The transferability of perturbed parameters across different regions and the effects of perturbed parameters on climatic extremes

Uncertain physical parameters in climate models give rise to an important source of uncertainties in earth system modelling. Parameter tuning often involves many choices and decisions that could significantly influence model results, and the parameter space that lead to improved model performance over a certain region could deteriorate model performance in other regions. Using a PPE previously selected to reduce summer mean warm and dry biases in a regional climate model (RCM) over Western North America, this study applies these PPE in an RCM over Europe to explore transferability of the results. The study demonstrates that the same PPE is also effective in reducing model biases over the Europe domain, and explores the key processes responsible for such improvements. We also investigate how the higher moments of the distributions of different climatic variables respond to parameter perturbations, and demonstrate that biases in the high tails of distributions are only improved to the same extent of the seasonal mean. This result has great implications for parameter tuning in the model development process, because if the tuning is focused on model performance of the mean climate states, perhaps it's unrealistic to expect the representation of climatic extremes to improve drastically through new generations of climate models.