Interdisciplinary research in hazards and disaster risk

Abstract

This Progress Report reviews the geographical literature concerning environmental hazards and risk focussing particularly on areas that require and enhance interdisciplinary working between human and physical geographers. Although there are still substantial gaps between disciplinary siloes, there is a growing recognition that critical interdisciplinary work is vital. Key areas include early warning, urban planning, hazard and risk mapping, scientific advisory processes, risk communication and institutional geographies. We review some of this work, examine emerging theory and consider the opportunities for greater knowledge exchange between disciplines using critical physical geography and cognate approaches.

I Introduction

Disaster risk crosses many kinds of borders – spatial, temporal and conceptual – at a range of scales from international and nation-state to household and even individual (Ansell, Boin and Keller 2010; Mukherji 2019; Olsson 2015; Sapat 2019). It is complex and strongly geographically specific. It also requires holistic and collaborative work from across a range of disciplines – and particularly across the human–physical geography interface.

Physical and human geographers approach disaster studies very differently, and sometimes in opposing ways – with the physical sciences typically taking a realist approach and the social sciences a constructivist approach. There are benefits to the disaster research community, pragmatically, in describing disasters as socio-physical: the argument that disasters are socially constructed remains central, but the inclusion of the physical, material elements enables clearer explanation beyond human geography and disaster sociology and hence collaboration with the physical sciences. For example, the physicality and temporality of the hazard and its (semi-) autonomy from human ‘causation’ work against a pure social construction argument and yet the hazard interacts with the social through its own multiple constructions – as hazard, as event, as shapeshifter, as it is interpreted, manoeuvred and responded to (Cloke, Dickinson and Tupper 2017; Coates 2022). Hazards reconfigure the social and physical strata in complex ways that change through time. Thus, while the geomorphology of a place changes with a landslide, so does the social, more-than-human lifeworld, as people and animals and plants renegotiate, relearn and relocate around the newly configured land (Clark 2011; Donovan 2021a; Yusoff 2013). Human and physical geographers may use different spatial heuristics – for example, while hydrologists may approach an area in terms of catchments and volcanologists think about the volcanic edifice, human geographers more commonly examine administrative units or urban areas. It can be challenging to integrate these diverse approaches and their underlying epistemologies in hazard or risk assessment and risk reduction activities.

The debates around the Anthropocene have brought many of these issues to light. The recognition that the geology of the Anthropocene is inherently political (Castree 2014a, 2014b) raises a number of important questions for hazards geography: it further opens up the realm of hazards to social study (Bobbette and Donovan 2019; Clark 2011), not only through science studies of hazard and risk assessment processes (Demeritt et al. 2013; Morss et al. 2005) and scientific institutions (Donovan and Oppenheimer 2015; Ogra et al. 2021) but also in the everyday and embodied experiences of hazardous environments (Coates 2022; Donovan 2021a; Gordillo 2021). The development of agent-based modelling and machine-learning techniques has also, conversely, led to a greater drive within the physical sciences to model social processes (Costa, Haukaas and Chang 2021; Ghaffarian et al. 2021; Wang and Zhang 2019). At the same time, there has been significant growth in interdisciplinary research projects seeking to enhance collaboration between social and physical scientists, recognising that the complexities of disaster risk require biophysical, technological and sociological assessments (Filippi et al. 2023; Hicks et al. 2014; Kendra and Nigg 2014).

The broad concept of the ‘geographical imagination’ has been very widely used in geographical literature, with a range of different but connected meanings (Daniels 2011; Gieseking 2016; Gregory 1995, 1994; Harvey 2006). It is a liminal idea, in that it holds together, as Daniels (2011) states, ‘the factual and the fictional, the subjective and the objective, the real and representational’ (p182). As such it also can encompass the uncertainties that emerge from pluralities of ontological and epistemological positions and provide a helpful way of thinking between human and physical geographies (Bergmann and Lally 2021). We review here several different themes that are emerging between the dominant domains of ‘hazard’ and ‘vulnerability’ research in disaster studies, and that to varying degrees involve collaborations between human and physical geographers – often along with other disciplines and with practitioners and communities. We suggest that taking a ‘geographical imaginations’ approach to disaster studies involves attention not only to hazard and vulnerability but also to issues of environmental justice, the imaginaries and values of people-in-places, the role, remit and imaginaries of science in societies, and the historical and cultural contexts in which disasters may be engendered – in and between sites. We are not ‘adding terms’ to the traditional risk equation; we are pointing to areas that are emerging as topics in the field of disaster studies and that overlap with both vulnerability and hazard (McGowran and Donovan 2021; Pelling et al. 2022). This includes recognition and respect of different epistemological approaches, holding them in tension. We propose that critical, interdisciplinary approaches require careful formulation that enables the creation of discursive epistemic spaces, in which explicit conversations about the nature of knowledge production in different disciplinary contexts may take place. This requires finding shared values between human and physical geographers and embracing a flat epistemological approach that accepts plural forms of knowledge-making.

