Safer and resilient schools in seismic regions: a systems perspective

ABSTRACT Sustainability of school infrastructure and resilience of educational communities to natural hazards are of paramount importance to provide the safety and protection against various natural and man-made threats children face. The 2015 Nepal earthquake and its impact on schools confirms this approach at multiple levels, since schools are built as structures, supported as financial entities by the state, and function as social clusters within each community. In the literature, there are various guidelines and frameworks for safety and resilience of schools. Notable amongst these, the UNISDR’s Comprehensive School Safety Framework consists of three pillars including safe school facilities, school disaster management and risk reduction education. A systems perspective requires a holistic approach to resilience of schools incorporating the safety of school buildings, governance and funding, provision of supporting infrastructure, school curriculum and more importantly, the resourcefulness of educational communities. The purpose of this paper is two-fold: first, to present a quantifiable, holistic framework for the resilience of schools, and second, to demonstrate how the framework is rooted in a systems thinking perspective. The novelty of the proposed framework lies in resilience assessment using a participatory, multi-disciplinary and holistic approach tailored to the salient features of a low-income country. This work is part of a Special Issue on Systems Perspectives: Clarity through Examples (see Dias 2023).


Introduction
Children are the future of a country, and their education is a priority for any government.Hence, appropriate policy measures are taken to create and maintain the education infrastructure.The local authorities usually take responsibility for the primary and secondary schools in their area.This includes, amongst others, ensuring enough classrooms, qualified teachers and safe buildings.The safety of school buildings is of most concern to civil engineers.However, there are many examples of school buildings being damaged due to earthquakes or becoming inaccessible due to flooding, thus affecting education.As an example, Gorkha (Nepal) earthquake in 2015 damaged over 19,000 classrooms and displaced an estimated 3.2 million children (Government of Nepal -National Planning Commission 2015).While school buildings may be assessed for their safety, or emergency funds may be made available for the repair of damaged school buildings, more fundamental questions are: 'what are the various aspects of school functionality' and 'how can it be affected due to any internal or external threats?'.Taking a humancentric view (UNICEF n.d.), the purpose of a school is to provide uninterrupted quality education in a safe environment.Hence, all attempts should be made to mitigate the potential impacts of various threats on education.Lessons from various disasters (e.g.Ranghieri and Ishiwatari 2014) show that communities' role is instrumental in times of threats as it helps them coordinate their physical and human components towards mitigation efforts.
In this context, we developed, as part of SAFER Nepal project (safernepal.net),a holistic approach for the resilience of schools that includes various aspects from physical infrastructure to educational communities themselves (Agarwal et al. 2023;Parajuli et al. 2020).The purpose of this paper is two-fold: first, to present a quantifiable, holistic framework for the resilience of schools to natural hazards, and second, to use the Nepal case study to demonstrate the underlying systems concepts.The novelty of the proposed framework lies in resilience assessment using a participatory, multi-disciplinary and holistic approach tailored to the salient features of a low-income country.The existing efforts are too focused on school infrastructure or refer to resilience frameworks that are applicable in developed countries.To the best of our knowledge, it is the first time that such a comprehensive, yet quantitative approach has been developed and applied in this context.
The paper is organised as follows: Section 2 provides a literature review on resilience with a particular emphasis on school communities; Section 3 presents the SAFER framework including the methodology, dimensions and indicators, resilience assessment process, and an example; Sections 4 demonstrates how the framework is rooted in systems thinking; Section 5 provides a discussion on the framework before concluding the paper in Section 6.

