Factsheet 19: CCAM – Conformal Cubic Atmospheric Model

What is CCAM?

The Queensland Future Climate Science Team has been using CCAM (Conformal Cubic Atmospheric Model), a climate model developed by CSIRO, to dynamically downscale global climate models (GCMs), to produce high resolution (10 km) regional climate projections. The team has downscaled the two latest generations of GCMs (i.e., CMIP5 and CMIP6). Dynamical downscaling is the process of running a climate model to improve the resolution of a coarse resolution model, such as a GCM. CCAM is an open-source climate model and its full scientific description of dynamical, physical and chemical components (Figure 1) is available online.

Dynamical downscaling is often accomplished using a regional climate model – a climate model that runs for a limited region, for example, Australia, at a high resolution. However, variable resolution global climate models, such as CCAM, can also be used to dynamically downscale. This approach avoids the use of boundary conditions and their associated errors, while reducing the dependence of results on the size of the domain.

CCAM has been developed as a global variable resolution model and can be used as stand-alone climate model and in downscaling. CCAM has a stretched grid (Figure 2), and over an area of interest, such as Australia, it runs at a high resolution (Figure 3).

CCAM model setup

The Queensland Future Climate Science Team has completed a comprehensive set of downscaled simulations for the Australasian domain following the CORDEX-CMIP6 experimental protocol using CCAM. CORDEX is an international downscaling experiment, similar to the CMIP6 experiment but for regional models. Participating in CORDEX allows our models to be used by researchers in Australia and internationally, and for our models to be compared to others available, which is vital to improve model performance, and to make sure we are using the best available science for Queensland.

We ran CCAM in atmosphere-only and ocean-coupled mode. Ocean-coupled mode means we used the ocean model in CCAM, and allowed interactions between the atmosphere and ocean. Atmosphere-only mode does not use the ocean model, and the only information from the ocean is the bias-corrected sea surface temperatures (SSTs) from the host model (GCM that is being downscaled). There are pros and cons to both modes – using the atmosphere-only bias corrected SSTs can reduce some biases from the ocean model, however interactions between the atmosphere and ocean are also important for things like tropical cyclones and ENSO. Using both modes allows us to get the benefit from both, and better understand the strengths and weaknesses of the model for Queensland, and improve future performance.

We ran CCAM used a stretched C288 grid, with approximately a 10km resolution over Australia (Figure 3). CCAM is a stretched grid model, so while we ran at a 10km resolution over Australia, the rest of the globe had a different resolution, defined by the C288 grid, as shown in Figure 3.

A climate model runs for the atmosphere as well as the surface of the earth. The model can be ran with different vertical levels in the atmosphere. We ran CCAM with 35 vertical levels in the atmosphere, and for the ocean-coupled models, 30 levels in the ocean.

CCAM was forced using the CMIP6 radiative forcings for the historical period and future climate change scenarios (SSPs) and by the bias-corrected sea surface temperatures (SSTs) and sea ice from the host CMIP6 GCMs. Radiative forcings include solar radiation, natural and anthropogenic aerosols, transient land use, ozone and greenhouse gases. For the atmosphere-only models, CCAM was run using bias-corrected SSTs. Bias correcting SSTs prior to downscaling has been found to improve model performance for temperature and precipitation biases. We also ran some models in ocean-coupled mode. For those models, spectral nudging was applied to bias-corrected SSTs from the host model over the ocean domain. Often, GCMs are downscaled using limited-area models. In these cases, information from the GCM comes into the high-resolution model every 3 – 6 hours at the boundary. CCAM however is a stretched grid model, and runs for the entire globe. The information from the host GCM comes into CCAM via spectral nudging which constrains CCAM to follow the host GCM at large scales, but allows the small scales to evolve freely. See Table 1 for a list of downscaled models, and whether they were run in atmosphere-only or ocean-coupled mode.

Table 1: CMIP6 GCMs and ensemble members selected for downscaling. Latitude and longitude grid size shown in resolution column. GCMs were downscaled using different CCAM configurations – atmospheric model only or atmosphere-ocean coupled.

CMIP6 Model	Model full name	Resolution	Ensemble member	CCAM configuration
ACCESS-ESM1.5	Australian Community Climate and Earth System Simulator, version 1.5	1.875 x 1.25°	r6i1p1f1 r20i1p1f1 r40i1p1f1	atmospheric atm-ocean coupled atm-ocean coupled
ACCESS_CM2_oc	Australian Community Climate and Earth System Simulator, version 2	1.875 x 1.25°	r2i1p1f1	atm-ocean coupled
CMCC-ESM2	Centro Euro-Mediterraneo sui Cambiamenti Climatici Earth System Model, version 2	0.9 x 1.25°	r1i1p1f1	atmospheric
CNRM-CM6-1-HR	Centre National de Recherches Météorologiques Coupled Global Climate Model, version 6.1, high-resolution	0.5 x 0.5°	r1i1p1f2 r1i1p1f2	atmospheric atm-ocean coupled
EC-Earth3	European Community Earth-System Model, version 3	0.8 x 0.8°	r1i1p1f1	atmospheric
FGOALS-g3	Flexible Global Ocean-Atmosphere-Land System Model, grid point version 3	2.5 x 2.5°	r4i1p1f1	atmospheric
GFDL-ESM2M	Geophysical Fluid Dynamics Laboratory Earth System Model, version 4	1 x 1°	r1i1p1f1	atmospheric
GISS-E2-2-G	Goddard Institute for Space Studies Model E2.2G	2. x 2.5°	r2i1p1f2	atmospheric
MPI-ESM1-2-LR	Max Planck Institute Earth System Model, version 1.2, low resolution	1.9 x 1.9°	r9i1p1f1	atmospheric
MRI-ESM2-0	Meteorological Research Institute Earth System Model, version 2.0	1.125 x 1.125°	r1i1p1f1	atmospheric
NorESM1-MM	Norwegian Earth System Model, version 2, 1 degree resolution	1 x 1°	r1i1p1f1 r1i1p1f1	atmospheric atm-ocean coupled

Data from the simulations is available in gridded netCDF format on NCI. This data also forms the basis of the Queensland Future Climate Dashboard, and data can be downloaded from there in csv format.

CCAM land surface model

We ran CCAM with the CABLE (Community Atmosphere Biosphere Land Exchange) land surface model. CABLE supports vegetation tiles, changing land-use and a carbon cycle, enabling representation of landscape heterogeneity within grid cells.

Using the CMIP6 land use change scenarios has caused some small artifacts in some areas of Queensland, which are most obvious when looking at 10-meter wind speed. In some areas the height of the vegetation has increased, which has caused wind speeds to slow down. This has minimal effect on indices other than wind speed and evaporation.

Further information

Further details on CCAM are available in:
Chapman et al., 2023: Evaluation of Dynamically Downscaled CMIP6-CCAM Models Over Australia,
Schroeter et al., A Comprehensive Evaluation of Mean and Extreme Climate for the Conformal Cubic Atmospheric Model (CCAM) and
Conformal Cubic Atmospheric Model (csiro.au).