II Socio-physical Disasters (Are Not Natural!)

This section reviews some emerging approaches to risk in hazards geography and disaster studies. It recognises that much of the theory, including nascent ideas from assemblage theory, Foucauldian readings of disaster and systemic risk, are dominantly based on the work of Western (European and North American) philosophers (including Deleuze, Foucault, Derrida and others). However, as Gaillard has argued, these writers seek to identify latent power dynamics, which includes those that lead to the dominance of Western thought and thereby allow its deconstruction (Gaillard 2021). We seek to show the emergence, out of conventional disaster studies, of critical disaster studies – which focus on power dynamics, geographical inequalities of resource, knowledge and wealth, and on the diverse constellations of institutions, methods, ideologies, uncertainties, epistemologies and individuals that emerge in any disaster context. We suggest that these emergent approaches forward a critical human and physical geography of disaster studies that can engage with issues of social and environmental justice, scale and diverse worldviews, ensuring that the uptake of technological and scientific developments to reduce risk is effective and appropriate in situ. We then think through some of the practical implications of this for hazards and disaster geography.

Disaster risk has been conceptualised in plural ways, from a technical–rational, ‘objectivist’ perspective that essentially reduces risk to the probability of harm (Woo 1999) to a fully socially constructed approach that suggests that disasters would exist regardless of hazardous processes (Oliver-Smith 1999; Tierney 2019). The latter approach typically focusses on social vulnerability, but not all work focussing on vulnerability marginalises the role of hazards – hazardous environmental processes, with or without human amplification or exacerbation, interact with and are part of social worlds in complex ways that are often highly contingent on development status, of which vulnerability is a complex function (Bankoff, Frerks and Hilhorst 2013).

During the twentieth century and building on historical accounts, geographers transformed the global approach to disaster risk through their detailed work on vulnerability (Adger 2006; Cardona 2004; Hewitt 1983; Pelling and Uitto 2001; Wisner and Luce 1993; Wisner et al. 2004). However, some have also critiqued the concept (Bankoff 2001; Marino and Faas 2020): labelling a community as ‘vulnerable’ can act as a means of ‘othering’, for example, and a push towards ‘resilience’ can be used by neoliberal governments as a tool of governmentality (Grove 2013, 2014; Reid, 2012). This aligns with recent calls for a locally led approach to research practice in disaster studies that not only is critical of the Western origins of much of disaster studies to date but also privileges the knowledges and experiences of researchers in context (Gaillard 2021). The notions and epistemes of disaster studies itself arise from the economic advantages of the West and must be critiqued as such: they can perpetuate the inequalities and injustices that they originally sought to expose (Gaillard 2019).

Risk is thus a complex and multiple entity. Although the United Nations office for Disaster Risk Reduction (UNDRR) emphasises the importance of moving towards a model of ‘systemic risk’, it also calls for the use of diverse knowledges. One of the challenges here is that the concept of ‘systemic risk’ emerges from Western economics – a very particular and quantified way of understanding an environment that risks oversimplifying social and cultural actants – particularly those that are not materialised (UNDRR, 2022a). Although systemic risk helpfully emphasises the interconnected nature of risk and the need for large-scale system change to combat threats like climate change, it also has a range of meanings in different contexts (e.g., Aven and Renn 2020; UNDP, 2021). In the 2021 United Nations Development Programme report on the social construction of systemic risk, for example, the authors argue that viewing systemic risk as socially constructed enables a focus on risk drivers rather than on hazard triggers – something that disaster risk literature would generally agree with. The issue, though, is the underlying assumption that if risk drivers were better understood, governments would then identify and help the most vulnerable: there are normative implications of this paradigm that are undertheorised and underdiscussed. Although systemic risk and other conventional approaches to disaster risk reduction have enabled significant advancements in understanding both the physical processes of environmental hazards and the dynamics of social vulnerability, including the key role of poverty and inequality in driving risk, there remain important questions about how these knowledges can be engaged effectively to reduce risk (Gaillard and Mercer 2013). Poverty reduction, for example, requires political and therefore ideological will. Ultimately, values underlie many of the factors involved in reducing risk, and values themselves are underlain by collectively held beliefs and stories which need to be understood for any scientific approaches to be used effectively.

