Rethinking climate engineering categorization in the context of climate change mitigation and adaptation


  • The copyright line in this article was changed on 26 August 2014 after online publication.


The portfolio of approaches to respond to the challenges posed by anthropogenic climate change has broadened beyond mitigation and adaptation with the recent discussion of potential climate engineering options. How to define and categorize climate engineering options has been a recurring issue in both public and specialist discussions. We assert here that current definitions of mitigation, adaptation, and climate engineering are ambiguous, overlap with each other and thus contribute to confusing the discourse on how to tackle anthropogenic climate change. We propose a new and more inclusive categorization into five different classes: anthropogenic emissions reductions (AER), territorial or domestic removal of atmospheric CO2 and other greenhouse gases (D-GGR), trans-territorial removal of atmospheric CO2 and other greenhouse gases (T-GGR), regional to planetary targeted climate modification (TCM), and climate change adaptation measures (including local targeted climate and environmental modification, abbreviated CCAM). Thus, we suggest that techniques for domestic greenhouse gas removal might better be thought of as forming a separate category alongside more traditional mitigation techniques that consist of emissions reductions. Local targeted climate modification can be seen as an adaptation measure as long as there are no detectable remote environmental effects. In both cases, the scale and intensity of action are essential attributes from the technological, climatic, and political viewpoints. While some of the boundaries in this revised classification depend on policy and judgement, it offers a foundation for debating on how to define and categorize climate engineering options and differentiate them from both mitigation and adaptation measures to climate change. WIREs Clim Change 2014, 5:23–35. doi: 10.1002/wcc.261

Conflict of interest: The authors have declared no conflicts of interest for this article.

For further resources related to this article, please visit the WIREs website.


The concept of large-scale environmental engineering is not new and a number of ideas have been proposed over the last century (see Ref [1] for a historical perspective). The idea to engineer the environment to specifically counter anthropogenic climate change can be traced back at least to the 1960s with a US report calling for research on ‘possibilities to deliberately bringing about countervailing climatic changes’ to that of carbon dioxide.[2] The term ‘geoengineering’ itself was coined by Marchetti[3] to discuss the idea of injecting CO2 into the ocean to reduce its atmospheric burden. Since then, the term has evolved and now encompasses a broad and ill-defined set of approaches that aim to deliberately alter the climate system on a large scale in order to alleviate the impacts of climate change.[4] Due to the strong emphasis on the climate modification aspect, we favor the use of ‘climate engineering’ (CE), which we use throughout this study (with the exception of a brief revisiting of the term ‘geoengineering’).

The Royal Society[5] categorized climate engineering methods into two broad classes. Solar radiation management (SRM) refers to the intentional modification of the Earth's shortwave radiative budget to reduce anthropogenic climate change. Injection of stratospheric aerosols[6] and cloud brightening[7] are two examples. Carbon dioxide removal (CDR) refers to approaches that aim to reduce atmospheric CO2 concentration by either increasing natural or engineering new carbon sinks. CDR methods can involve ocean, land, and technological systems, examples include iron fertilization, large-scale afforestation, and direct capture of carbon dioxide.[8] However, this classification does not capture all CE methods which have been proposed, for example, the proposal to increase outgoing longwave radiation by seeding cirrus clouds with ice nuclei.[9] Nor does it provide a clear distinction between climate engineering and other policy responses like climate mitigation and adaptation.

There has been considerable attention on CE since the publication by Crutzen.[6] Various studies have compiled and assessed CE schemes,[4, 8, 10, 11] and several national assessments have also been conducted.[5, 12, 13] Yet there is a lack of common understanding of what CE refers to and existing studies do not provide an up-to-date and complete taxonomy of climate engineering methods. How to define and categorize climate engineering options has indeed been a recurring issue in both public and specialist discussions of climate engineering.[14, 15] This study will (1) critically discuss how climate engineering fits into the wider portfolio of responses to anthropogenic climate change and (2) develop a more robust taxonomy to categorize these responses. It should be viewed as an opinion article whose role is to initiate and foster debate on the subject.

We stress that any such categorization will depend on its aim—here we attempt a primarily climate-science-based categorization focusing on spatial and temporal scales and Earth system processes impacted. We recognize, however, that our categorization involves nonuniversal cultural attitudes, as well as certain norms, value judgement, and political choices. We only address peaceful applications of climate engineering and omit any potential military dimension of climate engineering. Neither do we attempt to rank the effectiveness or appropriateness of climate change responses in any manner.


