Systematic temperature and precipitation biases in the CLARIS-LPB ensemble simulations over South America and possible implications for climate change projections

SILVINA A SOLMAN

Lund

Workshop; 3rd International Lund Regional-Scale Climate Modelling Workshop 21st Century Challenges in Regional Climate Modelling; 2014

Lund University and Baltic Earth

Within the framework of the CLARIS-LPB EU Project, a suite of 7 coordinated Regional Climate Model (RCM) simulations over South America were performed. The ability of this set of RCMs (driven by the ERA-Interim reanalysis) in reproducing the observed climate conditions was evaluated in a recent study (Solman et al. 2013) and several systematic biases were identified. In particular, most of the RCMs showed a systematic temperature overestimation and precipitation underestimation over the La Plata Basin (LPB) region. These systematic biases were shared by every RCM, suggesting a common shortcoming of the models. In this study we analyze the bias in the mean climate but for the present climate simulations driven by CMIP3 GCMs. We demonstrate that these systematic model errors are more dependent on the RCMs rather than on the driving GCMs. Moreover, we demonstrate that model biases are not invariant, but a temperature-dependent temperature bias and a precipitation-dependent precipitation bias were identified for the LPB region. A similar behavior was also found for other regions of the world (e. g. Christensen and Boberg, 2012). Under warmer and wetter climate conditions for the future, these biases may amplify the climate change signal. A bias correction method based on a quantile-based mapping was applied to model projections in order to account for the identified systematic biases.