Explore the vast range of titles available.

Spinning up a highly complex, coupled Earth system model (ESM) is a time consuming and computationally demanding exercise. For models with interactive ice sheet components, this becomes a major challenge, as ice sheets are sensitive to bidirectional feedback processes and equilibrate over glacial timescales of up to many millennia. This work describes and demonstrates a computationally tractable, iterative procedure for spinning up a contemporary, highly complex ESM that includes an interactive ice sheet component. The procedure alternates between a computationally expensive coupled configuration and a computationally cheaper configuration where the atmospheric component is replaced by a data model. By periodically regenerating atmospheric forcing consistent with the coupled system, the data atmosphere remains adequately constrained to ensure that the broader model state evolves realistically. The applicability of the method is demonstrated by spinning up the preindustrial climate in the Community Earth System Model Version 2 (CESM2), coupled to the Community Ice Sheet Model Version 2 (CISM2) over Greenland. The equilibrium climate state is similar to the control climate from a coupled simulation with a prescribed Greenland ice sheet, indicating that the iterative procedure is consistent with a traditional spin‐up approach without interactive ice sheets. These results suggest that the iterative method presented here provides a faster and computationally cheaper method for spinning up a highly complex ESM, with or without interactive ice sheet components. The method described here has been used to develop the climate/ice sheet initial conditions for transient, ice sheet‐enabled simulations with CESM2‐CISM2 in the Coupled Model Intercomparison Project Phase 6 (CMIP6). Plain Language Summary Experiments with Earth system models typically use the preindustrial (1850 CE) climate as a reference point when examining the climate response to a given experiment scenario. A preindustrial simulated climate state is therefore important to develop and represent consistently, which often requires long and computationally expensive spin‐up or equilibration simulations. The latest generation Earth system models include time‐evolving ice sheet components, which complicate the task of generating a self‐consistent simulated preindustrial climate. Additional complexity arises because ice sheets interact with the rest of the climate system through complex processes and feedbacks and respond slowly to climate change over many millennia. This equilibration timescale is computationally intractable using traditional spin‐up/equilibration simulation techniques. To circumvent this challenge, we present a novel method for generating an internally consistent climate state that is suitable for models with interactive ice sheet components. This method uses fewer computational resources than traditional simulation methods, while generating a climate consistent with more expensive methods. We demonstrate the viability of the method by generating the preindustrial control climate in the Community Earth System Model Version 2 (CESM2), which includes an interactive Greenland ice sheet. Key Points We describe a computationally tractable, iterative procedure for spinning up a coupled Earth system‐ice sheet model Equilibrium state from the iterative procedure is similar to a more expensive traditional model spin‐up with prescribed ice sheets The procedure is used for developing initial conditions for transient, fully coupled simulations in the Coupled Model Intercomparison Project phase 6

The possibility that Arctic sea ice loss weakens mid-latitude westerlies, promoting more severe cold winters, has sparked more than a decade of scientific debate, with apparent support from observations but inconclusive modelling evidence. Here we show that sixteen models contributing to the Polar Amplification Model Intercomparison Project simulate a weakening of mid-latitude westerlies in response to projected Arctic sea ice loss. We develop an emergent constraint based on eddy feedback, which is 1.2 to 3 times too weak in the models, suggesting that the real-world weakening lies towards the higher end of the model simulations. Still, the modelled response to Arctic sea ice loss is weak: the North Atlantic Oscillation response is similar in magnitude and offsets the projected response to increased greenhouse gases, but would only account for around 10% of variations in individual years. We further find that relationships between Arctic sea ice and atmospheric circulation have weakened recently in observations and are no longer inconsistent with those in models. The degree to which Arctic sea ice decline influences the mid-latitude atmospheric circulation is widely debated. Here, the authors use a coordinated multi-model experiment to show that Arctic sea ice loss causes a weakening of the mid-latitude westerly winds, but the effect is overall small.

