Regulator Theory, Natural Hazards, and Climate Change

Abstract

Climate change is increasing variability in environmental conditions and the frequency and severity of natural hazards such as hurricanes, floods, and wildfires. In this paper, we use general systems theory to describe how disaster management systems are composed of four types of system regulators (aggregation, passive, error control, and anticipation) that are deployed to provide protection from natural hazards. We argue that climate change, by changing causal relationships in the environment and thereby reducing the predictability of related hazards and altering exposure to them, is likely to require that disaster management systems be restructured by changing the combinations of system regulators that are employed to prevent or mitigate disasters. This leads to the conclusion that one of the keys to developing effective policies to support adaptation to climate change and to promote sustainability hinges on understanding how disaster management systems can be interpreted as mechanisms for regulating exposure and vulnerability to minimise the threats from natural hazards. Consequently, developing methods for interpreting and modelling system regulators in disaster management systems is an important next step.

1. Introduction

There is a consensus that, principally because of climate change, the frequency, severity, and complexity of natural hazards, such as floods, wildfires (also called ‘bushfires’ in Australia), and droughts, are increasing, while tropical cyclones (variously called ‘cyclones’, ‘hurricanes’ or ‘typhoons’, depending on location) are increasing in severity and reducing in frequency [1,2,3]. Since hazards are an emergent property [2,4,5,6], these changes in natural hazards mean that the natural disasters they create through their interaction with interrelated environments (bio-physical, social, economic, etc.) are changing. Consequently, the adequacy of disaster management systems is likely also changing.

Novelty in the hazard-generating behaviour of natural environments (which are systems in themselves) implies a change in their structure [7], and a change in structure means that those responsible for intervening to regulate the impacts of hazards, the dimensions of disasters, must modify their interventions if they are to control, as much as is practicable, the impacts of changes in the structure of causal environments [8]. In this context, ‘modify’ means changing the combination of interventions (regulators) that constitute the disaster management system, as it is these that are used to influence (regulate) and optimise the relevant behaviour of the various environments that interact with hazards to generate natural disasters.

This suggests that one of the keys to developing disaster management policies that promote sustainability by fostering suitable adaptation to changes in the environments creating natural hazards lies in having a clear understanding of the conditions under which regulators in the disaster management system can operate successfully. We suggest that general systems theory can provide that understanding [7,9,10,11,12,13]. We will use general systems theory and system regulator theory in particular to characterise the interventions that are relied on in disaster management to prevent or mitigate the impact of natural hazards and draw some implications for disaster management policies. To the best of our knowledge, the potential for regulator theory [7] to offer new insights that would assist lawmakers and policymakers in improving sustainability by means of disaster mitigation and preparedness regarding natural hazards has not been explored previously.

2. Management Context

We have written this paper in the spirit of contributing to Parker’s [14] observation that the science of disaster resilience faces considerable challenges, and we hope that this paper contributes to the challenge of improving risk management planning by supporting the identification and selection of suitable, concrete prevention and preparedness measures [15]. We will use the concept of system regulators from general systems theory as a means for framing the mechanisms that are employed to prevent or mitigate natural disasters [11,16,17] and, in so doing, show how the concept of system regulators can be applied to the management of various specific kinds of natural disasters in isolation.

In principle, we see the ideas presented here as contributing to meeting all four priorities of the Sendai Framework for Disaster Risk Reduction 2015–2030 [18] and, thereby, providing a holistic analytical frame capable of generating meaningful insights into the demands on policy if it is to be adequate to the risk management task. The value of such an approach is grounded in the role that systems fundamentally under human control play in defining vulnerability to hazards arising naturally. As the dimensions of natural hazards evolve in the ways noted above, the implications for the dimensions of disasters simply need to be understood as fully as possible if effective responses are to be identified and evaluated.

This is, arguably, particularly the case with respect to disasters that proceed to cascade, intersecting with and making demands upon increasing numbers of our substantive human-designed systems. In decision theory jargon, Nature is ‘upping the ante’ in our ‘game’ with it: the scale of disaster outcomes is increasing in our existing economic, institutional, social, and built environment contexts. Understanding the risks and designing valid responses to them relies on a comprehensive model of the capacities of alternative responses.

The domain and reach of natural disasters are diverse, ranging from the initially small-footprint, bounded consequences of the manifestation of a natural hazard, such as a wildfire ignited by a lightning strike, to the far-reaching, multidimensional impacts of a tropical cyclone passing over densely populated, coastal urban areas. The disaster management demands are also diverse. ‘Resilience’ is difficult to define as a workable construct due to this diversity [19], and the complexity of disasters [20] is likewise challenging to those seeking to intervene to ameliorate disaster impacts.

Our focus is on those aspects of disaster preparedness related to immediate responses to manifest natural hazards. The objective assumed is to optimise the suite of available responses and their deployment upon the actual or expected arrival of a threatening natural phenomenon. ‘Optimisation’, like ‘resilience’, is not a simple criterion: the economic and social consequences of the greater or lesser impedance of the impacts of a hazard are unlikely to be known with confidence, especially where cascading disasters are a threat.

To make alternative responses commensurable, it is necessary to apply an analytical frame that can be applied to a threat from any natural hazard and which identifies the degree of control available in physical terms and, thereby, the residual (physical) threat. Ideally, this should provide a physical representation that can be optimised in economic terms by the contemplation of the relative costs of regulation (that is, physical control) and the residual threat. At the same time, a focus on physical control should provide insight as to the rational timing of control interventions.

In the next section, we present regulator theory [7], describe the distinction between passive and active regulators, and outline the differences among various kinds of passive regulators [21]. We also outline the concept of defence-in-depth [22]. We then classify a range of mitigation and preparedness measures for earthquakes, floods, wildfires, tropical cyclones, and droughts in terms of types of system regulators. These classifications, together with the concept of defence-in-depth, provide the foundation for making inferences about the advantages and disadvantages of investing in various mitigation and preparedness measures. These inferences have implications with respect to disaster management policy and the sustainability of disaster management systems in the context of climate change.

