Confronting the abatement paradox Integrating aerosol cooling within climate change mitigation policy
Justin Wood BSc, BAppSc, MSc
is thesis is presented for the degree of Doctor of Philosophy of Murdoch University
2013
Discipline of Physics and Energy Studies School of Engineering and Information Technology I declare that this thesis is my own account of my research and contains, as its main con- tent, work that has not been previously submied for a degree at any tertiary education institution.
Supervisors: Prof. Philip Jennings and Adam McHugh, with assistance from Dr August Schläpfer.
Produced in the open-source LYX editor and Zotero bibliographic management soware. Typeset in LATEX and XƎTEX with the open-source fonts Linux Bolinum, Linux Libertine, and Fontin Sans. My sincere gratitude to the LYX user community, Jürgen Spitzmüller in particular, and to Petr Šimon for LyZ, his invaluable Zotero–BIBTEX–LYX integration tool.
e author’s original work is licensed under Creative Commons. All other material remains copyright of the authors and their publishers. Acknowledgements
My sincere gratitude and appreciation goes to my principal supervisors, Philip and Adam, for their wisdom and their forbearance. August provided welcome advice and enthusi- asm. I thank Prof. Tom Lyons for early review of the physical science material (remain- ing mistakes or omissions are mine alone). I can scarcely thank Dr Cecily Scu enough, for without her support, guidance, concern, and when needed, cajoling, this thesis would simply never have been completed. And my dear friend Dr Beth Gouldthorp, for her unending encouragement and understanding. I wish to recognise those that led or helped me to begin this doctorate, oen unbe- knownst to them: David Spra, whose book Climate code red seeded the idea that we need to talk about aerosols; James Hansen, for his indefatigable courage, determination, and insight; RealClimate.org, for teaching us all; and to my friend Glenn, who once told me that he had never regreed any decision to just take the leap.
And finally, to my friends and family, for puing up with me thus far.
Contents
Abstract 1
1 Introduction and overview 5 1.1 e aerosol dilemma ...... 5 1.2 On avoiding dangerous interference ...... 8 1.3 Research questions ...... 14 1.4 Methodology ...... 16 1.4.1 Research approach ...... 16 1.4.2 esis structure ...... 17 1.4.3 Nomenclature ...... 18
2 Aerosol effects on climate 19 2.1 Physical characteristics ...... 20 2.1.1 Formation and composition ...... 20 2.1.2 Mixing state ...... 20 2.1.3 Atmospheric residence time and emissions ...... 21 2.1.4 Atmospheric distribution ...... 22 2.1.5 Perturbation baseline ...... 23 2.2 Optical and radiative properties ...... 24 2.2.1 Particle effective radius ...... 24 2.2.2 Absorption, scaering, and extinction coefficients ...... 25 2.2.3 Asymmetry parameter ...... 26 2.2.4 Single scaering albedo ...... 26 2.2.5 Aerosol optical depth ...... 27 2.3 Radiative forcing ...... 29 2.3.1 Forcing efficacy ...... 33 2.4 Species ...... 34 2.4.1 Natural aerosols ...... 35 2.4.2 Anthropogenic aerosols ...... 35 2.5 Climatic effects and influence ...... 46 2.5.1 Direct radiative effect ...... 47 2.5.2 Indirect cloud albedo effect ...... 51 2.5.3 Indirect cloud lifetime effect ...... 54
i Contents
2.5.4 Semi-direct effect ...... 57 2.5.5 ermodynamic effect ...... 62 2.5.6 Glaciation effect ...... 63 2.5.7 Biomass fertilisation effect ...... 64 2.5.8 in Arctic cloud ...... 65 2.5.9 Longwave absorption effect ...... 66 2.5.10 Further effects on climatic processes ...... 66 2.5.11 Effects on surface irradiance ...... 68 2.5.12 Summary of effects ...... 73 2.6 Synthesis and implications ...... 75 2.7 Observation and modelling ...... 79 2.7.1 Observation and measurement ...... 79 2.7.2 Aerosol modelling ...... 83
3 Climate response and associated risk 95 3.1 Climate response to forcing ...... 95 3.1.1 Radiative forcing revisited ...... 96 3.1.2 Equilibrium climate sensitivity ...... 98 3.2 Planetary energy imbalance and climate response ...... 99 3.2.1 antitative estimates of energy imbalance and aerosol forcing . 103 3.3 Commied warming ...... 109 3.3.1 Derived climate response function ...... 110 3.4 Dangerous anthropogenic interference ...... 113 3.4.1 Aerosol-induced temperature change ...... 114 3.5 Toward appropriate metrics ...... 118 3.5.1 e problem with GWP and CO2-e ...... 119 3.5.2 Alternative methods ...... 122 3.6 Aerosol emission rates ...... 130 3.6.1 Aggregate emissions ...... 131 3.6.2 Emissions by source ...... 134 3.6.3 Future emissions ...... 136 3.7 Policy implications ...... 141
4 Comparative analysis of mitigation policy 143 4.1 Pollution taxonomy ...... 143 4.1.1 Emission and assimilation ...... 144 4.1.2 Atmospheric mixing ...... 147 4.1.3 Environmental impact ...... 148 4.1.4 Applying the taxonomy ...... 150 4.1.5 A new term for emission sources ...... 151
ii Contents
4.2 Policy ontology and form ...... 152 4.2.1 Less is beer ...... 152 4.2.2 Less is worse ...... 154 4.2.3 e policy challenge ...... 156 4.3 A revised ontology ...... 159 4.3.1 Clarifying effect and impact ...... 159 4.3.2 Expanded classification schema ...... 161 4.4 Accounting for aerosols in climate change policy ...... 166 4.4.1 Linkage synergy and co-benefits ...... 166 4.4.2 Linkage barriers and conflict ...... 167 4.4.3 Are they included currently? ...... 170 4.4.4 Could they be? ...... 174 4.5 Confounding complexities ...... 178 4.5.1 e problem of CO2 non-equivalence ...... 178 4.5.2 Correlated externality feedbacks ...... 179 4.6 Unintended consequences: the abatement paradox ...... 184 4.7 Dangerous omissions ...... 186
5 Aerosol-integrated mitigation policy criteria 193 5.1 Policy objective ...... 194 5.2 Discussion and scope ...... 196 5.2.1 Definition of terms ...... 197 5.2.2 Plausible compensative measures ...... 198 5.2.3 Which emissions, which sources? ...... 203 5.2.4 Measurement and reporting ...... 206 5.2.5 How much and for how long? ...... 208 5.2.6 Temporal and spatial considerations ...... 210 5.2.7 Compensative masking services ...... 212 5.2.8 e question of accounting ...... 213 5.2.9 Domestic implementation ...... 216 5.3 Implementation model evaluation ...... 218 5.3.1 Loss-liability model ...... 219 5.3.2 Cost-neutrality model ...... 223 5.3.3 Funding and ultimate liability ...... 226 5.3.4 Central authority model ...... 228 5.3.5 Conferred value model ...... 230 5.4 Ramifications for abatement pathways ...... 236
6 Scheme implementation: the balancing market 241 6.1 Restricted direct value model ...... 241 6.1.1 Overview and rationale ...... 242
iii Contents
6.1.2 Carbon price mechanism integration ...... 243 6.2 Scheme implementation ...... 244 6.2.1 Licensed compensative measures ...... 246 6.2.2 Coverage and reporting liabilities ...... 247 6.2.3 antification metric ...... 249 6.2.4 NFA emission baseline ...... 250 6.2.5 e residual masking license ...... 252 6.2.6 Market demand ...... 253 6.2.7 Reverse auction mechanism ...... 255 6.2.8 Contract funding ...... 260 6.3 Market dynamics ...... 265 6.3.1 Price determinants ...... 265 6.3.2 Supply shortfalls ...... 271 6.3.3 Effect on carbon price and abatement pathway ...... 275 6.3.4 Supply excess ...... 291 6.4 Interaction with air pollution regulation ...... 295 6.4.1 Apparent conflicts ...... 296 6.4.2 Operational price dynamics ...... 298 6.4.3 Targeted regulation ...... 301 6.5 Model evaluation ...... 302
7 Conclusions 305 7.1 Research questions revisited ...... 305 7.2 Summary of key findings ...... 306 7.3 Limitations and further research ...... 310 7.4 A final statement ...... 312
References 315
Glossary 331
iv List of Tables
2.1 DRF (Fa) by anthropogenic species reported by AR4 ...... 49 2.2 Summary of aerosol effects on climate ...... 74
3.1 Estimates of planetary energy imbalance and inferred aerosol forcing .. 108 3.2 Comparative GWP and GTPP at time horizon H years for BC, OC, and SO2 125 3.3 Specific forcing pulse for black carbon and organic maer ...... 128 3.4 Comparative RF and 20 & 100 year GWPs for BC and OM, derived from SFP ...... 129 3.5 Comparison of selected characteristics of alternative metrics ...... 130 3.6 Proportional anthropogenic emissions of major aerosol species by sector 135
4.1 Comparative dominant pollutant characteristics ...... 150 4.2 Comparative pollutant effects ...... 155 4.3 Co-benefits in climate and air quality policy linkage ...... 168 4.4 Conflicts in climate and air quality policy linkage ...... 169
5.1 Categories of plausible compensative measures ...... 199
6.1 Sealed-bid first-price reverse auction types ...... 257 6.2 Two balancing market funding Options ...... 262
v List of Figures
1.1 Global CO2 emissions and carbon intensity ...... 10
2.1 Characteristic AOD for combined anthropogenic aerosol loading ..... 28 2.2 Illustration of RF calculation methodologies ...... 32 2.3 Range of forcing efficacies reported in AR4 ...... 34 2.4 Illustrations of aerosol influence on climate ...... 47 2.5 AR4 DRF at TOA by anthropogenic species ...... 50 2.6 Indirect cloud albedo effect ...... 52 2.7 Aerosol atmospheric forcing at altitude ...... 58 2.8 Semi-direct effect ...... 59 2.9 Aerosol effects on mixed-phase clouds ...... 63 2.10 Surface dimming due to all anthropogenic aerosols ...... 70 2.11 AR4 climate forcings by major component ...... 76 2.12 Aerosol and GHG forcing PDF ...... 77 2.13 Time dependence of aerosol optical thickness and effective forcing .... 91 2.14 Temporal evolution of surface and radiative forcing ...... 91
3.1 Illustrative sketch of the response to positive energy imbalance ..... 100 3.2 Cumulative planetary energy budget since 1950 ...... 103 3.3 Indicative planetary energy imbalance over time ...... 107 3.4 Dangerous anthropogenic interference for a range of tipping elements . 115 3.5 Illustrative response to air pollution and climate change mitigation policies 119 3.6 Modelled aerosol burden over time ...... 132 3.7 Total annual emissions of major aerosol species since 1850 ...... 133 3.8 Global SO2 emissions since 1850 by source and by end-use sector .... 136 3.9 Projected anthropogenic aerosol emissions 2000 to 2100 under all RCP scenarios ...... 139
4.1 Illustrative pollution classification: aerosols as local air pollutants .... 164 4.2 Illustrative pollution classification: GHG effects on climate ...... 164 4.3 Illustrative pollution classification: GHG effects on climate and oceans . 165 4.4 Illustrative pollution classification: aerosol effects on forcing ...... 165 4.5 Illustrative pollution classification: aerosol effects on forcing and hydro- logical cycle ...... 165
vi Contents
4.6 Illustrative pollution classification: all forcing agent effects on climate .. 166
6.1 Restricted direct value model ...... 259 6.2 Illustrative carbon price tolerance of two representative facilities ..... 279 6.3 Shutdown curve for two representative technology groupings, period t=0 280 6.4 Shutdown curve for two representative technology groupings, period t=x 281 6.5 Shutdown curve for two representative technology groupings, period t=y 281 6.6 Economy-wide shutdown curve ...... 283 6.7 Illustrative carbon price tolerance of two representative facilities, shied by balancing market revenue under funding Option 1 ...... 284 6.8 Shutdown curve for coupled NFA (A) versus non-NFA emiing (B) tech- nology groupings under funding Option 1, period t=y ...... 285 6.9 Relative tolerances are reordered by balancing market revenue under funding Option 1 ...... 286 6.10 Balancing market revenue causes a shi to a new abatement pathway under funding Option 1 ...... 286 6.11 Illustrative carbon price tolerance of two representative facilities, shied by levy costs under funding Option 2 ...... 288 6.12 Shutdown curve for coupled NFA (A) versus non-NFA emiing (B) tech- nology groupings under funding Option 2, period t=y ...... 289 6.13 Relative tolerances are reordered by levy costs under funding Option 2 . 290 6.14 Levy costs causes a shi to a new abatement pathway under funding Option 2 ...... 290 6.15 Minimum required offer price coupled emier shutdown curve ...... 294 6.16 Scheme interaction with air pollution regulation through price dynamics 300
vii List of Boxes
2.1 e role of SO2 pollution regulations ...... 37 2.2 Aerosol effects on the hydrological cycle ...... 60 2.3 Missing satellite observations ...... 80
3.1 Is climate response faster than previously thought? ...... 111 3.2 Australian aerosol emissions ...... 137
4.1 Perverse incentives as an impediment to discourse? ...... 172 4.2 Beware Black Swans ...... 192
5.1 Geoengineering can never substitute for real GHG abatement ...... 200 5.2 e polluter pays principle ...... 219 5.3 e ethical dilemma of aerosol cooling ...... 238
6.1 Ethics and environmental effectiveness ...... 254 6.2 e risk of Scheme collapse ...... 275
viii Abstract
Anthropogenic climate change is a problem more confounding than commonly appreci- ated. For while greenhouse gas (GHG) emissions lead to a positive climate forcing and hence rising global mean surface temperature, parallel emission of particulate aerosol species does much the opposite, in parallel to their deleterious effects as local air pollut- ants. rough direct radiative interaction with incoming sunlight and complex indirect microphysical effects on cloud, anthropogenic aerosols exert a significant negative for- cing — a counteracting cooling influence. Global temperature rise to date is thereby substantially less than would have occurred due to past GHG emissions alone. In fact, painstaking investigations find that aerosols mask between about 35 % and as much as 50 % of current GHG warming.
What is more, where GHGs remain in the atmosphere for decades to centuries aer emis- sion, aerosols are washed out within days to weeks. is large residence time asymmetry means that the ‘protection’ afforded by aerosol masking depends entirely on their con- tinual emission. But that masking comes at the expense of well-documented damage to natural systems, physical infrastructure, and human health. Crucially, many activities that produce GHGs also emit aerosols and their precursor gases, fossil fuel combustion prominent among them. Climate change mitigation policies that abate GHG emissions therefore simultaneously reduce aerosols — not only their emissions, but actual atmo- spheric loading. Yet due to the residence time asymmetry, the positive forcing of existing GHG concentrations is lile affected in the short term, while the aerosol mask weakens
1 in direct consequence of those lost emissions. is weakened mask constitutes a ‘neg- ative abatement feedback’ of mitigation efforts.
If unmasking is large enough, near the full effect of existing GHG forcing will be ex- posed, leading to potentially strong and rapid increase in temperature. As thresholds for dangerous climate change are quickly approaching, including a range of climatic tip- ping points, a lost aerosol mask as an unintended side effect of genuine GHG abatement may hence, paradoxically, exacerbate rather than mitigate against dangerous anthro- pogenic interference. e current trajectory of ever-rising global GHG emissions and unambitious, ineffective international mitigation efforts therefore suggests that a policy response to lost aerosol cooling will be needed when abatement finally begins in earnest, as it must. However, critical assessment of current policy making and instrument design finds that this systemic risk is largely unrecognised, and that normative ontological bias in the conception of pollutant effects hinders adequate recognition of aerosols’ full role.
e objective of this thesis is then to investigate policy approaches by which that risk may be contained.
To do so, I construct a set of policy design criteria that must be met by any effective mitigation framework that explicitly confronts the aerosol abatement paradox. Chief among these is the need to avoid distortion of underlying real GHG abatement; to ac- count for aerosol emissions via metrics that accurately represent their negative forcing character; to properly recognise temporal constraints on their removal; and to ensure that withdrawal of aerosol masking is ‘managed’. Analysing a range of conceivable im- plementation models uncovers further critical insights for effective design, in accordance with the physical science.
2 From these foundations, I propose the ‘balancing market’ mechanism, a prototype policy instrument compliant with those criteria, integrated with a pre-existing carbon price mechanism under an assumed, and necessary, future international agreement. e cent- ral feature of the underlying implementation model is to prioritise abatement pathways in which GHG-emission sources without coupled aerosol emissions are shutdown first. e balancing market does this by paying for the quantity of negative forcing deemed required until mitigation efforts restore a ‘safe’ atmospheric configuration. In a stark illustration of the abatement paradox’s unprecedented challenge, this translates as pay- ment for continued aerosol emissions during a transitionary period while alternative compensative technologies are developed — likely including carbon dioxide removal, and forms of geoengineering.
It is my hope that this proposal and the analysis on which it is founded may assist in expanding the policy discourse so as to take full and unvarnished account of all com- ponents of anthropogenic interference in the climate system — and the danger they represent. e ethical implications are enormously confronting, but such is the nature of the age of consequences.
3
1 Introduction and overview
1.1 The aerosol dilemma
Anthropogenic climate change is a physical reality in the twenty-first century (eg, IPCC
2007a). Greenhouse gases (GHGs) such as carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), and the halocarbons, generated by a range of human activities, absorb and re-radiate thermal energy emied by the Earth’s surface. In doing so, these gases in- crease retention of energy within the Earth-atmosphere system, perturbing planetary energy balance. e global mean surface temperature rises in response until energy balance is restored at a new radiative equilibrium. Anthropogenic greenhouse gases thereby warm our planet. As the Intergovernmental Panel on Climate Change (IPCC) state clearly in their 2007 Fourth Assessment Report (AR4), the evidence of that warming is now ‘unequivocal’ (IPCC 2007b, 5).
Yet there exists another class of climatically active substances emied by human activity in parallel to greenhouse gases. Aerosols also exert significant influence on the climate system, though less well-known than greenhouse gases — and far less understood. While greenhouse gases heat our planet, in short, aerosols are cooling it. e cooling effect of these microscopic solid and liquid –phase atmospheric particles is caused by a complex array of direct and indirect processes that, on balance,1 enhance the effective planetary
1 Black carbon complicates this picture by exerting a significant net warming influence (see sec- tion 2.4.2.3).
5 Chapter 1 Introduction and overview albedo, reducing the quantity of solar radiation absorbed at the surface (see Chapter 2). is reduction results in a negative planetary energy imbalance. e continuing net positive imbalance is the sum of the negative aerosol term and the larger positive term of GHGs. Aerosols thereby mask as much as half of GHG-induced warming, though the true strength is far from certain.
e countervailing aerosol mask means that global temperature rise to date is markedly less than would have occurred in the absence of anthropogenic aerosols.2 Nonetheless, a significant positive transient energy imbalance remains, and the inexorable rise of atmospheric GHG concentrations continues to strengthen it — further global warming is hence ‘in the pipeline’.
at aerosols have slowed the onset of climate impacts can rightly be regarded as a blessing. Yet these same species are simultaneously major local air pollutants, caus- ing well-documented damage to natural ecosystems, human infrastructure, and human health. eir conflicting roles in climate (masking warming) and local air and water quality (damaging pollutants) then gives rise to a dilemma: actions that target these aerosol species for abatement as pollutants will also lead to increased global warming (Crutzen 2006; Ramanathan and Feng 2009).
is framing is the common expression of the aerosol dilemma, but it does not give the full picture. For beyond the negative consequences of air pollution regulations, the reality of aerosol masking also constitutes a potential foil for any eventual serious co- ordinated climate change mitigation effort, for two fundamental reasons.
First, anthropogenic aerosols are emied by a range of activities, including industrial practises, biomass burning, agriculture and land use change, and in particular the com- bustion of fossil fuels.3 Many of these same activities, especially fossil fuel combustion,
2 Other effects of increased GHG emissions are not offset by aerosols, ocean acidification chief among them. 3 is applies overwhelmingly to coal and oil; natural gas combustion produces few aerosols (Ramanathan and Feng 2008).
6 1.1 e aerosol dilemma are also the principal anthropogenic GHG emission sources — aerosols and GHGs are oen emied simultaneously from a single underlying source. Aempts to curtail GHG emissions from these shared-source activities will then necessarily also curtail aerosol emissions.
