Download PDFOpen PDF in browserMulti-Structure Hydrological Ensemble to Improve Flow Daily Prediction in the Sumapaz River Basin, Colombia8 pages•Published: September 20, 2018AbstractHydrological ensembles have gained importance for prediction and forecasting in water cycle variables. In spite of this, the relevance of the individual models in the ensemble is not usually established, in terms of the ensemble structure (i.e. their members) and the performance this structure exhibits through different climatic conditions (intrannual variability, for example). This analysis accounts for the uncertainty in the structure of the models and their responses (e.g. outputs), in comparison to the observed data. In this regard, the research here described attempts to determine the incidence of the ensemble members built for each month of the year, in the prediction of daily flows, through the use of the Bayesian Model Averaging (BMA) method. Moreover, using BMA calibrated parameters as inputs, an uncertainty analysis is carried out for the calibration period, and in monthly average terms, obtaining finer uncertainty bounds. This analysis was implemented in the Sumapaz River basin, part of the Magdalena Cauca Macro- Basin (MCMB) in Colombia. Results showed differences in ensemble structures and performance according to its original performance criteria, and better results when using a monthly BMA for the uncertainty analysis.Keyphrases: bayesian model average, hydrological ensemble, modelling In: Goffredo La Loggia, Gabriele Freni, Valeria Puleo and Mauro De Marchis (editors). HIC 2018. 13th International Conference on Hydroinformatics, vol 3, pages 62-69.
|