3. Methods: System Regulators

In this section we use general systems theory to describe how system regulators are used to stabilise open systems. We describe three fundamental types of system regulators and describe how they contribute to the design of disaster management systems and to the capacity of those systems to adapt to climate change.

A system is a set of components that are linked together by relationships to form a whole [11,23,24]. The focus in general systems theory is to understand the behaviour of the system rather than understanding the behaviour of its components or relationships. General systems theory is about understanding the behaviour of systems as the outcome of interactions among components and relationships [11,25].

3.1. Open Systems

Open systems interact with their environment [9,11,12,24,25,26]. Consequently, changes in the state of its environment will change the behaviour of an open system [27]. The behaviour of open systems can only be understood given the possible states of the environment with which they interact [25,28].

Disaster management systems are, by design, open systems. They have a structure consisting of a set of components and the relationships among them. For example, in the case of flood management systems, components might include levees, reservoirs, floodways, monitoring devices, and evacuation centres. Practices, protocols, procedures, and policies link components together in relationships. Disaster management systems interact by design with natural, economic, and social environments. Disaster management systems have a purpose [29], which is to protect people and their social, cultural, and environmental assets [2]. Disaster management systems are, then, open systems that are managed for the purpose of protecting society.

Changes in the frequency, severity, and complexity of natural hazards indicate that the possible states of the natural, economic, and social environments within which disaster management systems operate have shifted. Consequently, disaster management systems must be modified structurally if they are to continue to regulate the behaviour of these environments to protect society to acceptable degrees as currently perceived.

3.2. Stability and Adaptation in Open Systems

When the behaviour of any open system lies within acceptable limits with respect to the anticipated range of states for its environment, then the behaviour of the system can be described as stable [24,25,26,30]. The limits to acceptable behaviour are defined by the purpose of the system. When, because the environment enters an unexpected state [12], system behaviour falls outside what is acceptable, the structure of the system must be changed to restore acceptable behaviour. Hence, stability and structure are related, and stability can only be defined with respect to a given system structure [31,32].

For example, the capacity of a disaster management system to absorb the effects of climate change may be defined by the extent to which it continues to offer acceptable protection from disasters without requiring any change to its structure.

When a change in structure is necessary to maintain stability, this is referred to as ‘adaptation’ in general systems theory, so long as the achievement of the system’s core purpose(s) is preserved [24]. For disaster management systems, then, adaptation to climate change means changing the structure of these systems while preserving the achievement of their purpose (that is, they continue to offer protection within acceptable limits). If the environment changes so much that the system cannot be restructured to achieve this, the system must transform into a new system or fail [24].

3.3. System Regulators

System regulators are essential for maintaining the stability of a system by enabling it to absorb changes in its environment [8]. The choice of which regulators to use depends on two key factors: the ability to predict environmental variability (cause, timing, and extent) and the relative cost of regulating [7]. While regulators help the system adapt to changes, they come with costs, such as the need to process environmental information and take regulatory actions. Note that adding regulators to a system may constrain the potential for the system to achieve a specific outcome [25,33] and that introducing new regulators into a system will create switching costs (for example, investing in engineering works to prevent or mitigate flooding may reduce the attractiveness of retreat and relocation [34]).

There are two types of system regulators: unconditional and conditional [7]. ‘Unconditional regulators’ are passive and operate continuously, regardless of necessity. ‘Conditional regulators’ are more dynamic, operating only when needed, based on specific conditions.

3.4. Unconditional System Regulators

The type of regulator used to manage a system depends, to begin with, on whether the cause, timing, and extent of variability in the state of an environmental input into the system can be predicted (see Figure 1). An unconditional regulator must be used if this variability cannot be predicted. In other words, an unconditional regulator must be used when, given the characteristics of appropriate system response(s), there is not sufficient time to find out more about the cause, timing, and extent of variability before action must be taken [7].

Unconditional regulators function most efficiently when the state of an environmental input changes frequently and along a single dimension, such as quantity [7]. When the state of the environmental input only changes intermittently or along two or more dimensions, such as quality and quantity, then a conditional regulator may be more efficient [7].

Aggregates are one basic type of unconditional regulator [7]. Aggregates operate on the principle of redundancy in that the system is insulated from variability in the state of an environment by holding a sufficient stock or store of relevant inputs or satisfactory substitutes (i.e., an aggregate).

Passive regulators are another principal type of unconditional regulator [7] and, for our purposes, can be classified into four broad categories based on their manner of operation. Adapting from Nayak and Sinha [21], these categories are as follows:

• Regulators that do not require external signals about the environment, or external power sources, or any moving mechanical parts. This category of passive regulator relies on a fundamental design principle to function. Conservative design, high quality in construction, and high quality in maintenance are important in preventing the failure of these kinds of regulators. Examples include levees, floodways, and fire breaks.
• Regulators that do not require external signals about the environment or external power sources but have moving mechanical parts. This category of passive regulator also relies on a fundamental design principle to function. Again, conservative design, high quality in construction, and high quality in maintenance are important in preventing the failure of these kinds of regulators. Note that, compared with Category A, the reliance on moving parts creates an additional source of potential failure for this category of the regulator. This may place greater demands on monitoring and maintenance to ensure moving parts function correctly.
• Regulators that do not require external signals about the environment but do rely on an external power source. Note that, compared with Categories A and B, the reliance on external power could place even greater demands on monitoring and maintenance as these tasks must extend to the source of the power and the mechanism for transmitting the power to the regulator. The reliance on an external power source creates new sources of potential failure in that the connection to the power source and the power source itself may fail. This may be offset by building redundancy into the power system; that is, adding an aggregation regulator by, for example, installing connections to alternative power sources.
• Regulators that do require external signals. Note that, compared with Categories A and B, and possibly C, the reliance on an external signal could place even greater demands on monitoring and maintenance as these tasks must extend to each of the sources of the signal, the mechanism for transmitting the signal to the regulator, and the mechanism for translating the signal and actuating the regulator. The reliance on an external signal creates another source of potential failure in that the connection to the signal and the signal itself may fail (or be delayed). Again, this may be offset by adding an aggregation regulator and building redundancy into the signalling system.