Second, critically, while most GHGs have atmospheric residence times of decades to centuries, aerosol lifetimes are orders of magnitude smaller, measured in days to weeks. Atmospheric aerosol loading is therefore directly proportional to their emission rates at any point in time, hence so is their cooling effect. Conversely, atmospheric GHG con- centrations are a function of cumulative emissions, the particular gas species’ properties, and the response of the Earth-atmosphere cycles which remove them over time.4 Indeed, the changes already wrought by existing atmospheric carbon dioxide concentrations will likely persist for a millennium (Solomon et al. 2009) — even immediate cessation of all anthropogenic emissions would not ‘undo’ those changes, quite apart from the role of aerosols.
is residence time asymmetry means that abatement of fossil fuel GHG emissions will in general simultaneously diminish anthropogenic aerosol loading, not simply their emission rates. Aerosol cooling will thereby reduce immediately in response to any fall in emissions, while the GHG perturbation of climate continues, subject to the re- moval rate of the carbon and other cycles. Any loss of aerosol cooling therefore induces a matching increase in positive planetary energy imbalance: effectively an instant in- crease in solar radiation incident at the planetary surface. Global mean surface temper- ature responds rapidly to such sudden changes in perturbation (Brasseur and Roeckner 2005; Mahews and Caldeira 2007; Ross and Mahews 2009); a precipitous removal of aerosols may therefore induce a rate of temperature rise beyond any yet experienced (see Chapter 3).
e dilemma posed by anthropogenic aerosols is hence more perverse than first stated
4 Principally the carbon cycle, but also others such as that of nitrogen.
7 Chapter 1 Introduction and overview above. In achieving transformative action to significantly reduce anthropogenic GHG emissions, aerosol masking is simultaneously diminished; yet the rates of natural at- mospheric GHG removal processes are orders of magnitude slower than of aerosols, so near the full warming effect of existing GHG concentrations is exposed. In seeking to arrest GHG warming, we may in fact strengthen it for a decade or more, with potentially dire climatic consequences for human and natural systems. e aerosol dilemma hence renders anthropogenic climate change a threat more complex and confounding than indicated by consideration of temperature trends and GHG emissions alone. Hansen (2009, 98) describes the situation starkly as ‘the Faustian climate bargain that humanity inadvertently entered into’.5
e aerosol cooling dilemma, properly understood, has profound implications for cli- mate policy. It is the underlying contention of this thesis that those implications are to date largely unrecognised in the public and political discourse regarding anthropo- genic climate change, and the policy mechanisms designed to address it. e role of an- thropogenic aerosols may ultimately come to dramatically undermine mitigation policy mechanisms that remain blind to them.
1.2 On avoiding dangerous interference
e ultimate objective of this Convention … is to achieve … stabilization of greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system. Such a level should be achieved within a time frame sufficient to allow ecosystems to adapt naturally to climate change, to ensure that food pro- duction is not threatened and to enable economic development to proceed in a sustainable manner. (United Nations 1992, Article 2)
e need for systemic abatement of anthropogenic GHG emissions is now long known.
5 Hansen explains that no maer what action humanity may take, fossil fuels must eventually run out, and at that point our payment comes due as aerosols disappear.
8 1.2 On avoiding dangerous interference
e ever-strengthening scientific understanding of climate change and the danger it poses has been periodically reviewed and articulated by the IPCC since their First As- sessment Report in 1990. e United Nations Framework Convention on Climate Change (UNFCCC) was adopted in 1992, with its o-cited objective of avoiding dangerous an- thropogenic interference, reproduced above.
Aerosols’ role in anthropogenic climate change was broadly understood by the early 1990s, if not the finer details or full magnitude. e IPCC’s First Assessment Report dis- cussed aerosols along with GHGs, but noted their net effect on climate forcing was as yet unclear. By 2001, the IPCC’s 2001 ird Assessment Report (TAR) devoted a full chapter to the effects of aerosols, and this work was substantially updated in AR4.6 In general how- ever, aerosol cooling is not readily apparent as a factor in the evolution of climate change mitigation policy over the past two decades. Given the state of climate science from the start of that period, GHG emission rates, and perceived time available to properly enact changes to socioeconomic systems, that omission is understandable and not unreason- able. ough in a 1991 Nature article, Wigley raised the potential for increased warming as an unintended side effect of declining sulphate7 aerosol emissions as a consequence of fossil fuel abatement. His warning that ’[r]ather than acting as a panacea, the pos- sible effects of fossil-fuel-derived sulphate aerosols should be seen as further reason for implementing controls on fossil-fuel use’ (p. 506) has evidently gone unheeded.
Continued omission of aerosol cooling from mitigation policy discourse is no longer reasonable — all of the factors listed above have now changed dramatically.
GHG emission rates have surged since the UNFCCC was ratified, accelerating in par- ticular from 2000 to grow at about 3.4% per year8 (Allison et al. 2009; Richardson et al.
6 Indeed, AR4 clearly aributes the substantial uncertainties surrounding aerosols as the principal cause for the widening of the net anthropogenic forcing probability distribution compared to that of GHGs alone (see section 2.6). 7 Sulphate is an important aerosol species; see section 2.4.2.1. 8 Fossil fuel derived CO2 emissions were some 40% higher in 2008 than in 1990.
9 Chapter 1 Introduction and overview
2009); in fact, emission trends continue to track the upper bound of the most fossil fuel intensive scenario considered by the IPCC in the TAR or AR4, A1FI, though the airborne fraction appears to be remaining steady. Even the demand destruction effects of the global financial crisis of 2007 to 2009 did lile more than momentarily suppress GHG emissions, as shown in Figure 1.1.9
Figure 1.1: Global CO2 emissions and carbon intensity (Fig. 1a, Peters et al. 2012). CO2 emissions from fossil fuel combustion and cement production on le axis (black line), carbon intensity of world GDP in US$ (2000) on right axis (red line). Reprinted by permission from Macmillan Publishers Ltd: Nature Climate Change, copyright 2012.
Global CO2 emissions from fossil fuel combustion and cement production recovered in 2010, growing 5.9 % (compared to the 2009 fall of only 1.4 %), an increase of 0.59 Pg C to 9.1 ±0.5 Pg C — the highest ever annual jump, beating the previous records set in 2003 and 1979 (Peters et al. 2012). e IEA reported a further, though smaller, 3.2 % rise in
9 Peters et al. (2012, 2) state that ‘For recent decades, the growth in global CO2 emissions can be ex- plained mainly by the growth in economic activity corrected for decreases in the fossil-fuel carbon intensity (FFCI) of the global economy’. However, the downward trend in FFCI between 1980 to 2000 of 1.4 %/yr shrunk to 0.9 %/yr from 2000, which the authors interpret as indicating ‘the positive trend of improvements in carbon intensity reversed’ (p. 2).
10 1.2 On avoiding dangerous interference
fossil fuel CO2 emissions in 2011 (International Energy Agency 2012).
Knowledge of the climate system has also increased enormously in the past two dec- ades. Improved understanding and ongoing observations increasingly indicate climatic impacts more severe and more rapid than expected even a few years earlier, and with remaining uncertainties surrounding some important aspects of climate science point- ing only toward more deleterious outcomes (Allison et al. 2009; Steffen 2009; Richardson et al. 2009; Garnaut 2011). e time frame for effective mitigation action contracts due to both the persistent lack of any decline in emissions, and continual downward revisions of emission levels at which truly dangerous climate changes will occur. Perhaps most importantly of all, critical and far-reaching shis in climate science and in observations have occurred since the literature reviewed in AR4 was published. Prominent examples include:
• e precipitous decline in Arctic sea ice extent and volume, far exceeding worst- case AR4 projections, with consequent feedback reduction of ice albedo and markedly enhanced regional warming (Lindsay et al. 2009; Eisenman and Welaufer 2009; Allison et al. 2009; Comiso et al. 2008; Kwok and Rothrock 2009).
• Accelerating rates of mass loss in both the Greenland and Antarctic ice sheets (Rignot et al. 2011; Svendsen et al. 2013), beer understanding of ice sheet dy- namics, and strong upwards revisions of projected sea level rise (Rahmstorf 2010; Vermeer and Rahmstorf 2009).
• Firming consensus that cumulative emission budgets are more deterministic of eventual climate changes than year-on-year emission rates, or even atmospheric GHG concentrations (Zickfeld et al. 2009; Meinshausen et al. 2009; England et al. 2009; Allen et al. 2009).
• e paleoclimatic evidence suggesting that ‘safe’ atmospheric GHG levels have already been surpassed (Hansen et al. 2008), and the related concept of tightly-
11 Chapter 1 Introduction and overview
coupled planetary boundaries within which humanity must operate (Rockström et al. 2009).
e implications of these shiing dynamics for the continued relevance of the o-cited 2 ℃ target are emblematic. As a ‘guard rail’ against dangerous anthropogenic inter- ference in the climate system, limiting global warming to an increase of 2 ℃ beyond preindustrial levels10 has become a widely-nominated goal in international mitigation discourse in recent years (Richardson et al. 2009; WBGU 2009; Anderson and Bows 2011). Yet the progression of climate science makes that goal now decidedly problematic. Not only does a 2 ℃ limit realistically require serious mitigative action with net declines be- ginning by 2020 at the latest11 (WBGU 2009; Meinshausen et al. 2009; England et al. 2009), the target itself ‘now more appropriately represents the threshold between “dangerous” and “extremely dangerous” climate change’ (Anderson and Bows 2011, 20).
e looming risks of triggering multiple points of nonlinearity underscore re-framing of the problem. An array of ‘tipping points’ within the Earth system exist as probability distributions along a continuum of temperature increases, and some may be triggered before the 2 ℃ threshold is breached (Schellnhuber 2009; Lenton et al. 2008; Kriegler et al. 2009; Rockström et al. 2009; Ramanathan and Feng 2008). Smith et al. (2009) up- dated the iconic ‘burning embers’ illustration of assessed dangerous interference from the TAR, concluding that comparatively smaller increases in global temperature now result in ‘significant or substantial consequences’ in all five of the assessment frame- work’s ‘reasons for concern’. A further example of these seemingly constant revisions toward impacts occurring both faster and with greater severity is given by Robinson et al. (2012). eir recent fully coupled model simulation of Greenland ice sheet stability finds previous assessments systematically understate its sensitivity to climate change:
10 Global mean surface air temperature has risen by ≈0.8 ℃ Hansen et al. (2010); AR4 reported the increase from 1850–1899 to 2001–2005 is 0.76 ℃ (IPCC 2007b). 11 And even then with probabilities around 66%.
12 1.2 On avoiding dangerous interference an ice-free state may be initiated at global mean temperature rise of 1.6 ℃.12
Indeed, climate changes so far manifest with a rise of ‘only’ 0.8 ℃ are already damaging human and natural systems. Human health and mortality is increasingly being affected by anthropogenic climate change, especially by extreme weather and disruption to wa- ter supplies (Richardson et al. 2009). A wide range of natural ecosystems are negatively affected, and by the rate of climate changes as well as their magnitude. Ocean acidi- fication is a major threat, now progressing at a rate beyond any comparable extinction period in Earth’s history (Richardson et al. 2009). e intensity and frequency13 of ex- treme weather events is growing (Richardson et al. 2009; Smith et al. 2009), and anthropo- genic climate change has almost certainly substantially enhanced their risk (Schiermeier 2011; Garnaut 2011). Assessing the past decade of extreme events, Coumou and Rahm- storf (2012, 5) conclude that ‘anthropogenic, unprecedented heat and rainfall extremes are here — and are causing intense human suffering’. e extreme and unprecedented weather conditions in south east Australia on February 7, 2009, coincided with the ‘Black Saturday’ bush fires, the country’s worst natural disaster in a century (Karoly 2009).14 e scenario-based speculative investigation of Dyer (2010) into the consequences for international security and human welfare — and human warfare — throughout this cen- tury as climate change progresses makes for ominous, distressing reading.15
International negotiations under UNFCCC auspices to cra a successor to the modest 1997 Kyoto Protocol foundered badly in Copenhagen in 2009 and fared lile beer in Cancún in 2010, nor since. ere are few indications that growth in global GHG emis- sions will be arrested in the near future; an ultimate emissions peak before 2020 is far from assured. Negotiations also suffer from a growing disconnect with the underlying 12 is is the best estimate of Robinson et al. (2012), but with a 95 % CI range from 3.2 ℃ to as lile as 0.8 ℃. 13 ough not necessarily both in all instances. 14 On that day, parts of the region recorded fire danger indications of 190 or more, on a scale intended to have a maximum of 100, calibrated for the previous most deadly event of 1939. 15 It should be noted that Dyer’s material oen cites military planners and strategists, including the Pentagon.
13 Chapter 1 Introduction and overview science: talks continue to frame policy decisions in terms of distant emission reduction targets, despite the science pointing to cumulative emission budgets as the appropriate framework (Anderson and Bows 2011). Assessment by the UN Environment Program showed that the sum of international abatement commitments under the Copenhagen Accord have effectively zero probability of holding to the 2 ℃ limit (den Elzen et al. 2010). Updating that work, den Elzen et al. (2012) find that for a ‘likely chance’ of restraining temperature rise to 2 ℃, the total abatement pledges confirmed at Cancún leave an emis- sion gap in 2020 of between 7 Gt CO2 to 11 Gt CO2 above the approximately 46 Gt CO2 maximum required — 2.5 Gt CO2 greater than their earlier estimate. In the words of New et al. (2011, 6), a 2 ℃ target is now ‘extremely difficult, arguably impossible’. In fact, re- cent modelling of the A1FI scenario finds a likely temperature rise of 4 ℃ by the 2070s, and possibly by the early 2060s if carbon-cycle feedbacks turn out to be at the higher end of uncertainty (Bes et al. 2011). e geophysical realities of a world 4 ℃ warmer than the preindustrial16 would be truly catastrophic (Garnaut 2011).
GHG emissions continue without relent. International mitigation negotiations have to date produced lile of consequence. Further warming is in the pipeline due to the current planetary energy imbalance, regardless of any emission reductions. Damaging climate change impacts are already apparent, and worsening. And climate science advances rapidly, new understanding consistently raising the level of threat and shortening the time remaining to forestall dangerous anthropogenic interference. It is in this context that the aerosol mask must be considered.
1.3 Research questions
e principal objective of this research is to answer the overarching question, ‘Do an- thropogenic aerosols represent a threat to climate change mitigation efforts, and if so
16 Which involves significant regional temperature variations beyond that mark.
14 1.3 Research questions what is an appropriate policy response?’ I hold that the missing account of anthropo- genic aerosols in mitigation policy needs to be addressed. It may be that their likely role under advancing mitigation efforts is sufficiently benign that recognition alone is satis- factory. However, if a full understanding of the role of anthropogenic aerosols necessit- ates material changes to the policy response, those changes must be clearly identified. A further set of research questions therefore commend themselves:
1. What are atmospheric aerosols, what are their sources, and what is their role in anthropogenic climate change, both qualitatively and, to the extent it is possible to determine with current knowledge, quantitatively?
2. How might aerosol climatic effects be measured in a policy-relevant manner, ana- logous to global warming potential and carbon dioxide equivalence?
3. To what degree do existing major climate change mitigation policy instruments account for aerosol effects?
4. Is there a distinct need for climate change mitigation policy instruments to separ- ately and directly account for aerosol effects?
a) Given that model-derived projections of unmitigated future climate change have commonly been driven by SRES scenarios, and that these scenarios in- clude broadly declining aerosol emissions, is there a need for aerosol emis- sions to be explicitly considered by policy instruments?
b) Do explicit mitigation scenarios, such as those based on cumulative emission budgets, similarly include declining aerosol emissions, and to what extent?
c) Do such scenarios adequately reflect the current state of both climate sci- ence and emission trends, especially the pace and impact of climate changes already underway?
5. What design criteria should aerosol-aware mitigation policy instruments meet to be effective?
15 Chapter 1 Introduction and overview
a) Is it possible to amend existing policy instruments to address these criteria?
b) Are there particular circumstances or consequences pertaining to aerosol effects that must be identified and managed within such mitigation instru- ments?
6. What novel mitigation policy instruments or policy frameworks, plausibly meet- ing these criteria, can be identified?
a) What are the pros, cons, and tradeoffs of multiple approaches?
1.4 Methodology
1.4.1 Research approach
My research approach consists of three components:
1. Review and synthesis of existing literature defining the physical science of aero- sols’ role in anthropogenic climate change. is synthesis aims to establish the geophysical basis of the subsequent implications for mitigation policy, including the relation between aerosol cooling and dangerous anthropogenic interference.
2. Critical analysis of contemporary mitigation policy approaches in light of these implications, seeking to assess their capacity to incorporate anthropogenic aerosol effects, and any complications that may arise.
3. Development of novel insights into the necessary form of a valid policy response to aerosol cooling. is component is inherently based on my underlying initial hypothesis that existing policy frameworks do not have capacity adequate to the task.
e first two components of this thesis are therefore predominantly qualitative in nature; however, importantly, the particular climatic properties and impacts of anthropogenic
16 1.4 Methodology aerosols are expressed and considered quantitatively where appropriate and to the extent possible. e third component is informed by policy analysis frameworks such as that articulated by Oikonomou and Jepma (2008), and employs relatively simple microeco- nomic constructs in developing the prototype policy response model given in Chapter 6.
1.4.2 Thesis structure
is Introduction defines the broad scope of the aerosol dilemma, siting it within the cur- rent geophysical and sociopolitical realities of anthropogenic climate change and its im- pacts. Chapter 2 establishes the physical science basis of aerosols’ role in climate, identi- fying and explaining their mechanisms of action; the current limitations and persistent uncertainties in forward modelling quantitative estimates are highlighted. Chapter 3 ex- amines alternative quantitative approaches based on planetary energy imbalance made possible by recent advances in measurement of ocean heat content, laying out the im- plications for dangerous anthropogenic interference. Chapter 3 also investigates altern- ative metrics that are more suited to measuring aerosol effects, and reviews current and projected aerosol emission data.
From this physical science basis, Chapter 4 presents a critical analysis of mitigation policy and the challenges posed by a full recognition of aerosol cooling. Chapter 5 con- fronts these challenges and establishes a revised and expanded aerosol-integrated mit- igation policy objective, specifying a set of design criteria that any implementing policy instruments must satisfy. Further investigation of possible implementation models in the second half of Chapter 5 generates additional principles that help frame a success- ful, workable policy solution. Chapter 6 then builds on these principles to develop a prototype model, analysing its implementation in detail.
Chapter 7 concludes the thesis, re-visiting the research questions given in section 1.3 and discussing potential further research, as well as the limitations of this work.
17 Chapter 1 Introduction and overview
1.4.3 Nomenclature
Relevant technical terms found in the literature are typeset in sans serif font of slightly larger size than the surrounding text when first used. New custom terms introduced in the thesis are similarly typeset and additionally italicised. A Glossary of acronyms and key terms is given at the end of the document, following the References list.
18 2 Aerosol effects on climate
Aerosols form an ‘integral part’ of the atmosphere’s radiation budget and hydrological cycle (Denman et al. 2007). eir effects on the climate system are wonderfully com- plex; the role of greenhouse gases in anthropogenic climate change are almost simple by comparison. Indeed, Andreae et al. (2005) remark that the ‘tremendous growth’ in scientific understanding of aerosol effects and parallel emergence of likely positive car- bon cycle feedbacks have ‘transformed the orderly picture of climate change of the early 1990s, dominated by GHG warming, into a complex mix of opposing effects’ (p. 1187). is complexity manifests in the diversity of aerosol species and emission sources; their atmospheric interactions and removal processes; their breadth of optical and radiative properties, sometimes conflicting and opposing one from the other; and the number of distinct climatic effects that have been identified.
is chapter summarises current scientific understanding of aerosols based on a review of the literature. Aerosol physical and optical properties are first examined in a gen- eral sense, leading into a detailed examination of radiative forcing, the central measure of climatic influence. Individual aerosol species are then catalogued, giving an over- view of their physical influences and relative importance. e array of radiative and mi- crophysical mechanisms generating these effects is then explained, including available quantitative assessments. e chapter concludes with discussion of the observational and modelling challenges posed by aerosols, and their consequent role as key sources of uncertainty in understanding climate response. A sequential order of these topics is
19 Chapter 2 Aerosol effects on climate necessary, but of course some aspects rely on knowledge of others, so cross-referencing is used to make these links clear.
