New climate simulations: Heading toward a 7°C rise?

By Florian Gallo, Senior Consultant

The news broke on September 17: the latest climate simulations conducted by various French research laboratories suggest a temperature increase of up to +6 to +7°C by the end of the century (compared to the preindustrial period). This is one degree higher than in previous simulations. What do these new projections mean? How do they differ from the previous ones? What does this mean in practical terms?

We offer an analysis of these latest scientific findings.

A quick reminder: the IPCC and its assessment reports

Before we talk about the models, it’s worth taking a moment to review what the IPCC is. The IPCC is the Intergovernmental Panel on Climate Change. This organization does not conduct any research of its own but aims to compile and assess the available scientific, technical, and socioeconomic information and to provide a synthesis of this information to the public and, more specifically, to policymakers. The IPCC publishes both technical reports on specific topics (such as last year’s Special Report on 1.5°C or last August’s Special Report on Land Use) and more comprehensive assessment reports, the latest of which (AR5) was released in 2014. The Sixth Assessment Report is scheduled for 2022, and the new simulations conducted by the CEA, Météo France, and the IPSL [1] fall within this broader framework.

Why new simulations?

Since 1980, the World Climate Research Program (WCRP) has provided a framework for research on climate change and climate modeling. Within this program, a major project is underway to compare the various climate models developed by research laboratories around the world (including the CEA, Météo-France, and IPSL). These projects, known as CMIP (Coupled Model Intercomparison Project), make it possible to coordinate and standardize the modeling efforts of the laboratories. Since 2007, these modeling projects have been organized to inform the IPCC’s assessment reports (each new generation of models is released a few years before the assessment report, so a significant portion of the literature analyzed in these reports is based on these models). AR5, published in 2014, thus synthesizes the results of the latest generation of models, known as CMIP5, which were released between 2010 and 2014.

So why not rely on these models? There are many reasons. First of all, The scientific understanding of climate phenomena is still evolving, and some processes are still being unraveled (while others are still poorly understood). It is therefore important to update the models regularly by incorporating the latest scientific advances.

In addition, Modeling capabilities—in terms of computing power, data processing, and data storage—are improving year after year. It is therefore possible to simulate past and future climates at an increasingly high resolution.

These new simulations require a considerable amount of computing power (for the French simulations, this amounts to a total of more than 500 million hours of computation for about 20 petabytes of data—one million billion bytes).

The new generation of models, CMIP6, will therefore serve as the basis for numerous scientific studies that will be included in the next IPCC report, scheduled for 2022. The approximately 20 climate modeling laboratories involved in the CMIP6 project have begun to release their initial results in recent months, and Early trends emerging from French models (from the CNRM [2] and the IPSL) were unveiled on September 17.

A larger-than-expected rise in temperature?

Preliminary results from the two French models suggest a greater increase in temperatures than the previous generation had projected. For the highest-emission scenario (SSP5 8.5, which corresponds to a continuous increase in greenhouse gas emissions through the end of the century, based on the trend observed over the past few decades), The temperature could rise by as much as 7°C (compared to the preindustrial period), as opposed to the 5–6°C increase previously projected. This shift in projections toward higher temperatures is particularly pronounced in the CNRM model, which is consistent with the IPSL model. The consistency between the two models is greater in this new generation of models. Furthermore, these results are also consistent with the initial analyses of other climate models around the world, presented in recent months.

Among the scenarios tested, Only the most optimistic scenario (SSP1 1.9) would keep the temperature increase below 1.5°C, but with a transitional phase at +2°C by mid-century. Furthermore, the realization of these optimistic scenarios implies a significant increase in negative emissions (CO2 capture) in order to remove CO2 from the atmosphere, in absolute terms, during the second half of the century. An average scenario, a scenario in which emissions plateau through 2050 and then begin to decline, without achieving carbon neutrality during the 21st century, would lead to a temperature increase of 3 to 4°C (half a degree higher than in previous simulations).

Some analyses have already been conducted on extreme events: based on an SSP3 7.0 scenario (warming of approximately +5.5°C by the end of the century), by 2050, one out of every two summers will exceed the temperature anomaly observed during the 2003 heat wave. By the end of the century, statistically speaking, every summer will exceed this threshold.

Why are there such differences?

Two questions arise when reviewing these results: 1) Are they reliable? and 2) What are the reasons for such differences?

The reliability of a climate model is measured by its ability to statistically reproduce the characteristics of the past climate (in this case, the 20th century for these simulations). The difference between past observations and the new simulations is narrowing, which demonstrates an improvement in the reliability of these new models.

The new simulations Therefore, they appear to be more reliable and to better represent the processes at work in the climate system. But that does not necessarily explain the projected rise in temperatures.

The main reason is an increase in climate sensitivity models. This climate sensitivity represents the climate system’s response to an increase in atmospheric CO₂ concentration (in other words, what is the corresponding temperature increase when CO₂ levels double?). Without going into technical details, the new simulations show a higher climate sensitivity than in the previous generation of models. In other words, in CO2 equivalent, higher temperature. This trend is, in fact, consistent with the results of other simulations conducted by other modeling centers.

However, the reasons for this higher climate sensitivity in the new models have not yet been explained: potential avenues of research into the role of water vapor or clouds in the climate cycle are discussed. This issue will be central to the analysis of the results over the coming months.

Comparison of CMIP5 (pastel colors) and CMIP6 (solid colors) simulations for the IPSL model (top) and the CNRM-Cerfacs model (bottom). For the highest-emission scenarios, a temperature increase of 0.5 to 1.5°C is observed between the two generations of models (source: CEA/CNRS/Météo-France)

Acronyms: [1] Pierre Simon Laplace Institute [2] National Center for Meteorological Research