Reducing social vulnerability and enhancing both social and environmental justice go together (Cutter 2012; Ryder 2017; Shrestha et al. 2019; Sotolongo, Kuhl and Baker 2021): they involve challenging the political and development decisions that lead to inequalities of risk (Arifeen and Eriksen 2020; Fraser et al. 2020). The complex drivers of risk are rooted and perpetuated through poor development practices, through exposure to hazard events and ongoing poor risk governance. These issues are both multiscalar and multidisciplinary, and they vary in space and time. In recent years, there has been a growth in multidisciplinary and interdisciplinary work around urban planning to understand the longer term processes, including scientific understandings of hazards in particular places, and how those feed into the creation (or not) of disaster risk (Borie et al. 2019a; Galasso et al. 2021). Understanding environmental justice requires analysing the values, ideologies and consequent imaginaries of groups in particularly places, making power dynamics and the complex, historically contingent layers of risk visible. Ultimately it involves combining our (incomplete, dynamic) knowledge of hazardous processes, social structures, political systems, community dynamics and the power distributions inherent within them with knowledge about value systems and imagined futures.

The vulnerability paradigm emphasises the importance of empowering the disenfranchised and points to the need for people-centred research (Birkmann, Sorg and Welle 2017; Wisner and Luce 1993; Wisner et al. 2004). Foucauldian approaches suggest that we should pay attention both to the loci of power (and disempowerment) and to the dynamics of power and knowledge (Fraser 2017; Grove 2014, 2013). Assemblage approaches incorporate principles of flexibility and hybridity – paying attention both to the material world and to expressive elements (including how tools and ideas are represented through artefacts such as maps and models) (Angell 2014; Donovan 2016; Grove and Pugh 2015; McGowran and Donovan 2021). Knowledges are multiple and epistemologies must be flexible, hybridised and flat, not privileging one approach to knowledge-making over another (e.g., physical geography over human geography). Similarly, political geology and some of the related literature in geographical theory point to the importance of paying attention to histories and trajectories of power-in-places – including the complex dynamics of socio-physical hazards and geopower (Bobbette and Donovan 2019; Clark 2014; Donovan 2016). Physical geographers and scientists wrestle with problems in the expression and calculation of uncertainties (Beven 2016; Morss et al. 2005), with the development of models to represent complex systems (Gerstenberger et al. 2020; Mudashiru et al. 2021; Thompson and Warmink 2016) and with the complex uncertainties and practical challenges of monitoring and mapping, for example (Ghaffarian, Kerle and Filatova 2018; Khan, Gupta and Gupta 2020; McGuire, Kilburn and Murray 2022; Zhong et al. 2020). A truly interdisciplinary approach requires holding these approaches in tension with each other, in spite of epistemological differences. It is very difficult to do this without open acknowledgement of these epistemological barriers: our tacit assumptions about what makes good knowledge have to be voiced. In the next section, we outline some areas where this is particularly critical – where hazard and vulnerability intersect.

III An Interdisciplinary Critical Hazards Geography

Geography provides many tools for the reduction of disaster risk that may be employed by governments. However, even within scientific institutions and among scientists, for example, there can be diverse approaches to scientific research (Brandmayr 2021; Graham et al. 2022; Yamada 2020). This can include different priorities, theories, methodologies, preferences and goals – again emergent from personal experiences combined with historical events, ideologies, educational opportunities or pathways and other factors. Scientists may hold different epistemic values (Douglas 2009), such as preferences for modelling over observation or complex models over simple models. Geographically, there are differences in training, funding and availability of courses for different subjects. There are also differences in the historical role that science, scientific institutions and scientific knowledges have in societies and governments; for example, science may be associated with a political elite as in Argentina (Donovan 2020), or scientists may be very highly trusted as in Montserrat (Haynes, Barclay and Pidgeon 2007a). Citizens in different countries test the claims of scientists and other experts in different ways (Jasanoff 2005).

The prevalence and significance of technology in the management, mitigation and understanding of natural hazards and disaster risk has increased greatly over the last century – yet there has been relatively little work in disaster studies on the relationships between societies, governments and technology in the conception and management of risk (Shaw 2020; Shi et al. 2020). Considerations of power inherent in the discussion of vulnerability and environmental justice are, however, also relevant to the scientific landscape – and this is what ‘critical’ means in interdisciplinary critical hazards geography: awareness of power dynamics within and beyond the interdisciplinary science itself. This section initially outlines some of the relationships between science, power and justice, drawing together vulnerability and hazard science as it does so. It then discusses the role of democracy and institutions within this, before focussing on the role of objects such as models and early warning systems. Finally, it discusses human and physical histories of disaster. Each of these topics sheds some light on the intersection between hazard and vulnerability in disaster studies, and the need for critical collaborations between human and physical geography.