2.1 Physical characteristics
Aerosols are solid or liquid –phase particles with an effective diameter of about 0.05 to 10 µm, though some sea salt or mineral dust particles may near 20 µm (CCSP 2009; Penner 2000). Most aerosol particles influence climate during their short lifetimes sus- pended alo in the atmospheric column, but some types such as black carbon have signi- ficant effects when seled on surface snow and ice. Unless otherwise stated, this thesis is concerned with anthropogenic aerosol emissions to the troposphere; periodic natural volcanic emissions to the stratosphere have significant effects, but anthropogenic inputs here are relatively minor (emied by aircra).
2.1.1 Formation and composition
A fundamental distinction in aerosol species relates to their composition and form of emission. Primary aerosols are those emied at source in particulate form directly. Whereas secondary aerosols are emied at source as precursor gases, then undergo secondary chemical and photochemical reactions in the atmosphere to form particulates — sulphates formed from gaseous sulphur dioxide are a prominent example of this. In- terestingly, processing of gaseous emissions within clouds (section 2.5.2 and 2.5.3) is an important pathway for aerosol gas to particle conversion.
2.1.2 Mixing state
Aerosol composition is also distinguished by the chemical particle mixing state. In- ternally mixed aerosols occur when various species undergo further chemical reac-
20 2.1 Physical characteristics tions to form new particles as some combination of their constituents. For example, precursor gases may condense around organic carbon particulates, creating an external layer which then affects the optical and radiative properties of the resultant aerosol.
Externally mixed aerosols occur when various species co-exist within the same loca- tion but remain unadulterated in their original form. at is, the distinct separate aerosol species do not undergo further chemical reactions.
2.1.3 Atmospheric residence time and emissions
e atmospheric residence time of aerosols — the time period within which they remain climatically active within the atmosphere — is dramatically different to that of most GHGs. At this point it is important to distinguish the subset of GHG species that are characterised as well-mixed and long-lived (LLGHGs), such as CO2, CH4,N2O, and the halocarbons. While these LLGHGs are most commonly the focus of discussion, other GHG species are also a concern. A prominent such short-lived species is tropospheric ozone (O3), a potent GHG but with a residence time of at most a few months and concen- trations localised near the site of atmospheric formation.1 Note that while water vapour is also a powerful GHG, vital to the natural greenhouse effect, it has a residence time of around 10 days and is a temperature-controlled feedback process rather than a forcing agent (Schmidt et al. 2010).
e atmospheric residence times of the LLGHGs are on the order of decades to centuries.2 In stark contrast, tropospheric aerosol lifetimes are measured in days and most persist less than a week before being removed by meteorological processes such as wet or dry deposition (eg, Textor et al. 2006; Schulz et al. 2006). Wet deposition is the primary
1 Ozone forms via photochemical reaction of precursor gases such as carbon monoxide, non-methane volatile organic compounds, and nitrogen oxides (Rypdal et al. 2005). 2 Around 20 % of any carbon dioxide slug emied to the atmosphere remains aer a thousand years, and indeed equilibrium climate response is still only 90 % at that time (Hansen et al. 2007a; Solomon et al. 2009).
21 Chapter 2 Aerosol effects on climate mechanism for removal, and so its efficiency depends on aerosols’ proximity to cloud (CCSP 2009). Aerosols resident within the stratosphere are ‘above the weather’ and so not subject to removal by wet deposition; they may survive a few years.3 is residence time asymmetry between aerosols and LLGHGs is a central complicating factor for effective climate mitigation policy, as examined in Chapter 4 onwards.
Partly as a consequence of their short residence times and partly due to their inherent optical and radiative characteristics, the physical location of emission is an important factor in aerosol effects on climate — they exhibit large spatial and temporal hetero- geneity. Aerosol forcing is significantly greater in areas of high emission concentration, reflecting their lack of mixing in the atmosphere compared to the LLGHGs. Most import- antly, atmospheric aerosol loading is fundamentally a function of emissions at any point in time (Andreae et al. 2005); in fact, historical cumulative emissions are of lile real con- sequence, so instantaneous emissions are generally the only useful metric (Andronova and Schlesinger 2004).4 Composites of model studies indicate that aerosol burden5 is nearly 50 % greater over land areas, as expected (CCSP 2009).
Regionality is then crucial for understanding and quantifying aerosol climatic impacts — global means may well obscure significant changes occurring in different regions (CCSP 2009). In contrast, LLGHGs are well-mixed throughout the atmosphere and global means are far more appropriate.
2.1.4 Atmospheric distribution
Aerosol mass loading, M, is a measure of the atmospheric burden, determined by the combination of emission profile, atmospheric transport, and removal processes (CCSP
3 Stratospheric aerosols of volcanic origin have an e-folding decay time of approximately 1 year (Hansen et al. 2007b). 4 ese physical characteristics are important for mitigation policy formation, as detailed in section 4.1. 5 Measured as aerosol optical depth, defined in section 2.2.5.
22 2.1 Physical characteristics
2009). It is simply the aerosol mass quantity per unit area, g m−2. Related terms include the aerosol particle number concentration, a count of particles per unit volume (Spracklen et al. 2010), or simply particle number, the count per unit length (eg, in a plume column). Particle number concentration, and the size distribution of particles for a given mass (see section 2.2.1), are an important factor for indirect effects on cloud through their relation to the cloud droplet number concentration, Nd (see section 2.5.2) (Spracklen et al. 2010; CCSP 2009).
e vertical profile of aerosol particle injection and dispersion in the atmospheric column can significantly affect resultant radiative properties. Altitude, and presence above, below, or within clouds are important observational parameters. AR4 notes that the vertical structure of aerosols and other forcing agents influences forcing efficacy (sec- tion 2.3.1), and other aspects of climate response such as regional and altitudinal tem- perature profiles and changes to the hydrological cycle (Forster et al. 2007).
2.1.5 Perturbation baseline
As forcing of climate (see section 2.3) is calculated with reference to some baseline con- figuration of the Earth-atmosphere system, the choice of that baseline is important for assessing the effects of aerosols on climate. In particular, is the baseline to be a total absence of aerosols, an absence of anthropogenic species only, or calculated relative to preindustrial loading? e preindustrial is the most commonly chosen baseline, com- mencing around 1750, though estimating preindustrial atmospheric aerosol loading is far from simple (see section 2.7.2). Comparisons to a hypothetical totally unperturbed atmosphere may also be useful in certain contexts.
23 Chapter 2 Aerosol effects on climate
2.2 Optical and radiative properties
e subsections that follow explain important aerosol optical and radiative properties. To a first order though, aerosols are characterised by their relative propensity for scat- tering or absorption of incoming solar shortwave radiation (SW), that portion of the solar spectrum occurring in the visible and near-visible wavelengths.6 e specific ef- fects are generally wavelength dependent, with the strongest aerosol interactions occur- ring with wavelengths similar to their diameter (Penner 2000). at is, involving Mie scattering,7 which is predominantly in the forward direction, especially when particle radius approaches light wavelength (Sturman and Tapper 2005). Reflection is a partic- ular form of scaering caused by particles or surfaces of larger size where the incident radiation is scaered backward.
Generally, scaering aerosols exert a negative forcing — a cooling influence. Absorbing aerosols are more complex, however — they may exert either positive or negative forcing, depending on local conditions such as the presence of cloud cover, or their location alo above bright reflective surfaces.
2.2.1 Particle effective radius
Aerosol particles will not have a uniform size in a given atmospheric loading. Particle size is represented by the effective radius of the lognormal aerosol radius distribution
(essentially a mean), denoted re (Tsigaridis et al. 2008; Hansen and Travis 1974). ´ r2n(r)rdr r = ´ (2.1) e r2n(r)dr
6 e terms sunlight, solar radiation, shortwave radiation, and insolation are used fairly interchange- ably; any differences will be clear by context. See the shortwave radiation Glossary definition. 7 Contrasted with Rayleigh scaering, where the particle, such as atmospheric gas molecules, is much smaller than the incident light wavelength (an order of magnitude) and scaering occurs in all direc- tions.
24 2.2 Optical and radiative properties
Where r is the monodisperse8 particle radius and n(r) is the particle radius distribution; integration is over the whole lognormal distribution range. Aerosol indirect effects on cloud are principally related to changes in aerosol re and particle number concentration.
2.2.2 Absorption, scaering, and extinction coefficients
ree related parameters characterise the fraction of radiant energy aenuated by ab- sorption and scaering; all are wavelength specific. Aenuation of incoming solar radi- ation in a direct beam to the surface by either absorption or scaering is referred to as extinction (Sturman and Tapper 2005). ough as discussed in section 2.5.11, scaered radiation may still reach the surface and affect climate.
e extinction coefficient is the total fraction of incident radiation extinguished by both absorption and scaering, with reference to path length through the atmosphere measured per unit distance (m−1, km−1, even Mm−1 ). e absorption and scattering coefficients are the fraction of incident radiation that is, respectively, only absorbed or scaered.
ese coefficients are also defined as specific quantities with reference to unit aerosol mass, expressed as unit area per mass (m2 g−1). e specific extinction coefficient kext is defined as (aer Haywood and Shine 1997; Tsigaridis et al. 2008).
3qext kext = (2.2) 4ρre
Where qext is the dimensionless specific extinction efficiency of the mixed aerosol (com- puted through Mie calculations across the lognormal particle size distribution), ρ is the −3 aerosol density (g m ), and re is the effective particle radius as above. e specific
8 Meaning, uniform.
25 Chapter 2 Aerosol effects on climate extinction coefficient is the sum of the specific scaering and absorption coefficients.
kext = ksca + kabs (2.3)
2.2.3 Asymmetry parameter
e asymmetry parameter, denoted g, indicates the angular character of aerosol particle scaering. It is the average cosine of the scaering angle weighted by the scaered light intensity, derived by integrating over the complete scaering phase func- tion (the angular distribution of scaered light) (Andrews et al. 2006; CCSP 2009). ˆ 1 π g = cos θP (θ) sin θdθ (2.4) 2 0
Where θ is the angle between incident light and scaering direction and P (θ) is the phase function. A value of –1 is entirely backscaering, 0 is isotropic (uniform scaering in all directions), and a value of +1 is entirely forward scaering. At a wavelength of 0.50 µm, dry aerosol particles have an indicative range for g between 0.64 and 0.83, and an average of 0.72 for all species; under conditions of high relative humidity, g increases to an average of 0.79 and a range 0.64 to 0.82 (Andrews et al. 2006).
e aerosol asymmetry parameter is used within climate radiative transfer models (ie, as a parameterisation) as computing the actual phase function is relatively inefficient and computationally expensive.
2.2.4 Single scaering albedo
Single scattering albedo (SSA; also denoted ωo) is a dimensionless property express- ing the fraction of incident radiation scaered by aerosols, defined as the ratio of scaer- ing to total radiation extinction (Tsigaridis et al. 2008). ‘Single’ refers to ‘small volume
26 2.2 Optical and radiative properties elements’ rather than the atmosphere as a whole (Hansen and Travis 1974).
ksca ωo = (2.5) kext
Where ksca is the wavelength-dependent mass-specific scaering coefficient. An SSA of 0 is purely absorbing, while purely scaering aerosols have a value of 1. Most natural aerosols have an SSA greater than or equal to about 0.75 at peak visible wavelength, 0.55 µm. Scaering aerosol species have SSA values around 0.9 or higher; strongly re- flective aerosol loadings have values above 0.95, and sulphates may be as high as 0.99 (CCSP 2009; Ramanathan et al. 2001).
2.2.5 Aerosol optical depth
Aerosol optical depth (AOD; also denoted τaer) is a key measure of aerosol optical properties. AOD indicates the optical ‘thickness’ of the atmosphere, integrated through- out the vertical column; AOD is hence also commonly referred to as aerosol optical thickness (eg, Yu et al. 2006). Optical depth is a wavelength-specific dimensionless value indicating the degree to which aerosols prevent the transmission of incoming sunlight; ie, the fraction of total light extinguished by scaering or absorption (Tsigaridis et al. 2008; CCSP 2009).
AOD = kextM (2.6)
Where M is the aerosol mass loading (section 2.1.4) and kext is the mass specific ex- tinction coefficient (section 2.2.2). An AOD value near 0 represents clear skies, while polluted or ‘thick’ skies have values approaching 1.15.
Importantly, light wavelengths are affected in different ways by the combined aerosol burden. Unless otherwise stated, AOD is assumed to refer to a wavelength of 0.55 µm,
27 Chapter 2 Aerosol effects on climate peak visible light. AR4 found good agreement between models and observations for an anthropogenic AOD fraction of approximately 21 % to 26 % (Forster et al. 2007); the later CCSP (2009) review similarly reported a fraction of 25 %, noting that this varies significantly over populated land masses.
Figure 2.1 illustrates the characteristic contemporary AOD for all anthropogenic aerosol species, as derived from model studies assessed in AR4 — the strong spatial heterogeneity and concentrations over land areas of high industrial (or biomass burning) activity are clearly evident.
Figure 2.1: Characteristic AOD for combined anthropogenic aerosol loading (Fig. 2.12 panel (a), Forster et al. 2007). Data are derived as the mean result for the nine AeroCom models assessed in original reference. Copyright the Intergovernmental Panel on Climate Change, 2007.
28 2.3 Radiative forcing
2.3 Radiative forcing
e most commonly employed means to quantify the climatic effects of an agent is ra- diative forcing. It is important, therefore, to clearly define this term and examine its particular methodologies before continuing further, basing that definition on the discus- sion of physical and optical properties laid out above.
A radiative forcing (RF) is a perturbation to the Earth’s radiative energy budget rel- ative to some base configuration of atmospheric conditions. at is, a change in the Earth-atmosphere system’s boundary conditions. Radiative forcing is a change in the net downwelling minus upwelling irradiance, of both shortwave plus longwave energy fluxes, measured in W m−2 averaged over the Earth’s surface area (Forster et al. 2007; Ramaswamy et al. 2001). Longwave radiation (LW) is that portion of the spectrum with wavelengths in the thermal infrared and larger; shortwave radiation was defined earlier (see Glossary). A positive radiative forcing indicates that additional energy is retained within the Earth-atmosphere system, either as increased incoming shortwave radiation or decreased outgoing longwave radiation (OLR) emied from the planetary surface and atmosphere. A negative radiative forcing is the opposite, where incoming short- wave radiation received at the surface is reduced or outgoing radiation exceeds energy absorbed.
e Earth-atmosphere system is in radiative equilibrium in the absence of such per- turbations: energy received and absorbed at the planetary surface as shortwave solar radiation — total incoming radiation minus the fraction reflected back to space by plan- etary albedo — is balanced by emied outgoing longwave radiation (Hansen et al. 2005a;
Pierrehumbert 2011). OLR is in turn a function of global mean surface temperature, TS, and the emissivity of the Earth, determined in large part by atmospheric composition. A radiative forcing is a disruption to this equilibrium. A positive RF causes a warming effect: additional energy absorbed into the land and sea surfaces and the atmosphere
29 Chapter 2 Aerosol effects on climate
9 causes TS to rise until OLR (which is initially suppressed ) is again sufficient to balance absorbed surface radiation. Conversely, a negative RF has a cooling effect due to the net reduction in energy.
Increased atmospheric greenhouse gas concentrations exert a positive RF through en- hanced atmospheric absorption of OLR, which thereby reduces radiation to space — emissivity of the Earth is reduced. TS will rise accordingly via a series of feedback pro- cesses until reaching a temperature at which OLR is again in balance, given the mod- ified emissivity. Conversely, increased atmospheric aerosol loading, through an array of mechanisms described in section 2.5, generally serves to enhance planetary albedo and hence exerts a negative RF by increasing shortwave reflection back to space. TS will fall in response, reducing OLR until it again equals the decreased shortwave radiation absorbed at the planetary surface.
Radiative forcing (RF) is linearly related to the change in global mean surface temper- ature (TS) by the equilibrium climate sensitivity parameter, denoted λ, as per eq. 2.7.
Climate sensitivity is hence defined as the resultant change in TS in kelvin per unit for- cing, giving units of kelvin per was per square metre (K(Wm−2)−1).
∆TS = λ RF (2.7)
To provide a more intuitively meaningful indication of Earth-atmosphere response, cli- mate sensitivity is also commonly expressed as equilibrium ∆TS in ℃ (or kelvin) caused 10 by a doubling of atmospheric carbon dioxide concentrations (2×CO2), which equates to an RF of approximately 4 W m−2. e value of climate sensitivity under current geo- logic conditions is a key area of uncertainty, in part due to the role of natural aerosols and dynamic cloud processes. A widely reported estimate is 0.75 ±0.25 K/W m−2, which equates to 3 ±1 ℃ for a doubling of CO2 (eg, Hansen et al. 2005a). AR4 cites a best es- 9 See detailed explanation in section 3.2. 10 Temperature responds to rising CO2 concentrations logarithmically.
30 2.3 Radiative forcing timate of 3 K, with an asymmetrical likely range of 2 K to 4.5 K (IPCC 2007b). In their review of the literature, Knui and Hegerl (2008) note that reducing uncertainty in cli- mate sensitivity is a serious challenge and that the range has not narrowed substantially since the first estimates of the late 1970s.
A perhaps surprising range of particular RF calculation methodologies exist, and defini- tions vary significantly. Major methodologies are illustrated in Figure 2.2. RF is defined with reference to related but distinct degrees of included temperature and atmospheric response, and to the location within the atmosphere at which the forcing is calculated — at the top of the atmosphere (TOA), at the tropopause,11 and at the planetary surface (see Hansen et al. 2005b, for detailed discussion). Given the importance of IPCC as- sessment reports to climate change policy, the methodology usually applied therein12 is worth making explicit: radiative forcing is the net change in irradiance at the tropopause aer allowing stratospheric temperature to adjust to equilibrium13 while holding sur- face and tropospheric state fixed at their initial, unperturbed values (Forster et al. 2007).
is is the stratospherically-adjusted RF,Fa in Figure 2.2. Stratospheric temperature adjustment occurs on the order of months, whereas adjustment of the tightly coupled ocean-troposphere requires several decades — tropospheric temperature change, calcu- lated at the tropopause, is hence the principal indicator of the ultimate climate response
(Hansen et al. 1997, 2005b) and Fa is employed for this reason.
However, Forster et al. (2007) recognised that Hansen et al. (2005b) and other studies argued that for some aerosol effects that are not initially radiative — because they in- volve feedback responses — the ‘zero-surface-temperature-change RF’ is a beer pre- dictor of the equilibrium surface temperature response than is Fa (pp. 134, 196). is is
11 e boundary of the troposphere and stratosphere. 12 e Fourth Assessment Report continued the same definition adopted in the ird (and earlier). 13 ough for most aerosols stratospheric temperature adjustment makes lile difference to the resultant RF (Forster et al. 2007, 134).
31 Chapter 2 Aerosol effects on climate
Figure 2.2: Illustration of RF calculation methodologies (Fig. 2, Hansen et al. 2005b). Reproduced with kind permission from John Wiley & Sons, Inc.
the fixed-ground-temperature methodology (Fg in Figure 2.2), which specifically permits the necessary changes in tropospheric state, including water vapour and temperature.
For absorbing aerosols in particular, Forster et al. (2007, 196) concede Fa could ‘fail’ as a useful predictor because it does not readily incorporate the semi-direct (section 2.5.4) and cloud lifetime effects (section 2.5.3). In fact, in their seminal study, Hansen et al.
(2005b) concluded that the fixed-sea-surface-temperature forcing (Fs in Figure 2.2) is the best predictor of ultimate climate change for all forcings.14 Nonetheless, Forster et al. (2007, 137) conclude that although the definition of RF employed by AR4 could have been modified to accommodate these processes, the stratospherically-adjusted RF (Fa) was retained for reasons of pragmatism and simplicity — relating to climate modelling response and their uncertainties, and cloud changes in particular — and these indirect effects are consequently not considered RFs.
A further crucial distinction concerns perturbations to the surface radiative energy budget. ese changes — a prominent feature of some aerosol effects (see section 2.5.4) — are termed surface forcing and should not be directly compared with radiative forcings as they ‘have quite different properties’ (Forster et al. 2007, 133). See section 2.5.11 for full discussion of surface forcing.
