A NEW COMBINED APPROACH FOR BEDROCK INCISION ANALYSIS: GEOCHEMISTRY AND HYDRAULIC MODELLING. APPLICATION TO LOIRE RIVER (FRANCE)

The Loire river (France) has narrowed in the twentieth century. Flow has been reduced considerably and actually as result the meandering of the Loire river is entrenched into bedrock in some sections as deep as 2.5 meters. We intend to trace the changes in the longitudinal profile of the Loire between Grangent and Villerest (53 km), for the last 150 years in order to understand this bedrock incision process. In this paper, we show how a new combined hydraulic and geochemical approach can be used. A unidimensional model was developed to simulate stream flow. Quasi steady-state flows are to be simulated. However, for gradually varied flows, backwater formulation is performed. Otherwise, the Saint Venant equations are solved supposing shallow water and unidirectional flow. Discharge Q (m3/s) and depth h (m) values are used to perform sediment transport calculations. This uses Meyer-peter Müller formula and the Exner equation for bottom level aggradation or degradation. An attempt has been made to simulate cross-sections of this river over time by determining the distribution of eroded and deposited materials. A McCormack numerical scheme is used for the resolution. Our model was firstly tested and compared to experimental values of Soni et al. (1980). It appears that numerical values and experimental data adequately agree. Then, an application of this model to a 2 km stretch of the Loire. River morphological changes are linked both to the quantity and the nature of its transported materials. Due to the empirical aspects of sediment transport formulas, and the uncertainty of in-situ measurements , we propose a geochemical approach to complete the hydraulic aspect. This one is based on material assessment: in the absence of erosion of the substratum and banks, the chemical composition of the sediment varies only in response to contributions from the riverside. Samplers were placed along the considered span. Because the effects of sorting are more complex than a simple grain size differentiation, it is preferable to carry out very dense sampling, to interpret them in terms of mixture among various populations of grains, to finally extract an estimate of the proportion of substratum incorporated in the solid load. Therefore the way in which it varies longitudinally, could be tested and compared with the predictions of the hydro-sedimentary model.