Abstract
Droughts are comprehensive and complex naturally occurring hazards in any climatic region around the world and often result in the loss of life and severe ecosystem damage. Drought monitoring is usually based on single-variables that may not represent the corresponding risk appropriately to its multiple causation and impact characteristics under current and future climate scenarios. In order to address this issue, the multidimensional copulas function, which is a flexible statistical tool, could be applied to develop multivariate drought indicators and solve the complicated and nonlinear associations. The aim of this paper is to develop reliable designing, monitoring and prediction indicators for the proper assessment and intervention of drought risk by nature-based solutions (NBS). Using a copula-based multivariate drought indicator (CMDI) that considers all possible variables related to meteorological, agricultural and hydrological droughts is essential for better drought risk assessment and intervention. The CMDI was developed by integrating univariate marginal cumulative distribution functions of meteorological (precipitation), agricultural (soil moisture) and hydrological (streamflow) variables into their joint cumulative distribution function. CMDI was then applied to the selected study catchment (Po Valley, Italy and Spercheios River, Greece) using hydro-meteorological data from gauging stations and ERA5 gridded data for the period 1979-2017. The result of CMDI showed moderate, severe and extreme drought frequencies in the two selected catchments. The constructed CMDI captured more severe to extreme drought occurrence than the considered single drought indicators. This proved that the CMDI could appropriately represent the complex and interrelated natural variables. The uncertainty analysis based on Monte Carlo experiments confirmed that CMDI is a more robust and reliable approach for assessing, planning and designing a nature-based intervention for drought risk. The findings of this research can provide a reliable way to develop approaches that can be used for assessing and predicting non-linearly related variables or any risk that may occur simultaneously or cumulatively over time.