Assessment of forced responses of the Australian Community Climate and Earth System Simulator (ACCESS) 1.3 in CMIP5 historical detection and attribution experiments.

Abstract

The Australian Community Climate and Earth System Simulator (ACCESS) coupled climate model version 1.3 participated in phase five of the Coupled Model Intercomparison Project (CMIP5) with an initial contribution of high priority experiments. Further standard experiments have since been conducted with ACCESS1.3, including an ensemble of three simulations for the historical period (1850–2005) forced with time-evolving natural and anthropogenic forcings. Additional ensembles of simulations have been conducted for the same period with subsets of known forcings, including with natural forcings only (‘historicalNat’) and with greenhouse gas forcings only (‘historicalGHG’). In this study, we describe this ACCESS1.3 contribution to CMIP5 and assess several key aspects of ACCESS1.3 forced responses in these experiments against observations and an ensemble of participating CMIP5 models, consisting of 40 realisations from ten models. Overall, ACCESS1.3 historical experiments demonstrate skill in simulating the global and regional metrics assessed that is comparable to the CMIP5 multi-model ensemble utilised. Global annual average temperature and precipitation trends simulated with ACCESS1.3 (0.05–0.07 K/decade; −0.007 to −0.0004 mm day-1/decade) largely lie within the CMIP5 ensemble window (0.06 − 0.18 K/decade; −0.01 to 0.009 mm day-1/decade) and near those observed (0.10 K/decade; −0.0007 to −0.001 mm day-1/decade) over the 1950–2005 period. For the ACCESS1.3 historicalNat and historicalGHG experiments, simulated temperature trends are also predominately within the CMIP5 multi-model ensemble range. Similarly, ACCESS1.3 (−0.07 to −0.12 K) and the CMIP5 models (−0.03 to −0.21 K) largely capture the composited observed decrease in global temperature (−0.04 K) following three major late 20th century volcanic eruptions. However, like all global climate models, ACCESS1.3 has deficiencies that should be considered. In particular, one of most notable features of ACCESS1.3 historical simulations is the reduced warming trend over the period 1950–2005 that is evident in all ACCESS1.3 realisations at the global-scale for Australia, relative to both observations and the CMIP5 multi-model mean. This appears to be related to the overly strong response to increases in anthropogenic aerosols. Overall, these historical period experiments using ACCESS1.3 with various forcings are useful for inclusion with other CMIP5 models for studies aimed at detecting and attributing climatic changes.