Definitions of Mitigation and Adaptation

Mitigation is defined by the Intergovernmental Panel on Climate Change (IPCC) Fourth assessment report's glossary[16] as ‘technological change and substitution that reduce resource inputs and emissions per unit of output’. It further specifies that ‘although several social, economic, and technological policies would produce an emission reduction, with respect to climate change, mitigation means implementing policies to reduce greenhouse gas emissions and enhance sinks’. This definition is consistent with that from the IPCC Third assessment report[17] and implies that methods aiming at reducing natural sources or enhancing natural sinks of CO2 and other greenhouse gases do qualify as mitigation policies. This definition is also consistent with the United Nations Framework Convention on Climate Change[18] (UNFCCC) which stipulates in its Article 4 that Parties ‘shall adopt national policies and take corresponding measures on the mitigation of climate change, by limiting its anthropogenic emissions of greenhouse gases and protecting and enhancing its greenhouse gas sinks and reservoirs’.

These definitions raise three important comments:

  1. There is an ambiguity as to whether mitigation includes emission reductions through a voluntary reduction of production and/or consumption per capita, i.e., so-called nontechnical measures. The above-mentioned definition and IPCC[14] refer to emissions reductions per unit of output or gross domestic product, but emissions reductions are often considered to contribute toward mitigation objectives irrespective of their origin. However, behavior changes can also contribute to emissions reductions and it is surprising that this is not made explicit in the definition. A related issue is whether emissions and emission reductions should be counted from a territorial (i.e., depending on their location) or a consumption (i.e., depending on who the ‘final user’ is) perspective.
  2. Carbon capture and storage (CCS), as a technology which aims to capture and sequester waste CO2 from large source points before it is emitted in the atmosphere, qualifies as a mitigation technology.[19]
  3. The IPCC and the UNFCCC definitions include the enhancement of greenhouse gas sinks as a mitigation option. This was designed to include sustainable forms of agriculture and forest management, reforestation and afforestation within the scope of the UNFCCC and the Kyoto Protocol. It is unlikely that UNFCCC negotiators and IPCC authors had in mind the full spectrum of CE methods that have now been proposed to modify the carbon cycle when they crafted these definitions. Heyward[15] discussed this issue and concluded that there are advantages of specifying CDR as a separate category rather than a subset of mitigation.

Adaptation is defined as ‘initiatives and measures to reduce the vulnerability of natural and human systems against actual or expected climate change effects’.[16] The IPCC further specifies that ‘various types of adaptation exist, e.g., anticipatory and reactive, private and public and autonomous and planned’ and provides some examples of adaptation measures such as ‘raising river or coastal dikes’ and ‘the substitution of more temperature-shock resistant plants for sensitive ones’. Central to the concept of adaptation is the idea to reduce the vulnerability of natural and human systems to climate change through a modification of these systems. It is not always clear, however, what the system boundaries are when talking about adaptation. For instance, one could adapt to the risk of fluvial flooding by building dykes, by diverting streams and rivers, increasing storage capacity, or theoretically if it were at all possible to, by preventing extreme rain events from occurring. There is potentially varying degrees of climate or environmental engineering in such adaptation options. It should also be noted that modifying natural and human systems may feedback on the local climate. For instance, it is now well understood that the climate can be modified locally by increasing green spaces in cities,[20] building large dams[21] or changing agricultural practices through irrigation[22] or the modification of land cover. The boundary between adaptation and climate engineering within current definitions can therefore be blurred. This was also noted by Heyward[15] who observes that SRM could be seen as preventative or responsive depending on ‘whether dangerous anthropogenic interference or dangerous climate change is taken as the referent’. It is worth noting that the definition of adaptation considered here is very specific to addressing climate change; it is also possible to adapt our environment to better suit our individual and community needs as a habitat, as has been done already for millennia.