Resilience
The term resilience has been used in many ways in different disciplines.For example, in ecology it refers to the persistence of systems to external actions and their ability to absorb changes including finding a new dynamical state (Holling 1973); in psychology it refers to an individual's capacity to cope with significant adverse change (Leipold and Greve 2009); and in physics it refers to an object's ability to return to its original state after deformation.In the context of infrastructures, Bruneau et al. (2003) related resilience to the ability of a system to absorb a shock and recover from it quickly.ISO (2009) defines resilience as the adaptive capacity of organisations in a changing environment.Rose (2004) in the context of economic systems and Godschalk (2003) that for social systems, noted the key features of resilience as an ability of the system to withstand the external influences and to recover from adverse state.UNISDR ( 2009) provided a more inclusive definition of resilience as the ability of a system, community or society exposed to hazards to resist, absorb, accommodate, adapt to, transform and recover from the effects of a hazard in a timely and efficient manner, including through the preservation and restoration of its essential basic structures and functions through risk management.
There are many more interpretations of resilience and a review of these definitions can be found in Hosseini, Barker, and Ramirez-Marquez (2016).
Figure 1 illustrates a conceptual representation of resilience highlighting two key aspects: the degradation in functionality after a disruptive event and the rapidity of recovery.Each of these has a threshold of acceptability but these cannot be uniquely defined in general, rather these depend upon the societal context and may vary with time.Recovery operations may lead to three possible system states: its original state, a better state often referred to as 'build back better' or partial recovery.The resilience can be improved either by reducing the impact (e.g. through hardening the system) or by a more efficient recovery scheme (e.g. by advance planning and resourcing).An assessment of resilience is important for (i) prioritising the allocation of limited resources and (ii) monitor improvements in resilience following any interventions.However, the main challenge is its quantification for decision-making purposes (Didier et al. 2018;Heresi and Miranda 2017;Kilanitis and Sextos 2019) because of the difficulty in assessing the impact due to the complexity and interdependency of the affected systems.Bruneau et al. (2003) proposed robustness, rapidity, redundancy and resourcefulness as four aspects to measure resilience.The UK government (Cabinet Office 2010) identified resistance, reliability, redundancy and response & recovery as four principal components to increase infrastructure resilience.Robustness was noted by Blockley, Agarwal, and Godfrey (2012) as a necessary but not sufficient condition for resilience.Noting the difficulties in quantifying robustness for any system, Agarwal (2015) argued for the identification and management of vulnerabilities to improve the resilience of a system.The presence of redundancies helps to eliminate the corresponding vulnerabilities also.Similarly, the rapidity of recovery is a function of the available resources.The concept of resourcefulness is in fact a richer concept and includes ingenuity and innovation during times of crisis.Hollnagel, Woods, and Leveson (2006) proposed resilience engineering as a way of thinking about safety to enhance the ability at all levels of organisations and to use resources proactively.Thus, it is difficult to arrive at an encompassing quantitative measure of resilience.Still, many resilience metrics have been suggested in the literature (see Hosseini, Barker, and Ramirez-Marquez (2016) for a review) which have tended to be proxies based on some discipline-specific performance measures.For example, Henry and Ramirez-Marquez (2012) proposed a generic time-dependent resilience metric as the ratio of recovery to loss and related it to reliability, vulnerability and recoverability.Francis and Bekera (2014) proposed a dynamic resilience metric which uses (a) the ratio of performance level immediately following the adverse event to the original state, (b) the ratio of performance level after recovery effort to that before recovery and (c) the speed of recovery.Cimellaro, Reinhorn, and Bruneau (2010) used measures of the quality of service before and after an event to quantify resilience.

Community resilience
A community is a group of individuals with shared interests and activities (Fellin 1995).Community resilience then refers to the potential of the group of individuals to absorb shock, recover from it and support new growth (Ungar 2011).Magis (2010) defined community resilience as the 'existence, development and engagement of community resources by community members to thrive in an environment characterized by change, uncertainty, unpredictability and surprise'.Berkes and Ross (2013) noted that the following characteristics of a community play important roles in developing community resilience: community infrastructure; knowledge, skills and learning; values and beliefs; social networks; people-place connections; engaged governance; a diverse and innovative economy; leadership; and a readiness to accept change.Elms (2015) proposed two complementary aspects for community resilience: the flows connecting the community with its surrounding environment and the resources a community needs.The latter can be grouped in human (such as their ability to work, health and knowledge), natural (such as land and water), physical (such as housing, energy and transport) and financial (such as income and savings).Elms (2015) opined that the body of required resources is complex with many interactions and a systems approach is required to understand the issues.Further, the difficulty in defining what forms a viable community makes it difficult to quantify resilience, apart from some selected aspects as proxies.IFRC framework [2016] fosters community resilience through participation, connection and partnerships.This has been broken down in distinct stages: national/state level engagement and connectivity through small groups, consultation with the whole community to understand risk and resilience, preparing a plan to strengthen the community, continuous learning, and assessment and reflections on successes and failures with the involvement of all community members.Sharifi (2016) reviewed several tools for assessing community resilience using the following six criteria: dimensions of resilience, cross-scale relationships, temporal dynamism, uncertainties, participatory approaches and action plans.They concluded that several challenges remain in addressing community resilience including consideration of the environment, better stakeholder participation, dynamics over time and across space, and uncertainties.

