Development of a methodology for extreme flood estimations in alpine catchments for the verification of dam safety

In the context of dam safety, reliable safety flood estimation methods are necessary to guarantee a sufficiently designed spillway capacity. Today’s state of the art allows to approach this topic under different angles. Statistical extrapolations are common practice, but do not grant to address all needs of dam safety verifications. Due to the evolution of informatics in the last years, computational power has been pushed to a level where hydrological modelling can be performed under optimal conditions. Practitioners are, however, not yet very familiar with hydrological modelling and its potential is not completely exploited. The present research work develops an approach combining statistics and hydrological modelling techniques to propose an innovative ready-to-use methodology to address the complex issue of extreme flood estimations. In order for this to be successful, some lacks of scientific knowledge had to be addressed before dwelling on the combination between statistics and simulations. Concerning the temporal rainfall distribution, it could be shown that a unique rainfall mass curve is admissible for the entire territory of Switzerland. Regarding a coherent combination of temperature and extreme precipitations, linear relations between the duration of the precipitation event and the zero degree isothermal altitude could be determined. In the context of hydrological modelling, the influence of initial conditions for extreme flood simulations have been assessed and methodological recommendations for the choice of initial conditions for extreme flood simulations could be formulated. Furthermore, the maximum admissible spatial expansion of the PMP events derived from the Swiss PMP maps, elaborated during the CRUEX project, could be estimated to be 230 km2. Finally, the combination of the simulation results with the approach of upper bounded statistical extrapolations could be shown to be advantageous: the sample sensitivity is reduced and the plausibility of the extrapolations is enhanced compared to conventional statistical distributions. Besides these scientific challenges, the methodology has to be pragmatic and ready-to-use as it is destined to engineers. The methodology was developed to be easily communicable. Therefore, its development has been undertaken under respect of common practices. Ultimately, a holistic methodology for extreme flood estimations could be formulated. The main advantage of the methodology is that it allows to estimate extreme flood hydrographs using hydrological simulation and to plausibly attribute a return period to the simulated peak discharge taking a deterministically determined upper discharge limit into account by referring to the PMP-PMF approach. The methodology combines thus the possibility of flood attenuation estimations with the knowledge of the occurrence probability of the peak discharge of the event, that is normally a reference quantity in flood safety guidelines. The application of the developed methodology to three catchments with different characteristics could prove its ease of utilization and, more importantly, its advantages compared to conventional approaches. A computer tool could be developed to make the CRUEX++ methodology accessible and readily applicable.

Schleiss, Anton
Lausanne, EPFL
Other identifiers:
urn: urn:nbn:ch:bel-epfl-thesis7307-4

 Record created 2017-01-25, last modified 2019-01-18

Download fulltext

Rate this document:

Rate this document:
(Not yet reviewed)