Definitions of Climate Engineering

As discussed above, there is no definition of ‘geoengineering’ or ‘climate engineering’ which is agreed on by the research, policy and civil society communities at large. The IPCC working group III defined it as ‘technological efforts to stabilize the climate system by direct intervention in the energy balance of the Earth for reducing global warming’[23] and later more generally as ‘a broad set of methods and technologies that aim to deliberately alter the climate system in order to alleviate the impacts of climate change’,[14] while the Royal Society[5] refers to as ‘the deliberate large-scale intervention in the Earth's climate system, in order to moderate global warming’. Recognizing the lack of agreed definition, the Convention on Biological Diversity[24] stated in its decision X/33 that ‘Without prejudice to future deliberations on the definition of geo-engineering activities, understanding that any technologies that deliberately reduce solar insolation or increase carbon sequestration from the atmosphere on a large scale that may affect biodiversity (excluding carbon capture and storage from fossil fuels when it captures carbon dioxide before it is released into the atmosphere) should be considered as forms of geo-engineering which are relevant to the Convention on Biological Diversity until a more precise definition can be developed’.

There is a broader naming issue surrounding the concept of CE. Prior to the more recent usage of the term ‘geoengineering’ in relation to counteracting global warming, the term has historically been used as an abbreviation for ‘geotechnical engineering’, which is ‘a branch of civil engineering concerned with the engineering behavior of earth materials’ (Wikipedia), or ‘a science that deals with the application of geology to engineering’ (, or ‘the branch of engineering concerned with the analysis, design and construction of foundations, slopes, retaining structures, embankments, tunnels, levees, wharves, landfills and other systems that are made of or are supported by soil or rock’ (The Electronic Journal of Geotechnical Engineering). There is also a Journal of Geoengineering (published by the Taiwan Geotechnical Society) which ‘covers various topics in geotechnical engineering.’ This is quite different from what is understood under ‘geoengineering’ in the context of climate change. Due to the imprecise nature of ‘geoengineering’ and the potential for confusion with geotechnical engineering, many climate scientists have instead begun to use the term ‘climate engineering’, which is more specific and for which the intent is immediately more apparent.[8, 11-13, 25] Climate engineering is the most widely used alternate terminology, although many other terms have been proposed, such as ‘climate intervention’, ‘climate management’, ‘climate remediation’, and ‘novel options for addressing climate change’. In the rest of this study, we refer to the topic as climate engineering (sometimes abbreviated CE), for the reasons of greater precision, although we recognize that the term ‘climate engineering’ is not without a past: it has sometimes been used to refer to air conditioning technologies, as opposed to the large scale climate control being considered here. The word ‘engineering’ itself may also be misleading as we are considering here methods that try to influence one aspect of a complex system rather than control or ‘engineer’ the full system.

Finally, there is considerable discussion within the community about whether an umbrella term should be used at all, given the vast range of techniques that it subsumes. We will not attempt to address this issue here. We rather develop an improved categorization for what generally falls under the umbrella term. However, for future discussions of the issue, we hope that our improved categorization will allow people to be more comfortable either addressing a specific category of climate engineering and using terminology which makes that clear, and also to be able to specify what categories they are subsuming under the umbrella term.

Issues and Overlaps with Existing Definitions

There is also potential overlap with both mitigation and adaptation. Responses to climate change can be categorized according to where they take place in the chain of processes between anthropogenic drivers of climate change and the impacts of climate change.[15, 26] However, this is not enough to resolve all of the ambiguities.

Defining climate engineering as ‘engineering the climate system’ requires one to specify what is meant by climate system. It is usual to define the climate system as the sum of and the results of the interactions between the atmosphere, the ocean, the cryosphere, and the biosphere. It is sometimes considered to include the crustal lithosphere and the ‘anthroposphere’ of human population, technology, activities, and consciousness. Normally, however, the definition excludes sediments and other geological reservoirs that are nevertheless known to play a role on long climatic timescales. This justifies a posteriori why CCS[19] is usually not considered as CE as it reduces CO2 emissions at the source (i.e., capturing CO2 from flue gases of power plants or other CO2-emitting industry) rather than removing CO2 from the atmosphere. However, CCS is required for some forms of carbon dioxide removal from the atmosphere, notably biomass energy with CCS (BECCS), and direct capture through chemical engineering and storage.[8, 27] BECCS has the potential to remove CO2 from the atmosphere and will have an impact on the biosphere if performed on a large scale. It should therefore be considered as climate engineering even though bioenergy and CCS on their own are usually considered as mitigation options as they individually contribute to reduce atmospheric emissions of CO2.

