Identifying Decision Support Needs for Emergency Response to Multiple Natural Hazards – an Activity Theory Approach

Planning and decision making in emergency responses systems face new challenges due to climate changes and the increased risk of multiple or compound natural hazards. This is especially the case in areas with inexperience of such events. The aim of this paper is to identify important activities concerning planning and decision-making during responses to natural hazards, and their potential need of decision support. The knowledge base of the study consists of interviews with 12 representatives from the Swedish emergency response system, supplemented by documents covering policies, operations, and responsibility. Thematic coding was applied to the interview data for identi�cation of important planning and decision activities. Needs of decision support were identi�ed by applying activity theory to the identi�ed activities. We found needs of decision support connected to eight identi�ed key activities concerning consequence analysis, national reinforcements, and resource management. The results illuminate a lack of technology to support response activities during both single and multiple natural hazards. The �ndings can inform policy makers in emergency response of where to concentrate the development of tools for collaborative preparedness and response work to cope with future challenges from natural hazards.


Introduction
Climate change with an increase of the global mean temperature will likely lead to glacial melt and disruptions in the global ocean streams (IPCC 2014).Other phenomena include increased frequency of extreme weather events, leading to higher risks of additional natural hazards, e.g., landslides and forest res.Such a chain of events can be called multiple (Gill and Malamud 2014) or compound (Zscheischler et al. 2020) natural events or hazards, where a primary natural hazard directly triggers or, by changing the environment, increases the probability of secondary hazards.The increased risk of multiple natural hazards brings challenges to the planning and decision making in emergency response (ER) systems, especially in areas with limited experience from this, e.g., Sweden and other Scandinavian countries.
Multiple natural hazards will be even more challenging in terms of command, control, and coordination of the responses due to their complexity.To avoid potential devastating consequence, there is a need to improve decision support for ER systems to cope with these kinds of hazards (Chacko et al. 2014).To coordinate a single large crisis with many involved actors is di cult, not the least in a decentralized crisis management structure (Grottenberg and Njå 2017).Sweden has, like many countries, such a decentralized ER system with no single national agency responsible for overall emergency management (EM).Instead, most responsibility falls to the municipalities.To understand how technologies could support the management of multiple natural hazards, it is important to understand the interactions and activities of the studied ER system.This calls for an exploration of the current EM to increase the understanding of factors that in uence the response to multiple natural hazards, and how planning and decision support tools can help to improve this.

Aim and objectives
The aim of this paper is to deepen the understanding of ER to natural hazards by identifying and examining key planning and decision activities in the Swedish ER system.This is done by (i) identifying key planning and decision activities used in practice in the response to natural hazards, (ii) systematically describing the activities using activity theory, and (iii) using activity theory to examine the activities in search of planning and decision support needs.
The paper provides knowledge to policy makers and practitioners of where to concentrate the development of tools for collaborative preparedness and response work to cope with future challenges of natural hazards.The paper will also inform the EM research community by applying activity theory which provides a theoretical understanding of ER activities.

Related work
While there are previous studies developing planning and decision support tools to improve ER to natural hazards (e.g., Guth et al. 2019;Duan et al. 2020), there is a lack of studies on the actual needs of the ER systems to manage these types of events.This creates a gap between the science documented and the practice used in ER.However, several studies examine issues, needs, and challenges of ERs to large-scale single natural hazards, and some provide suggestions or directions for improvements connected to planning and decision support.Theodora (2020) suggests that natural hazards require the integration of advanced digital technology to support information gathering, evaluations, prioritization, visualization, and monitoring.According to Simonovic et al. (2021), decision tools based solely on risk are insu cient, and new technology needs to consider complex multi-hazards.They suggest a holistic approach, considering the larger societal system affected by the hazards, all costs and bene ts, and alternative solutions, using multi-objective optimization and simulation tools.However, user acceptance is a challenge for all technological systems, and can be an obstacle for joint sense-making between response actors (Lu and Xue 2016), which is a crucial aspect for responses to large-scale emergencies.
Another challenge identi ed in the literature is public awareness, both regarding the risk of hazards and the subsequent ER.According to Tingsanchali (2012), who studied EM for urban oods in Thailand, public awareness can be increased by community participation in risk assessments and planning of ood hazards.A participatory process will also increase the success and effectiveness of warning systems and communication with the public, by increasing the public's knowledge of risks and make them more familiar with natural hazard responses (Tingsanchali 2012;Bird et al. 2018).The involvement of communities and social organizations can be crucial in the management of natural hazards (Sim and Yu 2018), as they often constitute rst response and local resilience actors (Genovese and Przyluski 2013).Bird et al. (2018) identi ed several success factors of interactive crisis communication during the Eyjafjallajökull eruption in Iceland, e.g., pre-eruption town hall meetings, the establishment of information and media centers, and increased dissemination of risk data.More complex natural hazards in the future will also require more creative responses, which need to be accepted by the public through improved interactive public communication (Steelman and McCaffrey 2013).
According to Miao et al. (2013), typical issues of emergency resource management are ine cient communication and a lack of interaction, cooperation, and integration.In a retrospect analysis of four disasters in Asia and the U.S., they suggest a more comprehensive approach to address the issues, using comprehensive information systems with cross-network integration.Comprehensive EM should cover all phases of the EM cycle (Greiving et al. 2006;Motamedi et al. 2009), and be combined with planning and operational control systems (Pesigan and Geroy 2009).Also, Bosomworth et al. (2017) suggested that to overcome the challenges, EM needs to be a component in community disaster risk reduction and a network governance approach should be adopted for comprehensive cooperation and easier dissemination of information.This, to overcome two core challenges of strategic EM in Australia: community expectations, and effective use of information systems and social media.
The contribution of this paper is the identi cation and analysis of key planning and decision activities in the Swedish ER system, addressing the research gap on determining the actual needs of planning and decision support tools in ER systems when managing multiple natural hazards.Since Sweden has limited experience of such events, we address this gap by studying the current ER system and how it would function during multiple natural hazards.

