The study focused on three of the most prominent engineered carbon removal methods now moving from laboratory to real-world deployment. Direct air carbon capture and storage, or DACCS, strips carbon dioxide from ambient air using chemical sorbents and locks it underground. Bioenergy carbon capture and storage, known as BECCS, grows biomass that absorbs carbon, burns it for energy and captures the resulting emissions. Enhanced rock weathering, or ERW, spreads crushed silicate minerals on land, accelerating natural chemical reactions that pull carbon dioxide from the atmosphere and eventually lock it away in oceans and soils. Each carries a distinct risk and cost profile, from land-use conflicts and energy demands to mining and transport burdens, yet the researchers discovered something remarkable: the technology itself mattered far less than the way it was implemented.

To isolate what actually drives public acceptance, the team ran a multifactorial vignette experiment embedded in surveys conducted in Brazil, Malaysia, Saudi Arabia, Italy, Norway and the United Kingdom. Roughly 1,800 respondents per country were assigned to one of the three technologies and then evaluated five hypothetical projects out of 36 possible variants. Each project description differed along four attributes grounded in a multidimensional conception of justice: who implemented the project, whether the decision-making process involved consultation, how profits were distributed, and how much carbon the project could remove. Respondents rated each project on a ten-point scale from strong opposition to strong support, generating more than 54,000 individual evaluations for statistical analysis.

The headline result is unambiguous. Across every country and every technology, procedural fairness exerted the strongest effect on support. Compared with projects approved by authorities without any consultation, opening the decision-making process to local communities or scientific experts increased support by roughly 1.0 to 1.7 points on the ten-point scale. The effect was largest in Brazil, where consulting the affected population raised support by 1.68 points, and smallest but still substantial in Saudi Arabia, at just over 1.0 point. Notably, consulting local communities and consulting scientific experts produced nearly identical gains, suggesting that what matters to the public is the principle of being heard, not the specific identity of who is asked.

Distributive fairness came a close second. When developers retained all profits, projects fared worst. Redistributing revenues to affected local communities increased support by between roughly 0.5 and 1.25 scale points, with the largest effect again in Brazil, followed by the United Kingdom and Italy. Not-for-profit arrangements, in which carbon removal operates as a public service without generating profit, also raised support everywhere, though less dramatically in Saudi Arabia. By contrast, whether a company or a regional government carried out the project made little difference in most countries, with only a modest penalty for corporate implementation in the United Kingdom, Italy, Norway and Malaysia. The lesson for the emerging carbon removal industry is uncomfortable: the public appears far more interested in how benefits flow than in who owns the machinery.

Perhaps the most striking finding concerns trade-offs. The researchers tested whether a project promising high removal capacity could compensate for unfair governance, and the answer was a clear no. Across nearly all countries, the effect of increasing carbon removal was remarkably similar regardless of whether projects consulted anyone or shared profits. Respondents rewarded projects that were both effective and fairly governed, but these two considerations operated independently. A high-capacity project run without consultation and with privatized profits still lost the support that fairness would have generated. The reverse also held: progressive governance could not rescue a project with meager removal performance. From the perspective of public legitimacy, neither technical excellence nor ethical design alone is sufficient.

Support levels themselves varied across countries in ways the authors interpret cautiously. Brazil and Saudi Arabia showed the highest average support, with means between roughly 7.1 and 7.3, while European respondents were more moderate, with Norway recording the lowest scores, averaging around 5.1 to 5.4 depending on the technology. The researchers warn that cross-cultural differences in survey response styles, such as tendencies toward acquiescence or avoiding extreme answers, may partly explain these gaps, and they emphasize the experimentally varied attributes over absolute levels. Within Europe, DACCS consistently received lower support than the other two methods, while in Malaysia no significant differences emerged among the technologies at all.

When asked directly what level of public involvement would be fair, respondents revealed a strong appetite for engagement. Allocating decision-making authority to local communities was judged the fairest option for permitting projects by around 44 percent of respondents in the United Kingdom and Italy, 43 percent in Norway, and between 34 and 39 percent in Brazil, Saudi Arabia and Malaysia. Only a tiny fraction, between 2 and 9 percent, thought local communities should have no involvement at all. Strikingly, political orientation played almost no role in shaping support, a sharp contrast with the well-documented polarization of climate attitudes more broadly. Carbon removal, the data suggest, has not yet been absorbed into partisan identity, and public judgments rest instead on risk perceptions, trust and effectiveness.