14 See also the Fg and Fs Glossary definitions.
32 2.3 Radiative forcing
2.3.1 Forcing efficacy
e concept of forcing efficacy, developed aer the TAR, allows comparison of the ultimate temperature response (equilibrium ∆TS) for a given forcing agent in relation to carbon dioxide, thereby addressing some of the ambiguity in RF definition. Efficacy, E, is defined as the ratio of the climate sensitivity for the specific forcing agent i to that of
CO2 (Hansen et al. 2005b; Forster et al. 2007).
λi Ei = /λCO2 (2.8)
Efficacy can in turn be used to define an effective RF for a given agent:
RF = Ei RFi (2.9)
Because efficacy is derived from climate sensitivity — which inherently encompasses all elements of the ultimate climate response — calculations of E for stratospherically- adjusted RF are able to incorporate cloud-aerosol interactions beyond that of the indir- ect cloud albedo effect, explained in section 2.5.2. Forcing efficacy depends primarily on the spatial structure of the forcing agent (refer to section 2.1.4) and its projection onto climatic feedback mechanisms (Forster et al. 2007). e geographical and vertical dis- tributions are of most significance: studies have found that high-latitude forcings have higher E values than those in the tropics.
33 Chapter 2 Aerosol effects on climate
Figure 2.3: Range of forcing efficacies reported in AR4 (Fig. 2.19, Forster et al. 2007). Copyright the Intergovernmental Panel on Climate Change, 2007.
Forcing efficacies in the literature were surveyed in AR4 and are summarised in Fig- ure 2.3, where each leer represents a specific study. e Hansen et al. (2005b) study is a prominent work in the field, represented by leer a (see Fig. 2.19 of Forster et al. (2007) for full details). e efficacies for direct and indirect aerosol effects are discussed in section 2.5.
2.4 Species
A diversity of individual aerosol species exists, with marked variation in optical and radiative properties, emied by a wide range of sources. Natural aerosols are initially outlined below, followed by a detailed examination of anthropogenic species, the par- ticular focus of this thesis.
34 2.4 Species
2.4.1 Natural aerosols
Naturally occurring aerosols are a fundamental component of the Earth-atmosphere sys- tem, acting as part of planetary albedo directly and, perhaps most importantly, as con- densation nuclei for cloud formation (see section 2.5.2). ough anthropogenic aerosol emissions have had significant impact on climate since the industrial revolution (and some such as black carbon and biomass burning, long before), they constitute around 10 % of the current overall atmospheric loading (CCSP 2009) — natural aerosol species remain the clear majority. e principal natural aerosol species are listed below (Den- man et al. 2007):
• Particulate sea salt is the most prevalent natural aerosol, ejected into the mar- ine troposphere by breaking ocean waves. It plays a key role in the formation of clouds.
• Windborne dust, ejected from land surfaces, especially deserts — the Sahara is an important source.
• Sulphates are formed naturally from volcanic emission of gaseous sulphur diox- ide and, in particular, biological emission of dimethylsulphide (DMS) by marine phytoplankton.
• Organic and inorganic carbon compounds are produced by natural wildfires and other biogenic sources, including secondary formation from volatile organic com- pounds.
Natural aerosols tend to have larger particle size than anthropogenic species and shorter residence times, and are hence comparatively less radiatively active (CCSP 2009).
2.4.2 Anthropogenic aerosols
e most important anthropogenic species are examined in the subsections to follow.
35 Chapter 2 Aerosol effects on climate
2.4.2.1 Sulphate
Sulphate is a secondary aerosol formed from precursor gaseous sulphur dioxide (SO2).
15 Sulphate aerosols consist of sulphuric acid (H2SO4) particles present as liquid droplets or in partly crystallised form, as well as various solid phase SO4 compounds. Cloud processing is a key activation pathway for sulphate aerosol through aqueous phase re- actions with cloud water droplets, oxidation of SO2 via gaseous phase reactions, and by growth via condensation around pre-existing particles (Denman et al. 2007; Forster et al. 2007, p. 160). is cloud processing causes the particle size distribution to shi toward larger sizes, which enables easier subsequent aerosol activation.
AR4 reported proportional sulphate emissions of precursor SO2 by source as approx- imately 72 % from fossil fuel combustion, 19 % DMS from phytoplankton, 7 % volcanic
SO2, and 2 % from biomass burning (Forster et al. 2007). Of fossil fuels, coal and oil are the predominant sources, accounting for around 55 % and 25 % of global SO2 respect- ively in 2002, while natural gas contributed less than 1 % (Ramanathan and Carmichael 2008).16 Strong regional variation exists, with emissions in the US and Europe decreas- ing — largely driven by air pollution controls, examined in Box 2.1 — while those in Asia have increased in concert with rapid industrialisation over recent decades. Total global anthropogenic SO2 emissions (measured by mass of sulphur, S) decreased from approximately 73 TgS yr−1 in 1980 to 54 TgS yr−1 in 2000; Northern Hemisphere emis- sions declining from 64 TgS yr−1 to 43 TgS yr−1 over that period were partly offset by emissions rising from 9 TgS yr−1 to 11 TgS yr−1 in the Southern Hemisphere (Stern 2005, cited by Forster et al. 2007).17
15 Either partly or totally neutralised by ammonia. 16 55 % + 25 % gives 80 % of total, compared to the 72 % reported in AR4. 17 See section 3.6 for additional detail.
36 2.4 Species
Box 2.1: e role of SO2 pollution regulations
Sulphur dioxide emissions have another effect on the atmosphere, quite apart from any change to radiative flux: industrial emission of SO2 and nitrogen oxides (NOx) are also the ultimate cause of acid rain, or, technically, acid deposition (Schmalensee et al. 1998). Atmospheric reactions of these gaseous precursors produce sulphuric and nitric acid, which eventually precipitate as acid rain (or through dry deposition), causing harm to natural systems and human selements alike. Of the two, sulphuric acid deposition dominated across Europe and North America (Likens et al. 1996).
Historical SO2 emissions increased roughly by a factor of four from 1880 to the 1920s, and then a further 3.5 from the 1920s to the 1970s, primarily in the Northern Hemisphere mid latitudes (Shindell and Faluvegi 2009); ie, Europe and North America. e negative forcing in Northern Hemisphere mid latitudes over the 1931–1975 period is consistent with large increases in sulph- ate aerosol stemming from these emissions. In fact, the strong increase in aerosol loading aer WW II is likely an important factor in the slight cooling from the late 1940s to 1970s (eg, Wild et al. 2007). e combined effects of aerosols and GHGs therefore ‘likely explain the observed temperature evolution over mainland Europe since 1950’ (Philipona et al. 2009, 4). Aerosol loading was then ‘drastically decreased’ from the mid 1980s as emission control reg- ulations were implemented across the industrialised world in an effort to deal with acid rain and other air pollution damages (Philipona et al. 2009, 1). e US Clean Air Act was promul- gated in 1970, leading to curbs on SO2 emissions to redress acid rain (eg, Likens et al. 1996). In Europe, the Convention on Long-Range Transboundary Air Pollution (CLRTAP), established in 1979, implemented a protocol targeting SO2 from 1985 (EEA 2004). Regulatory activity in the US culminated with amendments to the Clean Air Act that established the world’s first large- scale emissions trading scheme commencing in 1990, aimed at cuing SO2 emissions from the electricity generation sector, the major source of acid rain precursors (Schmalensee et al. 1998). Damage from acid rain is now substantially lessened in North America and western Europe through these clean air policies, though far from eliminated; nitrogen deposition in the United States remains a problem, as one example (Talhelm et al. 2011). An unintended consequence of simultaneous reductions in sulphate and other aerosol species emissions across the Northern Hemisphere, however, was a diminished cooling influence — at the same time as emissions of carbon dioxide and other GHGs continued to accelerate. e result of this reduced masking effect are global in scope but particularly felt in the Arctic region, where sulphate decline may account for much of the estimated 1.09 ±0.81 ℃ aerosol contribution to a 1.48 ±0.28 ℃ increase in surface temperature during 1976–2007 (around 74 %) (Shindell and Faluvegi 2009). Indeed, the OECD countries can be regarded as the ‘dominant contributor’ to recent temperature increases from the combination of historical GHG emissions and strongly curtailed SO2 emissions post-1980 (Andronova and Schlesinger 2004). e net result of a reduced Northern Hemisphere aerosol cooling and ever-increasing atmospheric LLGHG levels is the strong increase in global mean temperature from the 1980s onwards, as the combined forcing of anthropogenic GHGs surpassed all other forcing agents (Forster et al. 2007). While the parallel surge in SO2 emissions in recent decades from countries such as India and especially China, discussed in section 3.6, may have reduced the net fall in aerosol loading, this has come at great cost to lives and infrastructure, as is now so starkly evident.
37 Chapter 2 Aerosol effects on climate
As is typical for aerosols generally, the paern of sulphate radiative effects is directly influenced by the geographic location of SO2 emissions; ie, sulphate aerosols manifest these effects principally in the location where they are emied. is is in part a function of their residence time: sulphates simply do not persist in the atmosphere long enough to become widely geographically dispersed.
Sulphate aerosol strongly scaers radiation across the whole solar spectrum, having an effective SSA18 of 1, though some minor absorption occurs at near-infrared wavelengths (Forster et al. 2007). is propensity for scaering is a function of sulphate’s common particle size distribution (refer to section 2.2.1): sulphate accounts for a significant frac- tion of the overall sub-micrometre aerosol mass, and is the largest anthropogenic com- ponent of AOD. However, sulphate is also ‘invariably mixed’ with other aerosol species — both internally and externally (refer to section 2.1.2) — commonly forming a ‘compos- ite aerosol state’ (Forster et al. 2007, 161).
2.4.2.2 Organic carbon
Organic carbon (OC) describes aerosols consisting of a ‘complex mixture of chemical compounds containing carbon-carbon bonds’ (Forster et al. 2007, 161). ese compounds may derive from primary particulates or form by condensation of organic gases of low to semi volatility, such as terpene. Sources include fossil fuel combustion, industrial processes and fugitive emissions, biofuel burning, and natural biogenic emissions (Pen- ner 1994). OC particles continue to change aer emission due to atmospheric chemical processing, further modifying their hygroscopic,19 chemical, and optical properties. In fact, hundreds of such OC types exist, making modelling of their climate effects a diffi- cult challenge. A subset of OC is termed contained burning, approximately the sum of fossil fuel and biofuel emissions; ie, effectively excluding emissions from open burn-
18 Defined in section 2.2.4. 19 Affinity for water.
38 2.4 Species ing of biomass. Organic carbon is included with other organic element compounds (see section 2.4.2.7) and collectively referred to as particulate organic matter (POM) (eg, Dentener et al. 2006). As a whole, OC is the second largest contributor to total AOD aer sulphate (Forster et al. 2007).
Fossil fuel OC primarily scaers shortwave radiation, but does absorb more strongly in the ultraviolet and some parts of the visible spectrum (Forster et al. 2007). e specific absorption characteristics of fossil fuel OC aerosol are affected by source combustion temperature, with those formed at higher temperatures tending to be less absorbing (more scaering).
Fossil fuel OC is in effect always internally and externally mixed with other fossil fuel combustion products such as sulphate and black carbon (Forster et al. 2007). Coatings of OC on other aerosol species can alter the resultant particles’ radiative and hygro- scopic properties. e optical and hygroscopic properties of secondary OC differ with originating combustion process, ambient aerosol loads, and atmospheric chemical pro- cessing, and its representation in climate models has thus been ‘highly simplified’ with substantial uncertainty remaining (Forster et al. 2007).
AR4 quantified OC emissions as approximately 5 TgC yr−1 to 17 TgC yr−1 from biofuels versus 2.4 TgC yr−1 from fossil fuels (Forster et al. 2007).
2.4.2.3 Black carbon
Black carbon (BC), the absorbing component of soot, is a complex graphitic substance emied directly as primary aerosol from incomplete combustion of fossil fuel or biomass, much of anthropogenic origin (Forster et al. 2007; Conant et al. 2002; Ramanathan and Carmichael 2008). e complex chemical chain structures of BC tend to collapse as the particles age, which modifies their optical properties. Black carbon and organic carbon (section 2.4.2.2) are collectively termed carbonaceous aerosols (Penner et al. 2001);
39 Chapter 2 Aerosol effects on climate though Bond et al. (2004) caution that classification of carbonaceous aerosol depends on measurement method. e climatic influence of black carbon is perhaps the most complicated of the anthropogenic aerosol species — this section is somewhat complex as a result, and necessarily involves cross-referencing to a number of later sections for explanation of the radiative and physical mechanisms.
Unlike most other aerosol species which tend to scaer, black carbon strongly absorbs solar radiation. Because this absorption occurs alo, BC increases energy in the atmo- sphere (the semi-direct effect, section 2.5.4) while reducing it at the surface (decreased ir- radiance, section 2.5.11)(Ramanathan and Carmichael 2008). e localised atmospheric heating can in turn modify the thermal structure of the atmosphere and surface (Con- ant et al. 2002). e net positive forcing at TOA is the sum of this positive atmospheric heating and the smaller negative surface forcing (Bond et al. 2011). BC thereby affects atmospheric energy distribution, removing energy at the surface but absorbing a greater amount in the atmosphere. For these reasons, BC’s vertical profile in the atmospheric column is particularly important among anthropogenic aerosols.
e microphysical and meteorological properties of BC particles govern their horizontal and vertical distribution, dictating their residence time. When emied, BC particles tend to be largely hydrophobic, around 80 % (Conant et al. 2002; Cooke et al. 2002). As they age, these particles tend to ‘acquire a hydrophilic coating’, which eventually allows them to act as cloud condensation nuclei (CCN, detailed in section 2.5.2), forming cloud droplets on activation under supersaturated conditions (Conant et al. 2002). e cloud processing of sulphate aerosol also influences BC atmospheric residence time by enhan- cing the conversion of hydrophobic carbon to a hydrophilic state (Denman et al. 2007). e enhanced hydrophilic fraction then increases the rate of wet deposition through pre- cipitation (Cooke et al. 2002); black carbon would therefore be removed less rapidly if that conversion is dampened by a reduction in SO2 emissions.
Emissions of BC are subject to large uncertainties. AR4 reported one global fossil fuel
40 2.4 Species
BC estimate of 5.8 TgC yr−1 to 8.0 TgC yr−1; another puts the total at 8.0 TgC yr−1 but with an uncertainty factor of two, comprising 4.6 TgC yr−1 from fossil fuel sources and 3.3 TgC yr−1 from open biomass burning (Forster et al. 2007). Emissions of BC decreased in the second half of the twentieth century in the UK, US, Germany, and ex Soviet states, but increased in Asia — India and China account for some 25 % to 35 % of current global BC emissions (Ramanathan and Carmichael 2008).
antification of BC forcing is best established for the direct effects of absorption. Here it should be noted that many studies refer to the mechanisms described in section 2.5.4 as direct effects rather than semi-direct. at is, those studies treat the atmospheric heating influence of BC explained in section 2.5.4 under the single category of direct effects (section 2.5.1), as absorption occurs through direct interaction with solar radi- ation. AR4 reported a best estimate of +0.20 ±0.15 W m−2 for BC direct RF at TOA (see section 2.5.1). A later review study of BC’s effects on climate by Ramanathan and Carmi- chael (2008) concluded a markedly higher TOA BC direct forcing of +0.9 W m−2, with a −2 −2 range of +0.4 W m to +1.2 W m , placing it second only to CO2 in warming influence
−2 20 (as much as 55 % of CO2’s 2005-level RF of +1.66 W m ). e model study of Bond et al. (2011) found net anthropogenic TOA RF of +0.40 ±0.18 W m−2, which includes enhanced absorption due to internal mixing with other aerosol species.
As fossil fuel combustion is a major anthropogenic emission source for both BC and sulphate, which have in general opposing influences (warming versus cooling), the ratio of BC-to-sulphate mass concentration affects their combined interaction with shortwave radiation (Ramana et al. 2010). A higher ratio will tend to produce a more absorbing aerosol mixture than a lower ratio.21 Ramana et al. (2010) infer from their observational
20 Ramanathan and Carmichael include biomass burning emissions of BC (see section 2.4.2.5), stating these account for about 40 % of the total, so the comparison to AR4 is not entirely clear cut. 21 us the usually low-quality fuels used in commercial shipping (high sulphur content) tend to have low BC:sulphate ratios, making their aerosol emissions relatively less absorbing (Ramana et al. 2010, citing Bond et al. 2004). ese factors will be relevant if shipping is eventually brought under internationally- agreed mitigation frameworks.
41 Chapter 2 Aerosol effects on climate study that fossil fuel derived BC is more strongly absorbing than that of biofuel origin for the same BC:sulphate ratio, supporting the conclusion from model studies that fossil fuel BC is a more effective warming agent. e implication for mitigation policy, the authors note, is that the fossil fuel BC:sulphate ratio should itself be a target for reduction, along with total individual species emissions.
Although BC is regarded as a strong net positive forcer through shortwave radiation absorption and hence suggested as a target for mitigation action, studies by Chen et al. (2010) and Spracklen et al. (2011) caution that its role as CCN in cloud formation (outlined above) is not well understood, nor evaluated on a global scale (see section 2.5.2 and 2.5.3). In fact, the model-based study of Chen et al. (2010) found that abatement of BC aerosol actually led to a net positive increase in TOA RF due to reduction in global CCN concentration and subsequently cloud droplet number concentration; ie, cooling caused by BC effects on cloud cover was found to be greater than its warming effects. In their study based on observations of CCN combined with a global aerosol model, Spracklen et al. (2011) find that anthropogenic carbonaceous aerosol generates a negative forcing of –0.23 W m−2 through the first indirect effect on cloud (section 2.5.2).22 e authors note that while the anthropogenic contribution is about one third of the total by mass,23 their small size means that they account for two thirds of the total indirect effect. Spracklen et al. (2011) suggest that studies that propose reductions of BC relative to other aerosol compounds by virtue of their direct warming effect (section 2.5.1 and 2.5.4), such as that of Ramana et al. (2010) discussed above, do not consider the importance of particle size distribution for the first indirect effect. While these two assessments do not fully accord with the majority of BC studies, they have important implications for mitigation policy24 that should be extensively researched.
22 No uncertainty range is given for this mean estimate. 23 at is, sources such as wildfires account for ⅔ by mass. 24 Chen et al. (2010) also cite other studies that indicate that the specific type of emission control tech- nology used strongly affects the resultant emission particle number and size distribution — potentially both increasing or decreasing them.
42 2.4 Species
e recent major study of Bond et al. (2013) aims to provide a comprehensive and quant- itative assessment of BC’s influence.25 Bond et al. (2013) conclude that the net effect of
26 BC is a strong warming, second only to CO2 in the present-day atmosphere. ey find an updated best estimate of BC forcing in the current industrial era of +1.1 W m−2 with a 90 % uncertainty range of +0.17 W m−2 to +2.1 W m−2. is assessment captures all known forcing mechanisms, including indirect cloud effects and cryosphere melting (see p. 61 in section 2.5.4). is estimate is substantially larger than those given above, though in part the difference may be due to omission of indirect effects in those previous studies.
2.4.2.4 Nitrates
Nitrate aerosol consists predominantly of ammonium nitrate particles (NH4NO3), formed from nitrogen radicals when sulphate is fully neutralised and excess ammonia (NH3) is available (Forster et al. 2007; Feng and Penner 2007). Nitrate is therefore sensitive both to atmospheric concentrations of ammonia and to emissions of NOx precursors, the laer subject to photochemical reactions (oxidation of NOx to nitric acid, HNO3)(Skeie et al. 2011). As nitrate and ammonium aerosol are highly hygroscopic, they affect resultant aerosol radiative and optical properties by modifying the amount of water uptake for a given relative humidity (Feng and Penner 2007). e highest nitrate aerosol concentra- tions are found in highly industrialised locations; concentrations are correspondingly low in rural areas (Forster et al. 2007).
Nitrate aerosol scaers in the visible spectrum (Forster et al. 2007). Feng and Penner (2007) note that model and observational studies indicate nitrate may exert a similar
25 However, as Bond et al. (2013) first became available as this thesis was nearing completion, only a limited examination of its principal conclusions is possible here, in part due to its substantial length (online pre-publication is 232 pages). 26 It should be noted that the studies of Chen et al. (2010) and Spracklen et al. (2011) were considered in this effort.
