Multi-Model Grand Ensemble Hydrologic Forecasting in the Fu River Basin Using Bayesian Model Averaging

Abstract

Statistical post-processing for multi-model grand ensemble (GE) hydrologic predictions is necessary, in order to achieve more accurate and reliable probabilistic forecasts. This paper presents a case study which applies Bayesian model averaging (BMA) to statistically post-process raw GE runoff forecasts in the Fu River basin in China, at lead times ranging from 6 to 120 h. The raw forecasts were generated by running the Xinanjiang hydrologic model with ensemble forecasts (164 forecast members), using seven different "THORPEX Interactive Grand Global Ensemble" (TIGGE) weather centres as forcing inputs. Some measures, such as data transformation and high-dimensional optimization, were included in the experiment after considering the practical water regime and data conditions. The results indicate that the BMA post-processing method is capable of improving the performance of raw GE runoff forecasts, yielding more calibrated and sharp predictive probability density functions (PDFs), over a range of lead times from 24 to 120 h. The analysis of percentile forecasts in two different flood events illustrates the great potential and prospects of BMA GE probabilistic river discharge forecasts, for taking precautions against severe flooding events.

Performance evaluation of sub‐daily ensemble precipitation forecasts

“…For example, the construction of a grand ensemble using the BMA via combining the UKMO, NCEP, ECMWF and CMA models showed that the UKMO and ECMWF had greater skills, while the grand ensemble of the four centres was recommended for heavy precipitation forecast in the study region (Liu and Xie, 2014). Qu et al (2017) constructed a BMA grand ensemble, and through a distributed hydrologic model, they forecasted river discharge associated with 24-120 hr lead times and concluded that the BMA post-processing resulted in more reliable probabilistic discharge forecast compared with relying on raw ensemble numerical precipitation forecasts. Overall, the results of most studies suggest that the BMA can enhance the probabilistic forecast skills (Vogel et al, 2018;Zhong et al, 2018).…”

Generating Coherent Ensemble Forecasts After Hydrological Postprocessing: Adaptations of ECC‐Based Methods

“…It must be noted that GAM have been necessary in a couple of cases only, due to the relatively long observation series used for specifying the distributions . As an alternative to the eNQT, the Box‐Cox transformation (Box & Cox, ) has been used by Duan et al (), Hemri et al (), Madadgar and Moradkhani (), and Qu et al (), but was found to perform slightly worse than eNQT in our case study.…”

“…For streamflow postprocessing, we applied the Bayesian model averaging (BMA) approach, which was introduced by Raftery et al (2005) and adapted by Fraley et al (2010) to deal with groups of exchangeable members. BMA has been considered for streamflow postprocessing in several past studies (Duan et al, 2007;Hemri et al, 2013;Madadgar & Moradkhani, 2014;Qu et al, 2017). The next paragraphs give a brief description of the procedure, but as it is not the core of the paper we encourage interested readers to read the above references.…”

Evaluation of ensemble precipitation forecasts generated through postprocessing in a Canadian catchment

“…The added advantage of using ensemble forecasts over deterministic forecasts have been addressed in many previous studies (e.g., (Abaza et al, 2013;Boucher et al, 2011) Medium-range Weather Forecasts (ECMWF), The Japan Meteorological Agency (JMA), The National Center for Environmental Prediction (NCEP), The Canadian Meteorological Center (CMC), etc.) seem to be a preferred choice Zsótér et al, 2016;Qu et al, 2017;Hamill, 2012).…”