Even self-described opponents of hosting carbon removal near their homes responded to fairness cues. Dividing respondents into opponents, moderates and supporters of local projects, the researchers found similar preference structures across all three groups, with opponents reacting as strongly or more strongly to governance attributes than anyone else. Opponents, in other words, do not categorically reject carbon removal; they remain open to projects developed through consultation and benefit sharing. This conditional openness offers developers a genuine path forward, but also a narrow one, since perceived injustice can trigger the protests, litigation and delays that have stalled energy infrastructure around the world.

The authors caution that today’s moderate support and weak polarization reflect the novelty of these technologies and low public familiarity, not a stable consensus. As real projects take shape and costs, trade-offs and distributional questions become tangible, controversies are likely to intensify, and spillovers from adjacent debates over biofuels, carbon storage, fracking or nuclear waste could reshape perceptions rapidly. Their prescriptions are correspondingly structural: open decision processes to public and expert scrutiny, treat carbon removal as a public good rather than a profit engine, and experiment with regional, upstream engagement at the planning stage rather than leaving participation to developer-led consultations after key choices are made. With national carbon removal governance still nascent in most of the world, the window for embedding justice into deployment pathways remains open, but the study’s message is that it will not stay that way for long.

Subject of Research: Public support for novel carbon dioxide removal technologies and the role of procedural and distributive fairness in their governance

Article Title: Public support for novel carbon removal hinges on procedural and distributive fairness

Article References: Fritz, L., Losi, L., Merk, C., Boldrini, M., Bosetti, V., Baum, C. M., & Sovacool, B. K. (2026). Public support for novel carbon removal hinges on procedural and distributive fairness. Nature Climate Change, 16(10), 1152-1161. https://doi.org/10.1038/s41558-026-02741-7

Image Credits: AI Generated

DOI: 10.1038/s41558-026-02741-7

Keywords: carbon dioxide removal, public perception, procedural fairness, distributive fairness, direct air capture, BECCS, enhanced rock weathering, climate governance, survey experiment, social acceptance, climate policy, benefit sharing

The core of the problem lies in a mismatch between two very different modeling worlds. Integrated assessment models, or IAMs, simulate the global economy, energy systems, agriculture, and land use, and they decide where and when carbon removal technologies should be deployed based on costs, resource demands, and policy assumptions. Earth system models, or ESMs, simulate the physics, chemistry, and biology of the atmosphere, ocean, and land surface in fully coupled detail. In the current state-of-the-art setup, used for the fifth and sixth phases of the Coupled Model Intercomparison Project and planned again for the seventh, the IAM simply hands the ESM a net emissions number in which the carbon removed by technologies like bioenergy with carbon capture and storage is folded in as a prescribed negative flux. The Earth system model never sees the bioenergy crops growing, never simulates the forests being planted, and never computes the ocean chemistry changes caused by dumping alkaline material into the sea. Everything the model does in response to that removal is therefore invisible.

Schwinger and colleagues argue that this prescribed approach must give way to what they call activity-driven simulation of carbon dioxide removal. Instead of prescribing a negative emission, the ESM explicitly represents the activity that draws down atmospheric CO2. For bioenergy with carbon capture and storage, that means growing the crops inside the model, harvesting the biomass, and routing a scenario-dependent fraction of the harvested carbon into geological storage. For ocean alkalinity enhancement, it means adding an alkaline agent such as slaked lime to the surface ocean and letting the model’s carbonate chemistry, circulation, and biology determine how much CO2 is actually sequestered, and how quickly. Only afforestation and reforestation have always been simulated this way, because the land-use patterns passed from IAMs to ESMs implicitly contain the forest transitions. Extending the same philosophy to other methods harnesses the full power of Earth system models: climate change effects on removal efficiency, from drought stress on bioenergy crops to ocean circulation changes, are computed rather than assumed.

But simulating the activity is only half the challenge. To measure what carbon removal actually achieves, scientists need a reference against which to compare, and this is where the study makes its most consequential conceptual contribution. The team defines two distinct metrics. The first is Process Carbon Removal, the gross amount of carbon a removal process takes up minus its own positive emissions, without any feedback effects. This is essentially what IAMs estimate, and it is the quantity relevant for carbon accounting, in the same way that emissions are priced per tonne regardless of how natural sinks respond. The second is Net Atmospheric Removal, the true difference in atmospheric carbon content between a world with the removal technology and a world without it, including every carbon-cycle and biogeophysical feedback the Earth system can muster. Crucially, the authors show that these two numbers can diverge substantially, because the planet’s carbon sinks respond to removal in ways that partially undo the benefit.

