Resumen de Assessing debris-flow hazard focusing on statistical morpho-fluvial susceptibility models and magnitude-frequency relationships. Application to the central-eastern Pyrenees
Guillaume Gerard Chevalier
Occurrence of debris flows has received little attention in the Pyrenees, probably due to the small risk faced by most of the debris-flow prone sites in this mountain range. Nevertheless, the event of Biescas, which occurred in august 1996 and causing 87 casualties, demonstrates the existence of high-risk spots in the region and justifies the elaboration of the debris-flow hazard assessment presented in this thesis. Five debris flows, which occurred in 2008, are selected; and site-specific descriptions and analysis, regarding geology and morphology, were performed. The results are compared with worldwide data and some conclusions on hazard assessment are presented. The preliminary analysis of some major Eastern Pyrenean debris flows represents the background for this thesis. The necessity of possessing an inventory of past occurrences is of crucial importance when assessing debris-flow hazard. Criteria of reconnaissance were thought to be visible from aerial viewing. 691 tracks through which debris flows are thought to have travelled have been revealed. Based on debris-flow inventories and using a geographical information system, the debris-flow hazard assessment presented in this thesis takes into account fluvio-morphologic parameters, gathered for every 1st-order catchment as well as every 2nd-order catchment. Mountainous headwaters are a common subject in geomorphological studies. Often investigated at local scale, the geomorphological context in which headwaters evolve has been poorly reported in the Central-Eastern Pyrenees or worldwide. A series of parameters obtained for Central-Eastern Pyrenean headwaters catchments consisting of 3005 1st- and 655 2nd-order catchments are presented. Acquired from a digital elevation model, these catchments have been digitalised, identified and attributed a value for each parameter. Previously reported parameters¿ ranges agree with those presented in this study. For the first time, the ranges of values give details about the Central-Eastern Pyrenees headwater catchments. Data mining techniques are used on the morphometric parameters, to calculate and test three different models. The first model is a logistic regression. The other two are classification trees, which are rather novel susceptibility models associated with debris flows. Results related to the training dataset show that the optimized model¿s performance lies within existing reported range although closer to the lowest end (near 70%). When the models are applied to the test set, the logistic regression seems to offer the best prediction, as training and test set results are very similar in terms of performance. Trees are better at extracting laws from a training set, but validation through a test set gives poorer results for a prediction at regional scale. The determination of magnitude of a historic event can be done by distinguishing its deposits. However this is not a trivial task in debris fans that accumulate deposits, corresponding to consecutive debris flows, especially if only a conventional geomorphological analysis is carried out. The event deposits can be mapped and, subsequently, trees damaged by the flows sampled for dating events. A magnitude-frequency relationship was prepared for El Rebaixader site, at local scale, and is compared to that of the Tordó creek. Moreover, a debris-flow inventory was created in the "Aigüestortes i Estany de Sant Maurici" National Park in the Central Pyrenees, Spain, and this regional magnitude-frequency relationship is compared to that of Rebaixader. Both curves include a strong rollover effect at about 2000 m2, and events larger than this magnitude can be represented by a power law, with an exponent between -1.5 and -1.9. This thesis is a first step toward the assessment of debris-flow hazard in the Central-Eastern Pyrenees. Although a lot of information is provided, more work is still to be done, in order to fully capture debris-flow importance in landscape evolution.