Resilience of schools
There are many threats to the continuity of education -some internal to a school (e.g.shortage of teachers) and others external (e.g.natural hazards).Some of these may occur more frequently (e.g.heavy rain) and others rarely during the life cycle of a school building (e.g.earthquake).However, a school needs to ensure continuity of education with minimum disruption while ensuring the safety of children.Hyogo Framework for Action 2005-2015 (UNISDR 2005) recognised the role of schools in building resilience of communities to disasters and launched various initiatives.As part of this, United Nations Centre for Regional Development produced its report (UNCRD 2009) on school retrofitting, disaster education, capacity building and raising awareness.IFC 'Disaster and Emergency' handbook (IFC 2010) emphasises school resilience through assessment and planning, physical protection and risk reduction, good practices towards capacity development, and processes for monitoring and improving.Further, it identifies school administrators and teachers for policy development and safety implementation.This regards the concept of resilience as a continuous process.A UNDP report (Winderl 2014) advocates the following elements in the measurement of disaster resilience: wellbeing before and after disaster, vulnerability, capacities, disaster-related losses and reaction to and recovery after disaster.OECD Guidelines on Earthquake Safety in Schools (OECD 2017) focus on (a) seismic safety policy, (b) accountability, (c) building codes and enforcement, (d) training and qualification, (e) preparedness and planning, (f) community awareness and participation and (g) risk reduction in schools.These also recognise the role the governments, policy makers, construction professionals, headteachers and teachers can play in designing policy and practice.Comprehensive School Safety framework (UNISDR 2017) was built upon three 'pillars' of (i) safe learning facilities, (ii) school disaster management and (iii) risk reduction and resilience education.These primarily relate to quality of school buildings, preparedness against hazards and risk mitigation measures, and increasing awareness of risks.This framework has recently been significantly revised and includes a cross-cutting 'foundation' consisting of enabling systems and policies (GADRRRES 2022).FEMA (2017) provides guidelines for improving school safety against natural hazards in the United States.A more recent study (Opabola et al. 2023) provides recommendations to foster a resilient school infrastructure drawing lessons from the recovery of earthquake affected communities in Central Sulawesi, Indonesia.Thus, it is evident that most school safety and resilience frameworks focus on school buildings and retrofitting, hazard awareness and contingency planning despite the research on community resilience.The SAFER framework, presented in this paper, has addressed the need to enhance school community resilience through a participatory, multi-disciplinary and holistic approach.At the same time, it has resulted in a tool which can be used by the school communities themselves to monitor their progress towards becoming resilient.