Study context
The Swedish ER system is decentralized, where the affected have the main responsibility of the response, i.e., the municipalities and the countyadministrations (CA).The actors involved in the response to natural hazards can be divided into three groups: response actors, support actors, and expert actors.
The main response actors are the re and rescue services (FRS), responsible for rescue missions during natural hazards, and the emergency medical services (EMS), providing aid to injured people.The FRSs are organized on municipal level or in federations covering multiple municipalities, while the EMS is organized on the county level with the ambulance service often procured from contractors.Depending on the circumstances, they can have support from voluntary response organizations, the Swedish Armed Forces, the Swedish Police, and private contractors.In this study, the EMS's potential involvement in the response has been excluded.
The supporting actors include the municipalities that have a "geographical area responsibility", which means that they should work for cooperation and coordination between societal actors during extraordinary events, e.g., natural hazards.The municipality is also responsible for providing information to the public.The CAs also have a geographical area responsibility on a regional level, concerning the coordination and cooperation between municipalities.The Swedish Civil Contingencies Agency (MSB) can provide support to FRS organizations in terms of resources, cooperation, and coordination.Other supporting agencies that may be involved during natural hazards are the Swedish Public Safety Answering Point (PSAP), the Swedish Church, business actors, and voluntary support organizations.
Expert actors can assist response actors with information concerning different natural events.For instance, the Swedish Meteorological and Hydrological Institute (SMHI) is one of the main expert actors, forecasting all weather-related events, monitoring all larger watercourses, and providing information about the current wild re risk.The Swedish Geotechnical Institute (SGI) has the role to inform and support other actors concerning geological hazards, e.g., landslides or debris ows.If a natural hazard has impact on critical infrastructure (CI), e.g., the road or rail network, the Swedish Transport Administration will be involved to support the response with information and expertise.
In the period 2006-2021, Sweden experienced several major natural hazards, mostly concerning wild res and ooding events.Data on the magnitude and consequences of some of these events are presented in Table 1.

Introduction To Activity Theory
Activity theory originates from psychology and the works by Vygotsky but was introduced internationally around 1980 primarily through the works of Leontiev (Kaptelinin and Nardi 2006).According to Vygotsky, an activity is individual and object-oriented, which means that the motive behind an individual's action is related to an object (Engeström 1999).Leontiev expanded on this idea and integrated the division of labor in his works, which introduced the phenomena of collective activity.During the 1990s, the theory had a dramatic increase of interest, primarily through the publications by Yrjö Engeström, who presented a new perspective of activity theory.Kaptelinin (2005) describes the differences between the original perspective and the new ideas where Engeström argues that activities are carried out by communities, and performed collectively, where the object is related to production, i.e., transformation of an object into the activity outcome.The reason behind activity theory is that individual action can only be understood when it is put into context, why the context needs to be put into the unit of analysis (Kuutti 1996).Thus, a research object concerning activity will always have a collective unit of analysis, although the phenomena of interest may be connected to individual action.Kuutti (1996) identi ed that the narrow research focus of human-computer interaction (HCI) was problematic since the context was not included in the unit of analysis, and that activity theory was a potential solution to the problem.Activity theory has since found multiple application areas within both HCI and information system design (e.g., Pilemalm and Timpka 2002), and in several domains of which one is EM (Chen et al. 2013).

Activity systems
According to Engeström (1999), an activity has six elements: the object, the subject, the mediating artifacts (tools), the rules, the community, and the division of labor.These six elements together with the outcome will form an activity system.A general version of such a system is presented in Fig. 1.
The subject is the group of individuals that carry out the activity.They work on the transformation of the object, which can be de ned as something that links individual action to the collective activity.The transformation from an object to an outcome is the motive of the activity and should bring meaning to everyone's action.The actions of the activity have support from a set of tools, which can help the transformation of the object.However, the actions also need to follow rules, e.g., laws, regulations, and directives.The community of the activity constitutes all actors that are involved in the activity and have interest in the outcome.Lastly, the division of labor describes the hierarchical structure of the activity and the connected community.

Externalization and contradictions
Adding to the activity system, two processes are recognized in activity theory: internalization and externalization (Engeström 1999).According to Kaptelinin and Nardi (2006), activity theory makes a difference between internal and external activities.Internalization means that an external activity is transformed into an internal activity, e.g., an individual who learns to type on a keyboard without looking at it.Externalization is the opposite, where an internal activity is transformed into an external activity.This is often needed when an action needs to be scaled but the human mind does not have enough capacity to perform it, for example a complex mathematical problem.Then, the activity needs to be performed external from the human mind, and it is transformed from an internal to an external activity.
Externalization can also be used to coordinate activities in collaboration between several people (Kaptelinin and Nardi 2006), and to support creative activities through computerization (Tikhomirov 1999) requiring less mental capacity of individuals.
A driver for change and development of an activity, through internalization or externalization, is the existence of contradictions in the activity system.This implies that the characteristics of an element of the activity system do not match the characteristics of another element (Kuutti 1996).This can lead to tensions in the activity system and constitute an obstacle for the activity to be carried out.It can eventually lead to an activity breakdown, where the outcome of the activity is unreachable.

Method
The method used has a case study approach (Yin 2014) with a single case and can be split into three steps: data collection, thematic analysis, and activity analysis.The single case in our study was the Swedish management of natural hazards, which was studied using one main and one supplementary knowledge base (Bowen 2009).The main knowledge base originated from interviews with practitioners of ER to natural hazards, which was supplemented by documentation from several Swedish ER actors.This, to strengthen the construct validity of the study (Yin 2014).The following sections will describe each step in more detail.

