Evaluation of bias correction methods in global climate models and their application in hydrological impact studies

MONTROULL, NATALIA; SAURRAL, RAMIRO; CAMILLONI, INÉS

Exeter

Workshop; 4th WGNE workshop on systematic errors in weather and climate models; 2013

Met Office

The hydrological cycle of the La Plata Basin (LPB) in South America is simulated using the Variable Infiltration Capacity (VIC) model and forced with atmospheric data from five climate models of the CMIP5 database to determine to what extent errors in temperature and precipitation fields impact the representation of the hydrology of the basin. As GCMs still have difficulties in representing the present climate at regional scales, bias correction methods are required to guarantee a correct representation of the hydrological processes. In this study, we evaluate the performance of two bias correction schemes (fitted histogram equalization function and quantile-based mapping) and compare the simulations of the water cycle of the LPB using the unbiased and biased meteorological data as input. The bias correction factors are derived using the 1981-90 observed (CRU v3.1) and simulated data and then applied to 1991-2000 simulations. Almost all climate models considered show too many wet days with low-intensity rain and an incorrect spatial representation of precipitation in LPB. The lack of accuracy in the meteorological data derived in and underestimation of the mean streamflows and, in some cases, also the seasonal water cycle is misrepresented. Validation against observed data shows that both statistical methods successfully remove the biases in all seasons improving the simulations of discharges. However, none of the statistical schemes shows the best performance in all GCMs. This study highlights the importance of removing systematic errors of GCMs when they are used as input data of hydrological models. Consequently, the better understanding of the sources of these biases could contribute to the improvement of hydrological impact assessment of climate change.