Should we correct biases in the diurnal cycle of climate model for hydrological studies?

With the growing importance of climate change risk assessment, the use of climate models as a tool to model the impact of a warmer climate on water resources has now become quite common. When working with climate model outputs, bias correction is considered an important and necessary step to ensure that impact models provide realistic simulations in the current and future climates. The past decades have seen continuous improvements in the spatial and temporal resolution of global and regional climate models. Climate model outputs are now available at the sub-daily temporal resolution and very few studies have looked at the need for correcting biases present in the representation of the diurnal cycles of model variables. This study has looked at the impact of correcting such biases on simulated streamflow over 133 North American catchments. The temperature and precipitation hourly outputs from a 50-member large-ensemble regional climate model (ClimEx) were used to model the impact of sub-daily bias correction on simulated streamflow using a hydrological model. To better understand the importance of diurnal bias correction as a function of the spatial scale, the impact of bias-correcting the diurnal cycle was evaluated on three classes of catchment area:  small (<500 km²), medium (500< area <1000 km²) and large (>1000 km²). Bias correcting the diurnal cycle resulted in small but systematic improvements in the representation of simulated streamflow, with an average bias reduction of 5%, likely due to a better representation of the daily evapotranspiration cycle.  The improvements were especially noticeable on the small catchments.