Eight-MODEL sub-set FOR APPLICATION-READY DATA

Eight of the 40 CMIP5 models assessed in this project have been selected for use in provision of application-ready data. This facilitates efficient exploration of climate projections for Australia.

Model selection

A number of steps were considered in the model selection process:

  1. Rejection of models that were found to have a low performance ranking across a number of metrics in Chapter 5 and in some other relevant assessments (see Chapter 9 of the Technical Report for a full description).
  1. Selection of models for which projection data were available for climate variables commonly used in impact assessments, for at least RCP4.5 and RCP8.5. Projections for other RCPs are included where possible.
  1. Amongst these, identification of models that are representative of the range of seasonal temperature and rainfall projections for a climate centred on 2050 and 2090 and RCP4.5 and RCP8.5 using the Australian Climate Futures software.
  1. Projections for wind were assessed separately from temperature and rainfall to ensure the CMIP5 range was captured. This is because the direction and magnitude of wind projections are not necessarily correlated with the temperature and/or rainfall projections.
  1. Availability of corresponding statistical or dynamical downscaled data.
  1. Consideration of the independence of the models (see Knutti 2013)

Selected CMIP5 models and reasons for their inclusion

SELECTED MODELSCLIMATE FUTURESWINDOTHER
ACCESS1.0 Maximum consensus for many regions.The model exhibited a high skill score with regard to historical climate.
CESM1-CAM5 Hotter and wetter, or hotter and least dryingThis model was representative of a low change in an index of the Southern Annular Mode (per degree global warming). Further, the model has results representing all RCPs.
CNRM-CM5 Hot /wet end of range in Southern AustraliaThis model was representative of low warming/dry SST modes as described in Watterson (2012) (see Section 3.6). It also has a good representation of extreme El Niño in CMIP5 evaluations (see Cai et al., 2014).
GFDL-ESM2M Hotter and drier model for many clustersGreatest increase This model was representative of the hot/dry SST mode as described in Watterson (2012) (see Section 3.6). It also has a good representation of extreme El Niño in CMIP5 evaluations (see Cai et al., 2014). Further, the model has results representing all RCPs.
HadGEM2-CCMaximum consensus for many regions.Greatest reduction This model has good representation of extreme El Niño in CMIP5 evaluations (see Cai et al., 2014)
CanESM2This model was representative of the hot/wet SST mode as described in Watterson ( 2012) (Section 3.6). It also has a high skill score with regard to historical climate and it increased representation of the spread in genealogy of models (Knutti et al., 2013). It also has good representation of extreme El Niño in CMIP5 evaluations (Cai et al., 2014).
MIROC5
(non-commercial use only)
Low warming wetter modelThis model was representative of a higher change in an index of the Southern Annular mode (per degree global warming). It also has good representation of extreme El Niño in CMIP5 evaluations (see Cai et al., 2014). Further, the model has results representing all RCPs.
NorESM1-MLow warming wettest representative model No wind dataThis model was representative of the low warming/wet SST mode as described in Watterson (2012) (see Section 3.6). The model also has results representing all RCPs.

This table can be found in Box 9.2 of the Technical Report

Model Details

MODELINSTITUTEOCEAN RESOLUTION (°)ATMOSPHERE RESOLUTION (°)
[km at Equator]
ACCESS1.0CSIRO-BOM, Australia1.0×1.01.9×1.2
[210×130]
CanESM2CCCMA, Canada1.4×0.92.8×2.8
[310×310]
CESM1-CAM5NSF-DOE-NCAR, USA1.1×0.61.2×0.
[130×100]
CNRM-CM5CNRM-CERFACS, France1.0×0.81.4×1.4
[155×155]
GFDL-ESM2MNOAA, GFDL, USA1.0×1.02.5×2.0
[275×220]
HadGEM2-CCMOHC, UK1.0×1.01.9×1.2
[210×130]
MIROC5
(for non-commercial use only)
JAMSTEC, Japan1.6×1.41.4×1.4
[155×155]
NorESM1-MNCC, Norway1.1×0.62.5×1.9
[275×210]

This information is extracted from Table 3.3.1 of the Technical Report

Page updated: 26th April 2017