Canada’s territory and the Arctic regions offer distinct challenges to Numerical Weather Prediction (NWP) and climate projection, due to complex processes and feedbacks between various components of the climate system. A better understanding of these regional climate processes and interactions is crucial to improving the quality of both climate projection and NWP for this region, and to better interpret and apply model results for use in weather and climate-change impact and adaptation studies.
The aim of the Network is to augment, evaluate and exploit the added value provided by regional models in climate and weather simulations. This added value is afforded as a result of the expected higher resolution, improved representation of physical processes, feedbacks and interactions through a Regional Earth System Model approach.
Over the past decade, the emphasis in climate-change studies has shifted from studying expected changes in the mean climate (mostly temperature and precipitation), to studying changes in the frequency distribution of weather and hydrological variables, including extremes (e.g. Sushama et al., 2010; Mladjic et al., 2011). In many cases climate extremes are simply the statistical footprint of high-impact weather events (e.g. flash floods); in other cases they reflect persistent anomalies (e.g. heat waves, droughts). Credible climate simulations require adequate representation of the weather that constitutes climate; conversely reducing systematic, climatological biases in forecast models is required to improve NWP. It is for this reason that the WCRP promotes ‘seamless prediction’ using models adapted to both NWP and climate simulations.
Global Climate Models (GCMs) coupling atmospheric, terrestrial, ocean and sea-ice components of the Earth System constitute the most comprehensive tools to make climate-change projections. GCMs have demonstrated considerable skill at reproducing the planetary and continental-scale features of the general circulation of the atmosphere and the ocean. Although there are developmental versions of GCMs that are tested with grid meshes of 20 km, operational GCMs use relatively coarse computational meshes compared to NWP models, due to their high computational costs resulting from the complexity of GCMs’ components, the slow adjustment of the deep ocean, and the need of ensemble simulations for statistical significance. As a result, the resolution of common GCMs is not adequate for representing several key regional and local climate processes (Laprise, 2008; Rummukainen, 2010), or to satisfy the expectations of the climate impacts and adaptation community. Dynamical downscaling using nested limited-area models (LAM) known as Regional Climate Models (RCMs) allows using meshes an order of magnitude finer over a region of interest, at an affordable computational cost. Hence RCMs are used to ‘add details’ to GCM simulations (Feser et al., 2011) and to study climate and weather processes occurring at finer spatial scales. Currently RCM grid meshes of 15 km are feasible over domains covering most of North America, and even finer meshes over smaller domains. At such resolution, geographical features such as valleys in the Rockies, the Great Lakes, the St. Lawrence River Valley and Estuary, James Bay, and the channels of the Canadian Arctic Archipelago, become resolved as well as the effects they induce on local weather and climate.
The focus of the proposed research will be on time scales ranging from a few seasons to several decades, covering the period starting around 1950 when reanalyses become available, up to 2100 under various Representative Concentration Pathways (RCPs). Through a series of numerical experiments and analyses of observations, the Network will exploit the added value of high-resolution modelling with an improved representation of land-surface heterogeneity and diversity of forcing, as reflected in the skill to represent specific weather and climate processes.