Data collection
In total, 10 semi-structured, virtual interviews with different actors of the Swedish ER system were conducted March 2020 -June 2021.The interviewees are presented in Table 2. Semi-structured interviews were chosen to enable exibility and the opportunity for follow-up questions (King et al. 2019).An interview guide was established with open-ended questions regarding the ER to multiple natural hazards, e.g., which actors that are involved in the response, what they do, and the challenges or factors in uencing the response given the speci c characteristics of the scenarios.The interview with the PSAP was a group interview with three interviewees, while the others were individual interviews.The recruitment of participants was mostly done through a mix of strategic selection and snowball sampling as described by King et al. (2019).Although there is a risk of bias applying snowball sampling, it was needed due to the low experience of multiple natural hazards in Sweden.
All interviews were held remotely via the meeting applications Zoom or Microsoft Teams.Remote interviews were chosen primarily because of the Covid-19 pandemic.In all interview sessions two researchers were present, one leading the interview and the other taking notes in terms of interpretations of the interviewee's answer and other re ections of the interview.All interviews were audio recorded.
As a supplement to the interviews, several documents (agency reports, action plans, policy documents) describing policies, operations, and responsibilities within the Swedish ER system were collected.The collection was limited to documents available at webpages of different actors in the Swedish rescue and response system.

Thematic analysis
A thematic analysis was done on the interview data to identify key planning and decision activities, with guidance from a manual by Saldaña (2013) describing a range of coding methods.An initial round of preparatory holistic coding was done by listening through each recording and code with a short phrase to get a grasp of the content and meaning of a longer sequence.A list of the holistic codes from each audio le was transferred to Excel as an individual sheet for a rst round of analysis coding, using a mix between descriptive and process coding methods.The rst round resulted in more re ned codes describing aspects and actions connected to the response.Through a second cycle of coding, using pattern and focused coding methods, the rst cycle codes were grouped and categorized according to similarities of what activities they described.The last step of the analysis was the establishment of superordinate themes, grouping similar second cycle codes in what could be called "categories of categories" (Saldaña 2013).
By using prede ned codes (Bowen 2009) established from the interview data analysis, supplementary contributions could be extracted from the collected documents.This, by looking for content describing the activity identi ed from the interviews.

Activity analysis
We used activity theory for the analysis of the Swedish ER to natural events.The theory has been frequently used within human-computer interaction, and technology and information systems development.However, it has been used more seldom within the ER domain, where two examples are Allen et al. ( 2014) and Mishra et al. (2015), both using activity theory as a conceptual or analytical framework for different aspects of EM.We believe the theory can contribute to the research area of EM, since the effects of system breakdowns due to contradictions are fundamental with increased risk of casualties.After identifying key planning and decision activities, they were all analyzed according to three steps.First, they were described and visualized as activity systems, including all six elements and the outcome (see Section 2.2), based on the interview data and collected documents.Then, contradictions were identi ed in the activity systems by analyzing the interview data.Primarily, we were looking for contradictions concerning the tools used, and which would imply a need for externalization of the activity to avoid potential breakdowns (see Section 2.3).The motivation behind such an analysis is that externalization would increase the operational load of the human resources in the activity system.Instead, they could use the time to analyze the results provided by a technological tool.However, it is important to note that externalization is not always a bad thing in activities but necessary, e.g., to create joint knowledge.But it will be an obstacle for the implementation of a standardized operating procedure, since it relies on the knowledge and skill from a limited group of individuals.Externalization is also timeconsuming, which is a limited resource during major emergencies and crises.

Results And Analysis
The thematic analysis regarding key activities resulted in three themes: consequence analysis, national reinforcements, and resource management.The following sections will describe the key activities under each theme, possible contradictions between each activity's elements, and suggest planning and decision support to improve the activity process.However, the focus here is to indicate the needs of support, and any suggestions of tools are speculative.

Consequence analysis
One important aspect before and during responses is to assess the risks of natural hazards.Most interviewees mention the importance of a consequence-based assessment.All municipalities and FRSs should do their own assessment and determine whether the identi ed consequences will be di cult for them to handle.To do quality risk assessments, several interviewees mention the need of support from actors with more knowledge of the event type, the surroundings, or operations affected.Such actors could be expert agencies, business actors, or affected landowners.In the theme consequence analysis, we have identi ed two key activities: the identi cation of critical infrastructure, and the assessment of resource requirements.
Forecasts and warnings regarding natural events in Sweden are mostly done by SMHI, which monitors and analyzes most meteorological, hydrological, and biophysical events.Some exceptions are avalanches and landslides, which are monitored by the Swedish Environmental Protection Agency and the SGI, respectively.During high risk of natural events, the responsible FRS may have daily contact with SMHI for status and forecast updates.
To determine the risk of wild res, most FRSs use the Fire Weather Index (FWI) provided by SMHI.The FWI is provided in a raster indicating the re risk in each cell (4 km 2 ) and is updated two times a day.At certain levels of FWI, the municipality and the CA can impose public re bans to limit the risk of ignition.
During high risks of wild res, regular surveillance by aircraft is done depending on the FWI and is procured by MSB from contractors.During ongoing wild res, surveillance takes place on several levels in the ER system.On a local level, the FRS monitor the re development to determine the spread, and how to distribute resources among the wild res.On regional and national levels, the PSAP does external monitoring to identify trends and deviations regarding wild res, while MSB gathers information about ongoing res to keep track on the national wild re situation.