A promising approach to improve cloud parameterizations within climate models and thus climate projections is to use deep learning in combination with training data from storm‐resolving model (SRM) simulations. The ICOsahedral Non‐hydrostatic (ICON) modeling framework permits simulations ranging from numerical weather prediction to climate projections, making it an ideal target to develop neural network (NN) based parameterizations for sub‐grid scale processes. Within the ICON framework, we train NN based cloud cover parameterizations with coarse‐grained data based on realistic regional and global ICON SRM simulations. We set up three different types of NNs that differ in the degree of vertical locality they assume for diagnosing cloud cover from coarse‐grained atmospheric state variables. The NNs accurately estimate sub‐grid scale cloud cover from coarse‐grained data that has similar geographical characteristics as their training data. Additionally, globally trained NNs can reproduce sub‐grid scale cloud cover of the regional SRM simulation. Using the game‐theory based interpretability library SHapley Additive exPlanations, we identify an overemphasis on specific humidity and cloud ice as the reason why our column‐based NN cannot perfectly generalize from the global to the regional coarse‐grained SRM data. The interpretability tool also helps visualize similarities and differences in feature importance between regionally and globally trained column‐based NNs, and reveals a local relationship between their cloud cover predictions and the thermodynamic environment. Our results show the potential of deep learning to derive accurate yet interpretable cloud cover parameterizations from global SRMs, and suggest that neighborhood‐based models may be a good compromise between accuracy and generalizability. Plain Language Summary Climate models, such as the ICOsahedral Non‐hydrostatic climate model, operate on low‐resolution grids, making it computationally feasible to use them for climate projections. However, physical processes –especially those associated with clouds– that happen on a sub‐grid scale (inside a grid box) cannot be resolved, yet they are critical for the climate. In this study, we train neural networks that return the cloudy fraction of a grid box knowing only low‐resolution grid‐box averaged variables (such as temperature, pressure, etc.) as the climate model sees them. We find that the neural networks can reproduce the sub‐grid scale cloud fraction on data sets similar to the one they were trained on. The networks trained on global data also prove to be applicable on regional data coming from a model simulation with an entirely different setup. Since neural networks are often described as black boxes that are therefore difficult to trust, we peek inside the black box to reveal what input features the neural networks have learned to focus on and in what respect the networks differ. Overall, the neural networks prove to be accurate methods of reproducing sub‐grid scale cloudiness and could improve climate model projections when implemented in a climate model. Key Points Neural networks can accurately learn sub‐grid scale cloud cover from realistic regional and global storm‐resolving simulations Three neural network types account for different degrees of vertical locality and differentiate between cloud volume and cloud area fraction Using a game theory based library we find that the neural networks tend to learn local mappings and are able to explain model errors

In theory, the same sea-ice models could be used for both research and operations, but in practice, differences in scientific and software requirements and computational and human resources complicate the matter. Although sea-ice modeling tools developed for climate studies and other research applications produce output of interest to operational forecast users, such as ice motion, convergence, and internal ice pressure, the relevant spatial and temporal scales may not be sufficiently resolved. For instance, sea-ice research codes are typically run with horizontal resolution of more than 3 km, while mariners need information on scales less than 300 m. Certain sea-ice processes and coupled feedbacks that are critical to simulating the Earth system may not be relevant on these scales; and therefore, the most important model upgrades for improving sea-ice predictions might be made in the atmosphere and ocean components of coupled models or in their coupling mechanisms, rather than in the sea-ice model itself. This paper discusses some of the challenges in applying sea-ice modeling tools developed for research purposes for operational forecasting on short time scales, and highlights promising new directions in sea-ice modeling.

A new collaborative organization for sea-ice model development, the CICE Consortium, has devised quality control procedures to maintain the integrity of its numerical codes’ physical representations, enabling broad participation from the scientific community in the Consortium’s open software development environment. Using output from five coupled and uncoupled configurations of the Los Alamos Sea Ice Model, CICE, we formulate quality control methods that exploit common statistical properties of sea-ice thickness, and test for significant changes in model results in a computationally efficient manner. New additions and changes to CICE are graded into four categories, ranging from bit-for-bit amendments to significant, answer-changing upgrades. These modifications are assessed using criteria that account for the high level of autocorrelation in sea-ice time series, along with a quadratic skill metric that searches for hemispheric changes in model answers across an array of different CICE configurations. These metrics also provide objective guidance for assessing new physical representations and code functionality. This article is part of the theme issue ‘Modelling of sea-ice phenomena’.

The increasing awareness of the many damaging aspects of climate change has prompted research into ways of reducing and reversing the anthropogenic increase in carbon concentrations in the atmosphere. Most emission scenarios stabilizing climate at low levels, such as the 1.5 °C target as outlined by the Paris Agreement, require large-scale deployment of Bio-Energy with Carbon Capture and Storage (BECCS). Here, the potential of large-scale BECCS deployment in contributing towards the 1.5 °C global warming target is evaluated using an Earth system model, as well as associated climate responses and carbon cycle feedbacks. The geographical location of the bioenergy feedstock is shown to be key to the success of such measures in the context of temperature targets. Although net negative emissions were reached sooner, by ∼6 years, and scaled up, land use change emissions and reductions in forest carbon sinks outweigh these effects in one scenario. Re-cultivating mid-latitudes was found to be beneficial, on the other hand, contributing in the right direction towards the 1.5 °C target, only by −0.1 °C and −54 Gt C in avoided emissions, however. Obstacles remain related to competition for land from nature preservation and food security, as well as the technological availability of CCS.