3.1 Justice in Environmental Science

Environmental justice provides a means of thinking through vulnerability that removes the potential for labelling the subaltern as vulnerable in a way that might actually disempower (Bankoff 2001; Sotolongo, Kuhl and Baker 2021). It also incorporates some of the deeply intrenched injustices that result from the hegemony that Western capitalist and neoliberal thought have had on how researchers from the global North interact with those in the global South (Gaillard 2019; Noxolo 2017). These deep historical injustices regarding the balance of resources and the values of governments, institutions (including universities) and the private sector continue to perpetuate cascading challenges for the developing world in facing disaster risk: it is a truism that if profit is the most important thing, then the choices that people in power make will be different from those that they would make if they valued equality more highly, for example. These differences can be made visible through studies of the diverse knowledges and values between groups (e.g., Balay-As, Marlowe and Gaillard 2018).

Critical human geography of disasters must carefully navigate this complex landscape to avoid exacerbating existing inequalities of power and influence. Such ethical considerations also apply to the ways physical geography is used in context. Research ethics has typically been viewed as the domain of the social and health sciences, but increasingly is recognised as relevant to physical geography too: physical geographers often carry out hazard assessments or monitoring among marginalised and disempowered populations and need to understand the complex institutional and cultural dynamics so that they can approach communities appropriately. There is also potential for hazard assessments undertaken in research projects to undermine the local agencies if they are not done collaboratively (Alexander 2014): the pressure of research impact metrics can sometimes lead to scientists publicising their research without thinking about the implications for the local institutions charged with managing risk, for example. Scientific research can also be extractive, requiring logistical support on the ground from local people but not recognising that contribution in publications or other outputs. Furthermore, the research process itself can reinforce or unbalance power dynamics between and within communities. Finally, ethics is of critical importance post-disaster, when external researchers can undermine local efforts by using up limited resources, for example (Gaillard and Gomez 2015; Gaillard and Peek 2019). Alongside environmental justice, then, goes environmental science justice: the reflexive consideration of justice and power dynamics within environmental (social and physical) science and also in its interactions.

3.2 Scientific Institutions and the Democratisation of Scientific Knowledge: The Politics of Expertise

Institutional contexts and cultures are also important in knowledge production and thus in the reduction of risk. The role and responsibilities of scientific institutions in disaster contexts are highly variable between nations and sometimes also within them (Brown 2009; Budimir et al. 2020; Donovan 2020; Ogra et al. 2021). Similarly, institutional geographies are highly variable. In some countries, for example, geological surveys are responsible for providing warnings for geohazards such as earthquakes, volcanic eruptions and landslides; in others, there are separate institutions for different hazards. Meteorological offices also often take responsibility for certain hazards, sometimes (e.g., in Iceland) including the geohazards. Numerous authors have called for transboundary cooperation in the spatial domain (Birkmann et al. 2021; Boin and Lodge 2016; Donovan and Oppenheimer 2019; Timmerman 2020). The dynamics of such cooperation are not straightforward. In a large area affected by a single hazard event, very diverse impacts may be visible, implying very diverse responses for example (Lavigne et al. 2011; Telford, Cosgrave and Houghton 2006). Here, the geography of science and its institutions is important: understanding each other across boundaries can be challenging (Donovan and Oppenheimer 2019) and typically requires work before a hazard event to build relationships. Scientific models, methods and resources can vary significantly between countries.

This complexity is compounded by the diversity in institutional structures and responsibilities. In some countries, such as the UK, institutions are typically responsible for issuing warnings, but political decision-makers would be involved in any evacuation calls, for example. However, in other contexts, the issuance of a warning can be tied into particular evacuation decisions or other civil defence measures – effectively meaning that the production of an alert is a policy decision made by scientists; and this is something many scientists are not comfortable with (Donovan 2019; Papale 2017). Indeed, there have been substantial debates in volcanology, for example, about the role of scientists in crises (Bretton, Gottsmann and Christie 2018; Giordano et al. 2016; Papale 2017). Advisors hold an unambiguously social and political role because knowledge travels and people question it: critical geographers and sociologists of science have demonstrated that a linear model of science in the public sphere does not work in practice (Beck 2011; Donovan and Oppenheimer 2014). Science cannot be separated out neatly into its own ‘objective’ space (Daston 1992; Jasanoff 2007; Porter 1996; Stirling 2008). This can create significant challenges for physical geographers and natural scientists in advisory roles because they feel that they are asked to go beyond the ‘scientific’ in providing advice (e.g., Papale 2017).