3.5. Conditional System Regulators

When enough is known about the cause, timing, and extent of variability in the state of an environment to predict whether regulation is necessary, conditional regulators can be employed [7]. There are two types of conditional system regulators: error control and anticipation. The choice between them depends on several criteria that characterise the conditions that suit each type of conditional regulator and also their relative cost to the system (see Figure 1).

Error control regulators are employed when the cost of unnecessary regulation is lower than the (expected) cost of failing to regulate when necessary [7]. These regulators are triggered by detecting small changes in the environment that may signal larger changes in the future [7,8,36]. Error control is based on two key principles: that small changes often precede larger ones and that small changes can be detected early enough to prevent or mitigate the larger changes. For example, the evacuation of communities based on the proximity of wildfires or tropical cyclones is an example of an error control regulator. In the absence of precise, reliable predictions about the direction in which and speed with which wildfires and tropical cyclones will travel, evacuations must be based on more conservative rules employing less reliable criteria such as threat proximity [37].

The weakness of error control regulators is that they are always triggered by small changes in an environmental input, even though small changes may not always foreshadow large changes [7]. This results in unnecessary regulatory actions (‘false alarms’) that may interfere with other regulatory systems. Thus, error control mechanisms are most effective when there is a high likelihood that small changes will lead to larger ones.

Anticipation system regulators are employed when the cost of regulating, when it is unnecessary, outweighs the cost of not regulating when it is necessary [7]. These regulators rely on understanding the causal relationships in the environment sufficiently well to predict changes in the state of the environment. Anticipation regulators are activated by detecting changes in variables that predict the future state of the environment, thereby allowing for early intervention to prevent or mitigate potential impacts [7,8,36]. These regulatory actions may involve altering the future state of the environmental input [7]. (For example, hail cannons operate as anticipation regulators that are intended to suppress the formation of hail [38]). The evacuation of communities based on predictions of the height and timing of flood events is an example of an anticipation regulator.

However, anticipation regulators have two key disadvantages: they may activate unnecessarily, leading to excessive regulation, or they may fail to trigger when needed [7]. Each of these can undermine future compliance with regulation by affected parties.

3.6. Defence-in-Depth

Given our framing of disaster management as the management of an open system through the deployment of system regulators, we can now consider how regulators contribute to the design of disaster management systems and their capacity to adapt to climate change. The concept of defence-in-depth (DID) is helpful here as it provides a way to frame the way regulators in a system should interact. The foundation of defence-in-depth is that a defender builds a series of defensive positions, and as an attacker advances, the defender falls back successively through these positions, resulting in the eventual defeat of the attacker through accumulated losses.

In its simplest form in the current context, defence-in-depth is realised by having a succession of diverse safety measures that are designed for the following [39]:

• Prevent undesirable incidents from occurring;
• Prevent these incidents from escalating should prevention fail;
• Mitigate or contain the consequences of incidents should they escalate.

In a safety management context, defence-in-depth means having more than one protective measure for a given safety objective such that the objective is achieved even if one of the protective measures fails. Protective measures can be anything from inherent safety features, use of multiple barriers, engineered safety features, principles, and procedures followed in design to the construction, operation, maintenance, and decommissioning of the system [40].

The effectiveness of the defence-in-depth protection is established through the principles of redundancy, diversity, segregation, separation, and single-point failure protection [22]. Effectiveness depends on having the capability to minimise failure at each barrier level and to minimise dependencies between barrier levels [40].

Conservative design and high quality in construction and maintenance are fundamental to minimising the likelihood of system regulators failing, whether they be active or passive. For example, with respect to flood management, the risk of levee failures depends on the quality of maintenance and inspection services as well as the quality of their design and construction (see

Table 1. Defence-in-depth.

Level of Defence	Objective	Critical Disaster Management Features
Level 1	Prevention of failures in system regulators	Conservative design and high quality in construction and operation
Level 2	Detection of failures in system regulators	Surveillance systems that monitor, but are independent of, the state of the system regulator
Level 3	Prevention of escalation once a system regulator fails	Engineered safety features and accident procedures. Complementary regulators and accident management.
Level 4	Mitigation of the consequences of regulator failure	Off-site emergency response

Adapted from [22].

). The same holds for active regulators like error control regulators such as flood control pumps and channel dredging but with the addition of rules that are, presumably, conservative with respect to timeliness regarding the conditions triggering their activation.

Surveillance systems that monitor the state of system regulators, such as levees or flood control pumps, should be independent of the state of the regulators they monitor. For example, to avoid creating a common source of failure, the source of power for monitoring and reporting the activation or performance of flood control pumps should be different from the source of power for the pumps.

A key principle of defence-in-depth is preventing escalation, should a system regulator fail, by the activation of additional passive or active system regulators. The sandbagging of an eroding levee is an example of the activation of an error control regulator to forestall further erosion and the collapse of the levee.

Another key principle of defence-in-depth is mitigating the consequences of the failure of a system regulator. For example, emergency evacuations to avoid loss of life from the flooding expected from the collapse of a levee are an example of an active regulator (an error control) being used to mitigate the consequences of failure. The diversion of floodwaters into sacrifice areas should a levee fail or be in danger of failing [41] is an example of a passive regulator being used to mitigate the consequences of the failure of a system regulator.

Note that a series of regulator failures does not comprise a cascade of natural disasters as such; cascades are characterised by one natural hazard triggering another hazard, natural or otherwise (see [20]). For example, the Tōhoku earthquake in 2011 created a cascade of an earthquake, a tsunami, and an uncontrolled radiation release from a nuclear plant.

