Disaster resilience in Pakistan: A comprehensive multi-dimensional spatial profiling

Highlights

• Pakistan's Disaster resilience is analyzed at the district-level for the first time.

• Spatial and statistical methods are integrated for resilience profiling.

• Around 20% of districts are significant hotspots of the least resilience.

• Sindh and Balochistan are comparatively less resilient compared with KPK and Punjab.

• Economic and institutional resilience need aggressive measures.

Abstract

Building disaster-resilient communities require operative resilience frameworks enabling factual decision-making and resource allocation at national and sub-national scales. While Pakistan is frequently hit by several natural hazards (i.e., floods, droughts, earthquakes, and extreme heatwaves) resulting in devastating impacts, no national-level higher-resolution disaster resilience information is available to provide references for informed planning. Hence, this study provides a, first of its kind, multi-level comprehensive disaster resilience evaluation in Pakistan. To do so, data on a customized list of indicators within three key resilience sub-components (i.e., economic, institutional, and social) are acquired to compute a resilience index. Frequency distribution and the Analysis of Variance (ANOVA) methods are employed to analyse the differences between different resilience indices and a cross-regional assessment is carried out at the sub-national level. Subsequently, an extensive spatial assessment is performed using geo-information models (i.e., Global Moran's I, Local Indicators of Spatial Association, and machine learning-based multivariate clustering) to explore the global and local geographies of the resilience. Based on ANOVA, significant differences between the resilience sub-components are found (95% confidence). The geographical distribution of resilience scores ascertains a large spatial heterogeneity across the study area with the least resilient regions belonging to Sindh and Balochistan provinces (95% confidence). As shown by the machine learning-based multivariate clustering, the least resilient regions particularly lack in economic and institutional aspects of disaster resilience. The findings provide important references to ensure resilience management-related cross-regional equity and justice. The rigorous analyses regarding the geographies of disaster resilience in Pakistan are important to support the country's disaster risk reduction efforts. While the results are useful for practitioners, decision-makers, and professionals in the risk management field, the study has important policy-relevant implications in the context of disaster risk mitigation strategies.

Introduction

The increase in the frequency and intensity of natural hazards (e.g., floods, earthquakes, droughts, and heatwaves) and associated costs over the past decades have gained significant attention of global leaders, policy and decision-makers, and professionals in disaster risk reduction field (Almutairi et al., 2020; Bukvic et al., 2020; Chuang et al., 2019; Kelman et al., 2015; Mysiak et al., 2018). The well-being and safety of communities can be adversely affected by these natural hazards with less resilient communities sharing a larger chunk of these impacts from disaster (Sajjad & Chan, 2019). With worsening impacts of natural hazards on communities, governments seek practical as well as comprehensive information in order to design possible ways-out to reduce disasters-related costs (Béné & Doyen, 2018; Reyers et al., 2015; Stringer et al., 2018; Yang et al., 2020).

Vulnerability assessment of communities has been a useful tool to formulate effective policies and to guide the priority actions in the context of disaster risk reduction (Fatemi et al., 2017; Frigerio et al., 2018; Rabby et al., 2019; Wang et al., 2019). However, a recent shift from disaster “vulnerability” to “resilience” has been witnessed as resilience is recognized as an optimistic and practical expression of community engagement in times of emergencies (Cutter et al., 2014; Meerow et al., 2016; Sajjad & Chan, 2019; Sweeney et al., 2020). Though some conceptualizations take vulnerability and resilience as an absolute opposite of each other, recent studies argue that these two might be correlated in some areas but not the absolute opposite of each other (Bakkensen et al., 2017; Sajjad, Lin, & Chan, 2020). Being a multi-dimensional and interdisciplinary concept, the definition of resilience depends upon the viewpoint of its application (i.e., the resilience of what? and resilience to whom?). Despite this situation, the ability to absorb and resist the disturbances (external/internal shocks), the competency of reorganization, fast recovery, and perform better in the future are the common features of resilience across its different definitions (Berkes & Ross, 2013; Cutter et al., 2008; Sajjad & Chan, 2019; Yoon et al., 2016). The assessment of disaster resilience on administrative scales (i.e., districts, counties, and cities) has now been well-recognized to aid prioritization process, informed decision-making, and smart resource allocation for disaster risk reduction (Cutter & Derakhshan, 2018; Frazier et al., 2013a; Moraci et al., 2018).

