Esplorazione delle reti fluviali dinamiche con strumenti empirici e teorici

Empirical evidence has shown that the flowing portion of many river networks does vary in time, owing to seasonal and/or event-based expansion-retraction cycles that mimic the unsteady nature of the underlying climatic forcing. Such rivers, commonly referred to as temporary streams, are believed to represent more than half of the global river network, and are observed in most climatic regions worldwide, from arid, to humid areas. The goal of this thesis is to provide a comprehensive hydrological analysis of temporary streams, by combining empirical data from 18 study catchments spread all over the world with theoretical analyses and stochastic models. A Bayesian framework is developed for the statistical description of the dynamics of the active network, linking the behaviour of the stream network to the underlying local properties of the constituting nodes. The local persistency of the nodes is found to be a key statistical index to quantify the probability of activation of each node, and dictates the interaction between different nodes and defines the spatial pattern of active network at any given time. The activation of the nodes during network expansion is found to follow a fixed order of decreasing persistency, while the deactivation of nodes during network contraction occurs in reverse order. This general behaviour, called hierarchical, is defined in a mathematically rigorous way within the Bayesian framework, and used to derive analytical expressions for the main statistics of the active length as function of the mean network persistency. The latter is then linked to the mean effective precipitation, thus providing a direct connection between climate and network dynamics. The Stream Length Duration Curve links each possible length of the active network to the duration for which that length is exceeded, and allows a quantitative description of the dynamics of a temporary stream. Under the hierarchical hypothesis, this curve is proven to be solely determined by the spatial distribution of the local persistency, providing a crucial link between the spatial and temporal dimensions of the problem. Temporary streams can be also characterized via their length regimes, providing an objective classification system based on the dynamic behaviour of the networks. A stochastic model for streamflow generation is also exploited in conjunction with the hierarchical activation scheme, to enable the spatio-temporal simulation of a temporary stream under a wide variety of climatic scenarios. This type of simulations require a small number of parameters and a limited computational effort, and allows a deeper understanding of the influence of climate on the origins and implications of channel network dynamics. A set of synthetic temporary streams are used as input for a dynamic metapopulation model simulating the occupancy of a target aquatic species within the network. This application reveals the fundamental role that network dynamics have in degrading the ability of a riverine ecosystem to support a metapopulation, particularly in drier climates, by significantly reducing the average occupancy and increasing the probability of extinction of the target species. Taking into account the dynamic nature of the active network represents a fundamental prerequisite for a correct assessment of many ecological and biochemical functions that are mediated by riverine systems. The work presented in this thesis offers a comprehensive analysis of dynamical river networks, providing a new theoretical framework and novel analytical tools and numerical models for the reconstruction and simulation of the dynamics of the active extent of a stream. Given the rising awareness of the impacts of human activities on the environment, and the sensitivity of temporary streams to a changing climate, the tools provided in this study will hopefully foster the development of better strategies for the management and protection of such systems.