Finding the Optimal Multimodel Averaging Method for Global Hydrological Simulations

Global gridded precipitations have been extensively considered as the input of hydrological models for runoff simulations around world. However, limitations hydrologic and inaccuracies precipitation datasets could result in large uncertainty forecasts water resource estimations. Therefore, it is great importance to investigate value a weighted combination driven by different datasets. In addition, due diversities members climate conditions, simulation watersheds under conditions may show various sensitivities combinations. This study undertakes comprehensive analysis multimodel averaging methods schemes (i.e., averaging) identify most skillful reliable application. To achieve this, four six were used members. The behaviors 9 11 tested over 2277 distributed regions results following: (1) multi-input consist one model multiple datasets) generally perform better than same dataset) each method; (2) use can improve performances. Six are found be necessary advisable, since using more only imrpoves estimation slightly, compared with all 24 members; (3) advantage modeling region dependent. methods, general, produced best warm temperate region, followed snow equatorial regions, while difference among arid arctic regions. mainly being affected extent poorly performed regions; (4) superensemble method (MMSE) recommended its robust outstanding performance climatic