The accuracy of modeled projections depends on the data beneath them, and climate modeling has historically been limited by relatively thin data on extreme events. Philippe Ciais and CALIPSO are quantifying the impacts of those events to give climate models a clearer view of Earth’s future.

To project what lies ahead for the planet, climate models track how the element carbon moves and where it stays. But models have so far largely ignored eventsScientists understand these processes, but quantifying them at the scale necessary for modeling has proven harder.

Philippe Ciais and his collaborators designed the Carbon Loss in Plants, Soils and Oceans (CALIPSO) project to change that. Funded by Schmidt Sciences’ Virtual Earth System Research Institute (VESRI), the effort aims to build these “carbon loss” processes into the models that shape our understanding of climate change—and inform global policy. 

CALIPSO brings together experts in terrestrial and ocean carbon modeling, two communities that, as Ciais puts it, “know each other, but don’t really talk much.” Their goal: translate the science of carbon losses into numbers models can use.

“It’s easy to say that trees suffer when you have a drought,” Ciais says, describing one example of these processes. “It’s more difficult to understand how many of them die, or if the big ones die instead of the small ones.”

To accomplish this, the team synthesized observations with theoretical work, then translated the results into lines of code. Their work is starting to fill gaps in climate models.

In one such project, they mapped carbon turnover—the length of time carbon atoms spend in a particular place—across the Amazon. When they incorporated this data into an advanced land carbon model, they saw improvements in its simulations of the amount of biomass, which stores and releases carbon, in the region. 

CALIPSO has also shed light on fire. The team discovered that roughly a third of northern Siberian fires burn peatlands, which have dried out in hotter summers, and, in some cases, smolder through winter. They have also built the first ocean ecological model to include viruses, which destroy microbes and release the carbon they contain. “It’s quite preliminary,” Ciais says of the ocean model. “But it’s also quite a step forward.”

Most climate models project decades ahead. CALIPSO treats carbon modeling more like weather forecasting by testing predictions against observations as events unfold, and so obtaining feedback to improve models’ performance. 

“We learn a lot from seeing the models exposed to variable climate,” Ciais says. For instance, by looking at observations from the recent drought in the Amazon and Africa, the team found that forest models overestimated carbon losses. 

Such real-world experiments aid CALIPSO researchers as they contribute to a global effort to devise a near-term carbon budget, which describes the amount of carbon humanity can still emit while capping warming.

Traditional grants ask researchers to fit their ideas to predetermined calls. VESRI, however, gave Ciais and colleagues the freedom to find their own topic. They settled on carbon losses, which Ciais calls “the most unifying and important blind spot” in the field. He and his colleagues handpicked CALIPSO researchers, assembling a “dream team” that was eager to collaborate. 

The project’s success extends beyond improving climate models. Three of its early-career scientists have secured permanent positions, and new collaborations have formed, including one with a team in China on soil modeling. Meanwhile, the team’s work on northern peatlands led them to receive a highly competitive $14.7 million European Research Council Synergy grant. With this new support, they plan to investigate the abrupt release of carbon from these landscapes as they thaw. 

“Without CALIPSO,” Ciais says, “this would not have happened.”