In disaster contexts, scientific institutions may, however, have very significant power: the advice that they provide to political actors can have huge impacts on the lives and livelihoods of citizens. Thus, there have been calls for greater engagement with the social contract of such scientific institutions (Pelling et al. 2022). A key area for future work in this field is in the integration of social scientific expertise into these institutions – they are charged with providing warnings and often with outreach around risk, and yet frequently leave this role to physical scientists, rendering the communication of risk top-down rather than based on understanding communities. Some institutions – such as GNS Science in New Zealand – manage this effectively (e.g., Doyle, Paton and Johnston 2015), but they are relatively rare. Others successfully collaborate with academic social scientists. Nevertheless, there is an important role for approaches like critical physical geography in this area (Lane et al. 2011; Lave et al. 2014). Understanding the cultures, assumptions and power of scientists working in nodal institutions (those that have responsibility for warnings) is a key element in understanding the challenges of risk communication – not only in how messages are conceived but also how they are received and trusted (or not) (Critchley 2008; Donovan, Borie and Blackburn 2019; Eiser et al. 2009, 2012; Haynes, Barclay and Pidgeon 2007a; Paton 2008). This is not purely work for human geographers: it is an area where much important work is being done by physical geographers too (Hulme 2008; Lane 2017; Lane et al. 2011).

Although much work has been done in geography and similar disciplines to emphasise the importance of democratising scientific knowledge and how it is used in society (Bijker, Bal and Hendriks 2009; Brown 2009; Fischer 2010; Jasanoff 2005; Rayner 2003), this remains in its infancy in disaster contexts. In contrast, a great deal of work has been done on integrating scientific knowledge with other kinds of knowledge (Bwambale et al. 2022; Cronin et al. 2004; Gaillard and Mercer 2013; Hadlos, Opdyke and Hadigheh 2022; Mercer et al. 2010; von Storch 2009), and the co-construction of knowledge through participatory approaches (e.g., Hendricks et al. 2022; Mercer et al. 2008). Again, there are links to be made here between the top-down and the bottom-up and also in recognising the ways in which scientific institutional remits, power and imaginaries influence how and whether other kinds of knowledge are valued in particular contexts (Šakić Trogrlić et al. 2021). This has to incorporate awareness of the spatialised origins of scientific knowledge itself, and its links to colonialism (Gaillard 2021). Ultimately, human geography can be the glue in understanding the limitations, contexts and environments where physical geography operates, but it needs to be allowed in Ogra et al. (2021). It also needs to engage clearly with physical geography, which can involve shifts in the theoretical language that is used. Approaches like critical physical geography allow those kinds of openings but are rare in practice (Lane et al. 2011; Lave et al. 2014): many of the arguments made by science studies and human geographers of science have been largely ignored and thus, for example, the COVID-19 pandemic very visually reproduced the challenges of past crises, albeit on a larger scale (Donovan 2021b; Pelling et al. 2022). Similarly, though, much human geographical theory fails to engage coherently with the language and challenges faced by physical geography. We return to some of the practical issues here in Section IV.

3.3 Models, Mapping and Early Warning Systems

There is considerable literature on the social life of climate models, and the potential impacts of propagating uncertainty on political decision-making – and indeed the space it opens for climate scepticism (Mahony and Hulme 2012, 2018; Shackley and Wynne 1996; Shackley et al. 1998; Wynne and Shackley 1994). Models can open up possible futures and themselves contribute to scientific imaginaries in a substantial way (Donovan, Oppenheimer and Bravo 2012a). They can also be challenging to represent with full acknowledgement of uncertainties, and indeed some forms of uncertainty are difficult to quantify (Spiegelhalter and Riesch 2011; Stirling 2007). This can create an ethical dilemma: uncertainty should be transparent, but scientists struggle to represent it and often users do not understand it. More interdisciplinary work is needed on this issue and we return to it later in this section.

In hazards geography, hazard mapping is an important tool for nodal agencies and for planners in establishing urban land use limitations. It is also widely used by the insurance sector. Here, again, ethics becomes a possible challenge, because scientific products are used to control populations – usually for their safety, but there is always the potential for misuse. There are also questions around the use of hazard maps in communication. Recent work has shown that people vary significantly in their ability to use hazard maps – or indeed other kinds of map (Haynes, Barclay and Pidgeon 2007b; Thompson, Lindsay and Gaillard 2015). Participatory methods have been used effectively in some communities, and there is also some evidence that cultural factors affect how people read maps (Gaillard and Maceda 2009; Lindell 2020; Thompson et al. 2021). Again, there is scope for further development of these methods, including using technologies (Borie et al. 2019b; Canevari-Luzardo et al. 2017; Taylor et al. 2020).

One of the key developing areas in understandings of environmental risk, often based on a combination of models and monitoring data, concerns the importance of equitable, multi-hazard early warning systems (MHEWS). The Sendai Framework for Disaster Risk Reduction emphasises the significance of scientific knowledge (broadly conceived in disciplinary terms but still viewed through Western categories) in the management of disaster risk (UNISDR, 2015). However, the UNDRR at its 2022 Global Platform noted that the global distribution of MHEWS is extremely uneven (UNDRR, 2022b), with least-developed economies and small islands particularly underrepresented. There are socio-economic inequalities involved: hazard modelling and transmission of messages are technologically and scientifically challenging, particularly not only for low probability hazards but also for meteorological hazards, where computing power is important too (UNDRR, 2022b).

