How scientists uncovered and corrected a specific bias in climate models

August 13, 2024 - by Santina Russo

Humanity faces major challenges due to climate change, such as floods in cities, mudslides causing fatalities in mountain villages, rising sea level threatening the homes of millions of people, and even wars fuelled by water shortages. And there could be future threats we may not even know of yet.

It is evident: We need to better understand current and coming challenges related to climate change, and for this we need climate prediction models that are as precise as possible. In addition to the processes in the lowermost atmospheric layer, the troposphere, this also involves processes in the next higher layer, the stratosphere, and at the border between these two layers.

In particular, today's climate models do not correctly represent the effects of water vapour in the stratosphere, as a research team headed by Dr. Timofei Sukhodolov from the Physikalisch-Meteorologisches Observatorium / World Radiation Centre (PMOD/WRC) in Davos, Switzerland, recently confirmed. Through simulations performed on CSCS’s supercomputer “Piz Daint”, the team not only helped to determine the source of the associated errors in climate simulations’ outcomes but also to find a method to improve the representation of stratospheric vapour — and thereby the overall accuracy of climate models.

How water vapour shapes the climate

For as inconspicuous as water vapour might seem, it is the most impactful greenhouse gas in the atmosphere: Via feedback processes, it considerably enhances the warming caused by CO₂ alone. This is due to the water vapour’s radiation absorption properties: In the troposphere, it has a similar effect as CO₂ and redirects ascending infrared radiation back to the Earth enhancing global warming.

In the stratosphere, the increased amount of vapour increases emission of radiation in all directions. It increases radiation emissions back into space, which cools the stratosphere, but at the same time, it increases downward radiation, which again contributes to the warming at the Earth’s surface.

And even though the stratosphere only contains up to 20,000 times less water vapour than the troposphere, this small amount has a large impact. An impact that will intensify in the future, as models predict an increase of water vapour in the stratosphere in response to climate change, particularly near the border to the troposphere. In fact, the stratospheric vapour increase will induce a climate feedback that could be responsible for about ten percent of the increase in global mean surface temperature. “This makes it all the more important to have an accurate representation of stratospheric vapour in climate models,” says Eugene Rozanov, senior scientist and former head of the climate modelling group at PMOD/WRC.

The illustration shows the cycles in the Earth's atmosphere, including the influx of water vapour into the stratosphere. At the boundary from troposphere to stratosphere, natural conditions change: In the troposphere, the temperature is high at the Earth’s surface and gets increasingly colder going upwards, reaching around minus 80 degrees Celsius at the border to the stratosphere. Above this boundary, temperatures rise again, since in the stratosphere, the sun’s UV radiation is absorbed by the ozone layer and converted into heat. Also, the radiation absorption processes that govern weather and climate change at this barrier. (Image: Wikimedia Commons, Phillipe Rekacewicz)

The issue at the border

Together with an international team of collaborators, among else from Research Center Jülich, Ludwig Maximilians University of Munich, Complutense University of Madrid, and the University of Cambridge, Dr. Eugene Rozanov and Dr. Tatiana Egorova from PMOD/WRC investigated the issue and identified the main source of the problem in the way current models misrepresent water vapour in the stratosphere.

This has to do with the drastic difference between the vapour content in the troposphere and the stratosphere and the transport scheme across this border used in the models, as Rozanov explains. “Most models adopt transport schemes that overestimate diffusion and do not accurately reflect reality.”

The troposphere and the stratosphere are characterized by very different conditions in terms of temperature and absorption of UV and infrared radiation from the Sun and reflected by the Earth. This means that the boundary between these layers, the so-called tropopause, represents a barrier preventing the transfer of water vapour to a certain degree. A diffusion scheme does not reflect this. “In current climate models, this results in excess diffusion of water vapour into the stratosphere causing persistent bias,” says Rozanov.

Bias exposed — and corrected

With the help of their simulations on “Piz Daint”, the team came up with a solution that better describes the process. “In reality, water vapour travels through the tropopause not by regular diffusion but by winds, so by a bulk motion, also called advection,” explains Tatiana Egorova, research scientist and co-author of the respective paper in Nature Communications. And advective transport is better captured by so-called Lagrangian transport schemes, as the work has shown.