Methodology
As noted in the literature review section, resilience of a school requires a consideration of several aspects including the school building, provision of supporting infrastructure, school governance, availability of resources as well as school community's ability to recover school functionality following a disruption.Building upon this knowledge, we worked with the school communities including the governing bodies through a series of engagement activities to develop a comprehensive set of indicators of school resilience and to produce the corresponding evidence-based measures.
Nepal, the country where we developed this framework, is divided into provinces and each province has several municipalities which are responsible for schools in their area.Schools closely follow the government guidelines and School Improvement Plan (SIP) is an important part of their interaction with the local municipality.Within a school setting, five categories of stakeholders were identifiedthese are School Management Committee (SMC), head teacher, teachers, pupils and parents/guardians (Figure 2).They have an interest in most aspects of the school including the provision of a safe environment, the quality of education and the wellbeing of the children.SMC runs the school and has an overview of the school's interaction with the parents/guardians and the wider community.An elected ward member or a community leader often acts as the Chair of the SMC.Following previous research (Fellin 1995), we define these stakeholders as a school community -they identify with one another, have a common culture and participate in shared activities.
School community's knowledge and experiences were gathered through questionnaires as well as direct engagement with them during field visits.A multi-stage process was followed, each stage incorporating feedback and findings from the previous one.The first stage consisted of visits to four schools followed by responses from over 60 schools, representing different settingsrural/urban and hills/plains, to a questionnaire with space for free-text responses.This questionnaire was aimed at gaining an understanding of the existing schools and identifying the challenges that may be encountered during recovery after hazardous events.The questions were arranged under the following headings: school building, governance and funding, use of school, travel to school, shocks, stresses, utilities, and preparedness and mitigation measures.This information was supplemented by classroom activities for children, meetings with school staff and parents, and themed discussions with members of school management committee.Activities with children were focused on identification of hazards, consequences and protection measures.Discussions with staff and parents were mainly around threats to education and potential mitigation measures.Discussions with school management covered additional aspects such as school resources and learning from past events.The school visits, engagement with the school communities and questionnaire surveys were arranged with participants' agreement and supported by our project partner Save the Children (STC) who have immense experience of engaging with the communities in local language and at the time (2017-2020) were supporting safer school construction programme in Nepal.Several potential variables for resilience of schools were identified using thematic analysis which enabled the project team to prepare a preliminary school resilience framework.
During the second stage, questions were arranged under the following headings: school and buildings; governance and funding (with emphasis on inclusivity in governance); social cohesion (including community culture and social connectivity); use of schools (including redundancy and exposure); shocks (including awareness of and recovery during past events); stresses (including environmental conditions), supporting infrastructure; preparedness, mitigation measures and training; resourcefulness and access; flexibility and creativity.Each question was designed with a possible set of responses for consistency and ease of data processing.This questionnaire was trialled in eight schools which led to a refined set of questions and/or response options.Discussions with the school management committee members informed the stakeholders' contribution factors for use in the data-processing algorithm.A consultation with the practitioners in international development was also undertaken to benchmark the outputs for selected schools.
During the third phase, the questions were arranged for ease of flow rather than being grouped according to the resilience indicator these gather evidence for.A few more questions were included to provide context for the subsequent questions, but these do not contribute to any indicator.This set of questionnaires tailored for different stakeholders, was built into a mobile application, offering both languages Nepalese and English, for data collection and automated processing.This has been used by over 150 schools in Nepal.

Framework dimensions
Figure 3 presents the SAFER framework for resilience of schools.At the level of a school, we arrived at four dimensions (inner circle) viz.infrastructure & environment, governance and funding, school curriculum and school community.Associated with each dimension are several indicators (outer circle) which contribute to the school resilience against a range of known and unknown hazards.These are determined based on school community's responses to a set of questions.The next few sections discuss the dimensions first and then the assessment process.

Infrastructure and environment
School buildings are one of the main components of the educational community and having information about their vulnerability to different hazards is important for resilience assessment and for undertaking any preparedness measures.Typical assessments of school buildings tend to be limited to visual surveys, often for maintenance needs, but preparedness for hazards such as an earthquake requires additional building characteristics including structural typology, building materials, design and construction quality, maintenance regime and current state of the building.SAFER project has developed a separate tool (Parajuli et al. 2022;Sextos and Mason 2021) to facilitate a more detailed structural assessment through a model of loads and material capacities but is not used here for consistency of detail with other indicators.
In addition, schools need to have essential services such as water and sanitation, electricity, communication networks, continuous access to resources, appropriate means to protect from severe cold or heat, etc.In many cases, these will be a part of the bigger systems to draw services from.However, these may be standalone services in remote areas or in developing states.
Having a school building is only good if it can be easily accessed by pupils and teachers.This may depend upon the size of the school catchment, the terrain and the means of transport to get to the school.Thus, a school's location in the relevant transport network becomes important.
Any extreme environmental conditions that can pose risks to the continuity of education should be considered and the available measures to mitigate their impact should be examined.These measures need not be limited to one-off major events (such as flooding or earthquakes) but smaller events (for example, heavy storm, noise pollution or public protests) with a potential to cause disruption to education are included.

Governance and funding
Good governance can make a huge difference to the quality of education and in dealing with threats from a range of sources.This may be reflected in, for example, having clear line of responsibilities, school improvement plans and efficient use of available resources.A participatory approach where all the school stakeholders are enabled to feedback their experiences and also contribute to school improvement plans is the necessary and advisable course of action.
The resources in this context are not limited to the funds provided by the local government, but also include school's ability to raise additional funds or the availability of volunteers to help with school activities in normal times as well as after a hazardous event.The availability of adequate funding will normally ensure that children can be taught in reasonable size groups and any teacher absences can be covered without damaging the quality of education.
The school governing body should have contingency plans, i.e. they should have an awareness of all hazards local to their school area and be able to plan well ahead for these.These may come from different sources and having a focal person could help in preparing relevant plans and training pupils and staff against these.
Schools may organise drills for fire, flood or earthquake ground shaking, for example, with due consideration of any special needs.In doing so, the idea is to reduce the scale of impact of hazards and increase the rapidity of recovery as a means to both prepare and act after a shock.

