Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur GraphSearch.
During the last decades public awareness of the limitations of traditional engineering practices and the imperative to conserve nature have led to changes in river management; including river restoration measures. The enlargement of the fluvial corridor is one of the often considered management measures. However, the high-pressure on land-use, the conflict of interests, as well as the uncertainty of vegetation and landscape development scenarios after restoration, can make their implementation difficult. In actual decision-making processes of large river restoration projects, no dynamic long-term modelling approach of potential riparian woody species development exists mainly due to the complexity of interacting driving-processes creating lateral and longitudinal gradients. So far, forest succession models applied to riparian areas are not conceived for river areas found in Central Europe and do not address explicitly environmental influences like nitrogen scarcity or drought stress important for certain riparian systems, nor they cover integrally the vegetation-hydraulics interaction. To support and enhance the decision-making processes in river restoration projects and to provide a better understanding of riparian forest dynamics and its driving-processes, the present thesis develops a coupled model of ecological and hydraulic processes to simulate riparian forest dynamics for Central European conditions, particularly for the case of enlarged fluvial corridors. The developed model RIFOD ('RIparian FOrest Dynamics') – a distribution-based forest succession model (i.e. ecological model) coupled to a quasi-2D hydraulic model – simulates short or long-term riparian forest dynamics at a yearly time step. The model, applied on a 10 times 10 m mesh grid, is spatially-explicit concerning the interactions of the ecological and hydraulic processes and integrates 65 Central European tree and shrub species. The ecological model is based on developments of different upland forest succession models, which were improved, adapted and complemented in regard to the ecological processes in riparian areas, for example concerning regeneration, nitrogen dynamics, soil water availability or flooding stress. At the basis of the modelling of physiological flooding stress response of plants, we carried out an in-depth review of the actual knowledge of the flooding stress response of Central European tree and shrub species. The review could highlight the main biotic and abiotic factors that influence species response and revealed the broad but still vague knowledge about physiological mechanisms and species-specific data of plant response. Based on the above findings, the fuzzy set theory was chosen to model flooding stress response integrating the main abiotic factors (e.g. flooding duration, -depth). The Central European tree and shrub species were classified into flooding tolerance classes by use of clustering analysis based on proxy-data, which allowed us considering indirectly the anatomical, morphological or physiological adaptations to flooding. To model mechanical flooding stress, existing mechanistic models simulating failure resistance to uprooting or stem breakage conceived for wind load studies have been adapted to the case of water flow. Required geometrical characteristics of trees and shrubs, such as crown width and crown heights, were estimated based on available field data, whereas rooting depths in dependence of the growth stage of an individual plant were simulated by developing a quasi-mechanistic vertical root growth model for Central European tree and shrub species. This root growth model allowed also a more realistic simulation of drought stress by calculating root water extraction in relation to the development stage of stand and determining species-specific and development dependent accessibility to groundwater – not integrated in the soil water balance so far. Compared to the situation in uplands, a more realistic modelling of nitrogen availability in riparian areas could be achieved by considering the loss of nitrogen via denitrifcation, as well as the loss of litter due to flooding. In opposition to existing riparian forest succession models, RIFOD considers riparian vegetation not as a purely dependent variable of flooding. Floods may affect vegetation but they are also affected by it, owing to the contribution of vegetation to hydraulic roughness. The coupling of the forest succession model to a quasi 2-D hydraulic model allowed considering this. Moreover, the quasi steady-state model approach allowed emphasizing on the ecological relevant lateral dimension and to make the model spatially explicit in the sense of vegetation-hydraulics interaction. The current version of RIFOD finds its application in riparian areas in which the geomorphological activity of the river is not a dominant process or in case of restoration projects, for widened fluvial corridors with morphologically stable stream channels. Model evaluation (validation and sensitivity analysis) revealed that RIFOD simulates plausibly the ecological gradients observed in the field and the resulting riparian forest dynamics. By applying the model at different lateral fluvial corridor designs at the River Rhone, the consequences of a restoration measure and the change of the hydrological regime for woody vegetation could be illustrated. From a management point of view, the model revealed for example that relative benefits become smaller as the width increases or that in absence of morphological activity (e.g. lateral bank erosion) the hydraulic processes alone are not sufficient for reinitiating riparian forest succession even for high energy streams such as the River Rhone. Moreover, the model allowed verifying and discussing current scientific concepts and hypotheses, as for example the intermediate stress hypothesis. Simulation results revealed that biological diversity is highest between the very low and very severe flooding stress levels confirming the intermediate stress hypothesis involving a trade-off between competitive dominant species which monopolise stable habitats and the few fugitive species that survive high levels of instability. The value of RIFOD relies in the capacity of displaying tendencies of riparian forest dynamics and associated characteristics in function of different fluvial corridor design variants. Moreover, it allows the understanding of processes and patterns in nature by allowing exploring the consequences of a set of explicitly stated assumptions that are too complex to explore by other methods. RIFOD is the first process-based riparian forest dynamics model for Central Europe and can be seen as a step forward into a more integral modelling of the riparian forest dynamics and its processes in view of a decision-aiding tool for large river restoration projects. A future integration of geomorphological processes will allow the application of RIFOD to quasi-natural river conditions.