There is a rapidly growing body of literature on EWS and MHEWS, much of it technical and focussed on single hazards (e.g., Alfieri et al. 2013; Cloke and Pappenberger 2009; Pappenberger et al. 2008). Where multiple hazards are included, they are typically multiple single hazards (Aguirre-Ayerbe et al. 2020; Khan, Gupta and Gupta 2020) rather than interacting hazards (Tilloy et al. 2019). A great deal of work has been done on community-based EWS (CBEWS) for single hazards (Fakhruddin, Kawasaki and Babel 2015; Gautam and Phaiju 2013; Macherera and Chimbari 2016) – though the meaning of ‘community-based’ varies from community operated (e.g Abon et al. 2012) to ‘in a community’ but techno-scientific (e.g., Thapa and Adhikari 2019). There are also some concerns about the potential for such systems to enable governments to shirk their responsibility to citizens (Gladfelter 2018): while the existence of warnings is positive, putting the onus on citizens can also undermine the social contracts of governments. On the other hand, they can provide a level of granularity that is not possible for larger scale systems and they can have a role in empowering communities: place-specific work is needed to understand these dynamics more. For example, Budimir et al. (2020) analysed the effectiveness of flood early warning systems in Nepal. They assessed the relationships between the interpretations of models that were produced by technical actors, and the ways in which locals were able – or not – to interact with the resulting warnings. Although scientific modellers had one particular imaginary of a flood – devised through a model – the community was not able to act on the basis of that alone. They needed practical instructions that matched their own embodied experiences of the floods in situ.

Uncertainty is a key issue for both human and physical geographies of hazard – in different but complementary ways. Physical geography and the wider environmental sciences typically refer to aleatory (or stochastic) uncertainty – the uncertainty inherent in the natural system and not reducible by knowing more – and epistemic uncertainty, which results from a lack of knowledge (of things that could be known) (Beven 2016; Beven et al. 2016; Donovan 2019). Although aleatory uncertainty can be represented statistically, epistemic uncertainty can be more challenging to represent – and can sometimes even increase as knowledge becomes available (because new ‘unknowns’ can become apparent). Much research has been done on ways to communicate uncertainty more readily (e.g., Brown 2010; Bruen et al. 2010; Friedman, Dunwoody and Rogers 2012; Harris 2015; Joslyn and Savelli 2010; Morss et al. 2005; Shackley and Wynne 1996), and many of these projects involve interdisciplinary collaborations. Human geographical analyses of the science–policy processes further reveal social forms of uncertainty that may or may not be readily quantifiable. In expert judgement exercises, for example, experts may make an estimate of their own uncertainty (Bamber and Aspinall 2013; Cooke 2008; Tadini et al. 2022) – but there are also more indeterminate sources of social anxiety within a decision-making group, between scientists and officials and other groups too – requiring a more substantial framework for understanding uncertainty in the context of ‘wicked problems’ (Donovan 2019; Stirling 2007; Sword-Daniels et al. 2016; Wynne 1992). Additionally, the use of belief-based probabilistic methods raises challenges for some scientists, who regard such methods as unscientific due to the heavy reliance on subjective judgements (Castanos and Lomnitz 2002; Donovan, Oppenheimer and Bravo 2012b; O’Hagan 2008).

The interdisciplinary nature of forecasts, EWS, CBEWS and MHEWS is plural: work is needed to understand the nature and limitations of the scientific and technological process itself, including the use of models, the framing of messaging and the social context of institutions in which science is done (Budimir et al. 2022, 2020; Demeritt et al. 2013; Morss et al. 2016, 2005; Neußner 2021; Potter et al. 2014). It is also needed to understand the communities where the warnings are to be used – recognising that such communities are themselves often hybrid, consisting of diverse groups with different needs. Other work has examined issues around standardisation of alert level systems at different scales, some arguing for and others against (Fearnley and Beaven 2018; Fearnley et al. 2012; Neußner 2021). There is a key role here for geographers of scientific knowledge in understanding the diversity of needs and how societies interact with these emerging technologies. Furthermore, collaboration is needed between countries with high computing power and those that lack it, particularly to run complex models. Ultimately, the development of EWS that are multi-hazard must also be appropriate to the communities that they serve – which, for example, have diverse historical experiences of warnings and diverse institutional histories.