School curriculum
A rich curriculum which addresses the needs of all pupils and prepares them for different hazards will improve their resilience.For example, various safety drills may be embedded in the curriculum rather than forming standalone activities.An exposure to different modes of delivery of education and a range of extracurricular activities will further prepare pupils for the challenges they may experience during shocks and stresses.

School community
The school community consists of actors identified in Figure 2 with the common interest being effective educational activities.Good working relationships amongst pupils, teachers, parents and community members will lead to a supportive learning environment.The more resourceful this community is, the more resilient the school will be.Such resourcefulness can appear in different forms, for example, their socio-economic state, health and wellbeing, rich culture and social bonding.Various researchers and organisations (for example , Fellin 1995;IFRC 2016;Koliou et al. 2020) emphasised a community's social capital and culturally embedded patterns of interdependence for its potential to recover from dramatic change, sustain its adaptability and support new growth integrating the lessons learned during a time of crisis.

Resilience assessment process
As mentioned earlier, a set of questions have been prepared for gathering evidence from the school communities (i.e. the stakeholders) corresponding to different resilience indicators.Each question is presented with possible responses and a few examples are shown in Figure 4.This not only helps in data processing but also provides an objective scale to the respondents.Figure 5 presents a flowchart to determine the resilience metrics.Qualitative responses from school community members are quantified using a pre-defined scale and aggregated for each category of school community.These are then used together with weighting factors to determine the resilience indicator scores and an overall resilience score.The weighting factors were pre-determined through consensus amongst experts.Note that the overall resilience score is not simply an average of individual resilience indicator score rather it takes into account the importance and quality of the evidence.
The resilience framework outlined above has been transformed into a mobile application and a web application for illustrating the data harvested on site and plot useful GIS-based statistics and reports.The mobile application enables the school participants to feed responses to the questions and the web application facilities the processing and presentation of the outputs at chosen level of geographical description i.e. individual school or schools in an area.Figure 6 presents the resilience assessment for a school in western Nepal and Figure 7 illustrates the assessment outcomes for schools in a western region.These results can be used to make improvement plans for individual schools or take policy decision at a regional level.

A systems perspective of SAFER school framework
There is a growing body of literature on systems, systems thinking and systems approaches (Dias and Jowitt 2020;Godfrey, Agarwal, and Dias 2010;Jowitt 2010).These have informed  the development of the SAFER school resilience framework presented in the previous section and the following sections illustrate many of these concepts.

System definition and state variables
A system is regarded as composed of a set of interacting components, objects or processes (Blockley 2020;Carmichael 2020).In this sense, a school has various interacting components, such as school building, teaching resources, sports facilities, water and sanitation services, teachers, pupils, governing bodies, health nurses, etc. (Figure 8).Some of these components are physical and some human.Many of these components are also a part of other systemsfor example, school building can be considered as a part of a building stock in a town, similarly, school children are a part of the local community.Hence, it is useful to define a boundary of school to clarify about what is within the school system and what is not.The introduction of this boundary means interactions occur across the boundary.Examples include incoming funding from the municipal government to maintain the school building, and pupils using their learning to contribute to their community.
One important input across a school boundary is the environmental action on the school as a whole or on its components.This may be in the form of hot or cold weather, storm, lightening or more severe actions such as floods or earthquakes.System behaviour emerges in response to such inputs and through the interaction of system components.Inputs such as earthquakes have huge potential to affect the function of the school and in SAFER, our focus has been on the system's ability to maintain its function when facing such disturbances or if degraded, recover quickly including through adaptation or transformation if necessary.
In systems language, it is useful to think in terms of attributes of the system and its components.In other words, these are state variables which get updated in response to external inputs or any internal changes.The school resilience framework (Figure 2) is a model to represent these attributes at this level of definition and how these may change in response to inputs.As an example, the typology of a school building is an important attribute for its capacity to respond to environmental inputs.The state of preparedness of school to deal with hazardous inputs is another attribute and the evidence to determine its state is gathered through a series of questions, some of these are shown in Figure 9.

