Integrated socio-environmental vulnerability assessment of coastal hazards using data-driven and multi-criteria analysis approaches

Abstract

Coastal hazard vulnerability assessment has been centered around the multi-variate analysis of geo-physical and hydroclimate data. The representation of coupled socio-environmental factors has often been ignored in vulnerability assessment. This study develops an integrated socio-environmental Coastal Vulnerability Index (CVI), which simultaneously combines information from five vulnerability groups: biophysical, hydroclimate, socio-economic, ecological, and shoreline. Using the Multi-Criteria Decision Making (MCDM) approach, two CVI (CVI-50 and CVI-90) have been developed based on average and extreme conditions of the factors. Each CVI is then compared to a data-driven CVI, which is formed based on Probabilistic Principal Component Analysis (PPCA). Both MCDM and PPCA have been tied into geospatial analysis to assess the natural hazard vulnerability of six coastal counties in South Carolina. Despite traditional MCDM-based vulnerability assessments, where the final index is estimated based on subjective weighting methods or equal weights, this study employs an entropy weighting technique to reduce the individuals’ biases in weight assignment. Considering the multivariate nature of the coastal vulnerability, the validation results show both CVI-90 and PPCA preserve the vulnerability results from biophysical and socio-economic factors reasonably, while the CVI-50 methods underestimate the biophysical vulnerability of coastal hazards. Sensitivity analysis of CVIs shows that Charleston County is more sensitive to socio-economic factors, whereas in Horry County the physical factors contribute to a higher degree of vulnerability. Findings from this study suggest that the PPCA technique facilitates the high-dimensional vulnerability assessment, while the MCDM approach accounts more for decision-makers' opinions.

Toward an Integrated Probabilistic Coastal Vulnerability Assessment: A Novel Copula‐Based Vulnerability Index

“…In the previous study, Tanim et al. (2022) outlined factors of quantification and aggregation for fine‐scale vulnerability assessment. In the current study, the CVI development and evaluation for integrated vulnerability analysis is extended at the census block level and enhanced using the earlier progress made by Tanim et al.…”

“…In the current study, the CVI development and evaluation for integrated vulnerability analysis is extended at the census block level and enhanced using the earlier progress made by Tanim et al. (2022). The approach applied in factor quantification includes both ways of aggregation and disaggregation: fine and large scale can be used for US national‐level CVI development.…”

“…Tanim et al. (2022) developed a comprehensive and multi‐dimensional CVI, combining physical, hydroclimate, socio‐economic, ecological, and shoreline characteristics of SC, which consisted of approximately 21 vulnerability indicators. In the present study, to estimate the vulnerability of SC's coast, the biophysical vulnerability factors were chosen from the hydroclimate and physical vulnerability indicators.…”

“…Vulnerability in the current study refers to the degree to which a socio‐environmental system is susceptible to natural hazards and the lack of coping capacity with coastal hazards or the adaptive capacities of coastal communities (Bevacqua et al., 2018; Sahin & Mohamed, 2014; Tanim et al., 2022; Zhang et al., 2021). The coastal vulnerability is defined in the present study as a function of biophysical exposure, the sensitivity of the coastal systems, and coastal communities' lack of adaptive capacities.…”

“…The main component in the spatial model developed in this stage consists with three main steps: (a) factor normalization using fuzzy functions, (b) determination of the spatial importance of each factor using objective weighting techniques as described in Section 2.1, and (c) aggregation of all fuzzy normalized spatial factors to a sub‐index (BPVI and SEVI) using the objective weighting technique. The main advantages of using the fuzzy logic‐based factor normalization approach are to determine the degree of vulnerability using a non‐linear data transformation function, the ability to reduce the uncertainty of discrete scale factor normalization, and precise indexing during the factor aggregation by employing the vulnerability functions of both deterministic and stochastic (Tanim et al., 2022). Details of the fuzzy functions are provided in Tanim et al.…”