Identi cation of critical infrastructure
The priority in any ER is to save lives.However, if no lives are at risk, the priority is to protect the public property essential for the society to function, here called critical infrastructure (CI).The activity system of the identi cation of CI is presented in Fig. 2.
The command and control (C2) function of the responding FRS is the main actor, hence, the subject of the activity.Given the projected area of the hazard's impact, the object of the activity, they will try to identify the CI in most need of protection.To do this, they have several tools at hand: maps, forecasts and warnings, and locations of the CI.There are several rules to consider in this activity.In Sweden, there is a law on the protection against hazards (LSO), which states that the municipalities have the responsibility to provide FRS in their geographical area, and that the FRS should be guided by an action program.The action program should include the goals of the FRS, the risks of hazards in the municipality, and the capability of the FRS to manage such hazards.Lastly, the activity is also guided by municipal policies and directives concerning the CI in the municipality.The community consist of all the actors that are involved in the identi cation of CI.Different expert agencies can be involved to support with information or consultation, e.g., SMHI or MSB.Business actors can also be involved, e.g., power companies operating in rivers can support the identi cation with information and water depletion data.The labor of the community is divided based on their respective responsibilities, a liation, and experience of natural hazards.According to action plans and agreements in the impacted area, each actor of the community will have responsibility based on their a liation.The responsibility may also be divided based on experience, for example where a business actor used to a certain type of natural hazard is expected to take larger responsibility for mapping and tracking of the hazard's development.
The outcome of the activity is a prioritization of CI to protect before or during an ongoing hazard.The quality of the outcome is dependent on the interplay between the elements of the activity system.Any contradictions between the elements will affect the outcome and can lead to a breakdown of the activity.For this activity, we have identi ed two contradictions.

Contradiction 1 -Insu cient tools
The rst contradiction exists between the tools and the object.Although the identi cation of CI is challenging and important, the tools to support it are insu cient.It is mainly done through dialog and cooperation between involved and affected actors.For slow-onset events, e.g., river oods, there are often pre-existing checklists to support the identi cation, since the rivers are monitored.However, for rapidonset events, e.g., wild res, the lack of support tools make the analysis more di cult.It gets even more complex in multiple settings of natural hazards, where simultaneous or cascading aspects need to be considered.This calls for a decision support tool that can identify CI under risk by calculating or using already existing hazard development projections.There are already web-based map tools in use by some FRSs, which could potentially be expanded to also include projections of the hazard development given the current risk index values.

Contradiction 2 -Lack of local knowledge in shared C2
The second contradiction exists between the object and the division of labor.A proposition from the government states that all FRSs should strive for a stronger and shared C2 to increase their capabilities for larger events.However, the current identi cation of CI depends on local level knowledge since there are no tools to support the identi cation by an external C2 function.With such a tool, the work for shared C2 functions can continue with decreased risk of losing knowledge about local CI.

Assessment of resource requirements
A central issue in all responses to emergencies, especially natural hazards, is the resources required for an effective response.The requirement can be either static, before a response, or dynamic, during a response.The activity system of the assessment of resource requirements is presented in Fig. 3.
The subject of the activity is the C2 of the responding FRS, as it performs the actual assessment.The object of the activity is the resource requirement for a response to a natural hazard.There are several tools present in the activity.One is the resource management system of the responding FRS, where the available or responding FRS resources are visible.The assessment also requires information about the hazard, in terms of development projections based on forecast and warnings from expert agencies, and what-if analyses done by the FRS or the municipality.The rules that in uence the activity are the FRS's principles, their action plan, and the LSO.Apart from the FRS, there are other actors indirectly involved in the activity, constituting the community.Different expert agencies can support the activity with information and consultation.The municipality can support the FRS in the assessment with their what-if analyses.The community is divided based on their domain of expertise concerning the resources required or the hazard's type and development, as well as their responsibility in the response based on agreements between actors in the activity system.
The outcome of the activity is either a changed static resource requirement to a typical natural hazard, or a changed dynamic resource requirement during an ongoing response.Effects of a changed resource requirement are changes of the preparedness to other natural hazards or emergencies, an important aspect in multiple natural hazards scenarios.One contradiction of the activity was identi ed.

Contradiction 1 -No tools to meet complexity
The contradiction exists between the tools and the object.Resource requirement analysis is a complex task, especially concerning dynamic requirements in an ongoing response, but necessary in a multihazard scenario.Despite the complexity and the future projections of increased natural hazards, the FRS does not currently have su cient technical tools to support the analysis.

National reinforcements
During larger natural events, the responsible FRS may get a shortage of resources.Then, reinforcement resources can be requested from the national resource pools.MSB is the broker of the pools, of which there are three main types especially relevant for natural hazards: wild re, ood, and aerial re ghting.
The wild re equipment includes, e.g., hoses, spreaders, All Terrain Vehicles, chainsaws, and water tanks.
The ood equipment includes barriers, pumps, and sand-lling machines.Available aerial re ghting resources are helicopters and scooping airplanes, procured from several contractors.However, the ying resources are only available during the re season as determined by MSB and SMHI.In the theme national reinforcements, we have identi ed two key activities: the assessment of reinforcement requests, and the allocation and relocation of national reinforcement resources.

