Unfolding Multivariate Drought Risk in Large River Basins accounting Onset Seasonality and Event Magnitude

Abstract

The frequency and severity of droughts are expected to increase in the warming climate. Understanding mutually interacting drought properties, such as their severity (deficit volume) and time to onset, is crucial for managing reservoir operations and low flows. Previous studies have performed bivariate drought frequency analysis considering drought severity and duration across different climate regions. However, little is known about the role of drought seasonality in shaping drought severity. This study aims to investigate the dependence between onset time (i.e., directional occurrence date) and deficit volume and evaluate the impact of drought seasonality on the deficit volume distributions in disparate climate regions across the global tropics. Leveraging streamflow observations from representative catchments in the northern and southern hemispheres and considering the nonlinear dependence strengths between onset time and deficit volume, we implemented a multivariate drought frequency model that yields a conditional probability of drought severity given the timing of peak drought intensity. We consider multiple univariate probability functions for modelling drought deficit volume, whereas drought onset time is modelled using von Mises distribution. Further, the joint dependence between drought onset and deficit volume is modeled using a bivariate Archimedean class of copulas. First, we show temporal variations of exceedance probabilities of drought deficit volume and their seasonal clustering behavior during dry/wet phases and then explore any possible shift in the risk of peak drought intensity based on its seasonality. Finally, employing a flexible multivariate probabilistic tool, we demonstrate different scenarios of drought characteristics combinations and a seasonality-informed drought probability model, aiding understanding complex processes of drought propagations across disparate climate regimes and assessing possible climatic shifts to drought frequency.