hazards in Bangladesh. Environmental Science & Policy, 27, S44–S59.