3.4 The Importance of the Past

Geological studies of environmental hazards are frequently based on detailed, meticulous stratigraphic work, looking at the deposits of past eruptions, landslides and flood events, for example. Similarly, human geographical and social scientific studies of disasters increasingly draw on historical information, such as through the forensic investigation of disasters (FORIN) (Oliver-Smith et al. 2016) and through the exploration of memories, colonial histories and institutional histories (e.g., de Guttry and Ratter 2022; García-Hernández, Ruiz-Fernández and González-Díaz 2019; Moulton and Machado 2019). The recent geological strata are themselves associated with human historical processes, violence and learning – sometimes made visible in archaeological studies (e.g., Plunket and Uruñuela 2006), and sometimes through analysis of oral histories or literature (e.g., McKinnon 2019). In some cases, too, the interplay between human and physical geological histories may be highly complex, as in the case of repeated flooding (e.g., Di Baldassarre et al. 2013).

Stratigraphy is a historical science rather than an experimental one (Frodeman, 1995). Historical sciences are faced with the fundamental challenge that history is received in a fragmentary manner: not everything survives. This problem is faced across historical studies in any disciplinary context, as is the related challenge of identifying and making inferences from dependencies between entities and events (Cleland 2002; Currie 2017; Frodeman 2014). Historical sciences are also of central importance within the modelling sciences because they provide inputs for models. They provide baselines for the observational and experimental sciences, feeding into expectations of the kinds of data and event that might be expected. Yet they are also different in nature from these sciences in certain respects – particularly the lack of experimentation and overt hypothesis testing. The nature of reasoning is that it emphasises interpretation, evidence and place: while the search for ‘golden spikes’ looks for global markers, most stratigraphic studies are small-scale and place-specific. There are a great many ways in which such studies are more similar to the social sciences and humanities than the modelling and experimental sciences (Frodeman 2014, 1995). Frodeman (1995), for example, highlights the significance of narratives in geological reasoning. Narratives are a key means of making sense of the past – both the human/more-than-human past and the geological past. They also, perhaps, provide a means to enhance interdisciplinary working through the exchange of stories and questions about the production of knowledge.

This section has highlighted some of the areas that fall between hazard and vulnerability – where there is overlap, interchange and a real need for a critical interdisciplinary approach. The success of such approaches, however, requires the acknowledgement and open discussion of the different epistemological positions of different disciplines and sub-disciplines. Geography itself is an inter-discipline: it incorporates a plurality of epistemologies and methodologies and holds them together. The challenge is getting them to speak to each other.

IV Discursive Epistemic Spaces

This report has already referenced the significance of disciplinary boundaries in how risk is understood – in the simplest formulation, between the objectivist and subjectivist schools – and the importance of involving multiple disciplines across humanities, social sciences, health sciences and physical sciences in the reduction of disaster risk. Interdisciplinary thinking starts to dissolve these boundaries, but it encounters obstacles in so-doing – particularly not only diverse epistemological positions but also the structures of the academy (which includes a great many discipline-specific idiosyncrasies in publishing, career progression and recruitment for example). These issues can underlie and indeed undermine the approaches taken by researchers in interdisciplinary projects but are often tacit until a conflict emerges (Sapat 2021). The significance of interactional expertise (Collins and Evans 2007) in interdisciplinary working has perhaps not been adequately recognised, though it is emerging in recent work (Donovan, Oppenheimer and Bravo 2012c; Gilligan 2021): having at least some researchers who are able to cross disciplinary boundaries can be critically important to avoid misunderstandings. However, interdisciplinary working in disaster studies is not sufficient, as the literature reviewed above demonstrates: disaster research should involve multiple epistemic communities and cultures, including local stakeholders and communities (Gaillard and Mercer 2013; Jasanoff 2003; Thompson et al. 2017).

Such transdisciplinary work is a particular challenge, not least because the goals of the researchers may differ fundamentally from those of the communities in terms of outcomes, and the complex range of ontological and epistemological positions that may be represented (Gaillard 2021; Sherman-Morris, Houston and Subedi 2021). Language may also be a significant barrier – the most ‘basic’ terms of disaster theory may be problematic (‘vulnerability’, ‘hazard’, ‘resilience’, etc can mean different things to different groups – if they mean anything) (Borie et al. 2019a).

Evidence suggests that these issues need to be made explicit in order to enhance interdisciplinary working (Filippi et al. 2023; Gilligan 2021; Sapat 2021). This is also beyond geography: disaster studies incorporates sociologists, anthropologists, earth and environmental scientists, engineers and medical scientists, among others. We argue that in the creation of discursive spaces to make explicit the differences between the norms for the production of robust knowledge, it is also important to hold shared values – particularly humility (in respecting that other disciplines hold their epistemic positions for good reasons; Jasanoff 2007); a shared goal (usually reducing risk, in this context); and shared and agreed on hybrid outputs. Reserving judgement and holding diverse epistemological positions in tension with one another collectively enables the production of interdisciplinary science. A geographical imagination can enable this process practically because it suggests deliberately paying attention to issues of institutional dynamics, power dynamics between/within disciplines, the imaginaries and values of communities in their historical context, scientific cultures and the issues of environmental justice that emerge in particular places. These are all issues that are marginalised by approaches that remain with ‘hazard’ or ‘vulnerability’. An additional problem that is rife is what has been described as ‘disciplinary extractivism’ (Fox et al. 2023): the use of approaches, methods and concepts from other disciplines without appropriate understanding due to a lack of training or collaboration within that discipline, and without respect for its epistemologies and practitioners. Strength of critical physical geography is that it is collaborative: it does not involve physical geographers doing human geography without being trained, or vice versa; it involves coherent and collaborative exchange through which both parties learn and create new knowledge together (Lane et al. 2011; Lave et al. 2014).

