Use an idealized protocol to assess the nesting procedure in regional climate modelling

Abstract. Newtonian relaxation allowing RCM to follow GCM is a widely used technique for climate downscaling and regional weather forecasting. A thorough assessment on effects of the relaxation procedure in an idealized framework is presented in this paper for both synoptic variability and long-term mean climate. LMDz is a global atmospheric general circulation model that can be configured as a regional model if the outside domain of the focused region is applied with a relaxation. It thus plays the role of both GCM and RCM in this paper. Same physical parameterization and identical dynamical configuration are kept to ensure a rigorous comparison between the two models. The experimental set-up that can be referred to as “Master (GCM) versus Slave (RCM)” considers the GCM as the reference to assess the behavior of RCM. A further simulation with RCM in a higher resolution configuration allows isolating the effect of relaxation procedure from that of mesh refinement. In terms of mean climate in GCM and RCM, there are noticeable differences, not only in the border areas, but also within the domain. In terms of synoptic variability, there is a general spatial resemblance and temporal concomitance between the two models. But there is a dependence on variables, seasons, spatiotemporal scales and spatial mode of atmospheric circulation. Winter/Summer has the most/least resemblance between the RCM and the GCM. A better similarity is noticed when atmospheric circulations manifested on large scales. No-correlation cases can be remarked when the dominant circulation of the region is at a small scale.