43 Chapter 2 Aerosol effects on climate forcing to that of sulphate on a regional basis. However, a decrease in sulphur emissions relative to NOx would increase the nitrate aerosol burden and consequent forcing — a declining sulphate forcing could be partially offset by a simultaneous increase in nitrates. Additionally, via a similar influence on cloud processes to that described for BC aerosol in section 2.4.2.3, sulphates also suppress nitrate aerosol residence time (Denman et al. 2007), further enhancing the nitrate forcing if sulphates decline.
AR4 noted that estimates of nitrate direct radiative effect (section 2.5.1) are subject to large uncertainty due to a lack of observational studies and significant uncertainty in assessments of nitrate optical depth (Forster et al. 2007). eir efficient scaering char- acteristic means that nitrates exhibit a similar negative forcing at the TOA and surface.
2.4.2.5 Biomass burning
Biomass burning aerosol (BB) is a combination of scaering organic carbon and inor- ganic compounds (such as nitrates and sulphates), and absorbing black carbon (Forster et al. 2007). ese aerosol species are usually grouped together and assigned a single RF as they result from the effectively uncontrolled combustion of biomass material. Much of the BB aerosol burden is natural in origin,27 with no significant change evident over the industrial era. Importantly though, climate change may affect BB emissions through modifying the intensity and frequency of wildfires.
e single scaering albedo (see section 2.2.4) of BB aerosol is in part dependent on the particular source biomass,28 and tends to increase as the particles age (on the order of hours), rising from 0.85 to 0.9 at 0.55 µm. is increase in SSA is aributed to con- densation of non-absorbing organic gases onto existing particles, thereby diminishing the magnitude of BC absorption within the mixture (Forster et al. 2007). e partially
27 Exemplifying the complexities and uncertainties of aerosol observation, CCSP (2009) note that satellite-based aerosol measurements may assume that all BB aerosol is actually anthropogenic, and all dust natural. 28 Boreal forest fires have a high SSA from the outset, greater than 0.9 at 0.55 µm.
44 2.4 Species absorbing nature of BB means that forcing at the surface and within the atmospheric column is greater than that at TOA.
BB aerosol generates a negative RF under clear sky conditions, but positive if overlying cloud due to reduced cloud reflectivity through absorption (Forster et al. 2007). Overall, BB may exert a net positive RF, including clouded sky conditions. e vertical profile of BB particle distribution is then an important determinant of its effect. For example, Hsu et al. (2003) found that biomass burning aerosol in South East Asia — present as smoke — darkened existing stratiform, low-altitude cloud during boreal spring, causing a reduction in cloud reflectance as large as –100 W m−2. ermal emission from the same particles generated positive forcing as strong as +20 W m−2. e sign of BB aerosol forcing is therefore not certain, and depends on particular atmospheric conditions and emission composition.
2.4.2.6 Mineral dust
Anthropogenic emissions of mineral dust derive mostly from agricultural practises (harvesting, ploughing, overgrazing), surface water drainage, and industrial practises such as cement production and transport (Forster et al. 2007). Mineral dust aerosol’s climatic effects are difficult to measure — the sign and magnitude of RF depends on SSA optical characteristics for shortwave radiation, but depends on the vertical dispersion profile for longwave. AR4 noted that newer in situ measurements suggest mineral dust is ‘considerably less’ radiatively absorbing than previously thought, with a central estimate for SSA of 0.96 at wavelength 0.67 µm (Forster et al. 2007). Such an SSA suggests that a positive RF from mineral dust in the solar spectrum is unlikely; however, absorption by dust from diverse mineralogical sources is not represented in models. e estimated anthropogenic fraction of mineral dust emissions is a maximum of 20 %.
45 Chapter 2 Aerosol effects on climate
2.4.2.7 Secondary organic aerosol
Secondary organic aerosol (SOA) forms from atmospheric reactions of precursor gaseous volatile organic compounds (VOCs) such as terpene; SOAs are regarded as a type of POM, along with OC (section 2.4.2.2)(CCSP 2009; Dentener et al. 2006). SOAs are thought to form through photochemical oxidation of these precursors by one of three electrophilic atmospheric trace gases: the hydroxyl radical (OH), ozone (O3), and the nitrate radical (NO3)(Jacobson et al. 2000). Aircra measurements indicate that POM is highly correlated with secondary anthropogenic gas-phase species, suggesting that POM is being formed from these secondary sources on time scales of a day or more (CCSP 2009). In general SOAs are poorly understood, particularly their emission sources, and in fact, along with nitrates are not yet simulated effectively by current generation climate models (CCSP 2009).
Note that AR4 does not examine SOA in its discussion of anthropogenic aerosols’ cli- matic impact. In their high-resolution aerosol model driven by observational data, Myhre et al. (2009) calculated an SOA RF close to –0.1 W m−2.
2.5 Climatic effects and influence
e range of aerosol effects and their influence on climate are detailed below, illustrated in overview by Figure 2.4. Note that these effects are not necessarily discrete and distinct — specific aerosol species (and combinations thereo) behave in different ways, poten- tially even changing from one effect to another under differing temporal, spatial, and local atmospheric conditions.
46 2.5 Climatic effects and influence
Figure 2.4: Illustrations of aerosol influence on climate (Fig. 2.10, Forster et al. 2007). Copyright the Intergovernmental Panel on Climate Change, 2007.
2.5.1 Direct radiative effect
e most well-known and relatively well-quantified aerosol climatic influence is the dir- ect radiative effect (DRE), so named for aerosol particles’ direct interaction with in- coming solar shortwave radiation. ose aerosol species characterised by a sufficient single scaering albedo, such as sulphates, scaer a significant proportion of incident shortwave radiation, increasing optical depth. Some fraction of shortwave radiation is hence backscaered to space — ie, reflected — effectively increasing planetary albedo and so reducing shortwave energy flux received at the surface. Reduced surface irradiance is a negative radiative forcing, leading to a relative reduction in global mean surface tem- perature, TS. is direct radiative forcing, abbreviated as DRF, is broadly the same when calculated for the surface, tropopause, or at TOA; although under some definitions that include reduced surface irradiance due to aerosol absorption (eg, Yu et al. 2006), for- cing at the surface can be up to twice or more that at TOA (see section 2.5.4 and 2.5.11). Anthropogenic emissions of sulphate, fossil fuel OC, fossil fuel BC, biomass burning, and mineral dust all contribute significant DRFs (Forster et al. 2007).
47 Chapter 2 Aerosol effects on climate
DRF is determined by a key set of aerosol optical properties including SSA, specific ex- tinction coefficient, and asymmetry parameter (Forster et al. 2007). All of these vary with light wavelength and relative humidity (RH), and are subject to wide ranges in at- mospheric loading, geographical distribution, and horizontal and vertical profiles of any particular aerosol species. Remote sensing platforms are able to combine observations of AOD with changes to TOA radiative flux to determine a net TOA DRF (CCSP 2009). Of all aerosol effects, calculating forcing in this way is most applicable to the DRE (see discussion in section 2.3, 2.5.11, and 2.7). In situ observational field campaigns have found that shortwave scaering as a function of RH tends to decrease as the mass frac- tion of POM increases, likely because of POM’s effects on the relative hygroscopicity of scaering aerosol species (CCSP 2009).
Regional factors affect the DRF. Shindell and Faluvegi (2009) show that sensitivity to the aerosol DRE is relatively reduced at high-latitudes because these overlay already highly reflective surfaces (ie, snow and ice). ough when indirect effects are included (see section 2.5.2 and 2.5.3), net aerosol sensitivity at high-latitudes is actually some 50 % greater than at the tropics.
Continually improving in situ and remote sensing observations serve to beer constrain the total anthropogenic aerosol DRF. ese estimates of aerosol RF are in turn extens- ively validated by models based on the whole vertical atmospheric column using op- tical depth measurements. As it encompasses the entire aerosol burden, AOD ‘implicitly provides an alternative procedure to estimating the RF uncertainty’ and may be more robust than propagating the uncertainties identified for individual aerosol components (Forster et al. 2007, 168). Moreover, AOD necessarily includes nonlinear processes such as mixing dynamics that modify the net radiative properties of the combined aerosol bur- den. ese nonlinear effects are also beer captured by estimating aerosol RF based on the combined output of multiple models; the result is more robust than per-component or single model estimates.
48 2.5 Climatic effects and influence
e mean model total DRF in AR4 was –0.20 ±0.20 W m−2 (Forster et al. 2007). However this is a low-end estimate as most model simulations neglected nitrate and mineral dust aerosol; adding these species takes the estimate to –0.40 ±0.40 W m−2 at the 90 % confid- ence interval. By contrast, satellite observations from three studies placed the central estimate at –0.55 W m−2, indicating a systemic model underestimation of 20 % to 40 %. e influence of absorbing aerosols is a large source of uncertainty in DRF estimates. Black carbon in particular can lead to large uncertainty in TOA RF due to the import- ance of particulate vertical position with respect to clouds (see section 2.4.2.3 and 2.5.4).
e total combined anthropogenic aerosol direct radiative effect for all species, relative to the preindustrial atmospheric loading, is estimated by AR4 as –0.50 ±0.40 W m−2. e DRF for individual anthropogenic species is listed in Table 2.1, showing the best estimate and 90 % confidence intervals; these values are ploed in Figure 2.5. Note that black car- bon and biomass burning aerosol produce a positive RF due to their degree of shortwave absorption; their effects are explained in section 2.5.4.
Table 2.1: DRF (Fa) by anthropogenic species reported by AR4 (data from Forster et al. 2007) Aerosol species Best estimate 90% CI Sulphate –0.40 ±0.20 Organic carbon (fossil fuel) –0.05 ±0.05 Black carbon (fossil fuel) +0.20 ±0.15 † Black carbon (all sources) +0.34 ±0.25 Biomass burning +0.03 ±0.12 ‡ Nitrate –0.10 ±0.10 Mineral dust –0.10 ±0.20
† Forster et al. (2007) Tab. 2.13, note k: mean of model results from fossil fuel, biofuel, and biomass burning sources. ‡ With large uncertainty.
However, satellite observational studies surveyed by CCSP (2009) give an anthropo- genic fraction of clear-sky DRF over ocean as –1.1 ±0.37 W m−2, about twice that of models at –0.6 W m−2. Globally averaged anthropogenic DRF ranged from –0.9 W m−2 to –1.9 W m−2, again substantially higher than model estimates of –0.8 W m−2. Con-
49 Chapter 2 Aerosol effects on climate 0.5 0.0 ) 2 − m
W (
DRF -0.5
sulphate OC (ff) BC (ff) BC (all) BB nitrate mineral dust TOTAL
Figure 2.5: AR4 DRF at TOA by anthropogenic species (data as per Table 2.1)
versely, Myhre (2009) found consistency between a global aerosol model and an adjus- ted observational method — coupled with a significant revision of estimated preindus- trial aerosol composition (discussed in section 2.7.2) — concluding a global mean RF of –0.3 ±0.2 W m−2, substantially lower than AR4. Such ongoing discrepancies and large uncertainties in estimating even the nominally ‘simpler’ DRF underscore the serious difficulty in fully quantifying the anthropogenic forcing of climate from aerosol emis- sions.
AR4 concluded that scaering aerosol forcing efficacy (section 2.3.1) is very similar to that of a change in solar constant (ie, incoming solar shortwave radiation), though stronger at high latitudes. Efficacy ranges from 0.7 to 1.1 for the DRE, with assessed medium confidence (Forster et al. 2007).
50 2.5 Climatic effects and influence
2.5.2 Indirect cloud albedo effect
Aerosols produce two principal effects on cloud. Both are termed indirect as they modify cloud optical properties, rather than interact with solar radiation directly. Cloud optical properties, in particular reflectivity of solar radiation, are a function of light wavelength and depend on key parameters such as droplet size distribution or ice crystal concentra- tion, and on cloud-type morphology (Forster et al. 2007).
e cloud albedo effect, originally described by Twomey (1977), was the first of the two to be identified, and is hence also known as the first indirect effect. An increased aerosol loading increases the number of particles available to act as cloud condensa- tion nuclei (CCN) — the subset of total particle population with a radius of 0.05 µm and above facilitate condensation of water vapour to form cloud water droplets (Rosenfeld et al. 2008; Forster et al. 2007). For a given and fixed cloud liquid water content (LWC), an increase in available CCN results in a greater number of water droplets, each of smal- ler re (effective radius; see section 2.2.1) — the same volume of water is now spread across more droplets but of smaller size than would otherwise be the case. is change in cloud droplet number concentration, Nd, and size distribution strengthens cloud optical brightness: the cloud is now more reflective to incoming shortwave radiation and so increases albedo. As is the case for the DRE (section 2.5.1), increasing planetary albedo in this manner induces a negative RF, leading to cooling. e cloud albedo effect is illustrated in Figure 2.6.
e relation of aerosol to cloud droplet number concentration is nonlinear, depending on the hygroscopicity and other characteristics of the aerosol species.
b Nd ≈ (Na) (2.10)
Where Nd is the cloud droplet number density, Na is the aerosol number concentration, and b indicates the aerosol particle hygroscopicity (Forster et al. 2007). e parameter
51 Chapter 2 Aerosol effects on climate
Figure 2.6: Indirect cloud albedo effect (Fig. 7.20 (top panel), Denman et al. 2007). Copyright the Intergovernmental Panel on Climate Change, 2007. b ranges widely, from 0.06 to 0.48, illustrating the sensitivity of cloud droplet activation to aerosol properties such as size distribution in particular, and to updra velocity; it further reflects the use of aerosol extinction as a proxy for actual CCN observations and the difficulty in quantifying the effect via remote sensing (Forster et al. 2007).
Determining the detailed effect of aerosol loading on CCN and droplet size distribution is challenging. Studies surveyed in AR4 indicate that aerosol particle size is the principal determinant in their efficacy as CCN, rather than composition, though local atmospheric conditions are also important (Forster et al. 2007). e observational and aerosol model synthesis study of Spracklen et al. (2011) similarly emphasised the large sensitivity of the effect to particle size. However results varied as to the potential sensitivity of liquid water path (LWP)29 to aerosol loading in the vertical column, changes to which in turn affect cloud optical depth. AR4 concluded that significant difficulty exists in observa- tional studies separating cloud albedo effects from other influences on cloud LWP and hence resultant cloud RF (Forster et al. 2007).
Modelling of aerosol effects is similarly difficult, and most general circulation models (GCMs) use parameterisation schemes to represent the relation between cloud droplet number concentration and aerosol loading (Forster et al. 2007). Considerable differ-
29 e mass of water available for cloud formation by unit area.
52 2.5 Climatic effects and influence ences in these parameterisations and other aspects of aerosol-cloud interaction have also largely prevented model result intercomparison. Large uncertainties in underlying aerosol and precursor gas emission rates and distribution — particularly in the prein- dustrial era (see section 2.1.5) — are a further limiting factor. Nonetheless, all models reviewed in AR4 estimated negative mean global RF for the cloud albedo effect, though with wide variation ranging from –0.22 W m−2 to –1.85 W m−2, and that this effect tends to be substantially larger over land than over ocean (Forster et al. 2007). Importantly, the review highlights the continuing weaknesses and biases in model simulation of cloud processes, which would affect these results.
Treatment of cloud droplet size distribution and an assumed ‘invariant’ droplet spectral shape not found under atmospheric conditions are further uncertainties: Liu and Daum (2002) estimated that a 15 % increase in the size distribution width reduced the cloud albedo effect between 10 % and as much as 80 %. AR4 termed this the dispersion effect (Denman et al. 2007). A widened droplet size distribution means that there are greater numbers of droplets both of smaller and of larger sizes, as well as an overall increase in number concentration Nd. Because larger size droplets are less reflective, this widening of the distribution in turn offsets some of the cooling effect of the enhanced Nd. AR4 found that subsequent studies confirmed that by not taking changes in droplet spectra into account, models overestimate the cloud albedo effect by some 15 % to 35 % (Forster et al. 2007).
Satellite observations are uncertain, both because of the satellite retrievals themselves (and assumptions in those processes), and because retrieval of aerosol and cloud proper- ties are not coincident, instead assuming that AOD can be linked to aerosol concentration below the cloud layer. Satellites are also unable to unambiguously distinguish anthro- pogenic from natural aerosols. However, the global coverage of satellite instruments makes these a highly useful observational platform and work continues to improve their resolution and capabilities.
53 Chapter 2 Aerosol effects on climate
e total estimated cloud albedo effect30 reported in AR4 from all aerosols is a best estim- ate median value of –0.7 W m−2, with a 5 % to 95 % range of –1.8 W m−2 to –0.3 W m−2 (Forster et al. 2007). McComiskey and Feingold (2008, 1) note that this uncertainty range ‘has remained nearly constant through time’ across the second and third IPCC assess- ment reports, underscoring the difficulty in quantifying and modelling aerosols’ complex influence on cloud microphysics. CCSP (2009) conclude a discussion of cloud indirect effects and radiative forcing by noting that ‘observational estimates … are still in their infancy’ (p. 49).
Returning briefly to the possibility for black carbon to act as CCN mentioned in sec- tion 2.4.2.3, Chen et al. (2010) state that two steps are required for this to occur. First, BC must be internally mixed with other hydrophilic aerosol compounds, such as sulph- ates or other organics; and second, the resultant particle size must be large enough (via condensation growth) to activate as a CCN, approximately 100 nm in diameter. In their assessment, ‘[u]nder essentially all reasonable atmospheric conditions, the second step is the limiting one for a soot [ie, BC] particle to become a CCN’ (p. 2).
2.5.3 Indirect cloud lifetime effect
In 1989, Albrecht hypothesised a second indirect effect of an increased aerosol load- ing, the cloud lifetime effect (Albrecht 1989). As highlighted in section 2.3 on page 32, only the cloud albedo indirect effect is regarded by the IPCC in AR4 as ‘a purely radiat- ive forcing’ — other aerosol influences on cloud and precipitation, discussed here and in the following sections, necessarily involve feedback response processes (Denman et al. 2007, 559). ese cloud feedbacks remain one of the largest single sources of uncertainty in estimates of climate sensitivity.
e ‘distinguishing quality’ of the various hypotheses broadly referred to as the cloud
30 RF estimates are for liquid water clouds only, as knowledge of ice-phase clouds was too poor.
54 2.5 Climatic effects and influence lifetime effect is that aerosol loading affects cloud macrostructure by modifying the ef- ficiency of precipitation development (Stevens and Feingold 2009, 607). As in the first indirect effect, an increased aerosol loading in warm clouds tends to raise cloud droplet number concentration, Nd, while simultaneously decreasing individual droplet size, re, for a given LWC. e second indirect effect posits that cloud precipitation efficiency is a monotonically decreasing function of Nd, and hence increasing Nd will suppress precip- itation relative to the unadulterated state. Reduced precipitation thereby affects cloud macrostructure, extending its lifetime and so increasing net albedo over time (CCSP 2009; Denman et al. 2007; Stevens and Feingold 2009). e extension of albedo changes over time is what distinguishes the cloud lifetime effect from the cloud albedo effect: albedo is postulated to be a decreasing function of precipitation efficiency, though both indir- ect effects stem from aerosol-induced change in Nd (Stevens and Feingold 2009). Aver- aged globally over the long term, precipitation fundamentally must balance evaporation (Rosenfeld et al. 2008; Trenberth et al. 2009), which Rosenfeld et al. (2008) argue must in turn mean that suppressed precipitation in shallow cloud is compensated by enhanced precipitation in deeper cloud (including possible changes in cloud height).
Decades of research have struggled to determine the extent of any cloud lifetime effect; it remains controversial, in no small part due to a complex range of challenges relating to observations of aerosol optical depth and cloud amount, and distinguishing between the two (Stevens and Feingold 2009). Deriving observational methods to separate the cloud albedo effect from changes to cloud lifetime is therefore extremely difficult, and stud- ies have commonly evaluated the effects in combination (Denman et al. 2007). Indeed, the studies surveyed in AR4 point to, at that time at least, a distinct lack of agreement between indications from satellite observations and model results in the scope and role of aerosol indirect effects on cloud. One such study of warm convective cloud (Jiang et al. 2006) found that although precipitation was suppressed with enhanced aerosol concentration, cloud lifetime was not extended — this was likely due to the competing
55 Chapter 2 Aerosol effects on climate effect of enhanced evaporation of smaller, more numerous cloud droplets. Jiang et al. (2008) used near-simultaneous satellite observation of ice cloud and an aerosol proxy over South America to conclude that polluted ice cloud did cause a decline in ice particle radii and subsequent precipitation efficiency suppression, though weakly during the wet season. CCSP (2009) cite the Hansen et al. (2007b) model study finding that either aero- sol indirect effect will produce a similar reduction in tropical precipitation, as does the direct effect. From this, CCSP (2009) infers that the reduced precipitation is caused by decreased shortwave irradiance at the surface (see section 2.5.11) and the feedback rel- ative suppression of evapotranspiration that results; that is, not strictly due to indirect changes in cloud precipitation efficiency. Such results highlight the complex dependen- cies on cloud type, aerosol species, and atmospheric conditions for any resulting cloud feedback effects.
