Monthly & Seasonal Forecasts now on ClimateData.ca: Learn what’s new, Explore the map.

Differences between CMIP5 and CMIP6 model outputs

Are there major differences between CMIP5 and CMIP6 model results?

A recent Canadian study has shown that while the CMIP5 and CMIP6 future climate simulations are qualitatively similar for Canada, CMIP6 simulations exhibit larger temperature (mean and extreme) and precipitation (extreme) changes by the end of this century. In another global-scale study, extreme high temperature and precipitation events are projected to become more frequent and intense in CMIP6 simulations. 

These responses can be partially explained by differences in GHG forcing between the high emissions scenarios – SSP5-8.5 (CMIP6) and RCP8.5 (CMIP5) and partly by the fact that some of the CMIP6 models exhibit higher climate sensitivity, as explained below. 

For more details, see: 

Sobie SR, Zwiers FW, Curry CL (2021): Climate model projections for Canada: A comparison of CMIP5 and CMIP6. Atmosphere-Ocean 59: 269-284. https://doi.org/10.1080/07055900.2021.2011103

Li C, Zwiers F, Zhang X, Li G, Sun Y, Wehner M (2021): Changes in annual extremes of daily temperature and precipitation in CMIP6 models. Journal of Climate, 34(9), 3441–3460. https://doi.org/10.1175/JCLI-D-19-1013.1

Why is the climate sensitivity higher in some of the CMIP6 models than in the last generation of models (CMIP5)? How would this be different from assessments/projects using CMIP5?

Equilibrium climate sensitivity refers to the amount of warming after the climate system has stabilized in response to an instantaneous doubling of CO2 in the atmosphere. The increase in global mean temperature when this equilibrium state is reached is used as a metric to describe how models respond to changes in forcing. Models with higher climate sensitivity project more warming in response to the same forcing than those with lower climate sensitivity. 

Some of the CMIP6 models (including the Canadian Model, CanESM5) have a higher equilibrium climate sensitivity than the models used in CMIP5. This is partially due to modelling enhancements in CMIP6, such as higher spatial resolution and improved cloud physics. Further research is underway to investigate these findings, which is a normal part of scientific process. 

For more details, see: 

Sobie SR, Zwiers FW, Curry CL (2021): Climate model projections for Canada: A comparison of CMIP5 and CMIP6. Atmosphere-Ocean 59: 269-284. https://doi.org/10.1080/07055900.2021.2011103

Why do historical conditions simulated from CMIP5 and CMIP6 models differ from actual historical observations?

Climate model simulations of the historical period do not match actual observations because climate models are imperfect representations of the Earth system. Each model simulation will be different, and it is important to consider multiple model runs as each run is a possible representation of the climate and its natural variability. 

Importantly, however, monthly, seasonal and annual climatological averages and variations are similar between simulated and observed historical climates when regional or larger scale averages are considered. 

Because models and meteorological observations do not generally represent information at the same spatial scales, it is important to use modelled historical data when making direct comparisons with modelled future data.

Have the CMIP6 data been downscaled in the same way as the CMIP5 data?

The same method used to downscale CMIP5 data (known as BCCAQv2) was used to downscale CMIP6 data to produce CanDCS-U6. However, a new downscaling method (known as MBCn) was applied to the CMIP6 data to produce the current dataset on ClimateData.ca, CanDCS-M6. More information on the downscaling of global climate model data can be found at the Pacific Climate Impacts Consortium website here.

Does the release of CMIP6 data mean that the work I completed using CMIP5 data is obsolete and will need to be redone?

That depends. Although CMIP6 climate models generally include more climate processes, have been run at higher spatial resolutions, and use updated emissions scenarios, there is no such thing as a “perfect” global climate model. Nearly all  CMIP6 models build off those used in CMIP5, and CMIP6 model results exhibit many similarities to those of CMIP5. 

CMIP5 data are  still valid, and can be used to explore possible future climates. Both CMIP5 and CMIP6 rely on emissions scenarios with comparable levels of radiative forcing at the end of the century (2100).

It is recommended, however, that you use CMIP6 for new work requiring future climate projections and that you examine the similarities/differences with CMIP5. Comparing CMIP5 and CMIP6 will indicate if there are significant differences between the two sets of projections and if you need to update any of the work which used projections from CMIP5.