Assessment of reinforcement requests
During simultaneous natural hazards, e.g., several wild res, there may be several response actors in need of reinforcements from the national resource pool.If several requests reach MSB, they will assess the needs and decide how to allocate the resources, which until recently solely had to be done based on the needs provided by the requesting actors.However, during 2021, changes in the LSO gave them the mandate to prioritize and distribute available reinforcement resources.The activity system of the assessment of reinforcement requests is presented in Fig. 4.
The subject of this activity is MSB's on-duty o cial (TiB) and directions and priority function (IPF).The TiB receives the requests from the actors in need of reinforcements.The IPF is a group of representatives from different functions within MSB, who jointly take decisions regarding priority and distribution of available reinforcements.The subject actors work on the object, which is the reinforcement requests received.When actors request reinforcements, they must attach a checklist with essential information about their situation that can be used by the IPF in their evaluation.Another tool is the hazard development projections done by SMHI, with information about risks connected to an ongoing incident.
During the request assessment, the IPF needs to consider several rules, primarily the LSO, stating their mandates, and the government directives, guiding the overall work of the agency.Apart from the MSB and the requesting actors, the community in the activity system also constitutes contractors, who will perform some of the reinforcing operations.Regarding the division of labor, the agency function of an IPF representative determines the type of information and expertise the representative can bring to the IPF.
The responsibility of each function will determine the internal responsibility within the IPF and the assessment of reinforcement requests.Furthermore, the division is based on geographical location, since the contractors and FRS organizations operating the resource pools are spread all over Sweden.
The outcome of the assessment activity is the deployment of reinforcement resources to selected FRS organizations.The success of the activity, in terms of providing resources to the actors in most need of aid, is dependent on the IPF's ability to analyze and prioritize between the requests.We have identi ed one contradiction for the assessment activity.

Contradiction 1 -No tools for effective prioritization between requests
The contradiction is connected to the tools used to evaluate the requests.Since the requests concern immediate needs of aid, the evaluations must be quick without failing to identify the most pressing need of support.However, the lack of technology to perform and support the evaluations increases the risk of such failures, either concerning the time for the aid to arrive or the evaluation quality.A tool with technology to support prioritization between requests could make the evaluations quicker and possibly improve the quality of the analysis.The features of this kind of tool could range from functionality that enable comparison between cases, for instance ranking according to a given set of metrics, to functionality that performs the comparison and suggest how the requests should be prioritized.

Allocation and relocation of national reinforcement resources
To increase preparedness, some of the resources in the national resource pools are stationed at prede ned positions all over Sweden.The wild re equipment constitutes 24 depots divided between 14 locations in conjunction with an FRS station.The ooding equipment is centered at one location since it is used for hazards with longer and more predictable onsets.However, with the increased risk and experience of sudden cloudbursts in Sweden, like the one in the city Gävle during 2021, that might have to change.The aerial re ghting resources have a predetermined location with possible preparedness relocation according to the current hazard risk.The activity system of the allocation and relocation is presented in Fig. 5.
The activity subject is the MSB IPF, who takes the decisions about resource allocation and relocation.The objects of the activity are the national reinforcement resources, primarily the ones used during wild res.The motive of the activity is to increase the preparedness, by determining the location of resources.To do this, the IPF have support from the hazard risk assessments done by MSB and SMHI.Also, the IPF itself can be seen as a tool, creating a platform for collaboration between different agency functions.The activity is also guided by rules based on the agency principles, the LSO, and government directives.The community of the activity constitutes the actors involved in, or affected by, the IPF's decision.The FRSs, municipalities, and CAs will have an interest in the decisions since their preparedness for natural hazards will change with the relocation of national resources.The relocation will also affect the contractors, who will need to move the resources and be stationed at another location.As for the assessment of reinforcement requests (Section 4.2.1), the community is divided according to their agency functions, their responsibilities, and their geographical location.
The desirable outcome of the activity is an allocation plan of national reinforcement resources strengthening areas under increased risk of natural hazards.Relocation is mainly done for the aerial re ghting resources.The relocation of resources is dependent on the hazard risk projections done primarily by the SMHI.The uncertainty of the projection will in uence the outcome of the activity.We have identi ed one contradiction in the activity.

Contradiction 1 -No tool to support the complex allocation and relocation
The contradiction identi ed lies between the tools used and the object.The task to determine the best location of resources is complex, with many factors to consider.Despite this complexity, there is currently no technology used for calculations on the best location of the resources.A mathematical optimization approach will probably improve both the preparedness and the task completion time and decrease the need of externalization in the activity system.The main input data of an optimization approach for wild res would be the projections of FWI-values, preferably a few consecutive days.However, these data should be supplemented with other data able to describe the likelihood that a need for national reinforcements will occur (e.g., resource availability, hazard site reachability, CI).The problem could then be solved as an instance of the p-median (Daskin and Maass 2015) or the capacitated facility location problem (Bertsimas and Tsitsiklis 1997).

Resource management
Any ER involves resource management, from resource allocation before the emergency to resource coordination during the response.In Sweden, the main pool of human resources lies within the FRSs.
Their personnel all have the same basic training, to respond to all types of emergencies.Some also have special competences, e.g., high-altitude rescue missions or chemical hazards.The personnel constitute full time, part time, and voluntary resources divided between different stations.In the theme resource management, we have identi ed four key activities: the initial resource deployment, the task planning for aerial re ghting resources, the local area management, and the resource coordination during large natural hazards.

