Assessing the influence of bias correction of boundary conditions, spectral nudging and model parameterisation on model errors and climate change signals for regional climate model simulations

Abstract Some of the most important considerations when undertaking dynamical downscaling of global climate models (GCMs) using regional climate models are the choice of model physical parameterisations, the use of spectral nudging, and whether to bias correct the driving GCMs prior to downscaling. While each of these factors have been extensively examined, very few studies have compared the effect of all 3 on model biases against independent observations during the historical period, as well as the change in future climate. We carry out this analysis and focus on the CORDEX-Australasia domain with all simulations driven using a common GCM. We found that the choice of model parameterisaton schemes had by far the largest influence on model biases and the change in climate, especially for precipitation during summer. While bias correction reduced large systematic biases for some variables in some regions, it also increased biases elsewhere, and results were not consistent for all variables. Our results show that it is important to first assess the performance of non-corrected GCM-driven simulations against the reference re-analysis driven simulations, as bias correction may not be necessary if the GCM-driven simulation already performs well compared to the reference simulation. Spectral nudging had a limited influence on both model biases and the change in climate, except for summer precipitation in the tropics. While we only use a single RCM and a single GCM, our key finding is that given limited computational and data resources, regional climate modelling groups should prioritize a multi-physics ensemble of the RCM to better account for internal physics-driven variability, over the use of bias-correction or spectral nudging.