Via simulations of present-day climate, the team first examined the transport schemes used in two large climate modelling projects, CMIP6 and CCMI-2022, and compared the simulations’ results with data observed by satellites. In a second step, the team replaced the transport scheme of one of the models with the pure Lagrangian scheme from a chemical stratosphere model named CLaMS developed at the Research Center Jülich. “Coupling these two elements is not straight forward,” says Rozanov. “Climate models are based on grids, whereas for a Lagrangian scheme, you need to introduce billions of particles being moved by winds to then couple these back with the climate model’s grids.” This complicates and expands the entire process.

But the effort paid off: While the transport schemes used in CMIP6 and CCMI-2022 both showed severe bias resulting in an incorrect excess of stratospheric water vapour, the team could almost completely remove this bias using the Lagrangian scheme.

Moreover, the results indicated that the bias-related water vapour effects on atmospheric circulation are of similar magnitude as climate change effects. In particular, the scientists observed that stratospheric water vapour plays a crucial role in controlling local temperatures at the border to the troposphere. This, in turn, influences the subtropical jets, which are prominent and fast atmospheric winds flowing from west to east in both global hemispheres that strongly influence weather and climate. Consequently, when employing the corrected transport scheme, the scientists saw substantial positional shifts in these important jet streams as well as resulting impacts on regional climate.

The team’s analysis of a suite of climate model simulations out of the CCMI2 and CMIP6 projects shows substantial bias caused by the diffusive water vapour transport schemes employed in the models. The deviation in the simulations’ stratospheric water vapour content from satellite data (Stratospheric Water and OzOne Satellite, SWOOSH) is visible as blue colour. The black line denotes the border between the troposphere and the stratosphere, which is higher in altitude around the equator than at the poles. (Image: E. Charlesworth et al., DOI: https://doi.org/10.1038/s41467-023-39559-2)

This image shows a comparison between the diffusive scheme from the comprehensive climate model EMAC (left) and an adapted EMAC simulation, in which the transport scheme was replaced with a Lagrangian representation (right). As expected, EMAC showed severe deviations from the satellite data when it came to stratospheric water vapour content, visible as blue colour in the graph (left). In contrast, the simulation employing the Lagrangian scheme (right) results in hardly any deviation from the observed data. (Image: E. Charlesworth et al., DOI: https://doi.org/10.1038/s41467-023-39559-2)

This image demonstrates how the stratospheric water vapour transport scheme influences processes in the troposphere. In this difference map of the near-surface winds, which are crucial factors in weather and climate, the blue and red colour depict the deviations between simulations that use different transport schemes: diffusive vs. pure Lagrangian. Visible are extensive deviations in the range of ten to 20 percent of the signal, which means that the accuracy of the simulated water vapour content in the stratosphere indeed strongly influences the winds at the Earth’s surface. (Image: E. Charlesworth et al., DOI: https://doi.org/10.1038/s41467-023-39559-2)

The way to more accurate climate predictions

“The implications are, of course, that predictions of future climate responses also depend on the treatment of the atmospheric water vapour and its feedback processes, and on how well they can be reproduced in climate models,” states Rozanov. As differences in water vapour content in the lowermost stratosphere are associated with important atmospheric circulation effects, they will also impact the models’ response to increasing greenhouse gas levels.

Rozanov hopes that this work will be an inspiration to the community of climate scientists: “We present the first example of a Lagrangian transport scheme used to capture stratospheric water vapour, and we show that the method is feasible and that it leads to significant improvement,” he summarizes. If other climate scientists follow suit and use a Lagrangian scheme in their prediction models, this will reveal the impact of the improved representation of stratospheric water vapour on the accuracy of climate predictions.

(Cover image above: A part of the Earth’s atmosphere as seen from the International Space Station (ISS) at an altitude of 335 kilometres: Above the clouds of the troposphere, the stratosphere is visible as a blue layer due to the scattering of blue light by its gases. Image credit: NASA Earth Observatory)

Reference

E. Charlesworth, F. Plöger, T. Birner, et al.: Stratospheric water vapour affecting atmospheric circulation, Nat. Commun14, 3925 (2023), DOI: https://doi.org/10.1038/s41467-023-39559-2