Sediment transport modelling at the catchment scale in Alpine streams

Sediment management in river networks is of crucial importance for both environmental, productive and safety reasons. For instance, lost of sediment connectivity along river networks is recognized as a major cause of habitat degradation and biodiversity loss worldwide, and erosion/deposition processes may severely contribute to flood hazards. Also, artificial reservoirs may reduce their capacity to provide water in the quantity and time needed for the productive uses they were built, like irrigation and hydropower. In the last sixty years for example, dam-regulated rivers in Europe have witnessed a loss of reservoir capacity due to sedimentation. An average estimation across European reservoirs indicates that by 2022 their water capacity decreased by 18\% due to sediment deposition. In this context, new advances in sediment transport modelling acquire a relevant role, especially in mountain streams. The work of the present Ph.D. thesis mainly focuses on new methodologies to enhance sediment transport modelling at the catchment scale. In particular, the present work aims at the evaluation of the efficiency of the CASCADE basin-scale sediment-transport model and its sensitivity to the local hydro-morphological conditions; to evaluate the distance from the equilibrium condition of transport capacity using different methodologies; to propose a integrative algorithm of the CASCADE model to estimate the suspended load at the basin scale, accounting for the contribution to the sediment transport related to the local hydro-morphological conditions. The present work analyses three river case studies located in the Italian Alps (Trento Province, N-E Italy)): Travignolo, Grigno and Leno di Vallarsa creeks. The Ph.D. work addresses four main research gaps, and is structured into four corresponding research elements. The first topic concerns the selection of the input parameters to the CASCADE model, which computes the transport capacity and the deposited flux in every reach of a river network. One of the most relevant input parameters is the grain-size distribution (GSD) of river bed sediments, which is generally affected by uncertainties and difficult to collect. To distribute the information provided by the available samples, many authors adopted the concept of morphological equilibrium to estimate the sediment GSD for different reaches. This equilibrium condition imposes the equality between the upstream and downstream transport capacity, assuming the absence of any deposited material. The purpose of the work is to propose an alternative method to estimate sediment GSD using a different condition. In particular, the new method imposes the equality between the GSD of the deposited sediment flux and the bed composition, which implies that the bed sediment distribution is in equilibrium with the deposited material. A large number of sediment GSDs (23 sampling sites) have been collected to build a database for the application of the estimation method to the network of the Travignolo Creek. The main result is an increase in the number of reaches in which the GSD is known and the potential to be coupled with a bed-composition estimation algorithm based on the morphological equilibrium of the channel. The second element of the present thesis focuses on the calibration of parameters and outcomes validation of the CASCADE sediment-transport model. The main novel feature of the proposed method is being based on sediment-lithology data. The percentage fraction of a specific lithological class in the GSD sample is adopted to estimate the deposited fluxes. In general, river bed sediments are composed of rock fragments that come from specific sources in the catchment. Therefore, the rock lithology may help to identify the spatial origin of each sediment grain found in the bed material. The analysis of sediment lithology provides the identification of rock types (igneous, sedimentary, metamorphic), categories, and textures by describing macro-physical characteristics. From a macroscopic point of view, sediment-lithology recognition may be directly performed during the field-sampling campaign. In particular, in the case studies of the Travignolo and the Grigno creeks, lithology recognition has been performed simultaneously with the random counting of coarse sediments. The two selected case studies are significant from a geological point of view as the lithologies of rocks characterizing the river basin are well distinguishable. Through field-sampling campaigns, sediment lithology has been characterized in 15 different sites, 8 along the main stream of the Travignolo Creek and 7 along the Grigno Creek. The application of the CASCADE model, which tracks the path of each sediment cascade coming from a specific source, provides transported and deposited sediment fluxes in every reach of the two river networks. Therefore, the observed presence percentages of lithological classes are compared to the percentage contributions to deposited fluxes modelled by CASCADE in different river reaches. In the first case study (Travignolo), the validation of model results is performed by comparing the presence percentage of each lithological category and the contribution to deposited fluxes coming from the corresponding sub-catchment. Differently, in the second case study (Grigno), lithological information is used to calibrate the transport limitation parameter (trlimit in CASCADE), i.e., a multiplicative coefficient of the entrained sediment flux from the river bed, in the reaches of middle-course tributaries. The third part of the present work shows the parallel use of sedimentation data in reservoirs, empirical relations for sediment production from the catchment (available from the current literature), and the estimation of bedload at the basin scale to evaluate the equilibrium condition for sediment transport in the upper part of the Leno river network. The used sediment transport model is a modified version of the CASCADE model that includes the estimation of mean-annual sediment load by using a probability density function of water discharge. The mean annual sediment volumes produced using the empirical relations are used to calibrate the transport limitation parameter of the modified transport model, by providing a estimate of the distance from the equilibrium condition at the reach scale. The transport condition is distant from the equilibrium of the transport capacity in most of the sources and the internal reaches of the network, which show a low erodibility of the channel. The fourth part of the thesis aims to propose a basin-scale model that couples the estimation of bedload and fine sediment load along a river network, evaluating the influence of local hydro-morphological conditions on the suspended load. Generally, the fine sediment fraction originates from mountain slope erosion or river bank degradation and different models of soil erosion have been proposed to estimate the fine sediment production from upslopes or catchments. The proposed approach modifies the basic version of the CASCADE model by including the suspended load component. Suspended load is modelled as a solid flux that is transferred downstream, while exchanging mass with the bedload layer. The parameter that discriminates which sediment fractions are transported as suspended load and which as bedload is the ratio between the settling velocity and the local shear velocity. Most of the fine sediment mass is assumed to be related to the production from lateral slopes, which is based on the catchment characteristics such as the udometric coefficient and the extension of the bare soil areas. This model is applied to the case study of the Leno Creek. To calibrate the sediment-production function, a dataset of coupled fine sediment concentration in the water column and flow discharge has been collected at a specific section along the main stream of the creek. The main result is a preliminary estimate of the sediment production coefficient and the estimation of suspended load and bedload rates in every reach of the river network. In addition, the proposed model estimates the contribution to suspended sediment given by the entrainment of fine sediments from the bed and the sedimentation rate that from suspension to bedload.