Comprehensive drought risk assessment using structural equation modeling and objective weighting methods

Abstract

Study region: Gyeongsang province in Korea Study focus: Drought is a complex phenomenon influencing the natural, physical, and social sectors depending on the occurrence probability, regional characteristics, and the water supply and demand. To reduce the damage resulting from drought, it is necessary to assess the drought risk that can identify the impacts, causes, and vulnerability of drought. Previous drought risk assessment has usually been conducted by combining drought hazard and vulnerability, but such assessment is of limited value because the regional response capacity for drought is not considered. Moreover, it contains high uncertainty because indicators and weighting factors are determined by subjective methods. In this study, the comprehensive drought risk was assessed including the drought response capacity and with consideration of the regional water supply system. To remove the uncertainty in the drought risk assessment, this study employed partial least squares – structural equation modeling (PLS-SEM) to select effective indicators including the regional drought response capacity, and also applied objective weighting methods such as entropy, principal component analysis (PCA), Gaussian mixture model (GMM), and Bayesian networks to determine optimal weighting factors. New hydrological insights into the region under study: As a result of application to Gyeongsang province in Korea, PLS-SEM selected 10 indicators for drought risk assessment. Using the selected indicators and the objective weighting methods, this study determined that the drought hazard, vulnerability, response capacity, and risk were highest in GS26 (Ulleung), GS27 (Changwon), GS16 (Cheongsong), and GS28 (Jinju), respectively. The districts with large actual drought damage had high drought risk, indicating that the results of this study were reasonable and useful in the identification of the major impacts and risk of regional drought and may facilitate the decision-making process for selecting drought countermeasures to reduce drought risk.

Keywords

• Water supply system
• Drought response capacity
• Drought risk
• Structural equation model
• Bayesian network