Conversely, some climate engineering proposals may also be considered as adaptation options. For instance, introducing more reflective roofs or road/pavement surfaces is sometimes presented as a climate engineering method.[28, 29] But this technique can be seen as an adaptation method to mitigate the urban heat island effect.[30] Highly reflective building surfaces were already a common architectural feature in countries with a hot climate long before anthropogenic climate change started to become an issue. Likewise changes in cropland management practices such as increasing irrigation[22] or increasing crop albedo[31] may also help to adapt to a changing climate by cooling the Earth's surface locally and maintain crop productivity in a warmer climate.

While the distinction between SRM and CDR[5] is useful, it does not cover all potential climate engineering schemes that have been proposed to date. For instance, it has been suggested that the terrestrial radiation budget could also be artificially modified through changes in cirrus clouds.[9] Such a technique aims to increase outgoing longwave radiation but shares a lot of characteristics with SRM techniques in that it is nonpermanent, quasi-reversible and is only meant to ‘mask’ the warming effect due to greenhouse gases. Carbon dioxide is not the only long-lived greenhouse gas in the atmosphere, and removal or destruction of methane can also be envisaged either in the atmosphere[32] or before emissions from natural reservoirs take place.[33, 34]


We now discuss a number of attributes of CE techniques, which can help to classify these techniques, and differentiate such techniques from mitigation and adaptation.


There is general agreement in the scientific literature that climate engineering is per se an intentional attempt to counteract climate change by a method that does not seek to reduce anthropogenic emissions of warming agents. For instance, the emission of sulfate aerosols from burning fossil fuels, although responsible for a cooling effect, is not considered to be climate engineering because it is a by-product rather than a deliberate human action to cool the climate. Although in most cases, it is clear if an action is or is not a deliberate attempt to modify the climate, there are situations where this criterion may not apply clearly. A particularly poignant example is the emission of sulfur dioxide (SO2) from ocean-going ships. A putative deliberate decision to continue burning high-sulfur fuel over the open oceans to maintain the cooling effect of sulfate aerosols could be considered a form of climate engineering, if such a decision is made despite the availability of low-cost, low-sulfur fuel and known adverse impact on air quality.

Scale and/or Intensity

Climate engineering implies a certain scale and/or intensity of action and/or impact. Below some spatial scale a deliberate environmental change should not be considered as climate engineering. Planting trees on a small plot of land or whitening roofs in a small urban area do not constitute CE as their impact on the climate system will be negligible. Scale and intensity clearly distinguish climate engineering from weather modification or other sorts of ecological or environmental engineering that usually attempt to modify the environment on a fairly small scale. CE can be considered to start when there is a measurable climate impact at the regional or global scale. What constitutes a ‘measurable climate impact’ is subject to judgment although analyses of past events, such as the eruption of Mount Pinatubo, and model simulations could be used to quantify this. We envisage that a typical scale relevant to regional climate impact could be of the order of 300 km × 300 km ≈ 105 km2. Conversely, any action that mainly aims at modifying the climate at the local scale should be considered as an adaptation measure rather than as CE per se, as long as there is no measurable or detectable remote effect. Again this is subject to judgment and we envisage that a typical scale relevant to local climate impact could be that of a small city or approximately 30 km × 30 km ≈ 103 km2. There is a scale or intensity beyond which it is no longer possible to affect the climate locally without having a measurable effect remotely. For instance, Jones et al.[35] showed that a rather intense cloud brightening applied on a fairly small fraction of the world's ocean would have significant remote impacts in terms of surface temperature and precipitation. Such options therefore qualify as climate engineering. We recognize though that the border between local and regional climate modification is likely to be fuzzy and will require additional research and public debate conceivably on a case-by-case basis. For example, a small-scale intervention in the wrong place could have a larger impact than a large intervention in a less sensitive area.