The reason for that divergence is the buffering nature of the carbon cycle. Today, terrestrial and oceanic sinks absorb more than half of every tonne of CO2 humanity emits. Models project that under declining emissions these sinks will take up less carbon, and under net-zero or net-negative emissions they can even flip into carbon sources, releasing CO2 back to the atmosphere as the reversed feedbacks operate in reverse. The practical consequence is stark: the net atmospheric reduction achieved by any carbon removal portfolio will always be less than the gross amount removed. How much less is a question the new framework is designed to answer quantitatively, scenario by scenario and model by model.

Constructing the necessary no-CDR baseline turns out to be anything but trivial. The authors distinguish two fundamentally different approaches. The first is an ex-post adjustment: take an existing IAM mitigation scenario and surgically remove the carbon removal, for instance by freezing the land-use transitions associated with afforestation or by omitting the alkalinity flux to the ocean. This preserves the original scenario’s geography and allows regional attribution of effects, but it produces a socio-economically inconsistent world, one where land earmarked for bioenergy simply sits idle rather than being reallocated to food production. The second is a counterfactual IAM scenario: re-run the economic model in a world where carbon removal was never an option, letting the energy system, land use, and emissions evolve consistently without it. This counterfactual world is socio-economically coherent but so different in its land-use and emissions geography that regional biogeophysical effects of the original removal deployment cannot be isolated. The two approaches are complementary, and the authors recommend that IAM teams begin producing counterfactual no-CDR baselines for selected scenarios, something that, to their knowledge, does not yet exist in the scenario literature.

To demonstrate the framework in action, the team turned to the Norwegian Earth System Model NorESM2-LM and a deep mitigation scenario generated by the REMIND-MAgPIE integrated assessment model, expanded to include ocean liming as a removal method. The scenario reaches an end-of-century carbon budget of 500 gigatonnes of CO2 from 2020, corresponding to 1.5 degrees Celsius of warming in 2100 after a substantial overshoot, with a peak cumulative budget exceeding 1000 gigatonnes around 2060 before massive upscaling of removals pulls it back down. In the Earth system model, ocean alkalinity enhancement was simulated activity-driven, with alkalinity deployed within exclusive economic zones excluding polar regions, while bioenergy removals remained prescribed. Three ensemble members were run for each of the required simulations.

The results are striking and, for anyone banking on ocean-based removal, sobering. For a deployment of 4.93 petamoles of alkalinity between 2050 and 2100, equivalent to 182.5 gigatonnes of calcium hydroxide, the model simulated a Process Carbon Removal of 28.2 petagrams of carbon by 2100. Carbon-cycle feedbacks from both land and ocean then clawed back 7.9 petagrams, roughly 28 percent, leaving a Net Atmospheric Removal of just 20.3 petagrams of carbon, equivalent to a 9.7 parts-per-million reduction in atmospheric CO2. The ocean and terrestrial biosphere contributed roughly equally to the feedback losses, at 4.8 and 3.1 petagrams respectively, though the authors caution that the land feedback in particular may vary greatly between models. Equally revealing was the comparison with the IAM’s own estimate: REMIND-MAgPIE, which assumes a constant efficiency of 0.57 moles of CO2 per mole of alkalinity and instantaneous uptake, overestimated the removal by about 7.4 petagrams, roughly 20 percent, partly because real ocean chemistry takes ten to fifteen years to fully equilibrate an alkalinity addition.

Perhaps the most humbling finding concerns the noise floor of the climate system itself. Even a tiny initial difference in carbon removal deployment acts as a perturbation that causes the scenario and baseline simulations to diverge chaotically, exactly as separate ensemble members with different initial conditions do. Within a few years of the alkalinity deployment beginning around 2040, the El Niño-Southern Oscillation states of the two simulations had become completely uncorrelated. The resulting internal variability in calculated removals and feedback fluxes remained comparable in magnitude to the global annual mean signal, and even decadal averages carried a noise level exceeding 0.25 petagrams of carbon per year. The implication is uncomfortable: detecting the small signals associated with carbon removal and its feedbacks in emission-driven simulations requires either sufficiently large initial-condition ensembles or averaging over long periods, and for some applications, such as quantifying regional biogeophysical effects, stylized modeling approaches may simply remain preferable.