Hard and soft systems
The SAFER framework integrates both hard (physical) and soft (human) system components.Examples of hard systems include school building, teaching resources various supporting infrastructure such as water supply, electricity, transport, etc.Many of these components are part of a bigger network system, for example, water supply network with only a distribution node present in the school.Such components rely on the continued functionality of corresponding networks.In other words, the state of such components cannot be determined without the interactions.However, a school might have an off-grid component, for example, a borehole to pump water, thus not requiring interactions with other systems.
School teachers and pupils form the two main categories of soft systems.Figure 3 shows a model of interactions amongst the different soft system components (i.e. the school-level communities).The ones external to the system viz.local government and state government actors are part of the bigger education system.The degree and quality of interactions amongst soft components are evidenced through a series of questions as shown in Figure 10.Some of the system state variables are directly or indirectly influenced by the local communities the school serves, for example, knowledge about historical natural events or disaster risk reduction plans in the local communities.

Processes and interactions
There is a strong thrust in the SAFER framework on the processes associated with the hard and soft system components.We seek the state of the processes or their outcomes as part of the evidence to inform resilience indicators.For example, we look for answers to 'are the classrooms big enough to accommodate all the pupils?' rather than counting the number of classrooms.Similarly, we examine 'are pupils able to get to school within a defined duration?' rather than looking for mode of transport to get to school.For pupils' wellbeing, processes for health checkup, lunch arrangements in school, and their feelings are considered.Socioeconomic state of the school community is gleaned from donations to school on one hand and the occurrence of theft and violence on the other.Another example is the school management's awareness of the hazards and their capacity to integrate these hazards in the school governance processes and in the school curriculum.These examples also point to the interactions between the soft systems and hard systems components.

Hierarchical description
Figure 11 shows the hierarchy of a country's education system a school is an integral part of.At the core of this hierarchy lies a school with its various components and several such schools can be found in a local municipality.They all receive funding from the local government to whom they report.Local authorities receive their funding from the provincial and central government.At the same time, schools are expected to follow the policies formulated at the central or local government level.On the one hand, the resilience of these schools can be derived at various levels (as shown in Figure 7), although the focus has been at the level of a school (the inner-most region in Figure 11).On the other hand, to respond to the outcome of resilience assessment, it may be necessary to take up relevant actions including policy interventions at different levels.

Complexity and uncertainty
There is a great deal of uncertainty associated with complex socio-technical systems but the decisions must be made, in our case to make them resilient.The SAFER resilience framework has several indicators which are determined based on evidence collected through a series of qualitative and quantitative questions.It is worth highlighting that these variables encompass attitudes, skills and knowledge (Jowitt 2020) of the school community members as well as cultural, historic, social, economic, governance and business aspects (Masterton and Jeffrey 2020).It was recognised that the data gathered from school communities may be imprecise and incomplete.Hence, likely responses were identified through early engagements with communities, and these were used in the developed tool.As mentioned earlier, a quantitative assessment of structural health is possible, but resilience score is derived using the structural typology and evidence relating to design, construction and current state to maintain a consistent level of crudeness.
Figure 5 presents a schematic of the algorithm to determine the system state in the form of values of resilience indicators.This brings together evidence in support of resilience as well as against it.Some of it is objective but other is prone to subjectivity.An example of objective evidence is whether evacuation drills are organised or not.However, responses to questions relating to sufficient teaching and learning resources in the school may differ amongst different community members.This is a cause of uncertainty in the resilience scores and two mechanisms have been built in to address this divergence in responses.First, each question has descriptive response options (as in Figure 4) to enable a share understanding amongst the participants.Second, this can be reduced by increasing the number of participants where possible and using a mean or median value.

