CORDEX – An international climate downscaling initiative

Abstract

The World Climate Research Programme (WCRP) is backing an international initiative called the COordinated Regional climate Downscaling EXperiment (CORDEX). The goal of the initiative is to provide regionally downscaled climate projections for most land regions of the globe, as a compliment to the global climate model projections performed within the fifth Coupled Model Intercomparison Project (CMIP5). CORDEX includes data from both dynamical and statistical downscaling. It is anticipated that the CORDEX dataset will provide a link to the impacts and adaptation community through its better resolution and regional focus. Participation in CORDEX is open and any researchers performing climate downscaling are encourage to engage with the initiative. The model evaluation framework consists of RCM simulations performed using the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim re-analysis (Uppala et al., 2008) as perfect boundary conditions. These simulations will be evaluated through a set of regional benchmark statistics, against regional datasets, that are designed and assembled by regional diagnostic teams. The climate projection framework within CORDEX is based on the set of new GCM simulations currently underway in support of the IPCC fifth Assessment Report, referred to as CMIP5. CORDEX will focus on the GCM experiments using emission scenarios known as RCP4.5 and RCP8.5 which represent a mid and a high-level emission scenario. Ideally CORDEX simulations will span 1951 to 2100 though given the computational demand of such long transient runs a series of 30 year time slices is also possible. Each RCM should perform these simulations for multiple GCMs. While CORDEX is focused on using CMIP5 simulations for boundary conditions, which are still being created by the GCM groups, NARCliM uses the same AustralAsia and south-east Australia domains to downscale CMIP3 simulations. One of the unique aspects of both the CORDEX and NARCliM databases will be the presence of many three hourly fields. This will allow investigations of the simulation of diurnal cycles in RCM ensembles that have not been previously possible.