Periodic Reporting for period 1 - FORCLIMA (Forecasting climate surprises on longer timescales)
Reporting period: 2023-06-01 to 2025-11-30
Climate change can also trigger abrupt and unexpected shifts in the Earth system. Such climate surprises occur when key parts of the climate system, called tipping elements, respond in sudden and nonlinear ways to relatively small changes in conditions. Prominent tipping elements include the Greenland and Antarctic ice sheets and the Atlantic Meridional Overturning Circulation (AMOC), a large-scale ocean circulation that strongly influences regional and global climate.
If destabilized, these tipping elements could lead to dramatic and irreversible changes, such as rapid sea-level rise or large-scale disruptions of weather patterns. Understanding the risk of such events is critical for societies worldwide, but current scientific methods struggle to provide reliable forecasts on the long timescales—centuries to millennia—over which these processes unfold. As a result, projections of future climate change carry deep uncertainties about the likelihood, timing, and consequences of tipping events.
FORCLIMA addresses this challenge by developing innovative tools to robustly forecast climate surprises. Its aim is to advance our ability to anticipate abrupt climate shifts, reduce uncertainty in long-term projections, and improve our understanding of interactions between different tipping elements. FORCLIMA takes a twofold approach that moves well beyond the current state of the art. First, it relies on a new generation Fast Earth System Model (FESM) called CLIMBER-X. This model integrates the most up-to-date understanding of ice sheets, ocean circulation, and atmosphere–ocean–ice interactions. CLIMBER-X is fast enough to run very large ensembles of simulations, which are essential to assess the full range of possible futures and to capture low-probability, high-impact events. Second, a novel probabilistic framework will be developed within the project. To ensure realism, the new model will be constrained with results from the latest ESMs and observational datasets. This probabilistic approach will allow FORCLIMA to generate forecasts that account for uncertainty in key processes while remaining consistent with the best available science.
FORCLIMA will provide the first probabilistic forecasts of major climate surprises on multi-centennial to millennial timescales. It will also deepen our understanding of the risks of ice-sheet collapse and AMOC weakening or shutdown, and their potential combined effects on sea level, climate stability, and ecosystems. Moreover, it will deliver tools that can be used by the broader climate science community to explore uncertainties in Earth system projections. Finally, it will contribute critical knowledge for climate risk assessments, supporting policymakers and stakeholders in preparing for low-probability but high-impact events.
In a broader sense, FORCLIMA will strengthen Europe’s position at the forefront of climate science by providing unparalleled insight into the long-term impacts of climate change. By addressing one of the most profound scientific and societal challenges of our time—the possibility of abrupt and irreversible climate tipping—the project will help societies better anticipate and prepare for the climate of the future.
If destabilized, these tipping elements could lead to dramatic and irreversible changes, such as rapid sea-level rise or large-scale disruptions of weather patterns. Understanding the risk of such events is critical for societies worldwide, but current scientific methods struggle to provide reliable forecasts on the long timescales—centuries to millennia—over which these processes unfold. As a result, projections of future climate change carry deep uncertainties about the likelihood, timing, and consequences of tipping events.
FORCLIMA addresses this challenge by developing innovative tools to robustly forecast climate surprises. Its aim is to advance our ability to anticipate abrupt climate shifts, reduce uncertainty in long-term projections, and improve our understanding of interactions between different tipping elements. FORCLIMA takes a twofold approach that moves well beyond the current state of the art. First, it relies on a new generation Fast Earth System Model (FESM) called CLIMBER-X. This model integrates the most up-to-date understanding of ice sheets, ocean circulation, and atmosphere–ocean–ice interactions. CLIMBER-X is fast enough to run very large ensembles of simulations, which are essential to assess the full range of possible futures and to capture low-probability, high-impact events. Second, a novel probabilistic framework will be developed within the project. To ensure realism, the new model will be constrained with results from the latest ESMs and observational datasets. This probabilistic approach will allow FORCLIMA to generate forecasts that account for uncertainty in key processes while remaining consistent with the best available science.
FORCLIMA will provide the first probabilistic forecasts of major climate surprises on multi-centennial to millennial timescales. It will also deepen our understanding of the risks of ice-sheet collapse and AMOC weakening or shutdown, and their potential combined effects on sea level, climate stability, and ecosystems. Moreover, it will deliver tools that can be used by the broader climate science community to explore uncertainties in Earth system projections. Finally, it will contribute critical knowledge for climate risk assessments, supporting policymakers and stakeholders in preparing for low-probability but high-impact events.
In a broader sense, FORCLIMA will strengthen Europe’s position at the forefront of climate science by providing unparalleled insight into the long-term impacts of climate change. By addressing one of the most profound scientific and societal challenges of our time—the possibility of abrupt and irreversible climate tipping—the project will help societies better anticipate and prepare for the climate of the future.
Through the project, two themes have been pursued. First, to build innovative and flexible methods for probabilistic evaluation of climate model output for different time periods / boundary conditions; and second, to develop and apply the fast Earth system model CLIMBER-X for better understanding of tipping-element dynamics within the Earth system.