Impact on Global Commons and Remote Effects

Climate engineering techniques may also be categorized according to whether they intervene within ‘territorial’ regions and/or intervene in the global commons. By global commons we refer here to the Earth's nonowned natural resources, such as the atmosphere and the oceans, but purposely leave the global climate out of this concept. Global commons are therefore resources that are beyond national jurisdictions whose ownership is subject to interpretation (Antarctica) or held jointly (e.g., the atmosphere because it is perpetually in motion and crosses borders). SRM by stratospheric aerosols and marine cloud brightening are examples of climate engineering techniques based on the global commons, for which some remote effects are known and understood.[35, 36] The CE medium (e.g., stratospheric aerosols) will travel in the atmosphere and cross national boundaries. Likewise ocean fertilization,[37, 38] increased ocean alkalinity[39] and SRM by ocean foam or hydrosols[40, 41] rely on using the ocean and can hardly be contained to territorial waters; they should therefore be considered as trans-territorial. Land-based CDR methods can be considered as territorial in that they operate within or at least from within national boundaries. They are unlikely to have any measurable trans-boundary impact (beyond that due to decreasing atmospheric concentrations of CO2) if performed on a small scale. However, such methods should be considered as trans-territorial when their intensity or scale are such that they have significant trans-boundary impacts. For instance land-based biological CDR techniques will have a regional impact on the hydrological cycle if they are performed on a large scale. Likewise SRM by whitening roofs or increasing desert reflectivity can be considered as territorial as long as they do not have any measurable remote, trans-boundary impact. There is, however, an intensity beyond which such techniques will have remote effects.[42] An important characteristic of climate engineering therefore relates to the existence of remote side effects.

Degree of (Perceived) ‘Naturalness’

Some proposed climate responses to climate change may be seen as more natural than others, such as planting trees even if on a large scale, and it has been suggested that ‘unnaturalness’ and/or the recourse to technology should be a characteristic of climate engineering methods.[43] Indeed, this issue also surfaced in the public perception of climate engineering.[44, 45] Notwithstanding that what constitutes a technology is ill defined, especially when technology attempts to reproduce or enhance some natural processes, this raises the issue of the relationship between humans and nature, which is value-laden. Furthermore, humanity has interacted with nature in a number of ways, e.g., through large-scale forest clearing, development of monocultures and breeding of animal and plant species for commercial purposes, thus it is debatable to what extent ‘nature’ exists independently of human influence, or is just socially constructed as such.[46, 47] Overall, the degree of ‘naturalness’ or the reliance on ‘technology’ do not really help to differentiate climate engineering from mitigation and adaptation.[15] However, it might be helpful to consider if a climate engineering approach removes warming climate agents or instead adds cooling climate agents (e.g., stratospheric aerosols). When removing climate agents, it may often be constructive to make a distinction between whether this is done by (a) decreasing anthropogenic emissions to the atmosphere, (b) increasing ‘natural’ sinks, (c) decreasing ‘natural’ emissions, or (d) creating new engineered sinks. While approaches under (a) definitely fall under a strict definition of mitigation and approaches under (d) are forms of climate engineering, those under (b) and (c) could be one or the other, depending on what other attributes are considered.

Degree of Permanency

This relates to the timescale at which the system will come back to its initial state after the action stops. The permanence of climate engineering methods (but also mitigation and adaptation actions) can vary greatly. It can be either a positive or a negative trait of particular climate engineering methods depending on their mode of action and side effects.


Related to but distinct from the previous attribute, is whether the impact of a particular method occurs slowly or rapidly. This relates both to the characteristics of the method and to its potential rate of deployment. This can again by either a positive or negative trait depending on mode of action and side effects.


High leverage is when a rather small effort (e.g., in terms of energy input or financial investment) results in a rather large climate engineering effect. It is generally the case that high-leverage climate engineering methods (such as stratospheric aerosol injection) can be rapid but have a fairly low degree of permanency, while methods with a greater degree of permanency have much lower leverage.


These considerations lead us to propose a more general categorization of responses to climate change (Table 1). In doing so, it is important to be clear about our aim and motivation. As a group of climate scientists, we seek to clarify the wider societal discourse and action on responses to climate change by distinguishing methods according to the processes in the Earth system being altered and the corresponding spatial and temporal scales of action and effects. However, our categorization is also aware of the political context in that it distinguishes territorial and trans-boundary actions/effects, rather than speaking purely in terms of numerical metrics of scale. In other words, it is a primarily scientific categorization that is also somewhat rooted in political reality. Alternative approaches to categorization could be technology-based or more policy-oriented.