The timing of this work could hardly be better. The upcoming seventh phase of the Coupled Model Intercomparison Project has adopted emission-driven scenario simulations as a priority, yet its ScenarioMIP protocol still foresees only prescribed carbon removal and no no-CDR baselines, limiting what those simulations can say about removal effectiveness. The authors argue that the next phase of the Carbon Dioxide Removal Model Intercomparison Project should adopt their framework, and they issue concrete recommendations to the modeling community: IAMs should provide gross positive and negative emissions separately for each removal method, deliver spatially explicit information on bioenergy with carbon capture in future land-use datasets, and flag the difference between passive forest regrowth and deliberate afforestation. As nations and companies commit billions to carbon removal portfolios whose real-world performance remains deeply uncertain, this framework offers something the field has lacked: a transparent, physically consistent way to separate what a removal technology promises from what the Earth system actually delivers.

Subject of Research: Earth system modeling of carbon dioxide removal efficiency and feedbacks in mitigation scenarios

Article Title: Assessing Earth system responses in mitigation scenarios with activity-driven simulation of carbon dioxide removal

Article References: Schwinger, J., Merfort, L., Bauer, N., Bernardello, R., Butenschön, M., Bourgeois, T., Gidden, M. J., Gupta, S., Lee, H., Mengis, N., Moustakis, Y., Muri, H., Nieradzik, L., Peano, D., Pongratz, J., Sauer, P., Tourigny, E., & Wårlind, D. (2026). Assessing Earth system responses in mitigation scenarios with activity-driven simulation of carbon dioxide removal. Earth System Dynamics, 17(5), 1341-1363. https://doi.org/10.5194/esd-17-1341-2026

Image Credits: AI Generated

DOI: 10.5194/esd-17-1341-2026

Keywords: carbon dioxide removal, Earth system models, integrated assessment models, ocean alkalinity enhancement, BECCS, afforestation, carbon cycle feedbacks, no-CDR baseline, CMIP7, net-zero emissions, NorESM2, climate mitigation scenarios

The study, published in the Journal of Industrial Ecology by Sadiq Abubakar of South Bank Colleges and Coventry University’s Centre for Future Transport and Cities, together with Fernando Cepero Mejias, describes a natural-language interface bolted onto OpenLCA, one of the most widely used open-source life cycle assessment platforms. The key architectural decision is deliberately conservative: the large language model behind the interface never computes environmental impacts itself. Instead, it acts as a translation layer, converting a user’s prose request into structured parameter-edit proposals that are checked by a deterministic validator before anything is executed in the LCA engine.

That separation matters because of a well-known weakness of language models: hallucination, the confident generation of plausible but wrong information. In life cycle assessment, where results feed into carbon footprints, policy decisions and environmental claims, a hallucinated number is not merely embarrassing, it is potentially misleading at scale. The researchers therefore treat every output of the language model as an untrusted artefact. Proposed edits are parsed as JSON, checked against the editable parameters actually present in the loaded model, and rejected outright if they touch the model’s structural elements, processes, flows, exchanges or datasets, or refer to parameters that do not exist in the correct scope.

Under the hood, the workflow runs in four stages. A practitioner first builds and parameterises a base product system in OpenLCA, defining its goal, scope and system boundary. The framework then imports this model through OpenLCA’s inter-process communication interface, exposing to the language model only a reduced representation: parameter names, process identifiers and functional-unit information, but never the full biosphere inventory. A fixed system prompt, which end users cannot modify, restricts the model to supported tools and schema-compliant responses. When validation succeeds, the accepted parameter changes are applied as a scenario and executed deterministically in OpenLCA, producing a complete audit bundle with prompts, diagnostics, results and SHA-256 checksums.

The interface ships with a toolkit of analytically standard operations. One-at-a-time sensitivity analysis perturbs single parameters while holding the rest constant. A simplex-lattice tool generates systematic mixtures of components, a design long used in mixture experiments. A PDF-extraction tool lets the model pull parameter values from uploaded documents, with the backend scoring candidate tables by lexical overlap before returning them. An uncertainty-propagation tool accepts explicitly stated triangular distributions and runs Latin hypercube or Monte Carlo sampling in Python, never letting the model invent distributions of its own. Finally, a bounded goal-seeking tool locates parameter values satisfying environmental constraints, using interval scanning and bisection for one-dimensional thresholds and simplicial homology global optimisation for harder multi-parameter problems.

