Skip to Main content Skip to Navigation
Theses

Consolidation de l'information hydrologique disponible localement et régionalement pour l'estimation probabiliste du régime des crues

Abstract : To define the design flood, practitioners must often deal with only few data available. The aim of this work was to propose new classes of probabilistic models that are more accurate for this kind of applications. In this perspective, we propose three different models: two regional approaches and a fully local one. Unlike fully local models, the regional approaches include information from other gauging stations. Our results show that the proposed regional Bayesian estimators are more robust to the discordancy degree of the sites within the region. In addition, for larger quantile estimation (T >50 years), the concept of a regional parameter which controls the tail behaviour seems to be relevant. However, this concept has to be proposed and not imposed within the likelihood function. It is overwhelmingly clear that the most important information one disposes is the target site one. To this aim, we propose a third model that is fully local, i.e., which only uses the latest recorded data. This new model is innovative as the whole time series is involved in the estimation procedure; not only cluster maxima. Consequently, even with only a five years record length time series, the sample size becomes large. Our results show that, for flood quantile estimations, this model clearly outperforms the estimators conventionally used in hydrology. Furthermore, by definition, this model allows inferences on flood dynamics.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00232772
Contributor : Mathieu Ribatet <>
Submitted on : Friday, February 1, 2008 - 4:55:27 PM
Last modification on : Monday, October 19, 2020 - 11:07:10 AM
Long-term archiving on: : Thursday, September 27, 2012 - 5:48:05 PM

Identifiers

  • HAL Id : tel-00232772, version 1

Collections

Citation

Mathieu Ribatet. Consolidation de l'information hydrologique disponible localement et régionalement pour l'estimation probabiliste du régime des crues. Mathématiques [math]. Institut National Polytechnique de Grenoble - INPG; Institut National de la Recherche, 2007. Français. ⟨tel-00232772⟩

Share

Metrics

Record views

229

Files downloads

1087