Table 1. A Possible Categorization of Responses to Anthropogenic Climate Change Along with Their Attributes
Proposed Name and AcronymShort DefinitionMapping onto Previous TerminologyExamplesScale of ActionScale of ImpactsImpact on the Global CommonsaTrans-Boundary or Transnational Side Effects*Permanence of the Effect
  1. CCS, carbon capture and storage; BECCS, biomass energy with CCS; CDR, carbon dioxide removal; SRM, solar radiation management. RF, radiative forcing; BC, black carbon
  2. aBeyond any direct effects due to decreasing concentrations of greenhouse gases and other warming agents.
Anthropogenic emissions reductions (AER)Initiatives and measures to reduce or prevent anthropogenic emissions of warming agents into the atmosphereIncludes most forms of mitigation but excludes human-induced CO2 sink enhancementImproved energy efficiency, reduction in production and/or consumption of goods and services, introduction of renewable energies, nuclear energy, fossil fuel energy with CCS, reducing emissions from deforestation and forest degradation, emission reductions of BC and ozone precursorsGenerally a localized action or a sum of localized actionsGlobal through a decrease in the global-mean RF by greenhouse gases and other warming agentsNone expected. Expected to slow down the depletion of fossil fuel resources (except for CCS). Expected to slow down ocean acidification for CO2 measuresGenerally noneCommensurate to the atmospheric lifetime of the species being mitigated, longer if emission reduction is sustained
Territorial or domestic removal of atmospheric CO2 and other long-lived greenhouse gases (D-GGR)Removal of CO2 and long-lived greenhouse gases from the atmosphere operating within national jurisdictions and little consequences outsideIncludes territorial CO2 sink enhancement previously labeled under mitigation, with environmental side effects if any occurring within national jurisdictionsReforestation, biochar and other means of increasing storage of C in soils, small-scale afforestation, BECCS, CO2 air capture and storage in territorial (geological) reservoirs, enhanced weathering (without input of by-products into rivers or the oceans)Generally a localized action or a sum of localized actionsGlobal through a decrease in the global-mean RF by greenhouse gasesNone in the strict sense. May not slow down the depletion of fossil fuel resources. Expected to slow down ocean acidification for CO2 measuresPossible (e.g., through changes in evaporation, runoff, river flow, changes in biodiversity) but limitedCommensurate with the permanence of the storage medium
Trans-territorial or trans-boundary removal of atmospheric CO2 and other long-lived greenhouse gases (T-GGR)Removal of atmospheric CO2 and long-lived greenhouse gases from the atmosphere operating or having consequences partly or fully across or beyond national jurisdictionsIncludes the more environmentally disruptive CDR techniquesLarge-scale afforestation, ocean alkalinity, enhanced weathering (with input of by-products into rivers or the oceans), iron fertilization, injection of CO2 into the oceanLocalized but with significant remote effects and/or diffuse with diffuse effectsGlobal through a decrease in the global-mean RF by greenhouse gasesExpected (e.g., through changes to the water cycle, rivers, and/or the global ocean). Expected to slow down ocean acidification in the case of CDR methods not involving iron fertilization or ocean injectionTrans-boundary or transnational side effects expectedCommensurate with the permanence of the storage medium
Regional to planetary targeted climate or environmental modification (TCM)Intentional modification of the Earth's energy fluxes in order to offset climate change at the regional to global scaleEssentially what used to be defined as SRM, but excludes small-scale SRM. Excludes removal of long-lived greenhouse gases but includes other large-scale changes to the Earth's energy budget that do not involve long-lived greenhouse gasesInjection of stratospheric aerosols, marine cloud brightening, cirrus suppression, desert brightening on a large scale, ocean heat mixing, modification to Arctic sea iceLarge-scale and/or diffuse (even though the initial action can be local)Regional to global coolingYes (e.g., through the atmosphere and remote climate effects). Does not slow down ocean acidificationMeasurable trans-boundary or transnational side effectsShort as this happens through rapidly-responding components of the climate system
Climate change adaptation measures including local targeted climate or environmental modification (CCAM)Initiatives and measures to reduce the vulnerability of natural and human systems to the effects of climate change. Local risk managementEssentially what is usually considered as adaptation also includes small-scale SRM that only has small if any remote climate impactsRelocating urban or rural settlements, building dykes, air conditioning, agricultural crop choices, reflective crops, whitening of human settlements on a small scale, irrigationLocal with no or little remote effectsLocalNone expectedNone, limited to the local scale, or not detectable (i.e., ‘within the noise’). Some measures may however affect river flowCommensurate to the lifetime of the adaptation measure (typically months to decades)