Integrated modeling of Earth and human systems, accounting for feedbacks, is key to fully understand climate change consequences and ensuing adaptation needs. In some aspects of climate research, however, closing this loop has proved particularly challenging. A primary example is the generation and use of earth system model (ESM) simulations. Integrated assessment models (IAMs) are used to project socio‐economic activities, emissions and land‐use change. ESM projections, driven by these scenarios, are then used by impact models. According to this modeling chain, however, those impacts do not affect the emissions and land use driving ESMs. Whether this feedback is large enough to warrant explicitly accounting for, needs addressing. Two consequential possibilities, discussed in the literature, are that emissions and land‐use scenarios representing the high and low ends of the plausible range might be too extreme: the high‐end scenarios missing damaging impacts, which could reduce economic activity, therefore emissions; the low‐end scenario ignoring climate feedbacks that make nature‐based carbon removal less effective and hamper‐ the assumed mitigation. In this piece, we describe the challenges that implementing such feedbacks would face, while arguing that recent developments in impact research, data, human and Earth system modeling and emulation make the time ripe for a structured model intercomparison project (MIP). MIPs have benefitted climate modeling for decades. An IAM MIP focused on integrating impacts in emission and land‐use change scenarios could enable testing these feedbacks' implications and assessing whether closing the loop would significantly change our outlook on future climate changes and their consequences. Plain Language Summary Emission and land‐use scenarios produced by Integrated Assessment Models (IAMs) drive projections of future climate by Earth System Models, which drive impact models in turn. This modeling chain, however, does not represent feedbacks from climate impacts on societal activities and emissions. Some have argued that climate change impacts on economic activity will be so large as to reduce emissions, making the highest emission scenarios not plausible. At the other end of the spectrum, wildfires, droughts, or other climatic changes associated with warming may undermine nature‐based mitigation solutions assumed by the lowest scenarios, making them also implausible. Uncertainties in understanding and modeling affect our ability to measure the strength of these hypothesized but seldom‐tested feedbacks. A coordinated model intercomparison exercise among IAMs can robustly assess these effects, especially thanks to recent progress in data, modeling and scientific understanding of many aspects of the interrelated systems at play. Key Points We need to account for climate impacts when modeling scenarios of future emissions and land use to enhance consistency and plausibility Recent progress in understanding, data and modeling can help overcome long‐standing obstacles in integrating emissions, climate and impacts We call for a coordinated multimodel comparison effort to foster critical progress, assess uncertainties, produce more plausible scenarios

Land surface models have an increasing scope. Initially designed to capture the feedbacks between the land and the atmosphere as part of weather and climate prediction, they are now used as a critical tool in the urgent need to inform policy about land-use and water-use management in a world that is changing physically and economically. This paper outlines the way that models have evolved through this change of purpose and what might the future hold. It highlights the importance of distinguishing between advances in the science within the modelling components, with the advances of how to represent their interaction. This latter aspect of modelling is often overlooked but will increasingly manifest as an issue as the complexity of the system, the time and space scales of the system being modelled increase. These increases are due to technology, data availability and the urgency and range of the problems being studied.

Purpose of Review Assessment of the impact of ocean resolution in Earth System models on the mean state, variability, and future projections and discussion of prospects for improved parameterisations to represent the ocean mesoscale. Recent Findings The majority of centres participating in CMIP6 employ ocean components with resolutions of about 1 degree in their full Earth System models (eddy-parameterising models). In contrast, there are also models submitted to CMIP6 (both DECK and HighResMIP) that employ ocean components of approximately 1/4 degree and 1/10 degree (eddy-present and eddy-rich models). Evidence to date suggests that whether the ocean mesoscale is explicitly represented or parameterised affects not only the mean state of the ocean but also the climate variability and the future climate response, particularly in terms of the Atlantic meridional overturning circulation (AMOC) and the Southern Ocean. Recent developments in scale-aware parameterisations of the mesoscale are being developed and will be included in future Earth System models. Summary Although the choice of ocean resolution in Earth System models will always be limited by computational considerations, for the foreseeable future, this choice is likely to affect projections of climate variability and change as well as other aspects of the Earth System. Future Earth System models will be able to choose increased ocean resolution and/or improved parameterisation of processes to capture physical processes with greater fidelity.

The development of the climate and Earth system models has had a long history, starting with the building of individual atmospheric, ocean, sea ice, land vegetation, biogeochemical, glacial and ecological model components. The early researchers were much aware of the long-term goal of building the Earth system models that would go beyond what is usually included in the climate models by adding interactive biogeochemical interactions. In the early days, the progress was limited by computer capability, as well as by our knowledge of the physical and chemical processes. Over the last few decades, there has been much improved knowledge, better observations for validation and more powerful supercomputer systems that are increasingly meeting the new challenges of comprehensive models. Some of the climate model history will be presented, along with some of the successes and difficulties encountered with present-day supercomputer systems.