Initial resource deployment
The static resource requirement analysis, see Section 4.2.1, will impact the initial resource deployment to a natural hazard.i.e., the immediate response given an emergency call.The activity system of the deployment is presented in Fig. 6.
In Sweden, the PSAP is operated by a public-owned company, SOS Alarm, by order from the Swedish government.The PSAP is responsible for the initial resource deployment, with possible adjustments by the C2 function of the responding FRS, hence, both actors are subjects in the activity.The object is the immediate resource deployment to a natural hazard.The PSAP and C2 function have several tools at their disposal for the activity.The rst is information from the emergency call, indicating the incidence level and resource requirements for an e cient response.The resource management system can be used to identify available FRS resources for deployment.If there is a shortage of resources, several FRS organizations have an agreement on dynamic resource management.This means the PSAP have the mandate from the involved actors to deploy resources from external FRS organizations.Some FRSs have basic agreements, while others have entered larger coordination clusters with unlimited access to each other's resources and a joint C2 center.The activity is primarily in uenced by three rules: the FRS principles, the LSO, and the contracts set up between the PSAP and involved FRS organizations.One such contract is a tactic called increased resource deployment during periods of high wild re risk, where the PSAP deploys more FRS resources than normal for a strong initial response.The community of the activity constitutes the involved FRSs and the PSAP.The community is divided based on their authority, which determines what kind of decisions each actor in the community is mandated to take, and their experience level of response to natural hazards, which will in uence the success of the resource deployment.
The activity should lead to a su cient initial response to a natural hazard.The outcome of the activity is dependent on the capability of evaluating the resources required for the response.Contradictions between the activity elements can affect this capability and lead to a resource deployment failure, which will have impact on the outcome.For this activity system, we have identi ed two contradictions.
Contradiction 1 -Principles lead to exaggerated deployment The rst contradiction exists between the rules and the object.As mentioned earlier, for a strong initial response during periods of high wild re risk, the PSAP and FRS will use the tactic increased resource deployment.Although this tactic is important to decrease wild re spreading, it will sometimes lead to exaggerated deployment.Such deployment will decrease the preparedness for other potential emergencies, and an increased risk of harm and property damage.
Contradiction 2 -Insu cient tools to determine the initial response The second contradiction exist between the tools and the object.This contradiction is similar to the contradiction concerning resource requirements, however, here the responsible actor is the PSAP rather than the C2 of the FRS.Although the determination of initial response to a natural hazard is complex, there seems to be no suitable technology to support it.

Task planning for aerial re ghter
If a reinforcement request to the MSB leads to deployment of aerial re ghting resources, the aided FRS will be responsible for directives to the resources during the response.The directives include where to do water replenishment and drop off locations.The motive of the activity is to maximize the amount of dropped water, and the e ciency of the drops.The activity system of the task planning is presented in Fig. 7.
The subject of the task planning activity is the C2 of the FRS.They should give su cient directives on drop off and re ll points to the activity object, the aerial re ghting resources.The task planning is supported by hazard maps, visualizing the current incident area, and the location of water replenishment sources.The task planning is constrained by the resource characteristics; for example, a scooping airplane needs larger water replenishment points than helicopters.It is also constrained by the crews of the planes and helicopters, who have a limited operation time based on laws and the contracts between MSB and the contractor.There are several actors constituting the community.First, there is the FRS, as the responsible actor of the activity subject.The MSB will also have interest in the activity, as they provide the activity object (i.e., the resource) and has a supporting role for the FRS.Finally, the contractors are part of the community, as they will be the operating actor executing the planned tasks.The division of labor is based on each actor's a liation, which also determines their responsibilities connected to the task planning (e.g., operations, communications, or command).
The desired outcome of the task planning is su cient response directives for the aerial re ghting resources.The task is dependent on a thorough analysis of the incident surroundings to nd suitable water replenishment points for the resources while combatting the re at the desired location.A de cient analysis can lead to a task planning breakdown, where the proposed plan is impossible to execute for the aerial resources.We identi ed two contradictions of the task planning.
Contradiction 1 -Responsibility leading to unequal opportunities for effective task planning The rst contradiction exists between the division of labor and the object.When an FRS gets aid from aerial reinforcements, they are responsible for providing su cient response directives.However, since the Swedish ER system is decentralized, some FRS organizations have a stronger C2 function than others and will have better opportunities to manage the planning.This leads to an unequal response system, where the reinforcement response will be dependent on the individual FRS's planning capability rather than their need of aid.

Contradiction 2 -No support tools despite task complexity
Task assignment planning can be a complex problem, and the uncertainty and dynamic environment of a wild re incident area certainly makes it even more complex.However, we have not been able to identify any use of planning technology as support to solve the task assignment problems.The use of such technology would improve the task planning, increase the response directive quality, and increase the opportunities to reach the motive of the activity.

Local area management
During large with rapid onsets and high resource demands, like cloudbursts and storms, the FRS and PSAP can agree to depart from standard deployment procedures in a municipality.It means that the backend FRS operator will not listen in on calls and no resources will be deployed for non-life-threatening matters.Instead, these matters are forwarded directly to the C2 center, removed from the response management system, and managed separately using local area management.Since the FRS resources are busy with more urgent responses, no resources are available for non-life-threatening emergencies, like ooded basements or fallen trees.When the FRS resources become available again, the challenge is to sort and prioritize among the reported incidents, which is also the motive of the activity.The activity system of the local area management is presented in Fig. 8.
The subject of the local area management is the C2 of the responding FRS, however, sometimes with support from the municipality.The object of the activity is a limited area with many incidents, leading to a need of local area management.As this type of management is done ad-hoc, the tools to support it varies and include document editors and calculus programs to sort and prioritize between the incidents.The rules that in uence the activity are the FRS principles, stating when and how the activity should be done, the LSO, stating who is responsible during different emergencies, and municipal agreements, stating whether the municipality or the FRS is responsible given different circumstances.The actors with interest in the activity, forming the community, are the FRS organization and the municipality, who may need to be involved in the activity or are affected by the outcome.The labor of these two actors is divided based on their command, which means both who they take orders from and which resources they can give orders to.Furthermore, the labor is divided based on responsibility, which may be regulated in municipal policies and agreements, otherwise the law.
The outcome of the activity is a list of non-life-threatening incidents to respond to.The challenge is to determine the order of the list to get an e cient response in terms of time and the need for aid among the affected actors.Thus, the outcome is dependent on the C2's ability to analyze the information at hand to prioritize and sort among the incoming incidents.We have identi ed two contradictions for the activity.

