Uncertainty Analysis of CMIP5 Future Projection in Global Water Cycle

Abstract:

In the impact assessment of climate change, the uncertainty analysis of the model projection (e.g., ensemble spread) is an indispensable part, since it directly affects the reliability of hydrologic simulation and associated assessment. This study estimates relative uncertainty of hydrologic simulations under projected future climates. The uncertainty is classified into three types of 1) model structure, 2) climate projection scenario, and 3) bias correction method. A huge dataset with 160 ensemble members is prepared as the combination of ten CMIP5 GCMs with two Representative Concentration Pathways scenarios and three existing and a newly devised bias correction method with two reference dataset.

Ensemble simulations are performed over the 0.5 degree global grids using a global hydrological model H08 and the bias corrected forcing variables (i.e., temperature, precipitation, surface air pressure, air humidity, long/short wave radiation, wind speed and Albedo). We found the calculated river discharge shows the large spread of discharge between the different types of bias correction methods, which is comparable to the spread of the GCM projections. In some regions, difference of trend between 21 and 20 century is even located within the envelope of the uncertainty of bias correction methods.