Developing a Multivariate Probabilistic Framework to Model Onset Seasonality and Event Magnitude of Streamflow Droughts

The frequency and severity of droughts have intensified in recent decades, significantly impacting water availability and human and ecological systems. This growing trend highlights the need for a comprehensive exploration of drought characteristics and their interconnected dynamics, such as the timing of onset and severity. More often, streamflow drought onset time and deficit volume show nonlinear interdependencies. The seasonality of streamflow response is a widely used indicator to assess flood probabilities, catchment classifications, and even regional frequency analysis. However, understanding streamflow seasonality in influencing low flows across different climate regimes is mainly unexplored. This study investigates streamflow droughts considering daily observations from 1160 global catchments spanning disparate climate regions between 60°N and 60°S. Our analysis indicates that approximately 12% of sites demonstrate pronounced seasonality, significantly affecting drought severity with a dependence strength greater than 0.6. In particular, 50% of sites in the tropics, 11% in subtropics, and 9% in the temperate regime show substantial seasonal impacts on the drought severity, highlighting the diverse influence of seasonality across different climatic zones. Approximately 16% of sites show a significant trend (p<0.10) toward earlier onset, whereas 34% show delayed arrival in streamflow droughts, which indicates possible nonstationarities in low-flow seasonality, potentially impacting other drought properties, severity, and duration. Considering the nonlinear dependence strengths between onset time and deficit volume in a bivariate probabilistic framework, we attempt to investigate the severity of hydrological droughts, conditional to their onset seasonality. Examining representative catchments from each climate zone, we find that winter (Dec - Feb) droughts tend to be more severe than other seasons in temperate and subtropical climate regimes. In contrast, catchments in the tropics experience more severe droughts during the summer (Jun - Aug). While winter droughts are more persistent in the tropics and subtropical regions, summer droughts tend to be longer in temperate regions. The developed model offers a probabilistic forecast of seasonal droughts and helps to assess forecast uncertainty, aiding water management during extreme low-flow seasons and water years. This approach underscores the critical role of incorporating seasonality into drought hazard assessments to enhance water security adaptations in a changing climate.