Contradiction 1 -technology to support incident-intense events
One contradiction exists between the tools and the object.Although digital EM systems are used by most FRS organizations, they are not applicable for incident-intense events like cloudbursts or storms.A tool for automatic prioritization of the non-critical incidents would increase the preparedness for serious events, make the prioritization more robust in terms of identifying incidents in most need of aid, and decrease the workload for the C2 function.

Contradiction 2 -Unclear responsibility during incident-intense events
This contradiction exists between the subject, the community, and the rules of the activity.
According to the Swedish hazard laws, the FRS should respond to an incident if it is classi ed as a hazard.However, many incidents during cloudbursts and storms ends up in a gray area, since they often concern private property with limited risk of casualties.At the same time, the municipalities should have a preparedness for extra-ordinary events, like cloudbursts and storms, which raises the question of who should be responsible for the planning and response to non-life-threatening incidents.If the responsibility is not well de ned between the FRS and the municipality, it leads to a contradiction between the Swedish hazard laws and the subject, which should then be the municipality.Furthermore, this leads to confusion between the subject and the community, since it is unclear who should be responsible for these kinds of events, which may lead to a breakdown affecting the outcome.

Resource coordination during larger natural hazards
Due to decentralized structure of the Swedish ER system, a response to a large natural hazard will, apart from FRSs, often involve other actors, e.g., voluntary organizations, the MSB, municipalities, CAs, and the Swedish Armed forces.The coordination between all responding actors is a challenge and has impact on the quality of the response.
The coordination and cooperation between actors during such responses is often facilitated through a cooperation conference, commonly initiated by the CA or a municipality.Also, when managing natural hazards, response and support actors can activate a state of alert (SOA).The FRS can activate a SOA on different levels: at the emergency site, on local C2 level, and on regional federation C2 level.For larger events, the emergency site can be split into sectors led by an appointed sector leader.This implies that one emergency site can consist of several C2 points.If the event affects several municipalities, involved FRSs may activate a shared SOA with a joint C2, often led by several appointed on-duty rescue chiefs, one from each FRS.The activity system of the coordination between resources is presented in Fig. 9.
The subject of the response coordination is either the C2 of the responsible FRS or the shared C2 between several responding actors.The subject works on the activity object, in this case the responding actors in need of coordination during a response.The activity is supported by information from the established common operational picture (COP), the resource management system of the FRS, and hazard development projections.However, the actual coordination planning is primarily done manually, in document editors or whiteboards.The rules applied to the activity are based on the cooperation agreements between the responding actors, which in turn are based on the Swedish hazard laws and the FRS principles.The community constitutes the responding actors since they will be directly or indirectly affected by the coordination plan.For the division of labor, each actor has different mandates to command and coordinate the resources, which is based on their responsibilities stated by laws and agreements.Also, the level of an actor's expertise of a speci c hazard type or the environment will determine the amount of in uence the actor will have in the coordination.As the responding actors also includes volunteers, the division of labor is also based on professionality.
The outcome of the activity is a coordination plan guiding the response and all involved actors.The motive behind the activity is to increase the coordination, hence, the overall response capability during larger responses.A de cient coordination will decrease the capability and put involved actors and the affected community under unnecessary risk, which could lead to more casualties and damage.We identi ed two contradictions connected to technology.

Contradiction 1 -Lack of technology for coordination of ER
The rst contradiction concerns the tools and the object.Larger emergencies involve more actors in the response and increase the need of coordination, especially in a decentralized ER system where cooperation is a necessity rather than an option.However, we have not been able to identify any technology in use that can support the coordination, which makes it time-consuming and complex.A tool that supports even just parts of the coordination, e.g., the establishment of a COP, would decrease the workload of the C2 and make the coordination more manageable.However, a shared C2 system among all FRS organization would enable better cooperation and coordination during large-scale natural hazards.
Contradiction 2 -Utilization of reinforcements is limited due to a lack of tools and procedures The second contradiction lies between the community, the object, and the tools.During the interviews, an issue was brought up concerning the use of volunteers during natural hazards.Some FRS organizations were sometimes forced to refrain from voluntary assistance due to limited capability to organize volunteers.At the same time, the Swedish ER system are dependent on volunteers and will be even more dependent on them in the future.This illuminates a need to integrate external resources into the existing resource management systems and create appropriate routines and procedures for managing the volunteers.This would give increased opportunities to use the voluntary assistance that is offered.According to the current climate projections, most countries, including Sweden, will experience increased frequency and magnitude of natural hazards.Hence, response resources could become even more scarce in the future, why it is important to do detailed analyses about both the static and dynamic resource requirements, as well as the resource allocation, relocation, and coordination.However, our analysis of activities concerning national reinforcements and resource management identi ed contradictions concerning the non-existing tools to support them.A breakdown of these activity systems would imply that the requirement analysis fails.Too few responding resources would be unable to manage the hazard, which could lead to more casualties and property damages.On the other hand, sending too many resources, the ER system would have a decreased preparedness for other hazards.A tool to support these activities would enable a more precise requirement analysis and subsequent resource deployment, leading to increased balance between deployment and preparedness.A future increase of natural hazards, both in terms of frequency and magnitude, will lead to larger ERs, which will require increased cooperation and coordination between actors.Our analysis of resource coordination during responses to larger natural hazards suggest that the Swedish ER system sometimes refrain from using voluntary resources due to a lack of capability to organize them.This is aligned with the ndings of (Trias and Cook 2021) in their study of the Indonesian ER system.They suggest that problems in emergency operations during the Central Sulawesi earthquake were not due to resource capacity limits, but rather structural problems that limited the inclusion of local and regional actors.According to Miao et al. (2013), the issues of inclusion and emergency resource coordination can be enhanced through comprehensive information systems with cross-network integration between different actors.This in combination with planning and operational control tools (Pesigan and Geroy 2009) will enable easier dissemination of information between actors before and during responses, and enable more inclusion of non-traditional and voluntary resources.
The results illuminate a lack of tools to support planning and decision-making during ER to both single and multiple natural hazards.This includes even fundamental activities, for instance resource deployment and coordination of responses.There might be several reasons why this is the case.One of them is limited budgets due to the decentralization of the Swedish ER system, which results in separate budgets of each response actor.Although decentralization is bene cial for other reasons, the separate budgets for each response actor makes it di cult to afford acquisition, implementation, and maintenance of decision support tools.More efforts are needed to nd common grounds between several response actors, and jointly invest in shared decision support solutions.Another reason is user acceptance, which can constitute and obstacle for technological solutions (Lu and Xue 2016).The collected interview data indicate a reluctancy to use tools with too narrow purposes, for instance focusing on only one natural hazard type.However, separate tools for different hazards are a necessity to grasp the individual characteristics of the hazard types.Therefore, more effort is needed to implement modularized platforms which includes several tools for different hazards, which will hopefully increase the user acceptance.

