Modelling Human-Flood Interactions

A Coupled Flood-Agent-Institution Modelling Framework for Long-Term Flood Risk Management

Yared Abayneh Abebe

• Publisher: CRC Press
• Year: 2021
• ISBN:
  9781000368130
• Edition: 1
• Topic: Ciencias biológicas
• Subtopic: Ciencia medioambiental

The negative impacts of floods are attributed to the extent and magnitude of a flood hazard, and the vulnerability and exposure of natural and human elements. In flood risk management (FRM) studies, it is crucial to model the interaction between human and flood subsystems across multiple spatial, temporal and organizational scales. Models should address the heterogeneity that exists within the human subsystem, and incorporate institutions that shape the behaviour of individuals. Hence, the main objectives of the dissertation are to develop a modelling framework and a methodology to build holistic models for FRM, and to assess how coupled human-flood interaction models support FRM policy analysis and decision-making. To achieve the objectives, the study introduces the Coupled fLood-Agent-Institution Modelling framework (CLAIM). CLAIM integrates actors, institutions, the urban environment, hydrologic and hydrodynamic processes and external factors, which affect FRM activities. The framework draws on the complex system perspective and conceptualizes the interaction of floods, humans and their environment as drivers of flood hazard, vulnerability and exposure. The human and flood subsystems are modelled using agent-based models and hydrodynamic models, respectively. The two models are dynamically coupled to understand human-flood interactions and to investigate the effect of institutions on FRM policy analysis.

1
INTRODUCTION

1.1 MOTIVATION

Of all weather-related disasters in the last two decades, floods are by far the most common (47 %), affecting 2.3 billion people (CRED and UNISDR, 2015). The CRED and UNISDR report emphasizes that after storms and geophysical disasters, floods have been causing the third highest amount of economic damage (662 billion USD) over the past 20 years. The number of flood events has significantly increased, in which urban areas have been hit particularly hard (Jha et al., 2012). The risk associated with floods can be defined as the probability of negative impacts due to floods (Schanze, 2006). Flood impacts are mainly attributed to the extent and magnitude of a flood hazard which can be caused by one or a combination of fluvial, flash, pluvial, groundwater and coastal floods (Vojinovic and Huang, 2014). However, the negative impacts are also due to the vulnerability and exposure of natural and human elements such as individuals, livelihoods, economic and cultural assets, infrastructure, ecosystems and environmental resources (Vojinovic et al., 2016).

In his dissertation, Gilbert F. White (1945, p. 2) states: “Floods are ‘acts of God,’ but flood losses are largely acts of man.” One may argue that floods can be “acts of human” as much as they are “acts of God.” For example, a rainfall with certain intensity may cause flooding that disturbs livelihoods in an Ethiopian city due to poor drainage infrastructure while a Dutch city may not register flooding from an equivalent rain intensity. Nevertheless, White’s statement that flood losses are aggravated by human encroachment of floodplains is indisputable.

Furthermore, in an article entitled “Taking the naturalness out of natural disasters,” O’Keefe et al. (1976, p. 566) stated: “Without people there is no disaster,” asserting their argument that socio-economic factors contribute to disasters more than natural factors do. When “nature” is considered as the threat, the hazard component of a disaster becomes more influential, and risk reduction strategies focus on engineering structural measures and hazard-based risk awareness and warning systems (Gaillard, 2010).

But, such measures are not always effective. For example, structural measures such as dykes are designed based on return periods (e.g., a 100-year storm event), which are computed using statistical analysis of historical flood data in an area. This entails that dykes may fail or overtopped when a potentially higher flood event occurs or when peak flood of the design return period increases over time due to, for example, climate change (Pinter et al., 2016). Such scenarios are sources of residual risk that residents either unaware of or ignore. As a result, dykes creates a sense of safety by reducing residents’ flood risk perception (Ludy and Kondolf, 2012). Ludy and Kondolf conclude that residents become “involuntarily exposed to risk”.

Based on these arguments, comprehensive approaches forwarded to reduce flood impacts should include human adjustment to floods (White, 1945), and focus on human elements such as vulnerability, capacity and resilience, which are shaped by socio-economic factors (Gaillard, 2010; O’Keefe et al., 1976).

