Assessment of uncertainty in river flow extreme simulations for China using multiple forcings and hydrological models

Abstract:

The uncertainties in simulations of river flow extremes (RFEs) using multiple meteorological forcings and global hydrological models (GHMs) are examined for China during the historical (1971-2001) period. RFEs are defined by monthly maximum / minimum river flows and threshold-based (e.g. 95^th and 5^th percentiles) extreme flows. The uncertainties arising from the forcing data and models based on the hydrological simulations are quantified with and without human impact (water withdrawal) which is made available by the ISI-MIP2 project. The entropy theory is incorporated to measure the uncertainties transferred from forcings to the hydrological models. Firstly, monthly discharge and RFEs simulations with human impact are validated against observed station data in the major basins of China for all GHMs and forcings. Annual means of monthly maximum / minimum flows and frequency of threshold-based RFEs are computed, and their trends are estimated over the historical period. GHMs variance of RFEs are computed across all GHM simulations for each forcing individually and then averaged for all forcings, which is further used to characterize GHMs uncertainty in RFEs simulation. The separation of GHMs for different forcings is helpful to assess uncertainty sources in regional hydrological simulation. Then uncertainties are calculated in the term of entropy for simulations with and without human activities to see whether the inclusion of human impact simulation in GHMs reduces or increases the uncertainty. This research would be in favor of understanding of the uncertainty in assessment of hydrological extremes at regional scale in China and potential improvement of GHMs in future.