Indeed, the CCSP (2009) review, citing a major 2007 WMO report examining aerosol impact on precipitation, stated that ‘to date there is no statistically robust proof of sur- face rain suppression’ (p. 72), calling into doubt the reality of any substantial cloud lifetime effect. However, Rosenfeld et al. (2008) proposed a detailed conceptual model of cloud-aerosol precipitation effects, claiming to reconcile the WMO report’s principal finding with previous work that does indicate precipitation suppression.31 ey intro- duce the concept of aerosol thermodynamic forcing (TF), representing the ‘aerosol in- duced change in the atmospheric energy budget that is not radiative in nature’, which is the change in latent heat flux in the atmospheric column (Rosenfeld et al. 2008, 1312). e details of this TF-based model are beyond the scope of this thesis, but using it Rosen- feld et al. (2008) demonstrate reconciliation of the apparently dichotomous evidence for low CCN tropical cloud raining out too quickly to become long-lived, versus strongly polluted cloud evaporating most of their water content before precipitation can occur,
31 Rosenfeld et al. (2008) was not cited by CCSP (2009), but this likely simply reflects the inherent lag between publication date and the underlying research.
56 2.5 Climatic effects and influence if the cloud can form at all (see the semi-direct effect below also). eir model predicts an optimal aerosol loading regime — AOD around 0.25 — that in turn optimises cloud development, which has been borne out by observation in at least one case.
Stevens and Feingold (2009) reviewed the available literature and evidence for a cloud lifetime effect, including that of Rosenfeld et al. (2008), to arrive at more cautious conclu- sions. ey discuss the evidence for multiple likely cloud regimes (and regime-dependent aerosol effects), and the existence of buffering systems (internal feedbacks) at micro and macrophysical scales that serve to dampen the aggregate cloud response to aerosol per- turbation. Stevens and Feingold ultimately argue that knowledge of aerosol effects on cloud precipitation regimes is progressing, but more research is necessary.
AR4 concluded that forcing efficacy for the cloud albedo indirect effect is very likely similar to that of the direct effect. Including the cloud lifetime indirect effect indicates efficacy is greater than unity, but with high uncertainty. Overall AR4 assessed forcing efficacy of indirect aerosol effects as >1.0 (Forster et al. 2007).
2.5.4 Semi-direct effect
e semi-direct effect refers to a complex array of processes relating to absorbing aer- osols, the most prominent species being black carbon (section 2.4.2.3). Predominantly these processes involve aerosols suspended alo and a consequent localised heating of the atmospheric column, but some occur at ground level. Absorption of incoming short- wave radiation reduces energy received at the surface, exerting a negative surface for- cing (explained in section 2.5.11); a TOA forcing, either negative or positive, may be generated depending on the relative reflectivity of the surface or cloud below the aer- osol layer. Absorption of outgoing reflected shortwave radiation, conversely, necessar- ily reduces planetary albedo and generates a positive forcing at TOA (Ramanathan and Carmichael 2008). e localised atmospheric heating from absorption (at altitude) pro-
57 Chapter 2 Aerosol effects on climate duces a substantial positive atmospheric forcing within the troposphere, increasing air stability beneath relative to the surface and modifying the atmospheric temperat- ure gradient and hence the lapse rate32 (Denman et al. 2007). To further complicate maers, Forster et al. (2007) state that the semi-direct effect should not be considered an actual forcing due to atmospheric heating’s impact on the tropospheric hydrological cycle and other feedback mechanisms (discussed below); yet other authors treat at least some aspects as forcings proper. Absorbing aerosol atmospheric forcing is illustrated in Figure 2.7, and a schematic of semi-direct effects is shown in Figure 2.8.
Figure 2.7: Illustrative atmospheric forcing at altitude due to aerosol SW absorption (in W m−2 ±30 %), for the period 2001 to 2003 (Fig. 1 panel (b), Ramanathan and Car- michael 2008). Reprinted by permission from Macmillan Publishers Ltd: Nature Geoscience, copyright 2008.
If absorbing aerosols are located in a cloud-free sky (lemost in Figure 2.8), they likely produce cooling by inducing negative surface forcing (Forster et al. 2007). Further, that surface forcing may not produce a corresponding TOA forcing: radiative transfer at the top of the atmosphere (and indeed the tropopause) is unaffected because aerosol ab- sorption of shortwave radiation within the (clear-sky) tropospheric air column modifies neither outgoing shortwave nor longwave energy flux. at is, only at and below the
32 Rate of decrease in temperature with increase in height within the troposphere.
58 2.5 Climatic effects and influence
Figure 2.8: Semi-direct effect, cloud-free and within cloud (Fig. 7.20 (mid panel), Denman et al. 2007). Copyright the Intergovernmental Panel on Climate Change, 2007. point of absorption in the vertical profile is incoming shortwave radiation extinguished — the surface would have absorbed that radiation if airborne aerosols had not.
However if shortwave absorption occurs above highly reflective surfaces, the result may be either warming or cooling. Extinction of incoming shortwave radiation by absorption in the atmosphere may be less (warming) or more (cooling) than the surface reflection that would otherwise have occurred (Forster et al. 2007). At the same time, absorption of shortwave radiation that is reflected by the surface-atmosphere-cloud system — outgoing shortwave radiation originating beneath the aerosol layer — reduces planetary albedo, because that radiation cannot now exit to space (Ramanathan and Carmichael 2008). In both of these contexts, a TOA RF as well as surface forcing would occur; ie, outgoing energy fluxes are now modified.
Absorbing aerosols — black carbon in particular — also affect cloud fraction33 and cloud formation directly, though understanding of these processes remains poor (eg, Koch and Del Genio 2010).
33 at portion of the sky covered by cloud.
59 Chapter 2 Aerosol effects on climate
e effect of absorbing aerosols alo above cloud depends in part on cloud type (Koch and Del Genio 2010). Stratocumulus clouds34 may be enhanced through the strengthen- ing of atmospheric stability in the air above them, by reducing downwelling shortwave radiation, and by other effects on atmospheric processes. Conversely, studies indicate suppression of cumulus cloud cover through the same stabilising effects, and also from a reduction in moisture availability due to decreased evaporation rates caused by the re- duced surface shortwave irradiance (somewhat counteracting the enhancing effect for stratocumulus). ese perturbations to evaporation also have important consequences for the hydrological cycle (see Box 2.2).
Box 2.2: Aerosol effects on the hydrological cycle
ough this thesis predominantly addresses aerosol effects on climate through perturbations to the energy budget and albedo, their influence on the hydrological cycle is of no less a concern. By affecting cloud type and extent, surface energy balance and, consequently, evaporation rates, aerosols can in turn affect precipitation amount and location (eg, CCSP 2009). ose impacts are important, as significant modifications to the hydrological cycle will have serious implica- tions for local ecosystems and species, food production, water supplies, and the frequency and intensity of extreme weather events. Indeed AR4 notes that as anthropogenic aerosols may have a suppressing effect on precipitation, their removal would then have benefit, despite any consideration of radiative influence (Denman et al. 2007). In a worrying example, Koch and Del Genio (2010) note that several BC studies indicate the po- tential for large scale effects on Asian monsoons (though uncertainty remains as to the detail). Ramanathan and Carmichael (2008) highlight that the negative precipitation trend observed in the last 50 years over many tropical regions (Africa, South Asia, and northern China in particular) cannot be explained by GHG warming alone, and that along with natural variability anthropo- genic aerosols are ‘emerging as major players in the observed trends’ (p. 224). Aerosols further affect water vapour transport from the Indian Ocean to South Asia land areas (Ramanathan and Carmichael 2008). e ability of aerosols to affect climate in ways that — by definition — are not reflected in global mean changes in surface temperature or radiative forcing underscores that anthropogenic cli- mate change fundamentally encompasses far more than ‘global warming’.
Where BC is present within clouds (rightmost in Figure 2.8), its absorption of shortwave radiation may be sufficient to raise the temperature of droplets containing BC CCN such
34 Which are not yet simulated well by models.
60 2.5 Climatic effects and influence that they evaporate, ‘burning o’ the cloud and decreasing LWP (Conant et al. 2002; Denman et al. 2007). is evaporation decreases reflection of incoming shortwave radi- ation by reducing cloud fraction, exerting a positive forcing at the surface and at TOA. Ramanathan and Carmichael (2008) point out this cloud feedback increases the already significant BC forcing, though non-absorbing aerosol-induced increases in cloud cover dominated overall (except in heavily polluted areas).
When suspended below the cloud layer, absorbing aerosols have been found to enhance or even induce cloud fraction, and hence shortwave reflectivity over time, under some circumstances (Koch and Del Genio 2010). Enhancement of convective vertical air move- ments below the cloud from localised heating is the likely cause, as well as an increase in LWP. Changes to surface energy fluxes are an important factor in these processes, making physical surface characteristics a critical determinant.
When black carbon becomes mixed within or deposited on snow and ice it modifies the optical properties of these surfaces, reducing their reflectivity and increasing melt rates, hence perturbing albedo causing a warming feedback effect (Flanner et al. 2007; Ramanathan and Carmichael 2008). In fact, as ‘a straight-forward consequence of pos- itive albedo feedbacks and atmospheric stability at high latitudes’, the forcing of BC on snow and ice is nearly twice as effective as that of CO2 (Hansen and Nazarenko 2004); in- deed, BC deposition is implicated in Arctic sea ice retreat (Ramanathan and Carmichael 2008).
Incorporation of BC soot within snowflakes, in combination with atmospheric heating, is implicated as the cause for a substantial fraction of the accelerating melt rates of Tibetan Plateau glaciers in recent decades (Xu et al. 2009). Advection of BC-heated air from other parts of Asia over the Himalayas increases lower and mid tropospheric temperatures by about 0.6 ℃ averaged annually (Ramanathan and Carmichael 2008). And in yet another complex feedback mechanism, more severe forest fires produced by global warming- induced drought can increase levels of BC deposition on sea ice, further enhancing its
61 Chapter 2 Aerosol effects on climate decline (Ramanathan and Carmichael 2008). AR4 adopted a best estimate RF for BC on snow of +0.10 ±0.10 W m−2 (Forster et al. 2007).
AR4 did not determine a best estimate forcing efficacy (refer to section 2.3.1) for absorb- ing aerosols as a whole, though Arctic albedo sensitivity to black carbon snow deposition enhances its efficacy to 1.7 (Forster et al. 2007; Hansen et al. 2005b). For aerosols within a certain SSA range, a linear forcing-response relationship breaks down: despite a neg- ative RF the net result is a rise in mean global temperature increase, hence E is actually negative. Studies of increased BC atmospheric burden found large E (>1) when resid- ent within the boundary layer, but much smaller E (<1) when above it. is is at least partly due to semi-direct effects modifying the tropospheric temperature gradient and cloud fraction. AR4 concluded that available studies are of insufficient agreement to de- termine a consensus BC efficacy and, again, that this ‘may represent problems with the stratospherically adjusted definition of RF’ (Forster et al. 2007, 199).
In summary, the semi-direct effect encompasses a complex set of atmospheric processes induced by absorbing aerosols, with potentially conflicting radiative and surface for- cing implications. In all cases where absorbing aerosols are alo, a localised heating and resultant positive atmospheric forcing occurs. Nil, positive, or negative TOA RF, and positive or negative surface forcings — with varying combinations of RF and sur- face forcing sign — can all occur, highly contingent on aerosol vertical profile, surface geography and reflectivity, and the presence and type of cloud.
2.5.5 Thermodynamic effect
e indirect effects on cloud from an increased aerosol burden lead to smaller cloud droplet size (section 2.5.2). Where this occurs within super-cooled, deep-convective water clouds, the thermodynamic effect means that the freezing of these smaller droplets is delayed (Denman et al. 2007; Lohmann and Feichter 2005). e resulting
62 2.5 Climatic effects and influence suppression of precipitation causes the cloud top to extend to higher, colder altitudes; increases in lightning and the risk of severe weather may result (CCSP 2009). e ther- modynamic effect (and glaciation effect, described below) is illustrated in Figure 2.9.
Figure 2.9: Aerosol effects on mixed-phase clouds (Fig. 7.20 (boom panel), Denman et al. 2007). Copyright the Intergovernmental Panel on Climate Change, 2007.
e forcing due to the thermodynamic effect is unclear. Either a positive or negative forcing may be produced, depending on cloud optical properties and the specific geo- graphy of land areas now covered by cloud. In their 2005 review of aerosol indirect effects, Lohmann and Feichter found that no forcing estimate could be given; the level of understanding does not appear to have changed substantially since.
2.5.6 Glaciation effect
Increased loading of hydrophilic aerosol in the presence of mixed-phase clouds may produce a glaciation effect in temperatures low enough for those particles to act as ice nuclei (IN), analogous to that of CCN in water clouds (section 2.5.2). By acting as IN, certain aerosol particles such as BC and dust may induce condensation of water in super-cooled stratiform clouds,35 forming new ice crystals within the cloud (Denman et al. 2007).
35 Due to the difference in water vapour pressures over ice and water (CCSP 2009).
63 Chapter 2 Aerosol effects on climate
In distinct contrast to water clouds, however, increased ice crystal concentration within ice-phase clouds actually enhances precipitation efficiency: faster growing ice crystals — glaciation — aain precipitation-size sooner, thereby reducing mean cloud cover. is glaciation effect essentially competes against the cloud lifetime effect in warm clouds. e larger of the two will depend on the chemical characteristics of the particles in- volved, and the behaviour of continental versus maritime clouds are distinctly different (Denman et al. 2007). e glaciation effect is illustrated along with the thermodynamic effect in Figure 2.9.
AR4 concludes that the glaciation effect consequently reduces net cloud albedo, generat- ing a positive forcing at both TOA and the surface (Denman et al. 2007). However, CCSP (2009) state in their discussion of IN effects that while some observations of aerosol in- fluences exist, a clear link between aerosol loading, IN concentration, and ice crystal growth inducing precipitation ‘has not been established’ (p. 46).
2.5.7 Biomass fertilisation effect
Perhaps the most surprising aerosol effect is that of biomass fertilisation. High SSA aerosols scaer sunlight in all directions, not only as backscaer to space (reflection), thereby increasing the relative diffusion of sunlight reaching the surface. at is, a lar- ger AOD due to an increased scaering aerosol burden in otherwise clear skies results in an increased fraction of diffuse radiation at the surface (diffusion is further discussed in section 2.5.11). In general, plant photosynthesis increases with available shortwave irradiance, and its efficiency is enhanced under conditions of diffuse sunlight (Mercado et al. 2009). In recent years it has been realised that the role of aerosols would then be to increase net plant growth via shortwave scaering. For example, Niyogi et al. (2004) conducted field experiments comparing biomass carbon flux and the relative diffusion of solar irradiance that support this. Enhanced plant growth means that biosequestra-
64 2.5 Climatic effects and influence tion of atmospheric carbon dioxide is increased, strengthening the biomass carbon sink (Denman et al. 2007). e net result from an increased scaering-aerosol burden is a reduction in atmospheric CO2 concentrations via biomass fertilisation. As a negative feedback mechanism,36 biomass fertilisation removes the positive radiative forcing that the drawn-down CO2 would otherwise have exerted.
Modelling by Mercado et al. (2009) found that aerosol diffusion of sunlight caused, inter alia, by anthropogenic sulphate aerosol, has increased the biomass carbon sink by up to 25 % over the period 1960 to 1999. e authors note that removal of the sulphate aerosol burden would see this fertilisation effect decline rapidly, reaching near zero by the end of this century. Biomass fertilisation in this way is extremely poorly understood, and could potentially become an important factor as mitigation progresses.
2.5.8 Thin Arctic cloud
Optically thin, low-level liquid-phase Arctic cloud has been shown to exert a warming influence over the region via absorption and re-emiance of outgoing longwave radi- ation (Lubin and Vogelmann 2006; inn et al. 2008). Arctic aerosol concentrations37 are seasonal, increasing markedly in late winter and early spring; the dry troposphere at that time also results in lile wet deposition (inn et al. 2008). In the case of low-level Arctic cloud the aerosol first indirect effect (section 2.5.2) can actually lead to enhanced efficiency of longwave absorption, and hence an increase in downwelling longwave ra- diative flux. Indeed, the indirect effect of aerosols on cloud droplet effective radii is ‘particularly high’ in the Arctic region (Garre et al. 2004). inn et al. (2008) states that the shortwave indirect cooling effect will occur at high sun angles over dark ocean surfaces, but that longwave indirect warming effect dominates at low sun angles over bright Arctic surfaces, inducing a significant positive forcing. e effect of short-lived
36 Biomass fertilisation is not a forcing and hence cannot be expressed in W m−2. 37 Consisting of sulphate, nitrate, OC, and BC aerosol.
65 Chapter 2 Aerosol effects on climate pollutants, including tropospheric aerosols,38 has been posited as a possible cause for the accelerated rate of Arctic warming relative to global averages in recent decades (inn et al. 2008).
2.5.9 Longwave absorption effect
Certain aerosol species with particle size greater than 2 µm in diameter will absorb out- going longwave radiation and consequently cause a warming effect in essentially the same way that GHGs do (Penner 2000). For example, a hybrid observational (down- welling) and modelling (upwelling) study of longwave radiative flux in an urban Indian location by Panicker et al. (2008) found enhanced tropical surface forcing on the order of 6 W m−2 to 8 W m−2, depending on varying temperature and humidity profiles. High altitude clouds (such as cirrus) can also trap OLR, causing additional warming (CCSP 2009).
2.5.10 Further effects on climatic processes
Beyond their radiative and beer-known indirect feedback effects on cloud, aerosols may modify still other climatic processes.
By cooling the Earth’s surface (reflection) and warming the atmosphere (absorption), aerosols act to decrease the atmospheric lapse rate — the surface is now cooler than normal, relative to the upper troposphere. e modified lapse rate in turn suppresses the water vapour feedback mechanism (Denman et al. 2007). ese factors then affect local atmospheric stability,39 changes to which are strongly related to the altitude of black carbon absorption.
38 Note that this includes warming through semi-direct and ice albedo effects, as well as the indirect longwave effect on cloud described here. 39 Atmospheric stability is a function of the ‘environmental temperature profile’ or lapse rate and the adiabatic lapse rates (Sturman and Tapper 2005).
66 2.5 Climatic effects and influence
Aerosols have also been found to affect large-scale atmospheric circulation paerns. Transient model simulations surveyed in AR4 found that a substantial Northern Hemi- sphere aerosol cooling caused a southward movement of the Inter-Tropical Convergence Zone and associated tropical rainfall belt; other studies suggest that aerosol effects (in- cluding reduced surface irradiance and changes to sea surface temperature (SST) gradi- ents) may explain rainfall reductions experienced in the Sahel in the 1970s and 1980s (Denman et al. 2007; Ramanathan and Carmichael 2008).
is is reinforced by the recent ‘state-of-the-art Earth system climate model’ study of Booth et al. (2012). ey find that documented links between observed multidecadal variability in North Atlantic Ocean sea surface temperature and climate shis relating to hurricane activity, and droughts in the Sahel and Amazon, are significantly affected by aerosol emissions. e authors’ analysis shows that the combination of aerosol emission and volcanic activity accounted for 76 % of SST variation. Importantly, the authors note that ‘it is the inclusion of aerosol indirect effects that allows us to capture the magnitude and the temporal and spatial structure of SST variability’, concluding that their study ‘suggests that we need to reassess the current aribution to natural ocean variability of a number of prominent past climate impacts linked to [North Atlantic SST], such as Sahel drought’ (p. 231).
