Premium
Ensemble Skill Gains Obtained From the Multi‐Physics Versus Multi‐Model Approaches for Continental‐Scale Hydrological Simulations
Author(s) -
Fei Wenli,
Zheng Hui,
Xu Zhongfeng,
Wu WenYing,
Lin Peirong,
Tian Ye,
Guo Mengyao,
She Dunxian,
Li Lingcheng,
Li Kai,
Yang ZongLiang
Publication year - 2021
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/2020wr028846
Subject(s) - ensemble forecasting , statistical physics , data assimilation , independence (probability theory) , ensemble learning , ensemble average , anomaly (physics) , scale (ratio) , physics , canonical ensemble , meteorology , computer science , artificial intelligence , statistics , monte carlo method , mathematics , climatology , geology , quantum mechanics
Abstract Multi‐physics ensembles have emerged as a promising approach to hydrological simulations. As multi‐physics ensembles are constructed by perturbing the model physics, the ensemble members share a substantial portion of the same physics and hence are not independent of each other. It is unknown whether and to what extent this nonindependence affects the skill gain of the ensemble method, especially compared with the multi‐model ensemble approach. This study compares a multi‐physics ensemble configured from the Noah land surface model with multi‐parameterization options (Noah‐MP) with the North American Land Data Assimilation System (NLDAS) multi‐model ensemble. The two ensembles are evaluated in terms of the annual cycle and interannual anomaly at 12 River Forecast Centers over the conterminous United States. The ensemble skill gain is measured by the difference between the performance of the ensemble mean and the average of the ensemble members' performance, and the inter‐member independence is measured by error correlations. Results show that, due to the improved model physics, the Noah‐MP configurations outperform, on average, the NLDAS models, especially in the snow‐dominated areas. The Noah‐MP ensemble almost always obtains an outstanding member that performs the best among the two ensembles, reflecting its dense sampling of the feasible model physics space. However, these two performance superiorities do not lead to a superiority of the ensemble mean. The Noah‐MP ensemble has a lower ensemble skill gain, which corresponds to the lower inter‐member independence. These results highlight the importance of inter‐member independence, particularly when most hydrological ensemble methods have overlooked it.