Study transferability and limitations
The empirical sample of the 12 interviewees covers long-term experience of response to natural hazards and several roles in response organizations.The sample is also spread over several parts of Sweden with different characteristics and challenges connected to natural hazards, making it su cient to study the needs for improved decision support.A larger sample would arguably give more detailed insights on ER to natural hazards, but for a holistic perspective, the current sample was considered su cient.
Activity theory was deemed useful as a modelling and analysis tool, and although the study case is the Swedish ER system, other countries can gain insights from this study, since similar contradictions and gaps will likely lead to similar breakdowns.However, studies into the needs of other countries for improved response to natural hazards are encouraged, as it would provide deeper insights and enable comparison between cases.In this study we were interested in the lack of tools which increases the need of externalization in the key activities.Consequently, we did not identify all potential contradictions for the activities but focused on the ones concerning our study aim and scope.

Conclusions
In this paper, planning and decision activities in the Swedish emergency response (ER) system were identi ed and analyzed using activity theory.In a society where hazards will likely increase, not the least due to climate changes, the results show a lack of technology to support planning and decision-making during ER to both single and multiple natural hazards.Even fundamental activities, e.g., resource deployment and coordination of responses, are mostly done manually through document editors, or even verbally.The use of technology is not only a valuable support for decisions and analyses.It can also provide situational understanding and insights between response actors, with the potential to enhance and develop cooperation, coordination, and integration in the response system.The ndings in this paper can inform policy makers and managers in ER systems of challenges and needs to expect due to increased frequency and severity of natural hazards.The ndings can also inform the scienti c community of future research directions and encourage nding solutions to the identi ed needs.At a theoretical level, the paper shows how activity theory can be applied to identify contradictions in activities of ER, which could be used by scientists to conduct similar studies in other settings, and by practitioners in their improvement work.
An extension of this work, given the current lack of decision support in the Swedish ER system, is to develop tools covering parts of the identi ed needs.The future research plan includes prototype development of models and support tools for resource planning during responses to single and multiple natural hazards.

Declarations Figures
A general structure of an activity as described by Engeström (1999) Page 28/31 The activity system of the identi cation of critical infrastructure.The red arrows between the tools and the object, and the division of labor and the object, respectively, indicate identi ed contradictions between the activity elements Figure 3 The activity system of the assessment of resource requirements.The red arrow between the tools and the object indicates an identi ed contradiction between the activity elements Page 29/31 The activity system of the assessment of reinforcement requests.The red arrow between the tools and the object indicates an identi ed contradiction between the activity elements Figure 5 The activity system of the allocation and relocation of national reinforcement resources.The red arrow between the tools and the object indicates an identi ed contradiction between the activity elements Figure 6 The activity system of the initial resource deployment.The red arrows between the tools and the object, and the rules and the object, respectively, indicate identi ed contradictions between the activity elements The activity system of the task planning of aerial re ghting resources.The red arrows between the tools and the object, and the division of labor and the object, respectively, indicate identi ed contradictions between the activity elements Figure 8 The activity system of the local area management.The red arrows between the tools and the object, the subject, the community, and the rules, indicate identi ed contradictions between the activity elements

Table 3
The contradictions identi ed in the studied activities The analysis identi ed a contradiction in the identi cation of CI, where currently no tools are used that matches the complexity of natural hazards with faster onsets (e.g., wild res).A consequence is the dependency on human mental capacity to do the complex identi cation, which requires externalization and increases the risk of an activity breakdown.A breakdown in this case would mean that the C2 of the responding FRS, due to limited mental capacity, fails to identify CI under risk.Beside the increased risk of damage on the infrastructure, such a failure could potentially have cascading effects leading to human suffering and high cost due to disruptions of public services.Another important reason for using tools in this activity is to enable external command from another C2 center, without losing the local knowledge Guth et al. (2019)8;Fang et al. 2019vious studies developing tools for the identi cation of CI (e.g.,Lagadec et al. 2018;Fang et al. 2019), most of them concern vulnerability and resilience assessments for mitigation purposes, rather than the immediate identi cation of risk during ongoing responses.However,Guth et al. (2019)presents a promising concept to analyze the accessibility of emergency facilities during natural hazards, which could also be used for the identi cation of CI under risk.
While there are several previous studies on static resource requirements and allocation (e.g., Liu et al. 2012), and resource deployment (e.g., Amorim et al. 2018), less is done concerning the dynamic resource requirements during an ongoing event, where the study by Duan et al. (2020) is a good starting point for further research.