摘要：The stage-discharge measurements and rating curves accumulated over decades at hydrometric stations are a valuable source of information on the long-term evolution of river bed levels. However, the methodology to extract meaningful geomorphic information from such hydrometric data is not straightforward. We introduce an original method to estimate the parameters of successive rating curves by Bayesian analysis in sequence. These parameters reflect the physical properties of the channel features that control the stage-discharge relation: low-flow riffles, main channel, floodway (bars), floodplain, etc. The dates of rating changes are assumed to be known in existing hydrometric records. The uncertainty interval of each parameter is estimated, assuming, however, that no rating change has been ignored by the station manager. It is thus possible to clearly distinguish overall trends of the channel bed level from the local evolution of riffles and to evaluate whether the observed temporal changes are significant compared to the estimation uncertainties.
其他摘要：The stage-discharge measurements and rating curves accumulated over decades at hydrometric stations are a valuable source of information on the long-term evolution of river bed levels. However, the methodology to extract meaningful geomorphic information from such hydrometric data is not straightforward. We introduce an original method to estimate the parameters of successive rating curves by Bayesian analysis in sequence. These parameters reflect the physical properties of the channel features that control the stage-discharge relation: low-flow riffles, main channel, floodway (bars), floodplain, etc. The dates of rating changes are assumed to be known in existing hydrometric records. The uncertainty interval of each parameter is estimated, assuming, however, that no rating change has been ignored by the station manager. It is thus possible to clearly distinguish overall trends of the channel bed level from the local evolution of riffles and to evaluate whether the observed temporal changes are significant compared to the estimation uncertainties.
