Evaluation of CMIP3 and CMIP5 Models over the Australian Region to Inform Confidence in Projections

Abstract

Model evaluation is an important tool to help rate confidence in climate model simulations. This can add to the overall confidence assessment for future projections of the Australian climate. Additionally it can highlight significant model deficiencies that may affect the selection of a subset of models for use in impact assessment. Here we present results from an extensive model evaluation undertaken as part of the Natural Resource Management (NRM) Project in order to inform the newest set of climate change projections for Australia. The assessment covers mean climate skill over Australia as well as variability measures and teleconnections from up to 47 CMIP5 models and 23 CMIP3 models (for comparison where appropriate). Additionally, the skill in representing important climate features such as MJO, SAM, blocking and cut-off lows are also reviewed. Selected extremes are evaluated as well as simulations of two different types of downscaling simulations used within the NRM project. Finally, an attempt is made to synthesise this information in order to highlight a small group of CMIP5 models which show consistent deficiencies in representing the Australian climate and its features