Examining Changes in Nonlinear Interactions of Streamflow Drought Seasonality versus its Severity across Global Tropics

Abstract

Streamflow drought is addressed as below-normal water availability in large rivers and tributaries. Streamflow drought impacts several sectors, including irrigation, river ecology, hydroelectric potential, financial, and drinking water supply. Analyzing variability in streamflow drought timing and the nonlinear interactions between drought onset and severity is necessary not only for better understanding of drought predictability but also of its temporal change, which aids in developing climate adaptation strategies. Very few studies have assessed the seasonality of streamflow droughts, although a few analyses have been performed focusing on other hydroclimatic extremes, such as extreme precipitation and floods. However, little is known about understanding the shifting behaviour of streamflow drought onset patterns at a local or regional scale. Further, a few studies have assessed the severity of low flows at a global and local scales. However, most of these studies have either considered a constant threshold approach to delineate low-flow episodes or employed sub-seasonal (monthly) temporal scales to access streamflow droughts using standardized indices of precipitation or runoff. However, none of the studies have investigated the non-linear interactions between streamflow drought onset and deficit volume and how these bivariate interactions evolve over time across large river basins. Here we investigate the timing of the streamflow drought onset and its severity (i.e., deficit volume) over 472 catchments that are spatially distributed across 21 Intergovernmental Panel on Climate Change (IPCC) Special Report on Managing the Risks of Extreme events and Disasters to Advance Climate Adaptation (SREX) reference regions in the global Tropics. We identified those catchments with little or no potential anthropogenic influences and were selected based on a detailed quality assessment of continuous streamflow records and their proximity to dam locations. We implemented a daily variable threshold approach with an 80% exceedance probability of the flow record to identify streamflow drought episodes. Moreover, based on large streamflow records, we compare the potential shifts in the seasonality of streamflow droughts in the recent (1994-2018) versus the pre-1990s (1969-1993). We show a strong persistency in the timing of streamflow drought onset in the core monsoon-dominated regions. In the northern hemisphere, the mean onset is observed primarily during August and September, whereas in the southern hemisphere, the onset timing is temporally clustered around November to March. Our proof of concept analysis suggests that North-East South-America is the most vulnerable region, in which an earlier occurrence of drought is compounded by an increasing deficit volume, indicating a drying trend throughout. Furthering this, we investigate the non-linear interactions between drought characteristics, onset time, and severity to decipher the pattern of associations across disparate climate regimes, especially in regions with pronounced seasonal cycles. The obtained insights has important implications for water resources management in tropics, where seasonal climates dominates. The findings can inform drought monitoring, planning and improve drought resilience to multiple climate stressors.