Watershed

Oral

394234 - Enhancing ensemble streamflow forecast skill through model diversity

Thursday, June 7
10:30 AM - 12:00 PM
Location: Greenway IJ
Co-Authors: Sanjib Sharma, State College – The Pennsylvania State University

Multimodel hydrological forecasting has been a topic of substantial interest in both operational forecasting and academic research. In this study, we investigate the skill of multimodel ensemble streamflow forecasts relative to single model forecast. For this purpose, short-to medium-range (days 1 to 7) ensemble streamflow forecasts are generated from three different hydrological models: i) lumped, conceptual Antecedent Precipitation Index (API)-Continuous model, ii) Hydrology Laboratory-Research Distributed Hydrologic Model (HL-RDHM), and iii) Weather Research and Forecasting Hydrological (WRF-Hydro) model extension package. We also introduce the Quantile Regression-Bayesian Model Averaging (QR-BMA) method for combining bias corrected predictive distributions from different hydrological models. Results demonstrate that the multimodel ensemble forecasts have higher skill than the best single model. The nature of skill enhancement through the multimodel ensemble forecasts is further investigated using conditional mutual information theory. Indeed, we demonstrate that the enhanced skill in the multimodel combination is dominated by model diversity, rather than from increased ensemble size alone.

Alfonso Mejia

Assistant Professor
The Pennsylvania State University

Presentation(s):

Send Email for Alfonso Mejia


Assets

394234 - Enhancing ensemble streamflow forecast skill through model diversity



Attendees who have favorited this

Please enter your access key

The asset you are trying to access is locked. Please enter your access key to unlock.

Send Email for Enhancing ensemble streamflow forecast skill through model diversity