Suppression of precipitation by the indirect cloud lifetime effect, while potentially redu- cing rainfall in some areas, may reduce the frequency of extreme precipitation events. e negative forcing induced by aerosols is in effect ‘competing’ with GHG warming in determining changes to evaporation and precipitation paerns (Denman et al. 2007, p. 566). e scope of aerosol-cloud interactions in transient model simulations at the time of AR4 were insufficient to give conclusive findings as to the net effect on cloud.
67 Chapter 2 Aerosol effects on climate
2.5.11 Effects on surface irradiance
Aerosols can modify solar irradiance received at the surface in a range of ways, with complex implications. at modification can exert a surface forcing that differs markedly from radiative forcing at TOA (see discussion in section 2.3).
Recall that planetary albedo is by definition the fraction of incoming shortwave radi- ation reflected back to space, viewed at TOA — a strengthened albedo is a negative TOA RF. And to reiterate, in general an increase in cloud cover or aerosol optical depth (sec- tion 2.2.5) increases extinction of incoming shortwave radiation, enhancing albedo and reducing irradiance at the surface.40 e direct effect of scaering aerosol (section 2.5.1) is the classic case: increased aerosol burden enhances atmospheric scaer of shortwave radiation (increased AOD), reflecting some back to space (stronger albedo, negative RF) and thereby decreasing shortwave irradiance at the surface (negative surface forcing). Scaering also reduces the direct–to–diffuse surface irradiance ratio (with a potential biomass feedback relation, discussed in section 2.5.7). is is not necessarily the case for scaering aerosol in clear skies, however: scaering aerosol with an asymmetry para- meter approaching +1 — ie, mostly forward scaering (section 2.2.3) — still increases optical depth, as AOD is defined with reference to irradiance received in a direct beam to the surface. Yet the total surface irradiance may be largely undiminished (negligible backscaer), with radiation simply arriving at an array of angles of incidence. Negligible backscaer means lile change to shortwave flux at TOA, hence lile to no change to planetary albedo and RF. ough surface irradiance is more diffuse, the actual quantity of energy received (W m−2) is not substantially decreased, hence producing lile to no surface forcing.
Absorbing aerosols extinguish incoming shortwave radiation entirely — that fraction absorbed — but the extent to which this changes shortwave flux at TOA or surface irra-
40 A reduction in surface irradiance is the sum of that shortwave radiation absorbed and that scaered back to space.
68 2.5 Climatic effects and influence diance depends on the characteristics of underlying cloud and surface features (see sec- tion 2.5.4). In general though, an increase in aerosol absorption of incoming shortwave radiation at altitude within the troposphere will decrease irradiance at the surface and exert a negative surface forcing, oen much larger than any radiative forcing at TOA. In fact, under the right conditions TOA RF may be nil. Ramanathan and Carmichael (2008, 223) describe a change to surface irradiance without change to TOA shortwave flux as a ‘redistribution of the direct solar radiation between the surface and the atmosphere’, which globally can weaken ‘the radiative-convective coupling of the atmosphere and decrease global mean evaporation and rainfall’.
Given such complexities, Forster et al. (2007) noted that the surface forcing is ‘arguably a more useful measure of the climate response’ for absorbing aerosols than TOA RF, particularly when assessing their effect on the hydrological cycle (p. 196) (see Box 2.2). Moreover, a surface forcing induces corresponding changes in latent and sensible heat fluxes, separate to those for shortwave and longwave energy, which can be substantially different to RF values calculated at the tropopause or TOA. at is, RF by definition does not convey the energy balance perturbation of the surface-troposphere system (Forster et al. 2007, 196).
us the fundamental distinction between aerosol scaering and absorption of short- wave radiation points to the limitations of radiative forcing as a single measure of per- turbed energy flux in all cases.
e physics outlined above is best illustrated by specific empirical examples.
A substantial decline in surface irradiance was observed at multiple locations over the second half of the twentieth century, the effect eventually becoming known as global dimming.41 Estimates vary, but surface irradiance reduced on the order of 1 to 3 percent per decade to 1990, reaching around –7 W m−2 (Trenberth et al. 2007). Since 1990 sur-
41 Although further studies concluded this effect was ‘confined to large urban areas’; ie, not truly global (Trenberth et al. 2007).
69 Chapter 2 Aerosol effects on climate face irradiance has recovered at most observation sites, brightening by around 6 W m−2. AR4 states that in general this increase is in agreement with satellite and surface obser- vations of declining cloud cover (Trenberth et al. 2007, 279), only tentatively referring to an ‘open question’ as to a possible aerosol contribution (including induced changes in cloudiness), though noting that in China the direct aerosol effect likely played a key role. A more recent review of global changes in surface irradiance highlights the central role of aerosols in stronger terms: Wild (2009) concludes that the main causes of these variations ‘appear to be changes in cloud and aerosol characteristics and abundance, which may or may not be microphysically linked’, further noting that the relative im- portance of these factors (including aerosol-cloud microphysical interactions) ‘may not be uniform over the globe but varies from region to region’ (p. 27). e change in surface irradiance from the combined anthropogenic aerosol loading is illustrated in Figure 2.10, a composite of multiple observational sources.
Figure 2.10: Surface dimming due to all anthropogenic aerosols (surface forcing in W m−2 ±30 %) (Fig. 1 panel (c), Ramanathan and Carmichael 2008). Reprinted by permission from Macmillan Publishers Ltd: Nature Geoscience, copyright 2008.
Further, a study of aerosol effects on surface irradiance over Germany concluded that total aerosol extinction must have reduced from 1975 to 1990 in order to explain the sim- ultaneous increase in the direct:diffuse shortwave irradiance ratio and weakened aerosol
70 2.5 Climatic effects and influence forcing (Liepert and Tegen, 2002, cited by Denman et al. 2007). at is, decreased aer- osol emissions meant both a reduced extinction of shortwave radiation by scaering — hence a relatively weakened negative radiative forcing — and a simultaneously reduced extinction by absorption — hence a relatively higher surface irradiance. e combined result is that more shortwave radiation is reaching the surface directly, indicated by the larger direct:diffuse ratio.
ere are disagreements in the literature on this issue. Ramanathan and Feng (2009) make the following important observation.42
e major source of dimming is [aerosol] absorption of direct solar radiation. is is fur- ther enhanced by the reflection of solar radiation back to space by [aerosols]. is should be contrasted with the TOA forcing that is solely due to the reflection of solar radiation back to space. is distinction has been ignored frequently; as a result, the dimming has been mistakenly linked with surface cooling trends (e.g., Wild et al., 2004; Streets et al., 2006). e problems with this approach are the following: for black carbon, the dimming at the surface is accompanied by positive forcing at the top of the atmosphere (Ramanathan and Carmichael, 2008), thus it is erroneous to assume dimming will result in cooling. Further- more, as we will show later, the surface dimming due to … absorbing aerosols is a factor of 2–5 larger than the aerosol TOA forcing, and for many regions they can be even of opposite sign. (Ramanathan and Feng 2009, 41-2)
Bond et al. (2011, p. 1512) support this assessment, stating that ‘[e]ach gram of emied BC adds about 1 GJ to the Earth-atmosphere system, but that energy is distributed as 2.4 GJ of increased atmospheric absorption, counteracted by 1.5 GJ that does not reach the surface’.
GCM estimates of reduced surface irradiance due to aerosols vary between –1.3 W m−2 to –3.3 W m−2 averaged globally as reported in AR4, larger than TOA flux because of aer- osol absorption (Denman et al. 2007). ough in their later review paper, Ramanathan and Carmichael (2008) report annual mean surface forcing (for 2001–2003) of –4.4 W m−2. Further, these decreases are markedly larger over land, illustrating the spatial depend-
42 Citations in original.
71 Chapter 2 Aerosol effects on climate ency of aerosol emissions.
On the other hand, absorbing aerosols have a ‘well known’ capability to reduce TOA forcing, and even change the sign from one of cooling under clear skies to a ‘large net heating’ in the presence of low altitude clouds43 (Ramanathan et al. 2001, 28386). e principal mechanism for this is aerosol absorption of cloud-reflected, outgoing short- wave radiation, exerting only minor change to surface irradiance but significantly sup- pressing cloud albedo and hence decreasing negative RF (see section 2.5.4). Ramanathan et al. (2001) observed a reduction of natural cloud RF at TOA by 5 W m−2 under these conditions: the semi-direct effect of absorbing aerosol in clear skies produced a TOA RF of –7.0 ±1 W m−2, and a factor of ten greater surface forcing of –23 ±2 W m−2; in cloudy skies TOA RF fell to –2.0 ±2 W m−2, but surface forcing was still –20 ±3 W m−2, redu- cing the difference to a factor of 3. Podgorny et al. (2000) and Satheesh and Ramanathan (2000) reported similar findings. Haywood and Shine warned in their 1997 study that, as a result of such complexities, even the sign of absorbing aerosol RF44 cannot be inferred from surface irradiance changes alone.
Both transient and equilibrium GCM studies point to decreased surface irradiance from an increased AOD and aerosol indirect effects as more important for changes to the sur- face energy budget than GHG-induced forcing of surface temperatures (Denman et al. 2007). All components of the surface energy budget — radiant energy, sensible and latent heat fluxes — decrease in response to reduced irradiance. As mean global evap- oration must equal mean global precipitation, a reduction in latent heat flux (ie, evap- oration) ought to cause a matching reduction in precipitation, again underscoring the complex effects of aerosols on the hydrological cycle as well as radiative flux and global mean temperature. However, observations over the twentieth century do not appear to demonstrate this result, indicating that aerosol-induced declines in precipitation have
43 Below around 3 km. 44 e authors refer to this as the ‘direct forcing’ (p. 1924), but this is the semi-direct effect in the cat- egorisations herein.
72 2.5 Climatic effects and influence been overwhelmed by increased evaporation and specific humidity from GHG-warming.
e principal point here is that aerosol radiative effects may vary markedly from those of GHGs, exhibiting large variation in forcing at the surface, and within and at the top of the atmosphere. ese and other important effects on climate are not captured by assessing changes to TOA radiative flux alone, expressed as radiative forcing.
2.5.12 Summary of effects
Table 2.2 presents a summary of the aerosol effects on climate examined in the preceding sections. A synthesis of these effects follows, discussing current understanding of the overall role of aerosols within anthropogenic climate change.
73 Chapter 2 Aerosol effects on climate
Table 2.2: Summary of aerosol effects on climate (aer Tab. 7.10 from Denman et al. 2007, with updates and revisions based on other sources cited in this chapter)
Effect Climatic influence Meanism Direct effect Negative TOA RF, negative Direct scaering incoming SW surface forcing radiation Indirect cloud Negative TOA RF, negative Feedback enhancement of cloud albedo effect surface forcing albedo Indirect cloud Negative TOA RF, negative Feedback increase of cloud fraction lifetime effect surface forcing Semi-direct effect Nil, positive or negative TOA RF, Absorption of incoming or outgoing in clear skies positive or negative surface SW radiation forcing Semi-direct effect Positive or negative TOA RF, Feedback change to cloud fraction above cloud positive or negative surface forcing Semi-direct effect Positive TOA RF, positive surface Feedback reduction of cloud fraction within cloud forcing Semi-direct effect Likely negative TOA RF, Feedback increase of cloud fraction below cloud negative surface forcing BC on snow & ice Positive TOA RF Feedback reduction of surface albedo ermodynamic Positive or negative TOA RF, Feedback suppression of precipitation effect positive or negative surface forcing Glaciation effect Likely positive TOA RF, likely Feedback reduction of ice cloud positive surface forcing fraction
Biomass Reduction in CO2 TOA RF Feedback enhancement of biomass fertilisation effect CO2 uptake Effect on low Positive TOA RF, positive surface Feedback enhancement of cloud OLR Arctic cloud forcing absorption Longwave Positive TOA RF, positive surface Direct absorption and re-emiance of absorption effect forcing OLR
74 2.6 Synthesis and implications
2.6 Synthesis and implications
AR4 does not provide a clear and unambiguous statement of the aggregate influence of aerosols on climate. In fact, the two relevant chapters — Chapter 2 (Forster et al. 2007) and Chapter 7 (Denman et al. 2007) — offer final estimates of the net aerosol effect that, while numerically similar, are distinctly different in character.
AR4 Chapter 7 provides what could be interpreted as the most complete estimate, as it incorporates the direct (section 2.5.1), indirect cloud albedo (section 2.5.2), indirect warm-cloud lifetime (section 2.5.3), and semi-direct (section 2.5.4) effects: Denman et al. (2007) find total net TOA forcing from all anthropogenic aerosol species relative to the preindustrial is an average of –1.2 W m−2 , with a range of –2.3 W m−2 to –0.2 W m−2. Chapter 2 considers only the direct and indirect cloud albedo effects because, as noted previously, the other aerosol effects are not regarded as radiative forcings: Forster et al. (2007) conclude that the combined total net TOA forcing from these two effects is a median value of –1.3 W m−2 with a 90 % CI of –2.2 W m−2 to –0.5 W m−2.
AR4 assessed the total positive anthropogenic forcing of LLGHGs and tropospheric ozone as +2.9 ±0.3 W m−2 (Forster et al. 2007, 200). e Chapter 2 mean anthropogenic aerosol forcing (–1.3 W m−2) therefore gives an offset of ≈45 %, with a total possible range of ≈15 % to ≈85 %; using the Chapter 7 value (–1.2 W m−2), this offset is ≈41 % with a total possible range of ≈6 % to ≈88 %. In their review, CCSP (2009) cited the Chapter 2 range as constituting the IPCC AR4 value (p. 56).
Confusingly however, the AR4 Summary For Policymakers (SPM) (IPCC 2007b) does not report any combined value, instead restating only the individual direct and indirect cloud albedo effect RFs from Chapter 2.45 Figure 2.11 below, taken from the SPM,46 shows the aerosol component of radiative forcing in context (blue coloured blocks), from the
45 And the LLGHG plus tropospheric ozone forcing sums to –2.99 W m−2, not –2.9 W m−2. 46 e similar Fig. 2.20 in Chapter 2 does not actually show the RF values but estimates of climate efficacy instead (see section 2.3.1). Fig. SPM.2 appears in that form only in the SPM.
75 Chapter 2 Aerosol effects on climate preindustrial period to 2005. ese best estimates sum to –1.2 W m−2, the same numer- ical value reported by Chapter 7. But that numerical value refers to differing physical processes in Chapters 2 and 7, and Chapter 2’s stated net forcing is, again, –1.3 W m−2 not –1.2 W m−2. Indeed, adding the positive forcing of BC on snow (+0.10 W m−2) gives a net aerosol forcing of –1.1 W m−2.
Figure 2.11: AR4 climate forcings by major component, between 1750 and 2005 (Fig. SPM.2, WG I SPM, IPCC 2007b). Copyright the Intergovernmental Panel on Climate Change, 2007.
is lack of clarity in expressing the climatic influence of aerosols suggests two conclu- sions. First, the inherent limitations of using RF as a comparison measure for various climatic agents are underscored — if important aerosol effects cannot readily be quan- tified by RF methodologies, then other approaches may be needed. Second, and apart from the limitations of RF itself, the uncertainty ranges for aerosol effects are an order
76 2.6 Synthesis and implications of magnitude greater than those for LLGHGs (as noted by Ramanathan and Feng 2008). Such large uncertainty is the principal reason for the flaened and widened net anthro- pogenic forcing probability distribution function (PDF), shown in Figure 2.12. Moreover, such complexity may act as a disincentive — explicitly acknowledged or otherwise — for policy makers to grapple with anthropogenic aerosols in devising mitigation responses.
Figure 2.12: Aerosol and GHG forcing probability distribution functions, between 1750 and 2005 (Fig. 2.20 panel (b), Forster et al. 2007). Copyright the Intergovernmental Panel on Climate Change, 2007.
Importantly, unlike GHGs the aerosol effect PDF is non-Gaussian in character, meaning that the uncertainty ranges are asymmetric about the median (Forster et al. 2007): a significant proportion of the area under the aerosol PDF is for negative forcing values substantially greater than the best estimate of –1.3 W m−2. at is, while AR4 considered –1.3 W m−2 most likely, there is a non-trivial probability that the true value is greater, and the likelihood of it being greater than –1.3 W m−2 is stronger than the likelihood of it being less. Consequently, the total net anthropogenic forcing is similarly asymmetrical, with a best estimate of +1.6 W m−2 and a 90 % CI of +0.6 W m−2 to +2.4 W m−2, where the PDF is biased toward smaller net positive values due to the aerosol uncertainty (Forster et al. 2007). Hansen has repeatedly pointed out that, should aerosol forcing be close to –
77 Chapter 2 Aerosol effects on climate
1 W m−2 for net forcing close to +2 W m−2, a major reduction in aerosol emissions — say, half — would remove –0.5 W m−2, increasing net forcing by one quarter to +2.5 W m−2; yet if aerosol forcing is close to –2 W m−2 for net forcing close to +1 W m−2, that same reduction by half removes –1 W m−2, nearly doubling the actual net forcing (eg, Hansen 2009; Hansen et al. 2011).47
Other estimates derived from model studies since AR4 produce similar but not equival- ent results. Shindell and Faluvegi (2009) derive a global mean net aerosol forcing for the period 1890 to 2007 as the difference between observations and modelled changes due to GHGs, natural forcing, and ozone. e authors obtain net aerosol forcing of – 1.31 ±0.52 W m−2, which they note is ‘consistent’ with the –1.2 W m−2 of Forster et al. (2007).
In a review of the literature, Ramanathan and Carmichael (2008) determined total aerosol cooling (BC and non-BC) is –1.4 W m−2, masking as much as 50 ±25 % of GHG warming. Referring to this study, Ramanathan and Feng (2008) state that aerosols mask ≈47 % of anthropogenic warming, with a 90 % CI of 20 % to 80 %, contrasting this with a claimed IPCC value of ≈40 %.48 is increase in the assessed aerosol mask is more significant than it appears, because the assessed BC warming of +0.9 W m−2 is markedly larger than AR4 (see section 2.4.2.3), as is the net non-BC-aerosol cooling of –2.3 W m−2. Bond et al. (2011) suggest such high values for modelled BC forcing might indicate greater than expected emissions, not necessarily a stronger BC forcing per mass.
Conversely, the substantially smaller DRF estimate of –0.3 ±0.2 W m−2 from Myhre (2009) (see section 2.5.1) necessarily reduces that mask. In fact, because that downward revi- sion partly stems from an adjustment to the assessed preindustrial aerosol burden, the relative forcing of aerosol effects beyond the direct effect may logically be reduced also.
47 e presence of these asymmetric uncertainties and potentially large negative consequences suggest extreme caution in the mitigation policy response. See Box 4.2. 48 Which presumably refers to aerosol cooling relative to LLGHGs plus tropospheric ozone but excluding BC on snow.
78 2.7 Observation and modelling
Accordingly, only a crude assessment of a revised total aerosol cooling is possible.49 I estimate a value between –1.1 W m−2 and –1.0 W m−2, proportionally between ≈38 % and ≈34 %. While Myhre (2009, 190) notes that the smaller DRF value suggests it off- sets ‘only a modest 10% of the radiative forcing that is due to increases in well-mixed greenhouse-gas concentrations at their current concentrations’, this does not account for other aerosol effects, a substantial caveat.
aas et al. (2009) used ten separate GCMs to diagnose global mean total aerosol short- wave forcing via two methods. Combining models and satellite observations, they found a value of –1.5 ±0.5 W m−2, the highest of any discussed here. In an alternate method in- volving scaling Nd by satellite estimates of AOD, the authors found a total all-sky forcing of –1.2 ±0.4 W m−2, again broadly in line with AR4 estimates.
Alternative approaches for deriving total net aerosol forcing via planetary energy bal- ance studies are detailed in section 3.2.
2.7 Observation and modelling
e sections to follow elaborate on observational and model based assessment of aerosol climatic effects, particularly the limitations and persisting uncertainties.
2.7.1 Observation and measurement
AR4 highlighted that differentiating aerosols of anthropogenic versus natural origin in satellite retrievals is ‘very challenging’ (Forster et al. 2007, p. 155). is view is reiterated in the detailed CCSP (2009) survey of aerosol knowledge, and observational capabilities in particular. Part of the reason for these difficulties is the lack of appropriate observa- tional systems, briefly examined in Box 2.3.