4. Results: Characterising Regulators in Disaster Management Systems

Given our conceptualisation of disaster management as the management of an open system and the characteristics of different types of system regulators, we can now characterise, in these terms, the way various kinds of system regulators may be used to prevent or mitigate natural disasters such as earthquakes, floods, wildfires, droughts, and tropical cyclones. Note that the examples we provide are illustrative; they are not meant to provide a comprehensive or exhaustive description of all the system regulators relevant to each type of natural disaster.

4.1. Earthquakes and Tsunamis

To begin with, our limited understanding of the forces that trigger earthquakes and tsunamis means that their timing, magnitude, and scale cannot be predicted [42]. Consequently, disaster management is mainly limited to mitigating the effects of earthquakes and tsunamis using Type A passive regulators such as traditional [43,44] and mandatory design standards [45] for built structures in areas that experience earthquakes (see

Table 2. System regulators in earthquake and tsunami management.

Type	Examples	Comments
Aggregation	-
Passive Category A	Design standards for structures, emergency shelters, sea walls
Passive Category B	-
Passive Category C	-
Passive Category D	Evacuation warning system
Error control	Evacuation based on proximity
Anticipation	Evacuation based on prediction	Predictions of tsunami height based on earthquake magnitude

Notes: The data in the table are intended to be illustrative only and are not intended to provide a comprehensive or exhaustive description of the system regulators associated with each type of natural disaster. Category A regulators do not require external signals about the environment, external power sources, or any moving mechanical parts. Category B regulators do not require external signals about the environment or external power sources but have moving mechanical parts. Category C regulators do not require external signals about the environment but do rely on an external power source. Category D regulators do require external signals. (-) indicates example not provided.

).

The detection of an earthquake does allow the possibility of mitigating the consequences of a subsequent tsunami using an evacuation warning system. Emergency evacuations may be triggered based on proximity to the coast (error control) or, perhaps, on predictions based on the location and magnitude of the earthquake (anticipation). In the case of earthquakes and tsunamis, the idea of defence-in-depth has limited relevance.

4.2. Floods

Compared with earthquakes, we have a much better, though limited, understanding of the dynamics of floods. Consequently, although we cannot predict the timing and magnitude of floods in advance of the rainfall events that trigger them, we can predict the height and rate of movement of floodwaters through catchments once those events transpire. We also have some information on the frequency with which flooding of varying severity occurs in catchments. Consequently, the full range of system regulators may be employed to prevent or mitigate floods. A variety of examples of system regulators are presented in

Table 3. System regulators in flood management.

Type	Examples	Comments
Aggregation	Wetlands, vegetation planting, reforestation
Passive Category A	Terracing, dykes, levees, bunds, reservoirs, weirs, floodways, design standards for structures
Passive Category B	Mechanical barriers
Passive Category C	-
Passive Category D	Evacuation warning system
Error control	Sandbagging, inflatable tubes, channel dredging, flood pumps
Anticipation	Evacuation based on predictions of flood height

Notes: The data in the table are intended to be illustrative only and are not intended to provide a comprehensive or exhaustive description of the system regulators associated with each type of natural disaster. Category A regulators do not require external signals about the environment, external power sources, or any moving mechanical parts. Category B regulators do not require external signals about the environment or external power sources but have moving mechanical parts. Category C regulators do not require external signals about the environment but do rely on an external power source. Category D regulators do require external signals. (-) indicates example not provided.

.

First, aggregation regulators such as large-scale planting of vegetation in catchment uplands and the preservation or construction of wetlands in catchment lowlands may reduce or impede runoff. Second, information on the frequency with which flooding of varying severity occurs in catchments creates the opportunity to use a range of passive system regulators, such as levees, weirs, and reservoirs, to manage flooding up to a certain magnitude (see Table 3). The predictability of the movement of floodwaters through catchments means there may be sufficient time to implement error control regulators such as channel dredging and flood pumps. Third, predictions of the height and rate of movement of floodwaters through catchments may provide the basis for community evacuations (an anticipation regulator).

Defence-in-depth may be applied in a variety of ways with respect to flooding. For example, an error control regulator, such as sandbagging, might be employed to prevent escalation if a system regulator, such as a levee, begins to fail. The diversion of floodwaters to preserved or constructed wetlands is an example where an aggregation regulator might be employed to mitigate the consequences of the failure of upstream regulators, such as levees. Alternatively, emergency evacuation is an example of using an error control regulator to mitigate the consequences of regulator failure such as the collapse of a levee.

Note that, in the context of floods, defence-in-depth means that critical utilities such as power, communication, and transport systems should not be endangered by the failure to prevent the escalation. Also, the failure of a system regulator designed to prevent flooding should not, ideally, endanger the operation of system regulators intended to prevent escalation or to mitigate the consequences of failure.

4.3. Wildfires

We have some understanding of the dynamics of wildfires. Consequently, although we cannot predict the timing and magnitude of wildfires, we can predict the intensity and rate of movement of wildfires (to a degree) once they materialise. We do, however, have limited capabilities when it comes to controlling wildfires because of delays in detecting them and their rapid movement and scale. Consequently, we are constrained to employing passive and error control regulators to prevent or mitigate wildfires (see

Table 4. System regulators in wildfire management.

Type	Examples	Comments
Aggregation	Fire equipment and crews
Passive Category A	Emergency shelters, housing design standards
Passive Category B	Permanent firebreaks
Passive Category C	-
Passive Category D	Evacuation warning system
Error control	Controlled burns, fire suppression activities, wildfire detection systems, evacuation based on proximity	Suppression activities include firebreak construction, backburning, firefighting
Anticipation	-

Notes: The data in the table are intended to be illustrative only and are not intended to provide a comprehensive or exhaustive description of the system regulators associated with each type of natural disaster. Category A regulators do not require external signals about the environment, external power sources, or any moving mechanical parts. Category B regulators do not require external signals about the environment or external power sources but have moving mechanical parts. Category C regulators do not require external signals about the environment but do rely on an external power source. Category D regulators do require external signals. (-) indicates example not provided.

