A Global Analysis of Daily Streamflow Data to Unravel the Heaviness of Flood Distribution

Abstract

Hydroclimatic extremes, such as floods, present complex challenges in risk assessment due to their spatial and temporal compounding nature. This research aims to improve our understanding and modelling capabilities by investigating the complex interactions among record length, flow regime, and upper tail of floods. Our study resolves conflicting results in prior studies by utilizing a quasi-global peak-over-threshold (POT) analysis of flood with the Generalized Pareto (GP) distribution. Based on an analysis of 4,482 streamflow series over six different regime types with record lengths ranging from 30 to 213 years, our results show a strong relationship between the GP shape parameter and record length. The results show that the variance of the shape parameter of GP distribution diminishes with record length, and it eventually converges to a single value depending on the flow regime. We show that the shape parameter of snow-dominated streams is the lowest, whereas intermittent streams have the highest. Our research reveals regime-specific patterns in the impact of hydroclimatic and catchment controls on flood tails, underscoring the necessity of regime-specific strategies for flood risk management. Identifying catchments that are more likely to experience extreme flooding provides useful information for determining which mitigation measures to prioritize.