1.2 HUMAN-FLOOD INTERACTIONS

Floods and their impacts are not just nature-related. They rather are the result of meteorological and hydrological factors aggravated by human actions (APFM, 2012). Changes in the climate system and economic, social, cultural, institutional and governance factors are drivers of flood hazard, vulnerability and exposure (IPCC, 2012, 2014a). For example, in the context of urban flood risk, population growth and the associated urban expansion result in changes to land use and land cover. That leads to an increase in impermeable surfaces, which increases the flood hazard. When accompanied by inadequate planning and policies, urban expansion may occur in flood-prone areas increasing exposure; or it may occur in dense, low-quality informal settlements that contribute to a higher number of vulnerable people (Jha et al., 2012). For example, in the UK, the government acknowledged that the increasing demand for housing leads to more building in high flood risk zones (Department for Communities and Local Government, 2007), in which the proportion of new residential properties located in flood zones grow from 7 % in 2013–14 to 9 % in 2015–16 (Department for Communities and Local Government, 2016).

Moreover, the behaviour of individuals plays an essential role in flood risk. Based on their economic situation and risk perception¹, heterogeneous individuals living in flood-prone areas may implement their local measures to reduce hazard (e.g. nature based solutions such as green roofs or rainwater tanks (Vojinovic and Huang, 2014)) or vulnerability (e.g. dwellings with a non-habitable ground floor (Gersonius et al., 2008)). Further, individuals may insure their properties to avoid financial losses or to recover better, in the case of flooding. Currently, governments are reorganizing flood insurance policies changing individual behaviour (Dubbelboer et al., 2017). Individuals may also reduce exposure to flood hazard by relocating assets to less flood-prone areas and through evacuations (UNISDR, 2015).

1 Risk perception is a function of values, feelings, experiences and cultural perspectives (Schanze, 2006).

In flood risk management (FRM), the likelihoods of adopting and implementing measures that reduce flood hazard, vulnerability and exposure depend on changes in individual and institutional behaviour in response to the potential of flooding and the accompanying impact (Loucks, 2015). Therefore, on the one hand, FRM is dependent on the rules, regulations, policies and implementations that aim to reduce flood risk. On the other hand, it relies on how individuals react towards those aspects and adapt their behaviour. The factors, which shape the flood hazard and a community’s exposure and vulnerability to flooding, can be understood as institutions. Institutions are key elements in the social, economic and political makeup of human beings that define our interactions with the physical system. The importance of institutions as social structures that influence society as a whole, and in turn, are influenced by society has been repeatedly emphasized by prominent scholars in economics, political science, sociology and ecology among others (e.g., Hodgson, 1988; North, 1990; Ostrom, 1990; Young, 1986).

1.3 SYSTEMS PERSPECTIVE AND SOCIOHYDROLOGY

To strengthen FRM and to reduce flood risk, a holistic, interdisciplinary approach that integrates all components of risk is essential (Aerts et al., 2018). This approach should consider the interactions between human and physical subsystems² (Schanze, 2006; Vojinovic, 2015). The “human subsystem” consists of decision-making individuals, whose collective behaviour creates and is constrained by institutions such as norms, habits and laws. The human subsystem is embedded in and interacts with the “physical subsystem”. The physical subsystem includes drainage systems and dykes that might be affected by flood events, and the flood itself. With interactions across multiple spatial, temporal and organizational scales, and behaviour driven by imperfect information and bounded rationality, the coupled human-flood system is a complex system (see also Pahl-Wostl, 2015). Further, as individuals and organizations learn (Mitchell, 2009) from previous flood impacts, the human-flood system is a complex adaptive system (CAS).

2 The term “physical” in the physical subsystem is a generic expression. Depending on the coupled model we address, as in Chapter 2, it will be replaced by a specific term (for example, “environment” or “ecology” in socio-environmental or socio-ecological systems; “technology” in socio-technical systems; “water” in coupled human-water systems; and “flood” in coupled human-flood systems).

Human-flood interaction studies have been a subject of interest for decades. However, there has been resistance from hydrologists to include or couple models that capture the human dynamics with their hydrological models (Loucks, 2015). As a result, models used for policy decision support in FRM focuses on quantitative assessment of flood hazard and flood hazard reduction. Recently, modelling of the coupled human-flood system is getting more attention in socio-hydrology (also sociohydrology), which studies the co-evolution of humans and water explicitly by considering the possibility of generating emergent behaviours (Sivapalan et al., 2012). In socio-hydrology, the human subsystem is regarded as an endogenous part of the water subsystem, and there is a two-way interaction between the two subsystems.