).

The construction of emergency shelters, the installation of permanent fire breaks, and compulsory design standards for buildings and infrastructure in fire-prone areas are examples of passive regulators. Once wildfires are detected, a variety of error control regulators in the form of fire suppression activities such as firebreak construction, backburning, and firefighting are employed to prevent escalation. The unpredictability of wildfire movement means that error control, in the form of evacuation based on proximity, is the primary means of mitigating the effects of uncontrollable wildfires.

As the areas that are prone to wildfires are extensive and the fires can grow rapidly, to ensure rapid response to the detection of fires, stores of fire suppression equipment (and crews to operate that equipment) are distributed throughout fire-prone areas. This is an example of using a spatially distributed aggregation regulator.

The application of defence-in-depth to wildfires is limited by their unpredictable location and timing. For example, passive regulators such as building standards offer a means of mitigating the damage should error control regulators such as fire suppression fail. Emergency evacuation based on proximity to fires is another error control regulator that may be employed to mitigate the consequences of failing to suppress wildfires.

4.4. Tropical Cyclones

We have a good understanding of the dynamics of tropical cyclones, which are, relatively speaking, slow-moving phenomena. Importantly, we can, to some degree, predict the timing, magnitude, intensity, and path of tropical cyclones once the conditions for their formation materialise. We do not, however, have any capabilities when it comes to controlling tropical cyclones, not least because of their scale. Consequently, we are constrained to employing passive and error control regulators to mitigate the damaging effects of tropical cyclones (see

Table 5. Tropical cyclones.

Type	Examples	Comments
Aggregation	-
Passive Category A	Emergency shelters, housing design standards
Passive Category B	-
Passive Category C	-
Passive Category D	Evacuation warning system
Error control	Property protection measures, evacuation based on proximity Temporary shelters	Property protection measures include window boarding and sandbagging
Anticipation	Evacuations based on forecast intensity and path of tropical cyclone

Notes: The data in the table are intended to be illustrative only and are not intended to provide a comprehensive or exhaustive description of the system regulators associated with each type of natural disaster. Category A regulators do not require external signals about the environment, external power sources, or any moving mechanical parts. Category B regulators do not require external signals about the environment or external power sources but have moving mechanical parts. Category C regulators do not require external signals about the environment but do rely on an external power source. Category D regulators do require external signals. (-) indicates example not provided.

).

The construction of emergency shelters and design standards for buildings and infrastructure in tropical cyclone-prone areas are examples of passive regulators [44]. Once tropical cyclones are detected, a variety of error control regulators in the form of property protection measures, such as sandbagging, boarding up windows, tree trimming, etc., are employed to mitigate the damage caused by tropical cyclones. The damage created by tropical cyclones can be mitigated by evacuating communities. Evacuation may take the form of an error control regulator if they are based on proximity or an anticipation regulator if they are based on predictions of tropical cyclone movement and intensity.

Defence-in-depth does not apply to tropical cyclones because there is no alternative to mitigation.

4.5. Drought

Droughts are slow-moving phenomena, which means that although their timing, duration, and extent are unpredictable, in principle there is ample scope to implement measures to mitigate the effects of drought. This also means that there is scope to implement defence-in-depth. Drought, like tropical cyclones, cannot be prevented. Instead, regulation must focus on mitigation and a variety of aggregation and error control regulators (see

Table 6. Drought.

Type	Examples	Comments
Aggregation	Food reserves, cash reserves
Passive Category A	Vegetation planting, tree crops
Passive Category B	-
Passive Category C	-
Passive Category D	-
Error control	Distribute food, import food, foreign aid, drought-tolerant crops, evacuation	Requires transportation and communication infrastructure
Anticipation	Establish food reserves based on predictions of seasonal conditions

Notes: The data in the table are intended to be illustrative only and are not intended to provide a comprehensive or exhaustive description of the system regulators associated with each type of natural disaster. Category A regulators do not require external signals about the environment, external power sources, or any moving mechanical parts. Category B regulators do not require external signals about the environment or external power sources but have moving mechanical parts. Category C regulators do not require external signals about the environment but do rely on an external power source. Category D regulators do require external signals. (-) indicates example not provided.

).

Planting forage vegetation and tree crops are examples of passive regulators. Food and cash reserves operate as aggregation regulators. Cash reserves enable the importation of food, an error control regulator. The planting of drought-tolerant crops when seasonal conditions are dry is another example of error control. Drought-tolerant crops may also operate as an anticipation regulator when planting is based on forecasts of seasonal conditions.

Nesting of system regulators creates opportunities to implement defence-in-depth. For example, for a livestock operation, this may involve supplementary feeding using feed reserves (aggregation), selling non-breeding stock as conditions worsen (error control), purchasing supplementary stock feed (error control), selling breeding stock when conditions become extreme (error control), and drawing down cash reserves (aggregation) if extreme conditions persist.

5. Discussion

We have classified examples of the various means by which we seek to prevent or prepare for natural disasters into different types of system regulators. Several observations may be made regarding the characteristics of natural disasters and the types of system regulators that may be employed regarding them. The first is, not surprisingly, that differences in the nature, scale, and predictability of natural disasters lead to differences in the degree to which system regulators can be deployed to prevent them or mitigate their consequences and to differences in the types of system regulators that can be deployed.

5.1. System Regulators and Differences in the Characteristics of Natural Hazards

The nature and scale of tropical cyclones and droughts mean that the deployment of system regulators is restricted to mitigation. Differences in the rate at which these natural disasters unfold mean that defence-in-depth is feasible with respect to drought, whereas it is not for tropical cyclones. Relatedly, differences in the rate at which these natural disasters unfold mean that mitigation for tropical cyclones depends primarily on passive regulators, whereas error control regulators can be employed to mitigate the effects of drought. The sheer unpredictability of earthquakes, as well as their scale, means defence-in-depth is largely irrelevant, and mitigation is limited to passive system regulators.