Although our categorization scheme was developed independently, it shares some aspects of the typology recently put forward by Heyward.[15] However, it goes beyond Heyward[15] in that it broadens the SRM and CDR categories and revisits the boundaries of both mitigation and adaptation. The categorization proposes five main categories. A preliminary version of this categorization has been tested against a list of 47 possible responses to climate change by 10 scientists from the European Framework Programme 7 EuTRACE project involved in CE research. Because of the small sample of responses considered here, we did not seek to interpret the responses in a statistical sense but rather as expert elicitation to guide our study. Results from the questionnaire showed that in about one third of the cases, responses were unanimously attributed to a particular category by all 10 researchers, and in about two thirds of the cases it was unambiguous, with 8 of 10 agreeing; however, there were cases with multiple answers and/or some spread in the answers, with only about half of the responses agreeing on the ‘best’ choice. Most of the time the spread in answers was because of some ambiguity on the scale of the climate change response or because the primary intention of the method was not necessarily to address climate change. This exercise led to further improvements and refinements of the definitions and attributes of the five categories. We present the new categorization that resulted from these improvements in Table 1.

The first category, anthropogenic emissions reductions (AER), includes initiatives and measures to reduce or prevent anthropogenic emissions of warming agents into the atmosphere. This encompasses most of what IPCC and UNFCCC regard in their definition as mitigation, including CCS from fossil fuel combustion. AER excludes, however, CO2 sink enhancement techniques that are considered as mitigation by IPCC and UNFCCC. This category could be further divided into the mitigation of emissions of well-mixed (long-lived) greenhouse gases and the mitigation of emissions of short-lived radiative forcing agents such as black carbon.

The second and third categories, domestic and trans-boundary greenhouse gas removal (D-GGR and T-GGR, respectively), encompass all techniques aiming at removing CO2 from the atmosphere (i.e., CDR) as well as potential techniques to remove other long-lived greenhouse gases from the atmosphere. D-GGR includes techniques that are territorial, do not involve the global commons and are unlikely to have any significant trans-boundary impact (i.e., those techniques originally envisaged as being part of mitigation in a UNFCCC sense), while other, potentially more disruptive techniques belong to T-GGR. Whether the carbon is stored in national jurisdictions (e.g., in forests, soils, or geological reservoirs) or outside such national jurisdictions (e.g., in the ocean) matters as this would raise very different cultural, political, and legal issues (e.g., for verification purposes). While we focus here on the potential trans-boundary effects on climate and global commons, we also recognize that other trans-boundary transfers may be involved (e.g., fibers or financial flows).

The fourth category, targeted climate modification (TCM), encompasses all techniques aiming to affect the climate on the regional to global scale through a direct modification of radiative or energy fluxes, rather than indirectly through the removal of long-lived greenhouse gases. It includes but is not restricted to SRM techniques. Most of the proposed techniques that fall in this category are likely to operate rather rapidly, but they lack permanency, i.e., the climate system will return rapidly to its initial state if this type of climate engineering is stopped. We tentatively define the regional scale as covering an area of the order of 300 km × 300 km ≈ 105 km2 but recognize there is a judgement involved here.

Finally, the fifth category, climate change adaptation measures (CCAM), covers traditional adaptation measures, but broadens the concept to also include those relying on local targeted cooling, as long as they have no measurable remote effects. We tentatively define the local scale as being of the order of 30 km × 30 km ≈ 103 km2. The broadening of the adaptation concept does not necessarily require any changes in the currently accepted definition, but we nevertheless propose the new acronym CCAM to designate this expansion, and also to make clear that it is referring to adaption to climate change (since adaptation can also apply to other environmental and also social changes).

Figure 1 provides a flowchart that helps classify a particular action into the categories defined above. The question of ‘significant trans-boundary (remote) effects’ is likely to be the one that is most difficult to answer and where some fuzziness may arise between categories D-GGR and T-GGR, as well as between TCM and CCAM. Analyses of past events and regional climate modeling can be used to inform what the threshold scale/intensity of a particular method is for remote effects to become detectable. For completeness we can mention a sixth possible category, introduced and labeled as ‘rectification’ by Heyward,[15] which covers financial compensation and symbolic measures.