V Geographical Imaginations and Flat Epistemologies

The sections above have discussed some of the areas in which disaster research has been advancing as an interdisciplinary field. Although the traditional domains of ‘hazard science’ and ‘vulnerability’ remain important, there is a strong emphasis too on integration and plurality of ontological and epistemological approaches. This plurality enables the production of hybrid knowledges that can hold together (even if in tension) not only multiple academic disciplines but also diverse community knowledges and ontologies. Ultimately, alongside plurality of knowledges, effective interdisciplinary working in the hazard sciences also requires flat epistemologies: approaches that respect the diversity between disciplines in how robust knowledge is made. We suggest that this can be achieved through the creation of discursive spaces to make explicit those differences, and differences in areas such as terminology, outputs and expectations, early in a project. We further argue that much key interdisciplinary work occurs at the margins of ‘hazard’ and ‘vulnerability’, and intersects both of them.

Critical interdisciplinary hazards geography involves the production of scientific and social knowledges with reflexive awareness of their power dynamics – within and beyond the science. The diversity in the characterisation of what actually constitutes disaster risk also reflects disciplinary disparities in power and influence. This is part of the challenge for interdisciplinary hazards geography: it is often the physical sciences that are best placed to advise governments on disaster risk, and this reinforces the technical–rational imaginaries that pre-exist in many government approaches – a key concern of disaster studies for over 40 years (Burton, Kates and White 1968; Hewitt 1983; Wisner, Gaillard and Kelman 2012; Wisner et al. 2004) in emphasising that vulnerability is the major driver of disaster risk. Although the UNDRR has championed a broader approach, at national level there is still a great deal of work to be done: it is politically much more challenging to reduce vulnerability than it is to fund physical monitoring for example. Furthermore, the role of scientific institutions in decision-making around environmental hazards can afford huge power in a crisis (Donovan, Bravo and Oppenheimer 2012), and that power is open to abuse by governments seeking to ‘blame’ science for their decisions for example. Ulrich Beck wrote that ‘uncertainty lends power to perception’ (2006: 42): human interpretations fill the ‘gap’ that is created by the uncertainty on scientific models, maps, measurements and indeed decisions. It is here that the nexus of the intertwined human and physical geographies of disaster risk becomes clear: these dynamics are often spatially and temporally specific, and highly complex. Geography – an interdiscipline when at its best – is well placed to explore such issues in depth and with attention to power dynamics, cultural issues and knowledge politics.

Another side of this concerns the political pressure that scientists may face when asked for advice in a crisis: this can be substantial and can make scientists highly vulnerable, including to political misuse (e.g., politicians blaming scientific advice when things go wrong, as occurred at L’Aquila; De Marchi, 2014; Scolobig et al. 2014). Again, it is evident from recent cases that advisory science should always be interdisciplinary across the human–physical interface: physical scientists need social scientists to help them understand their audiences, the political and economic context and issues of framing and trust, for example. The social sciences also have to respect the epistemic judgements and products of the natural sciences – and willingly engage with them in clear, unambiguous and explicitly agreed language.

Interdisciplinary work is fundamental in moving disaster studies forwards: we need plural knowledges to combat the disasters of the Anthropocene. This requires reflexivity from researchers across all disciplines, along with some researchers with ‘interactional expertise’ (the ability to interact meaningfully with multiple disciplines; Collins and Evans 2007). Reflexivity refers to an awareness of positionality and the limitations afforded by background. It also involves openness to other ways of looking at the world, and an engagement with the historically and spatially constructed imaginaries of diverse groups – underlain by value systems and power dynamics. Interdisciplinarity involves critical approaches to both human and physical geography (Lave et al. 2014): scientific work incorporates, negotiates and influences power dynamics as it encounters communities. It also encounters other ways of knowing and other value systems. Effective interdisciplinary projects are those that engage in dialogue with those systems and learn from them. A geographical imagination is key here, in understanding the place-specific and multiscalar circulation of knowledges and technologies that underlie the production or reduction of disaster risk.