In contrast to tropical cyclones, earthquakes, droughts, floods, and wildfires, they tend to result in more localised natural disasters. Hence, there is potential to use system regulators to limit their escalation, if not prevent them. With respect to wildfires, limiting escalation depends heavily on error control regulators, with passive regulators being used for mitigation. With respect to floods, in contrast, passive regulators are employed to prevent or control escalation, with little in the way of error control regulators available to be used to prevent escalation and for mitigation.

The dynamics of floods are reasonably predictable; wildfires less so, perhaps. Certainly, the movement of floodwaters through catchments is more predictable than the movement of wildfires. The greater predictability of the dynamics of floods compared to wildfires creates greater opportunities to employ anticipation regulators for mitigation, such as community evacuations based on flood height predictions.

5.2. The Role of Historical Data in Designing Passive Regulators

A second observation is that passive regulators are fundamentally important in preventing floods and mitigating the consequences of floods, earthquakes, tropical cyclones, and wildfires. Passive regulators are constructed using historical information on the characteristics of natural disasters, such as their incidence, location, and severity. Their regulatory capacity is framed by past environmental behaviour rather than current environmental behaviour [7]. For instance, the size of an aggregation regulator and the thresholds governing the settings of passive regulators are determined based on expectations about the frequency distribution of the severity of disasters derived from experience or historical data. Consequently, the effectiveness of these regulators depends on the relevance of historical data to current conditions.

For example, flood levees are constructed using historical information on the distribution of flood events. Changes in key characteristics of catchments (deforestation, urbanisation) or in the climate may change the shape of these distributions, thereby altering the performance of levees. The same reasoning applies to other passive regulators, such as building standards for housing and infrastructure regarding floods, earthquakes, tropical cyclones, and wildfires.

Error control regulators require the setting of a threshold to trigger activation, as they are based on the principle that small changes portend large changes. The setting of these thresholds is, one way or another, based on expectations derived from historical data. Consequently, the effectiveness of error control regulators also depends on the relevance of historical data to current conditions. For example, error control regulators that rely on proximity thresholds, such as evacuations triggered by the nearness of wildfires or tropical cyclones, may fail because the changes in landscape or climate characteristics alter the rate at which wildfires or tropical cyclones move, intensify, and spread. Similarly, the deployment of error control regulators, such as periodic channel dredging to prevent or mitigate flooding, is based on expectations of water movements and may fail because changes in catchment or climate characteristics alter the volume and rate at which floodwaters move or the deposition of sediments in channels.

This suggests that, as local climates gradually shift from historical norms because of global warming, the thresholds that are currently used to trigger error control regulators are likely to become progressively outdated and ineffective. This implies that the costs associated with error control will escalate. The appropriate policy response is either to adopt more conservative thresholds or invest in passive regulators, or both. The choice between these alternatives will depend, to some degree, on the feasibility of establishing new thresholds and the capacity to respond in a timely fashion to breaches of those thresholds (see Section 4.3 and Section 4.4 for examples in relation to flooding and wildfire, respectively).

5.3. The Role of Historical Relationships in Designing Anticipation Regulators

Anticipation regulators are, by their nature, based on models of key causal relationships in the environment. As these models are based on historical data, changes in key relevant variables may alter causal relationships, thereby reducing the accuracy of these models and undermining the performance of the system regulator. For example, models of catchment hydrology may be used to predict floodwater movements, the predictions being used to trigger community evacuations to mitigate the consequences of flooding. Changes that occur in catchment hydrology [34] or in climate characteristics may mean the models cannot reliably predict the movement of floodwaters. If the volume of floodwaters is underestimated, then they may move more quickly and be more widespread than anticipated, and evacuations may not be triggered in a timely fashion.

For example, if the dredging of a river mouth to prevent flooding depends on expected river flows derived from rainfall forecasts [46], then dredging is an anticipation regulator. Changes in the predictability of the intensity and duration of rainfall due to climate change may reduce the accuracy with which river flows can be predicted from weather forecasts. Consequently, the threshold for triggering dredging may not be triggered in a timely fashion, resulting in unexpected flooding [46] and dredging as an anticipation regulator fails. In these circumstances, the appropriate policy response is to switch to using dredging as an error control regulator by changing the rule for setting the triggering threshold, for example, by switching to a calendar-based rule to ensure the river mouth is regularly dredged. This change in the rule for setting thresholds better aligns with the necessity to dredge the river mouth to prevent flooding with the capacity to dredge it.

Another observation is that, except in the case of earthquakes, community evacuations are a critical means of preventing loss of life from the natural disasters we have considered here. The failure to evacuate in a timely fashion can have catastrophic consequences. This suggests that evacuations based on error control regulators, such as proximity, should employ conservative thresholds and that evacuations based on anticipation regulators, such as predictions of floodwater flow and tropical cyclone intensity and movement, should be based on exceptionally reliable predictive models. Worryingly, the increasing pressure on timeliness in issuing evacuation warnings might increase the potential for more frequent failures of emergency evacuation warning systems.

5.4. Climate Change and the Choice of System Regulators

There are several implications for the use of system regulators if changes to the climate mean that historical data are not a reliable guide to setting thresholds for either error control or anticipation regulators, particularly if we have moved from a world where the climate could be treated as a placid random component in the environment to a world where it is a disturbed reactive (or even turbulent) component [28]. One implication is that anticipation regulators should be abandoned in favour of error control regulators (at least until any new predictive models prove themselves to be thoroughly reliable). This means, for example, switching from activating evacuations using anticipation regulators based on criteria such as predicted path of travel and intensity to activating evacuations using error control regulators based on a criterion such as proximity. We note that there is a heavy reliance worldwide on anticipation regulators in relation to, for example, flash floods [47].

