Qualitative Approach for Assessing Runoff Temporal Dependence Through Geometrical Symmetry

Abstract

Currently, noticeable changes in traditional hydrological patterns are being observed on the short and medium-term. These modifications are adding a growing variability on water resources behaviour, especially evident in its availability. Consequently, for a better understanding/knowledge of temporal alterations, it is crucial to develop  new analytical strategies which are capable of capturing these modifications on its temporal behaviour. This challenge is here addressed via a purely stochastic methodology on annual runoff time series. This is performed through the propagation of temporal dependence strength over the time, by means of Causality, supported by Causal Reasoning (Bayes’ theorem), via the relative percentage of runoff change that a time-step produces on the following ones. The result is a dependence mitigation graph, whose analysis of its symmetry provides an innovative qualitative approach to assess time-dependency from a dynamic and continuous perspective against the classical, static and punctual result that a correlogram offers. This was evaluated/applied to four Spanish unregulated river sub-basins; firstly on two Douro/Duero River Basin exemplary case studies (the largest river basin at Iberian Peninsula) with a clearly opposite temporal behaviour, and subsequently applied to two watersheds belonging to Jucar River Basin (Iberian Peninsula Mediterranean side), characterised by suffering regular drought conditions. Keywords: Causal reasoning, Theorem of Bayes, Temporal dependence propagation, Runoff time series, Water resources management