For the first theme, we have created several packages in Julia to serve as the basis for a broad set of climate analyses. One package is ModelWeights.jl which allows us to easily load climate-model data from different sources, like the CMIP database, and put them in a data-cube structure that represents the ensemble of data in a way that is easy to work with. This tool works well with Python-based packages like ESMValTool, but is meant to provide the user with an active set of tools for working with the data on any analysis of interest, as opposed to the recipe-style approach of ESMValTool. A core function of this package is to generate a set of weight associated with each ensemble member, in order to calculate weighted statistics that can more robustly represent the information contained within the ensemble. As a first application, we are working to reconstruct the Last Glacial Maximum climate state from a combined data-model approach.
For the second theme, we have performed numerous experiments with CLIMBER-X that help us to understand the ice-sheets and the ocean circulation response to climate change, and allow us to evaluate the realism of the model results against state-of-the-art ESMs. One study explores the emergence of multi-centennial oscillations that appear in the southern ocean under high CO2 scenarios on long time scales. Within a specific range of atmospheric CO2, convection in the Southern Ocean activates, mixing the water column and redistributing heat, then eventually shuts off again and repeats. This appears to be robust behaviour only seen in simpler models until now. However, we are developing a sub-grid convection scheme to help improve the resolution of the relevant model physics. This will be validated against ESMs for further use in climate change predictions. Likewise, another study focuses on the future evolution of the Greenland ice sheet under realistic CO2-emission scenarios. CLIMBER-X was coupled with the ice-sheet model Yelmo and calibrated to produce a reasonable present-day ice-sheet state. From there, we were able to test the sensitivity of the ice sheet to different climate-change scenarios and global climate sensitivities. Our results clearly indicate that the long-term and peak temperature changes in time are correlated and thus that the peak temperature over the next century is critical in determining the fate of the Greenland ice sheet.
For the first theme, we have created several packages in Julia to serve as the basis for a broad set of climate analyses. One package is ModelWeights.jl which allows us to easily load climate-model data from different sources, like the CMIP database, and put them in a data-cube structure that represents the ensemble of data in a way that is easy to work with. This tool works well with Python-based packages like ESMValTool, but is meant to provide the user with an active set of tools for working with the data on any analysis of interest, as opposed to the recipe-style approach of ESMValTool. A core function of this package is to generate a set of weight associated with each ensemble member, in order to calculate weighted statistics that can more robustly represent the information contained within the ensemble. As a first application, we are working to reconstruct the Last Glacial Maximum climate state from a combined data-model approach.
For the second theme, we have performed numerous experiments with CLIMBER-X that help us to understand the ice-sheets and the ocean circulation response to climate change, and allow us to evaluate the realism of the model results against state-of-the-art ESMs. One study explores the emergence of multi-centennial oscillations that appear in the southern ocean under high CO2 scenarios on long time scales. Within a specific range of atmospheric CO2, convection in the Southern Ocean activates, mixing the water column and redistributing heat, then eventually shuts off again and repeats. This appears to be robust behaviour only seen in simpler models until now. However, we are developing a sub-grid convection scheme to help improve the resolution of the relevant model physics. This will be validated against ESMs for further use in climate change predictions. Likewise, another study focuses on the future evolution of the Greenland ice sheet under realistic CO2-emission scenarios. CLIMBER-X was coupled with the ice-sheet model Yelmo and calibrated to produce a reasonable present-day ice-sheet state. From there, we were able to test the sensitivity of the ice sheet to different climate-change scenarios and global climate sensitivities. Our results clearly indicate that the long-term and peak temperature changes in time are correlated and thus that the peak temperature over the next century is critical in determining the fate of the Greenland ice sheet.
We have developed new methods for probabilistic evaluation of climate models. We combine ensemble model weighting with a Monte-carlo based data assimiliation approach to be able to arrive at a probabilistic estimate of the true climate state for a given set of boundary conditions. This will allow us, and others, to directly compare model output to a target set of climate variables instead of sparse observations or climate reconstructions that need interpretation.
Moreover we have set a new standard for regional ice-sheet modeling, by developing a new model for glacial isostatic adjustment called FastIsostasy that runs orders of magnitude faster than the best global models, but accurately represents the physics on a regional domain. This will allow much more realistic ensembles of ice-sheet model simulations and allow better quantification of feedbacks related to ice-sheet retreat under climate change.
We also derived an analytical solution to the transient 1D heat equation for studying the thermodynamic evolution of an ice column. This work combined previously available mathematical techniques to provide a flexible and more complete solution to this problem than was previously available. This new analytical solution will allow us to better understand the factors that influence changes in ice temperature under different boundary conditions and to validate the results from model implementations.
Moreover we have set a new standard for regional ice-sheet modeling, by developing a new model for glacial isostatic adjustment called FastIsostasy that runs orders of magnitude faster than the best global models, but accurately represents the physics on a regional domain. This will allow much more realistic ensembles of ice-sheet model simulations and allow better quantification of feedbacks related to ice-sheet retreat under climate change.
We also derived an analytical solution to the transient 1D heat equation for studying the thermodynamic evolution of an ice column. This work combined previously available mathematical techniques to provide a flexible and more complete solution to this problem than was previously available. This new analytical solution will allow us to better understand the factors that influence changes in ice temperature under different boundary conditions and to validate the results from model implementations.