Development of a Bayesian Copula-based Spatial-temporal Method for Multivariate Drought Risk Analysis

In this study, a Bayesian copula-based spatial-temporal analysis (BCSA) method is developed for multivariate drought risk analysis. BCSA can help identify drought variables and to build dependent structure of variables (duration-severity and severity-area). BCSA was applied to a real case in Central Asia (1918-2018). Some major findings can be summarized: (i) Central Asia suffered 72 droughts in 1918-2018, and droughts are mainly concentrated in Karaganda and Kyzylorda states of Kazakhstan; (ii) drought is more frequent in spring and summer than autumn and winter; (iii) due to interaction effect on variables, joint guarantee rate is less than design guarantee rate, and joint guarantee rate can provide a new reference for drought project design; (v) when design dependability is 0.99, 0.98, 0.95, 0.90 and 0.80, multivariate drought risk would be 3%∼7%, 6%∼14%, 13%∼27%, 23%∼39% and 35%∼52%, respectively. The above findings can provide decision support for drought mitigation and project management.