Figure 1.

Flowchart of the proposed categorization of climate change responses. Text in white boxes corresponds to policy questions; text in the green box corresponds to a legal question; text in pink boxes corresponds to questions that involve value judgements but can be informed by climate and related sciences. Yellow boxes to the right side list the five categories.

Although not universal, the new five-class categorization maps naturally on current usage. The combination of AER and D-GGR corresponds to the definition of mitigation by UNFCCC and IPCC. The combination of D-GGR and T-GGR encompass all methods of atmospheric removal of carbon dioxide (CDR) along with other long-lived greenhouses gases. Categories T-GGR and TCM collectively correspond to the more ‘disruptive’ forms of interventions, which corresponds to what is often envisioned under the term ‘climate engineering,’ and are likely to be the most controversial. Finally, categories TCM and CCAM have in common that they do not attempt to reverse the anthropogenic modification of the atmospheric composition, which is the main cause of climate change.

There remain some overlaps in our classification, for example, BECCS can both offset fossil fuel emissions (AER), achieve territorial removal of CO2 (D-GGR), potentially range across territories by involving import of biomass in the global economy (T-GGR), and create a land-use change carbon source (thus adding to the anthropogenic causes of climate change).[48] Such overlaps are inevitable in any attempt to fit the world into simple categories. They can be beneficial in the sense of encouraging broader thought about the consequences of particular actions. We suggest that particular methods such as BECCS can be assigned to a primary category in our classification, with the possibility of secondary categories where these are clearly identifiable.


The large and growing number of CE methods that have been proposed, irrespective of their feasibility, is causing ambiguity in the current definitions of mitigation and adaptation. The definition of climate engineering itself is unclear and has been a recurring issue in discussions. We have proposed a simple categorization of the strategies to respond to anthropogenic climate change in five distinct classes that alleviates some of the issues associated with the existing terminology.

In particular, we recommend that CDR techniques (except CCS directly at power and industrial plants) should no longer be considered as part of mitigation, but should instead form separate categories of responses to climate change. We also suggest that localized climate modification can be seen as an adaptation measure, as long as there are no measurable remote environmental effects. In both cases the scale and intensity of action are essential attributes from technological, climatic, and political viewpoints. Differentiating territorial or domestic removal of greenhouse gases (D-GGR) from trans-territorial or trans-boundary removal (T-GGR), and differentiating local adaptation via targeted climate and environmental modification (a component of CCAM) from regional and global modification (TCM) will require further work.

Our aim is to progress the debate on how climate engineering should fit (or not) in the portfolio of existing climate change policies. We hope that our proposed revisions to the terminology and necessary definitions of the boundaries between categories can help to foster a debate among stakeholders such as the research, policy, and civil society communities at large. We invite in particular further feedback from social scientists and policymakers on our preliminary categorization.

Finally, we have not tried to rank or assess policy responses to climate change, either within a category, or between categories. Multiple factors have to be considered when comparing these policy responses, including their technological maturity, affordability, effectiveness, scalability, timescale for implementation, risk, residual climate change, unintended consequences, degree of interference with the climate system, intergenerational and trans-regional ethical implications, the policy and governance challenges they pose.


This opinion article has benefited from discussions held at the IPCC Expert Meeting on Geoengineering that took place in Lima, Peru on 20–22 June 2011, which four of the authors attended. PMF and NEV would like to acknowledge support from the EPSRC/NERC for the IAGP project. OB, ML, TML, AM, and NEV would like to acknowledge partial support from the FP7 EuTRACE project grant agreement (306395). AM and ML would additionally like to thank the German Ministry for Education and Research (BMBF), the Brandenburg Ministry for Science, Research and Culture (MWFK) and the Research for Sustainability Platform (FONA). OB acknowledge supports from the Agence Nationale de la Recherche under the REAGIR project. NG acknowledges the financial support of ETH Zürich. The authors would also like to thank Jim Haywood, Thomas Leisner, Helene Muri, Rodrigo Riestein, Simon Shackley, Harald Stelzer for useful discussions and for responding to the questionnaire. Bénédicte Fisset helped with formatting the manuscript and references. We thank three anonymous reviewers for their insightful comments that have contributed to improve this opinion article.