Research update

Knowledge gains: Showcasing climate emulators, how they can be used, and how they compare to CMIP6 models

State-of-the-art tools

Climate emulators can be used to assess how the Earth’s climate responds to both human activities and natural factors. These emulators are considered to be “simple” climate models because they represent the most critical components of the Earth’s climate at a global scale.

In comparison, more traditional complex climate models, such as Earth system models, solve physical equations with a view to actually representing atmospheric and oceanic circulation in three dimensions. In other words, simple climate emulators directly calculate the Earth’s global mean climate, whereas more complex models have to first assess the Earth’s weather.

Time and cost saving

The main advantage of the simple climate models is that they can be run from a small personal computer, rather than a supercomputer, taking seconds instead of weeks.  This means that emulators are the best available tools to explore a wide range of climate scenarios, at a low cost and with instant results.

This is particularly helpful for assessing which emissions reduction pathways are compatible with the climate targets set by international agreements.  For example, the latest IPCC report on mitigation of climate change included hundreds of emulator simulations.

The figure below shows one hypothetical simulation (using the FaIR v1.6.3), where CO2 emissions from fossil fuels are stopped in 2023, showing that past CO2 emissions cause no further surface warming in the long term.


click on image to enlarge

Calibration is key

The skill of the emulator greatly depends on the climate models, scenarios, and climate metrics used to calibrate it. Emulators that are calibrated in a similar way are more consistent with each other.

Emulators are typically calibrated against a small number of climate model simulations, for instance those made in the sixth Coupled Model Intercomparison Project (CMIP6). These simulations include historical conditions (where climate models reproduce the observed past warming), standardised diagnostic experiments (e.g. with quadrupling atmospheric CO2 concentrations), or future climate projections based on shared socio-economic pathways with pre-defined emission scenarios.

In principle, a well-calibrated emulator should accurately reproduce the results of climate models for all these types of simulations. To test this, CONSTRAIN carried out a series of intercomparisons of similarly calibrated emulators.  The results generally agreed and although the emulators projected lower warming than CMIP6 models for a given scenario, this is because of the high proportion of CMIP6 climate models producing very strong warming.

Future developments in CONSTRAIN

We will continue to improve our ability to make regional temperature projections using emulators. As temperature changes on a local scale are often related to global mean temperature change, it is possible to predict regional warming patterns from emulator-based projections of global warming. This technique is referred to as pattern scaling, and improves the application of emulators to climate impact modelling.

read the full report