Construction of the Scale-Specific Resilience Index to Facilitate Multiscale Decision Making in Disaster Management: A Case Study of the 2015 Nepal Earthquake

Abstract

Many scholars have advocated for the use of empirical evidence to assess resilience across scales and over time. Accordingly, we conduct a case study using survey data on individual perceptions of disaster relief that were gathered each month from August to December 2015, shortly after the 2015 Nepal earthquake. We construct a scale-specific resilience index (SSRI) based on a set of variables that are validated separately at different spatial scales and over time against the survey data. The regression results show that the variables related to household structure, industrial diversity, community capital, and accessibility and emergency services are validated against the survey data at both the district and sub-district levels, the variables related to ethnic diversity and the capacity of emergency camps are validated only at the district level, and the earthquake experiences variable is validated only at the sub-district level. Consequently, to achieve optimal models, we use six validated variables to construct an SSRI at the district level and seven variables, including those related to the vulnerability of household property and the average elevation, to construct an SSRI at the sub-district level. The SSRI scores are validated via multilevel regression models against the surveyed relief scores after the 2015 Nepal earthquake. The results show that the SSRI scores based on the validated variables correlate favorably and as expected against the survey data at both district and sub-district levels, and outperform the composite resilience index, which considers all of the variables regardless of their individual validation results. The method used to construct the SSRI helps to identify the contributions of multidimensional resilience indicators across spatial scales and over time in real cases, and also provides index scores of scale-specific resilience that are easily understood and applicable to multi-scale decision-making processes.