A second implication concerns the advantage that anticipation regulators offer relative to error control regulators. Recall that regulation by anticipation is appropriate if the cost of acting (evacuating) when it is not necessary is judged to be greater than the cost of not acting (not evacuating) when it is necessary and that regulation by error control is appropriate if the cost of acting (evacuating) when it is not necessary is judged to be less than the cost of not acting (not evacuating) when it is necessary. If climate change means that the frequency and severity of a natural hazard will increase over time, the cost of not acting when it is necessary will increase over time. This means that, as climate change progresses, the cost savings that anticipation regulators offer over error control (and passive) regulators will diminish. See Su [34] for a historical account in relation to Taipei, Taiwan.

A third implication is that error control regulators should be activated using more conservative thresholds. For example, if tropical cyclones or wildfires are increasingly frequent, intensify more rapidly, and move more erratically or quickly, then proximity thresholds for issuing evacuation warnings should be more conservative. Similarly, the effectiveness of temporary tropical cyclone shelters, an error control regulator, might be improved by increasing the threshold for their construction from a few days before the onset of the tropical cyclone season to a few weeks before [48].

Another implication is that aggregation regulators and passive regulators based on conservative thresholds may provide more reliable and less expensive means of preventing or mitigating natural disasters. If climate change means that climate is becoming a turbulent component in the environment [28], then error control regulators based on conservative proximity thresholds may not offer acceptable degrees of protection. In these circumstances, investing in aggregation or passive system regulators may be an effective option. Investing in reforesting or terracing upland areas and reconstructing wetlands in lowland areas of catchments to reduce or impede runoff are examples of aggregation regulators that may prevent or mitigate flooding.

Alternatively, converting an error control regulator into an aggregation or passive system regulator may be an effective option. For example, setting the proximity threshold for evacuation at zero means permanently evacuating an area [49]. Effectively, doing so transforms evacuation from an error control regulator to a passive regulator. With respect to flooding, this means permanently relocating communities, housing, industry, services, utilities, and infrastructure away from flood-prone areas [34]. Presumably, relocations of this nature substantially reduce the costs of floods and may even mean other passive regulators such as dykes, weirs, and floodways may be dismantled. Note that the extent to which climate change alters the predictability of flooding, and therefore the effectiveness of different system regulators, depends on the characteristics of catchments.

With respect to wildfires, this reasoning raises the possibility of limiting the use of error control measures, such as fire suppression, to mitigate the effects of wildfires because they may be both less effective and more hazardous in extreme conditions [50]. This would mean relying more on defence-in-depth using passive regulators to protect lives and property, such as building standards, vegetation-free zones around towns, villages, and buildings, and fire shelters for communities and dwellings. Limiting the use of error control measures such as fire suppression requires a corresponding change in other error control measures, such as evacuation based on proximity. One possibility is to change the criterion for evacuation from indicators of proximity to wildfires (e.g., prominent geographic features [51]) to a criterion reflecting the risk of wildfire (see Yeo et al. [52]). Another is to invest in passive regulators such as community refuges of various kinds [53].

Switching to error control regulators, such as proximity-based evacuations, does mean that unnecessary evacuations may become more frequent, which may strain the public credibility of the evacuation authorities [37]. This suggests that the reasoning underpinning the rules used to order evacuations should be clearly communicated to the public. Relatedly, the decision to evacuate may be devolved from a central authority (compulsory evacuation) to the individual citizen (voluntary evacuation), assuming the individual is better informed regarding the state of their defences against hazards such as tropical cyclones, floods, and wildfires, though this may not always be the case [47,54,55]. It is worthwhile noting here that technological change influences regulator choice and design. For example, artificial intelligence might improve the accuracy with which hazards such as tropical cyclones are forecast, thereby enabling improved error control or even the use of anticipation regulators with respect to the issuing of evacuation warnings.

5.5. Climate Change and Defence-in-Depth

The idea of defence-in-depth also has implications in relation to sustainability, climate change, and disasters arising from natural hazards. The effectiveness of defence-in-depth depends on having the capability to minimise failure at each barrier level and to minimise dependencies between levels. The higher the category of passive regulators, the greater the scope for failure, at least in terms of the number of sources of failure (see Table 1). This suggests, in principle, that using higher-category passive regulators to mitigate failures in lower-category passive regulators should be avoided to minimise the potential for failure by compromising independence between levels. The same reasoning suggests that using anticipation regulators to mitigate failures in error control regulators should be avoided, and using error control regulators to mitigate failures in passive regulators should be avoided whenever possible. For example, the failure of a flood levee may result in the inundation of power sources for error control regulators such as flood pumps [34]. Similarly, the failure of a flood levee may result in the inundation of transport networks, rendering mitigation using error control regulators such as evacuation inoperable. If climate change increases the likelihood that defence-in-depth may be compromised, then aggregation and passive regulators may increasingly offer more effective disaster prevention and mitigation than anticipation and error control regulators.

Relatedly, and especially in terms of disaster recovery, the defence-in-depth criterion of independence would suggest that infrastructure critical to recovery, such as communication, transportation, and power systems, are protected (to the degree this is practical) from natural disasters using aggregation regulators; for example, by permanently locating infrastructure on high ground to avoid flooding and placing power systems underground to avoid wildfire and cyclone damage [56].

5.6. Disaster Mitigation and Disaster Recovery

Relatedly, as the impacts of climate change on the frequency and severity of natural disasters become increasingly apparent, pressure is mounting for change in the approach of governments to disaster management. To date, the typical response of authorities has seemed to focus much more on rescue and recovery than on pre-disaster policy. Sometimes viewed as mere grandstanding [57] (and, by definition, insufficient to the task), current government approaches to the increasing frequency and magnitude of climate impacts are evoking strong popular rejection (see BBC [58] for example).