The Sendai Framework for Disaster Risk Reduction 2015–2030 adopted by the United Nations member states also emphasizes disaster resilience at different spatial scales (e.g., national and sub-national) through an implementation of inclusive and integrated social, economic, institutional, technological, educational, and political measures (Marzi et al., 2019). This implementation can progressively inform planning and decision-making to reduce hazard exposure and strengthen resilience through mainstreaming disaster risk mitigation into sustainable development policies and reasonable investments in resilience (Mysiak et al., 2016). However, researchers have outlined that there is still a lack of a ubiquitous disaster resilience assessment framework/model (Almutairi et al., 2020). A well-adopted approach for resilience assessment is the capital approach where overall resilience is classified in different sub-components in a system-of-systems fashion (Cutter & Derakhshan, 2018). These sub-components primarily include social, economic, institutional, infrastructural, and organizational pillars based on expert opinion and subject to data availability. This implies that the classification is flexible and can be molded according to the situation in area of interest (Sajjad et al., 2019). To assess the relative disaster resilience of different geographical (administrative) units, indicator-based methods are used. In this approach, several representative indicators of different sub-components are used to compute a spatially relative composite resilience index, which is able to capture the significant facets influencing the overall resilience (Cui & Li, 2019; Frazier et al., 2013a; Marchese et al., 2018; Marzi et al., 2019). The indicator-based assessments provide rapid screening and early insights regarding several issues, which is helpful in designing appropriate strategic actions to overcome the shortcomings (Mysiak et al., 2018).

Pakistan, a country in South Asia (Fig. 1) with 212.2 million people, is one of the most at-risk countries to the impacts of climate change and disasters and is regularly cited as one of the top-10 countries vulnerable to severe consequences due to global warming (Kreft et al., 2016). For example, the worst wave of 2010 flooding in Pakistan affected approximately 20 million people along with major economic and infrastructural damage across the country (World Food Program, 2018). As climate change continues, intense disasters are expected to be more frequent and intense, which could result in manifold damages in Pakistan, who is currently ill-equipped with the coping mechanisms. Despite the fact that Pakistan is significantly vulnerable to disasters and climate change-induced uncertainties, no national-scale higher-resolution resilience assessment is available currently. Similarly, though there exist several resilience assessment frameworks as aforementioned, data availability, disparities in terms of coping capacities, and socio-economic conditions differ among different communities. Hence, it becomes imperative to develop area-/community-specific frameworks for the assessment of resilience.

In this context, the current study formulated and demonstrates a comprehensive framework to assess the Pakistan-wide disaster resilience in a geographic information systems (GIS) environment. Considering the aforesaid aspects of resilience, in this study, the resilience of communities is assessed against natural/environmental hazards. Serval statistical methods are used to systematically evaluate the overall resilience and its sub-components. Subsequently, different spatial information models and a machine learning-based algorithm are applied for extensive geographical profiling of resilience at the district level in Pakistan. This study is an initial, yet comprehensive, effort to benchmark disaster resilience in Pakistan and to answer which areas need to be prioritized in this regard to cope with worsening impacts of natural hazards. The findings from this study will serve as a preliminary disaster resilience baseline, which will be of particular interest to researchers, development actors, and risk management authorities in Pakistan. As the resilience assessment has emerged in many policy and disaster risk reduction contexts (Mikulewicz, 2019; Paganini, 2019), the results from this study will also be useful to guide decision-makers, practitioners, and professionals in disaster management field.

The overall work presented in this paper is structured as follows: data collection and management, a methodological framework for the assessment of disaster resilience, and the approaches used to analyse the resilience, comprehensively, are detailed in Section 2. The results are then systematically documented in Section 3. Section 4 provides the discussions on the findings, outlines important policy implications, and describes current limitations and the way-forward. Section 5 finally concludes the study with major findings.