“…Apart from the uncertainty due to the many empirical statistical techniques which have been developed to postprocess ESM outputs (Maraun, 2016), the quality and length of the reference observational dataset for bias correction remains a major issue (Reiter et al, 2016;Schoetter et al, 2012;Sippel et al, 2016). In Canada and other regions of North America, regional gridded datasets such as the combined Global Environmental Multiscale (GEM) atmospheric model forecasts (Yeh et al, 2002) and the Canadian Precipitation Analysis (CaPA; Mahfouf et al, 2007) have been found to perform comparably to ground observations, both statistically and hydrologically (Alavi et al, 2016;Boluwade et al, 2018;Eum et al, 2014;Fortin et al, 2015a;Gbambie et al, 2017;Wong et al, 2017). However, the duration of GEM-CaPA is too short to be used to directly correct ESM climate due to unsynchronized internal variability -the recommended minimum record length for bias correction is 30 years (Maraun, 2016;Maraun et al, 2017).…”