The framework was tested on three real parameterised models. The first, an electric vehicle charger model, was asked to explore interior concrete footing mixtures of cement, sand and gravel summing to 145 kilograms using a simplex lattice design, and to label each scenario with a compressive strength computed from Abrams’ law. The system correctly generated all six feasible interior mixtures, and the resulting life cycle impact results matched manual OpenLCA reproductions to within rounding. The same model also faced a multilingual test: a single prompt mixing Italian, Hindi, Chinese and Arabic instructions, asking respectively for a steel increase, a copper tripling, a transport-distance change and a specific aluminium mass. All four edits were extracted and executed correctly in the flagship demonstration.

The second case used the NETL carbon dioxide utilisation algae biodiesel model with a functional unit of one megajoule of combusted diesel equivalent. Asked to propagate uncertainty through four parameters, electricity input, carbon dioxide pumping demand, pond utilisation efficiency and recycle rate, the framework sampled 250 parameter sets with a fixed random seed and delivered percentile ranges for global warming, acidification and water-consumption indicators, all consistent with the specified distributions. A follow-up test uploaded a PDF containing four predefined scenarios; the system extracted the relevant tables, generated 22 parameter edits across the scenarios and executed all four calculations successfully.

Perhaps the most striking demonstration came from a carbon-negative BECCS power plant model. Asked in plain English to find the minimum carbon dioxide capture rate at which the plant remains carbon-negative per kilowatt-hour, the system completed 54 model evaluations and identified a threshold of roughly 0.887, which manual recalculation in OpenLCA confirmed with a result of essentially zero emissions. A harder multi-constraint task asked the system to maximise the allowable biomass transport distance while holding climate change in a narrow negative band and capping five other impact categories. The optimiser ran 880 evaluations and nearly tripled the feasible transport distance from the 193-mile baseline to about 553 miles, although manual verification showed three constraints were very slightly exceeded, leading the authors to describe the solution as near-boundary rather than strictly feasible.

The honesty of the evaluation is what sets this work apart from breathless AI announcements. In 175 scored diagnostic runs across five language models and seven prompts, reliability varied substantially. A single-parameter threshold task succeeded in all 25 runs across every model tested, and uncertainty-propagation requests were nearly as stable. But arithmetic labelling proved fragile: without a deterministic calculator tool, only four of fourteen accepted simplex-lattice runs produced strength labels within tolerance, and even with the calculator enabled, some models passed incorrect arguments, such as a water-to-cement ratio ten times too small, producing absurd strength values. Multilingual extraction stumbled on the Arabic aluminium instruction, and PDF-driven scenarios sometimes mapped values to the wrong parameter. Crucially, the authors distinguish schema validity, execution success and task correctness, showing that a payload can pass validation and execute perfectly while still representing the wrong scenario.

The researchers are careful about what their framework is and is not. It is a complement to scripting, not a replacement: for fully specified, repeated workflows, direct Python or IPC scripts remain more transparent and efficient. It cannot build product systems, choose allocation rules or judge whether the underlying model is methodologically sound, responsibilities that remain with the practitioner. Nor does the audit trail prevent selective interpretation of results. Yet the contribution is nonetheless significant: a single natural-language doorway to parameter-level scenario analysis, with deterministic validation, tool calling and cryptographic audit records making every accepted and rejected proposal inspectable. As language models and structured-output reliability improve, and as future work adds semantic retrieval and automated correctness checks, the vision of sustainability analysts interrogating their own models by conversation, without surrendering reproducibility, comes considerably closer to routine practice.

Subject of Research: Natural-language interfaces for parameter-level scenario analysis in life cycle assessment models

Article Title: A natural-language interface for parameter-level scenario analysis in parameterised life cycle assessment models

Article References: A natural-language interface for parameter-level scenario analysis in parameterised life cycle assessment models. (n.d.). https://doi.org/10.1007/s44498-026-00163-4

Image Credits: AI Generated

DOI: 10.1007/s44498-026-00163-4

Keywords: life cycle assessment, large language models, OpenLCA, natural-language interface, parameter scenario analysis, uncertainty propagation, sensitivity analysis, BECCS, tool calling, hallucination, deterministic validation, industrial ecology