Generating precipitation ensembles for flood alert and risk management
Génération d'ensembles de précipitation pour l'alerte aux crues et la gestion de risque
Résumé
Precipitation fields are crucial for hydrological forecasting. A forecasting system comprising a rainfall-runoff model uses observed and forecast precipitation to forecast discharges several hours to days ahead, which will be used for flood alert and risk management. Many sources of uncertainty play a role in a hydrological forecasting system, including uncertainties coming from the observed precipitation fields used as input data. One approach to take into account these uncertainties is to generate an ensemble of possible scenarios of observed precipitation. The aim of this study is to create an ensemble of precipitation fields merging information from rainfall radar and rain gauges. To do that, the turning bands method (TBM) adapted by Leblois and Creutin (2013) is applied in a new manner, eg with: 1) the generator parameters estimated from radar rainfall fields and 2) the generated fields conditioned on precipitation data measured by rain gauges. The case of a rainfall event observed in the Var region (southeastern France) is used to illustrate the method. Results show that ensemble simulation experiments conditioned on point measurements can be useful to quantify the uncertainties of precipitation fields observed by rain gauges. The proposed method could be a solution to merge radar and rain gauges information, while quantifying the observed precipitations uncertainties.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...