Discussion
From a disaster resilience perspective, schools and governing bodies face a number of challenges.The SAFER framework addresses this need by following a holistic approach.The previous section was aimed at clarifying the underlying systems aspects, which include many of the aspects appearing in the civil engineering systemsbody of knowledge (Dias and Jowitt 2020).On one spectrum, we present this as a 'framework' (Carmichael 2020) and on the other it is also a 'stance' (Elms 2020).The school resilience, as presented here, truly reflects the hard system embedded in soft ones (Blockley 2013).Whilst the school buildings are the main physical infrastructure in our case study, the schools can't achieve their functionality without the efforts of teachers, pupils, local communities and governing bodies.
Resilience is necessary for sustainability (Blockley, Agarwal, and Godfrey 2012) and it is changing with time.It is a dynamic process in modern as well in ancient communities (Xanthou 2018).Many of the questions in our resilience assessment rely on the notion of process (Blockley 2020).Although we have not attempted to create a 'true' process model here, our resilience assessment model is still capable of enhancing learning through feedback while tackling uncertainty.As noted in the literature, 'dealing with complexity and uncertainty are important features of systems approaches' (Dias and Jowitt 2020), the case study has highlighted the complex nature of the interactions and uncertainty associated with data collected for the purpose of resilience assessment.Some of the data contributes to more than one resilience indicators, so the individual indicator scores and the overall resilience score have their own merits.
The engagement with several schools suggests that a systems perspective has enabled the school communities and local governance bodies to increase awareness of various aspects likely to cause disruption to education not just during extreme events but during normal times also.These include, for example, communication amongst school community, spare teaching capacity, school meal arrangements, supporting less welloff pupils, transport routes to schools, etc.The need to have plans for alternative teaching space, richer curriculum enabling hazard preparedness and different teaching methods during extreme events was also evident.
Schools are the places where the data will come from, but the outputs will inform actions, for the benefit of schools, at multiple levels: schools themselves and the governing bodies at local, provincial and national levels.Thus, schools (i.e. the community through the headteacher and the School Management Committee) are the primary end-users.The simplicity of the assessment with the mobile app makes it suitable for 'self-assessment' (i.e.receiving feedback following the assessment thus permitting periodic evaluation of resilience metrics).The latter, being the first of its kind in such a context is and this is seen as an additional outcome since it empowers the local school community to assess and improve its resilience metrics in time.It also creates awareness to students, parents and teachers about potential hazards and how their individual and collective resilience can enhance by simple measures that are provided by the app upon assessment completion.The process of school communities using the tool to do self-assessment, learn from it and make changes before self-assessing again, will thus help create learning communities (Senge 2006).There are opportunities for the use of big data or crowd-sourced data (Sun, Bocchini, and Davison 2020) but that will require advanced skills for data processing.

Conclusion
There is a growing body of literature on resilience, but it has tended to be in specific contexts, for example, disaster resilience, infrastructure resilience, community resilience, etc. Further, low-income countries need resilience frameworks tailored to their salient features.This paper has presented a conceptual framework and a practical tool for resilience of schools from a multi-disciplinary perspective where school infrastructure and environment, hazard preparedness and governance, school community's resourcefulness, and school curriculum's richness are analysed under the same lens.Using a participatory approach, resilience is assessed based on the evidence gathered through a set of carefully designed questions.It also creates awareness amongst students, parents and teachers about potential hazards and how their individual and collective resilience can enhance by simple measures that are provided by the app upon assessment completion.The outputs can be used by the schools and the governing bodies to make policy decisions or interventions to enhance school resilience.The paper has also highlighted the underlying systems-thinking approach in arriving at the resilience framework.A school has many components, both physical and human, which interact amongst themselves as well with other systems for the purpose of uninterrupted education in a safe environment.

Figure 1 .
Figure 1.Conceptual representation of resilience through reduced impact and faster recovery.

Figure 2 .
Figure 2. A representation of the school community.

Figure 3 .
Figure 3. SAFER framework for resilience of schools.

Figure 4 .
Figure 4. Some examples of qualitative questions integrated in the mobile phone app used for assessing the resilience of educational communities.

Figure 5 .
Figure 5. Flowchart to determine the resilience scores.

Figure 6 .
Figure 6.Community resilience indicators for a school in western Nepal (overall score 59/100).

Figure 7 .
Figure 7. Community resilience assessment output for a region in western Nepal (a) individual schools, (b) clusters of schools.

Figure 8 .
Figure 8.A model of a school showing some of the system components.

Figure 9 .
Figure 9. Examples of hazard preparedness related questions.

Figure 10 .
Figure 10.Examples of evidence of interactions amongst school community members.

Figure 11 .
Figure 11.A hierarchical description of educational communities.