1.4 RESEARCH GAPS IN HUMAN-FLOOD INTERACTION MODELLING

Sivapalan and Blöschl (2015) identified two possible approaches to model coupled human-flood interactions. The first ones are called stylized models³, and they formalize the human and flood subsystems processes using a single differential equation. The second type of models are called comprehensive system-of-systems models, and they represent the subsystems by individual models that are based on well-established methodologies from the relevant disciplines.

3 Stylized models are also referred to as “system dynamic models” (Konar et al., 2019) and “conceptual models” (Troy et al., 2015b).

Stylized models such as those developed by Ciullo et al. (2017), Di Baldassarre et al. (2013, 2015) and Viglione et al. (2014) conceptualize the dynamics of settled floodplains as a complex human-flood system. These models are easy to use and flexible. But, as also pointed out by the above authors, the main drawback of the stylized models is that they neglect the heterogeneity that exists within the human subsystem. In addition, their conceptualization is based only on societal memory or experience of prior flood events as a link between the human and flood subsystems. The model conceptualization does not incorporate the institutions that shape the behaviour of humans in their interactions with their environment and flood.

The only stylized human-flood model that considered institutions is the study by Yu et al. (2017). Yu et al. studied human-flood interactions in polders of coastal Bangladesh by including institutions for collective actions. But, they also used stylized models that do not consider heterogeneity, and focused only on informal institutions for collective actions.

Conversely, studies such as (Dawson et al., 2011; Dubbelboer et al., 2017; Erdlenbruch and Bonté, 2018; Haer et al., 2016; Tonn and Guikema, 2017) developed system-oriented models that conceptualize and model human-flood interactions using agent-based models (ABMs) considering decision makings of heterogeneous actors. However, one of the main gaps in these studies is that they either consider flood as an exogenous element (for example, an agent’s flood experience is set initially and stays the same throughout the simulations) or simplify flood models. Another gap is that they do not methodically analyse institutions to study drivers of flood risk. Instead, they use simplified set of behavioural rules.

Votsis (2017) utilized a cellular automaton model to study the relationship between urbanization trends and FRM strategies. The study shows the effects of bottom-up, flood risk information-based housing market responses and top-down floodplain development restriction scenarios on urbanization. However, the study does not show if the flood extent and depth changes with the development pattern. It also focuses only on the exposure component of the flood risk.

In general, there are important initiatives to model human-flood interactions using a systems perspective. However, these efforts are fragmented and do not address either heterogeneity of actors or all components of the flood risk (i.e., hazard, vulnerability and exposure) in their modelling exercise. Besides, a systems approach which explicitly takes into account institutions as factors that shape the flood hazard and community’s exposure and vulnerability to flooding has not yet been sufficiently addressed in the literature. Developing a framework that integrates the human and flood subsystems and supports modelling decision makings of multiple stakeholders in FRM has also been a major challenge (O’Connell and O’Donnell, 2014).

1.5 RESEARCH AIM AND QUESTIONS

The aim of this dissertation is to develop a modelling framework and a methodology to build holistic human-flood interaction models that provide new insights into FRM policy analysis and decision-making. In this context, “holistic” refers to capturing both the human (i.e., communities’ vulnerability and exposure including the drivers) and the physical (i.e., flood hazard) components in a coupled model using knowledge from the respective disciplines. However, it should be noted that models are abstractions of reality and could not represent all aspects of each subsystem.

To realize the aim, we formulate the following research questions:

1. Which elements should be included to conceptualize the human-flood interaction?
2. How can we couple models that explicitly represent the human and the flood subsystems and the interactions between them?
3. How can coupled human-flood system models that incorporate institutions such as risk drivers advance FRM?

1.6 RESEARCH APPROACH

Based on the research gaps identified in modelling and studying human-flood interactions, in this dissertation, we investigate the merits of the CAS perspective and the integrated modelling approach to build holistic models for FRM. We first develop a modelling framework to decompose the elements that make up human-flood systems. The framework defines the coupled system as a CAS and conceptualizes the drivers of flood hazard, vulnerability and exposure as factors that shape the complex interactions between and within the component subsystems.

In the methodology that accompanies the framework, the human subsystem is modelled using the agent-based modellin...