The principal difficulty with the widespread emphasis on post-disaster responses is that climate hazards and disasters are worsening in scale, as warned repeatedly over decades by climate scientists. The typical government responses are becoming, by definition, less effective. Governments are perceived to be failing more often in the duty of care owed to their citizens.

The trajectory of climate change indicates that too few contributing nations are adopting adequate policies to moderate it. In this context, it is politically courageous for governments to fail substantially to respond as comprehensively as possible to the mounting threat. For example, the main emphasis in Australia’s National Strategy for Disaster Resilience [59] is on post-disaster responses, with some attention to mitigation via, for example, land management policies and better information management. The suggestion is that individuals and communities need to take greater responsibility for their own welfare.

There is little in the strategy to indicate a perception that the extent of change in the frequency and severity of disasters may imply a need for a qualitative change in disaster management whereby potential disaster impacts are optimised in advance to lessen reliance on post-disaster management to deal with the impacts. This seems a partial response to the Sendai Framework for Disaster Risk Reduction [18], the third priority of which is to invest in disaster risk reduction and the fourth, and last, to enhance disaster preparedness. See also Guo et al. [47] regarding incremental changes in disaster prevention regarding flooding and contrast Harman et al. [60] and de la Vega-Leinert et al. [61] with respect to coastal inundation.

Interestingly, government responses tend to reinforce the growing perception that climate change is beyond meaningful control: there is little any entity can do. One significant recent proposition [62] is that the changing economics of wind and solar energy and storage options imply that climate change is not a terminal threat and that peak global warming may occur at around +3 °C [63]. As serious as that level is, the possibility that it is a cap reinforces the notion that governments need to do better than offer more of the same and exhortations to ‘work harder’; if this proposition that this threat is bounded is correct, it should be susceptible to rational substantive responses.

Another, much more imminent trigger for action arises from developments in insurance markets. The need for a change in government approach may well be forced by the pace and direction of change in insurance markets. Claims for fire, water, and tropical cyclone damage in countries with high levels of private insurance have triggered upward pressure on insurance and significantly increased reinsurance premiums. That is, inadequacies in the capacity to prevent or mitigate disasters in response to climate change are leading to significant rises in insurance claims. That is, the variety of disasters that intersect with insurance systems is making ‘insurance disasters’ a looming component of various cascading disasters: the predictable impacts of rises in premiums will be greater rates of uninsurance and underinsurance, increasing the average exposure to subsequent disasters, and popular anxiety. This means that ‘more of the same’ is not viable and, importantly, that any option involving the government itself carrying insurance will have significant fiscal consequences.

Responding to the 2023–2024 Insurance Catastrophe Resilience Report [64], the CEO of the Insurance Council of Australia [65] said the following:

De-risking is the only sustainable way to reduce the pressure on premiums and close the protection gap: improved planning so no more homes are built in harm’s way, stronger buildings that are better able to withstand extreme weather, greater investment in public infrastructure to protect communities, and an ongoing program of home buybacks where no other mitigation is possible.

We suggest that the adoption of a comprehensive, integrated framework, such as the systems theoretic one we have outlined above, with the potential to minimise or, better, optimise the residual disaster impacts with which it is necessary to cope, would be a rational starting point for policymakers.

5.7. Further Research and Limitations

We have sought to show that disaster management systems can be conceptualised as complex mixes of various types of system regulators with implications for the design and management of such systems. As this is the first application of regulator theory to (natural) disaster management, we have limited ourselves to describing the theory and principles and illustrating the potential for their application. Testing the theory and principles by detailed application to specific disaster management systems is an obvious next step.

The implications for policy, flowing from conceptualising disaster management systems as being composed of system regulators, suggest a critical area for further work would be to develop methods for modelling system regulators in disaster management systems. These could be employed to test the conditions under which the different types of system regulators are preferred in disaster management systems and to compare the performance of different system regulators under the various environmental scenarios associated with climate change. Importantly, these methods could be used to assess the magnitude of the residual risk of a disaster remaining after all the relevant system regulators have been activated. Such assessments would help in providing an indication of the extent to which government investment in disaster preparedness offsets the need for government, communities, and individuals to invest in disaster recovery.

The above leads to the following preliminary suggestions for incorporating consideration of regulator theory into the design of disaster management systems:

• Identify and classify the suite of system regulators in a DMS.
• Trace inter-relationships and dependencies between system regulators.
• Evaluate inter-relationships and dependencies using the principles of defence-in-depth.
• Integrate a model of the system regulators with risk assessments of regulator failure (e.g., inadequate aggregation, insufficient capacity to respond to threshold breaches for error control and anticipation triggers) to assess the residual risk of disaster.
• Compare costs of unconditional and conditional regulators.
• Evaluate risks to the operation of regulators posed by changes in data used to define size for aggregation and passive regulators, thresholds and response times for error control regulators, and changes in data and causal relationships used to set triggers and response times for anticipation regulators.

6. Conclusions

The fundamental objective in creating a disaster management system is to design and operate the system so that it offers an acceptable and sustainable degree of protection over time, given expectations about the behaviour of the biophysical environment. The components and relationships that comprise a disaster management system can be interpreted as mechanisms that are employed to regulate the behaviour of environmental and human inputs into the system. Consequently, a disaster management system can be viewed as a complex mix of system regulators, each of which can be characterised in terms of the fundamental types identified in general systems theory. This suggests that the criteria that have been formulated in general systems theory for choosing system regulators should be applicable to the design of disaster management systems.

These considerations suggest that if disaster management systems are to be sustainable and adapt to climate change, adaptation will involve restructuring these systems by changing the mix of system regulators within them. The magnitude of change across environmental inputs entailed in climate change implies the need for a qualitative response rather than relying on scaling up responses that previously were satisfactory. This leads to the conclusion that one of the keys to developing policies to support the adaptation of disaster management systems to climate change hinges on understanding how these systems can be construed as mechanisms for modifying environmental inputs to create protective flows.