CHARACTERIZING THE FREQUENCY OF HIGH FLOWS IN RIVERS

Abstract:

Probability density functions and related cumulative distributions are widely used to describe magnitude and frequency of river flows at-a-station. They constitute important tools in hydrological and engineering applications for their ability to quantify in a condensed fashion the overall flow availability and variability, and properly describe the underlying hydrologic regime. The river flow regime is a pivotal driver of natural and industrial processes occurring in riverine environments, like e.g. riparian vegetation dynamics and hydropower production. Nonetheless, some processes are strongly influenced by specific ranges of streamflows, namely high flows, whose features are described in the right-tail of the probability distribution. For this reason, an accurate description of high flow frequencies represents an important task to study fluvial processes such as sediment transport and floods.

Recently, a physically-based stochastic model of streamflow dynamics has been developed and applied to a variety of catchments. The model provides an analytical expression for the streamflow distribution, which proved able to reproduce the frequencies of observed discharges and characterize in a meaningful way river flow regimes. This work focuses on the model capability to reproduce observed patterns in the tail of the flow distribution. In particular, a new method for the estimate of flow recession rates (based on analysis of single recessions) proved more effective in representing the cumulative distribution function, especially for high flows. At the same time, the model proves able to capture the emergence of heavy tailed distributions with divergent moments. The correlation between the peak flows and recession features, particularly accentuated in some cases, is extremely important for a correct representation of the tail of the streamflow distribution.

The results constitute a basis for a physically-based study of hydrologic and sediment transport regimes, and poses the basis for a mechanistic description of flood-frequency curves.