49 Simply by decreasing the AR4 Chapter 2 and 7 DRF values only.
79 Chapter 2 Aerosol effects on climate
Box 2.3: Missing satellite observations
Hansen (2009) explains that effective observation of aerosols requires satellite platforms de- signed for the purpose — the two key measurements needed are polarimetry of reflected sunlight, and interferometry of planetary thermal emission.a He then lays out the history of specific pro- posals to achieve the needed measurements in the early 1990s, requiring four instruments of moderate size and cost each (pp. 65–9). Unfortunately, Hansen believes the political dynamics of the US space program (desire for large, expensive projects) were a significant factor in these proposals failing to be adopted. NASA’s Glory satellite was set to improve aerosol measurement dramatically — a purpose-built platform to observe total solar irradiance, and an innovative new ‘aerosol polarimetry sensor’ (Mishchenko et al. 2007).b Tragically, Glory was lost when its launch vehicle failed in early March, 2011 (the same vehicle that failed in 2009, destroying the Orbiting Carbon Observatory spacecra).
a e physics of this was known in 1970 (p. 65). b NASA’s Glory web site is still available at http://glory.gsfc.nasa.gov/.
Discrepancies among different satellite products remain and uncertainties persist — in- deed systemic biases may exist. Retrieval of aerosol data over land is necessarily more complex, simply because of the natural variations in geography, topography, surface reflectance and so on; model-satellite integrations must commonly be relied on instead (CCSP 2009). Observations of aerosol loading and properties are generally limited to clear-sky conditions. As current satellite platforms do not measure chemical composi- tion, they have limited ability to distinguish the anthropogenic aerosol fraction; although retrieval of aerosol fine-mode fraction may be an avenue for deriving the anthropo- genic component of AOD. Satellite remote sensing is also not sensitive to particles less than 0.1 µm in diameter — which are a ‘significant fraction’ of CCN — with clear con- sequences for accurate measurement of indirect effects on cloud (CCSP 2009, 25).
In satellite-based estimates of aerosol direct radiative forcing, uncertainties in the an- thropogenic fraction are greater than for all species combined, especially over land (CCSP 2009). Nearly 80 % of overall uncertainty in anthropogenic DRF stems from five paramet- ers, each accounting for 13 % to 20 %: the fine-mode fraction and anthropogenic portion of it over both land and ocean, and AOD over ocean.
80 2.7 Observation and modelling
Ground-based in situ programs (such as those operated by the WMO) and remote sensing platforms (such as NASA’s global robotic AERONET system50) also have a vital role to play, particularly since these can observe aerosol behaviour below the cloud layer, which is normally opaque to satellites. In situ observations can provide much more detailed, localised measurement and at higher temporal resolution than remote sensing. CCSP (2009) notes that accurate observations of aerosol ‘extinction profiles’ (discussed in section 2.5.11) are ‘pivotal’ to improved calculation of radiative forcing and atmospheric responses. New lidar systems in particular are increasingly being applied to this need. In fact lidar systems can even observe local aerosol entering cloud formations, rather than relying on the column-integrated AOD view of satellites (CCSP 2009). Surface network measurements of aerosol optical depth have obtained an accuracy of 0.01 ~ 0.02, which is considered adequate for assessing clear-sky TOA DRF to within 1 W m−2 (CCSP 2009).
In situ studies at multiple global locations found that while increased aerosol burden does lead to increased Nd, the process is highly variable and always sub-linear, highlighting that cloud droplet number concentration is a function of numerous parameters and not only aerosol concentrations (CCSP 2009). One reason for the variability is the role of updra velocity as an important factor in the activation of aerosol particles to water droplets (such as that found in ‘vigorous’ cumulus clouds). Further, not all aerosols will increase Nd — certain ‘giant’ CCN around a few micrometres in diameter can lead to
‘significant suppression in cloud supersaturation’ and reduced Nd (CCSP 2009, 49).
Studies based on surface radiometers, reported in CCSP (2009), actually found that vari- ance in cloud LWP accounted for most of the variance in cloud optical depth. is means that detection of aerosol indirect effects becomes even more difficult, swamped as those changes are by the larger shis in water content. A number of post-AR4 studies ex- amining the complex interactions of aerosol particles with cloud, and discrepancies in
50 e Aerosol Robotic Network of ground-based remote sensing sun photometers and radiometers; see http://aeronet.gsfc.nasa.gov.
81 Chapter 2 Aerosol effects on climate various observational methods, ‘reinforced the importance of LWP and vertical velo- city as controlling parameters’ for particle radius and cloud opacity, and have begun to reconcile the underlying reasons for those discrepancies in observation (CCSP 2009, 47). McComiskey and Feingold (2008) note that improving the measurement accuracy of cloud droplet effective radius response to aerosol burden will strengthen estimates of RF and narrow their uncertainty range. Satellite observations showing a weak response were used as inputs to the GCMs that produced the lowest RF estimates for the indirect cloud albedo effect in AR4 — McComiskey and Feingold (2008) point out that in situ ob- servations find a stronger response, more in line with theory, and would consequently produce a stronger RF.
antifying aerosol effects on cloud precipitation efficiency is a complex task for satel- lite platforms: their instantaneous view does not readily allow causal relationships to be determined (CCSP 2009). is difficulty stems from the temporal aspect of the cloud life- time effect, in that aerosol induced changes to precipitation rates and cloud lifetime are self-evidently not observable by instantaneous snapshots. However, CloudSat, launched in 2006 and riding as part of the A-Train satellite group,51 provides substantial oppor- tunities for inferring aerosol precipitation effects.
Reconciliation and synergy of observational data with model simulations is another im- portant area of continued effort. Closure studies employ coordinated use of multiple instruments sampling regional aerosol properties combined with intensive in situ field experiments. One such study reviewed in CCSP (2009) determined uncertainties for clear-sky DRF estimates of around 25 % for sulphates and carbonaceous species and as much as 60 % for aerosol containing dust. Along with detailed in situ measurement of aerosol properties, closure studies can in part be used to constrain regional chemical transport model (CTM) simulations of aerosol DRF (CCSP 2009). CTM results can then be provided as inputs to climate models in place of assumed values for aerosol proper-
51 Glory was intended to join the A-Train (L’Ecuyer and Jiang 2010).
82 2.7 Observation and modelling ties; though CTMs are themselves subject to a range of uncertainties in emissions and complex aerosol factors. Satellite observations of global parameters such as AOD permit evaluation of large scale climate model simulation results, constraining aerosol contribu- tions. Integrating observational and model data aids the goal of reducing discrepancies between them and narrowing uncertainties in both.
2.7.2 Aerosol modelling
Modelling of atmospheric aerosols has developed rapidly since the start of the twenty- first century (CCSP 2009). e effort and complexity involved is a serious challenge, as models must account for atmospheric chemical transformation, transport, and re- moval processes, as well as the highly heterogeneous distribution of primary aerosols and secondary precursor emissions. Complex additional atmospheric processes such as hygroscopic growth of aerosol particles via condensation of ambient water vapour are commonly included as parameterisations (in this case as a function of relative humid- ity) (CCSP 2009). AOD (section 2.2.5) is the parameter in best agreement across models: CCSP (2009, pp. 59-60) reports that the median value for global annual mean AOD across 16 AeroCom models is 0.127, with a standard deviation of 18 %. is value also agrees with global AOD means from recent satellite observations.
However, in another section comparing a different set of five models with the MODIS and MISR satellite platforms, CCSP (2009, p. 38) reports that global annual mean AOD from the models is smaller than the results obtained from both satellites, and that these dif- ferences are substantially larger on regional scales. Integrated satellite-model products tend to fall somewhere in between model AOD projections at the low end and satellite observations at the high end, though they are in beer agreement with the ground-based AERONET. Agreement for total AOD among the AeroCom models also masks significant diversity in individual aerosol species’ AOD, mass loading, and extinction efficiencies.
83 Chapter 2 Aerosol effects on climate
On balance, these issues suggest that AOD alone is not sufficient to fully assess aerosol forcing, and that it may mask cancelling or compensating effects across species. CCSP (2009) highlights the parallel importance of the asymmetry factor (section 2.2.3), SSA (section 2.2.4), and other optical properties. Further, regional and seasonal diversities are much larger than global means, again underscoring aerosols’ temporal and spatial heterogeneity.
Successful reproduction of observed twentieth century climate is an important valida- tion of climate models. But to date models have included very uncertain aerosol for- cing (CCSP 2009). A survey of GCMs used in reproduction of twentieth century climate found a strong inverse correlation between climate sensitivity and total anthropogenic forcing: models with high climate sensitivity generated low total forcing and vice versa, so that all arrived at broadly the same reproduction of climate52 (Kiehl 2007). Uncer- tainty ranges for climate sensitivity and for total forcing are of the same magnitude, and the primary reason for the intermodel variation in total forcing is an almost threefold spread in aerosol negative forcing contribution. is spread among nine of the mod- els used by AR4 meant that though individual model climate sensitivities also varied threefold53 — between 1.5 ℃ and 4.5 ℃ — all reproduced the observed trends (CCSP 2009). For this reason, CCSP (2009) cautions that successful reproduction of past climate does not necessarily mean projections of change for the twenty-first century will be accurate. ough Kiehl (2007) argues that these uncertainties do not invalidate the use of (cur- rent) models for future projections, in part because future emission scenarios include large GHG increases, which would come to dominate the total anthropogenic forcing.
Writing in 2005, Andreae et al. made the implications of these relationships stark: if climate sensitivity is low then aerosol forcing is necessarily weak (and total forcing re- latively high); but if sensitivity is actually high then aerosol forcing is strong (and total
52 Recall ∆TS = λ RF , as described in section 2.3. 53 e variation in climate sensitivity is largely due to uncertainties in cloud feedback processes (Kiehl 2007).
84 2.7 Observation and modelling forcing relatively low) and we have ‘a worrying future that may bring a much faster temperature rise than is generally anticipated’ (p. 1188).
e climate modelling community employs two broad approaches relevant to aerosols: forward modelling aempts to explicitly calculate the forcing from aerosol particu- late and precursor emission inputs through models of aerosol physics and chemistry; the inverse approach assumes observed climate change is the result of known forcings, so with assumed climate sensitivity (or a range thereo), aerosol forcing can be inferred from the residual forcing of GHGs and other agents54 (CCSP 2009; Hegerl et al. 2007). As discussed in section 2.6, AR4 concluded from forward modelling studies that tro- pospheric aerosols offset ≈45 % of the positive forcing of GHGs and other agents, but with a large uncertainty range (15 % to 85 %). Studies of the inverse type examined by AR4 Chapter 9 (Hegerl et al. 2007) found an aerosol forcing range of –1.7 W m−2 to –0.6 W m−2, which brackets the possible extent of aerosol forcing without requiring detailed knowledge of complex aerosol processes (CCSP 2009). is inferred aerosol for- cing range is both narrower and contained within the ranges reported in AR4 chapters 2 and 7.
Aempts to estimate aerosol emissions are also part of the overall modelling effort. Nat- ural sources are actually less well quantified than anthropogenic, and can have dramatic spatial and temporal variation. A 2004 study, for example, concluded that available data allowed only an order of magnitude estimate of sea salt aerosol production rates (CCSP
2009). In contrast, secondary sulphate and nitrate aerosols formed from SO2 and NOx precursors have relatively well understood emission sources (CCSP 2009). In comparis- ons of observations with chemical transport model simulations, sub-micrometre sulph- ate and black carbon were found to be the best represented (CCSP 2009). Large discrep- ancies between model projections and observations were found for other species such as POM, and especially those of super-micrometre size such as mineral dust (in total,
54 e Shindell and Faluvegi (2009) study discussed in section 2.6 is an example of this.
85 Chapter 2 Aerosol effects on climate underestimated in models by a factor of three). Model projections of POM are a ‘severe problem’, in large part because of serious difficulty in representing the formation of secondary organic aerosols (CCSP 2009, 35). Models also struggle to represent organic aerosols’ vertical distribution, underpredicting their presence in the free troposphere (CCSP 2009). In fact, organic aerosols are one of the greatest challenges facing climate modellers, and even the distribution of well-characterised sulphates are not necessarily estimated correctly (CCSP 2009).
Surveying a range of historical emission inventory studies, CCSP (2009) concluded that the growth in primary aerosol emissions was ‘not nearly as rapid’ as the growth in CO2: heavy use of biofuels and a lack of technology and controls on coal burning led to relat- ively high late nineteenth and early twentieth century emissions of BC and POM; over the twentieth century much stronger particulate controls were introduced, and biofuel use declined. In the modern period, while small-diameter sulphate, BC, and POM con- stitute about 10 % of total aerosol mass, these species are primarily anthropogenic and so located in the most populated regions, hence their climate and air quality effects are of major concern (CCSP 2009).
Related to these inventories is the determination of a reasonable preindustrial aerosol baseline. is is no simple task given the paucity of available data and spatial coverage — CCSP (2009) notes that large uncertainties necessarily exist in estimates of the prein- dustrial aerosol burden. e relative abundance of the various aerosol species is also difficult to assess. It had been assumed that the preindustrial aerosol composition was broadly the same as that of the modern period, simply at lower quantities. But a recent revision by Myhre (2009) concludes that black carbon has in fact grown at twice the rate of other species, which has significant consequences for net aerosol effects, reducing total SSA (see section 2.2.4) from a preindustrial value of 0.986 to 0.970 today — about twice as absorbing.
Model calculations of DRF (section 2.5.1) are defined by three key aerosol optical para-
86 2.7 Observation and modelling meters, as a function of given wavelength: AOD, SSA, and the asymmetry parameter (g) (CCSP 2009). All parameters are influenced by relative humidity, and varying particle size will also change SSA (see definition in section 2.2.4). Model estimates of DRF tend to be smaller than observational estimates (by around 55 % to 80 %), and even more so on regional scales (CCSP 2009). In part this is because models usually use the preindus- trial atmospheric aerosol burden as their comparative baseline, whereas observational estimates are commonly calculated by comparison to the total absence of anthropogenic aerosols.55 Other factors also play a role, such as differences in AOD and SSA values, unperturbed surface albedo, model parameterisations, and radiative transfer schemes. At the same time, two of the major satellite platforms (MODIS and MISR) also diverge substantially in estimates of AOD, by 10 % to 15 % (CCSP 2009).
Model calculations of aerosol indirect effects are markedly more difficult — ‘a great deal of complexity’ remains unresolved within GCMs, particularly in their representation of aerosol-cloud interactions (and clouds in general), the principal context of indirect effects (CCSP 2009, 66). In fact, in their survey of the field, CCSP (2009) noted that most models had not incorporated the indirect cloud lifetime effect (see section 2.5.3).56 e presence of cloud-aerosol feedbacks make projection of global mean temperature response based on linear addition of RF values problematic once GCMs include aerosol- cloud interactions beyond the cloud albedo effect, though linearity can be restored if these feedbacks are treated as additional forcing terms (Forster et al. 2007, 197). CCSP (2009) contrasts the conclusions of Hansen et al. (2005b) regarding such feedbacks with the discussion in AR4 (Forster et al. 2007; Denman et al. 2007): Hansen et al. argue that the cloud lifetime effect (represented by time-averaged cloud cover) is actually more
55 Recall that Myhre (2009) found a substantial reduction in estimated DRF, in part due to the revision of preindustrial aerosol burden. 56 is no doubt reflects the enormous difficulty of doing so: ‘Deficiencies in the data record, and a poor understanding of what processes regulate the behaviour of cloud regimes, frustrate aempts to aribute real correlations among the aerosol, clouds and precipitation to lifetime effects’ (Stevens and Feingold 2009, 608).
87 Chapter 2 Aerosol effects on climate likely to dominate the cloud albedo effect (section 2.5.2), and that this dominance is supported by studies of both cloud-resolving models and satellite observations.57 An important issue in resolving aerosol indirect effect uncertainties is that satellite obser- vational platforms to date are not able to provide aerosol and cloud field retrievals with temporal resolution greater than once per day, and that higher level cloud obscures those at lower levels. CCSP (2009) summarises the situation by stating that aerosol indirect effects, particularly cloud lifetime, remain ‘to a large extent unconstrained by satellite observations’ (p. 69).
Because current generation GCMs and available computational resources cannot yet re- solve cloud microphysics nor cloud-aerosol interactions, modellers must represent aero- sol effects through parameterisation schemes — and doing so is itself extremely difficult (CCSP 2009).58 For the indirect cloud albedo effect, the main unresolved question is the degree to which water droplet concentrations increase with enhanced aerosol burden — b recall relation eq. 2.10 from section 2.5.2, Nd ≈ (Na) — and the relative importance of particle size distribution, aerosol composition, and updra velocity in affecting that con- centration. Aerosol number concentration and particle size distribution are believed to be the most significant; updra velocity becomes important when aerosol number con- centrations are high, but the inherent small scale makes this perhaps the hardest aspect to model (CCSP 2009). ough composition is generally regarded as the least influen- tial of these factors (represented by fairly simple parameterisations of the soluble and insoluble fractions), it cannot be ignored, as any substantial shi away from fossil fuels toward biofuels — triggering a decline in sulphate — could change aerosol composition in possibly significant ways.
e indirect cloud lifetime effect is represented in GCMs by a change to precipitation production and is commonly parameterised by cloud fraction (CCSP 2009; Hansen et al.
57 e two studies referred to were published in 2004 and 2005, respectively. 58 Given the pace of growth in computational power, however, this assessment may now be outdated as regards a resource constraint.
88 2.7 Observation and modelling
2005b). CCSP (2009) identified a range of complexities in aerosol effects on precipitation and cloud lifetime, which reinforce the overall conclusion that understanding of these processes remains an open challenge. e large eddy simulation study of Ackerman et al. (2004, cited in CCSP 2009) found that cloud water content may increase or decrease with additional aerosol loading in stratocumulus clouds, depending on the overlaying air’s relative humidity. Wang et al. (2003, cited in CCSP 2009) showed that increased aerosol concentration in stratocumulus cloud tends to reduce water content because smaller droplets more readily evaporate, inducing a feedback process that dilutes the cloud. In their recent large eddy simulation study, Xue et al. (2008) identify evidence of two distinct regimes for aerosol effects on cloud lifetime: the first, ‘precipitating regime’, exists for increases in Na up to some threshold level, where cloud lifetime is extended by precipitation suppression (increased ‘cloud condensate’ residence time); but in the ‘non- or weakly precipitating regime’, further increases in aerosol loading beyond a certain
Na level reduces the cloud condensate residence time through enhanced evaporation of the smaller droplets. Given these persistent complexities and sometimes conflicting results, CCSP (2009) observes tellingly that ‘[i]t could be argued that the representation of these complex feedbacks in GCMs is not warranted until a beer understanding of the processes is at hand’, and that ‘[m]oreover, until GCMs are able to represent cloud scales, it is questionable what can be obtained by adding microphysical complexity to poorly resolved clouds’ (p. 72).
In general, only a fraction of current generation models consider aerosol species and climatic effects other than sulphate DRF (CCSP 2009). e CCSP (2009) review goes so far as to state explicitly that model simulations used in AR4 did ‘not adequately account for’ aerosol forcing (p. 72); their summary table of AR4 models clearly shows the relat- ive lack of accounting for species such as BC, OC, or mineral dust, and even the indirect effects of sulphate. A further concern among models is that global average values that agree with other models and with observations mask serious differences in the spatial
89 Chapter 2 Aerosol effects on climate character and relative species contribution of aerosol loading. GCMs also tend to as- sume that multi-species aerosol loadings are externally mixed (see section 2.1.2), while observations indicate that internal mixing is actually more common, which can have non-trivial implications for the resultant radiative forcing (CCSP 2009). Some modelling groups are developing ‘sophisticated aerosol mixtures’ using detailed calculations of mi- crophysics, and these could well produce ‘very different’ forcing values for both direct and indirect effects (p. 78). Finally, CCSP (2009) notes that almost all model simulations are for all-sky conditions — because obtaining cloud-free scenes from GCM output is difficult — whereas satellite and remote sensing observations are all currently for clear sky conditions, as these platforms are not able to measure aerosol optical depth through cloud. Because of the effect of relative humidity on AOD, the actual value for AOD under cloudy conditions is expected to be higher.
e short residence time and geographically-based nature of aerosols mean there is an important time dependency in calculating the net forcing and consequent changes to global mean surface temperature (CCSP 2009). at is, the type and quantity of aerosol species has varied over time (especially the twentieth century), and this has a marked influence on net forcing and ∆TS. Figure 2.13 shows an illustrative example of this time dependence from 1850 to 2000. e arrested growth in sulphate optical depth as a result of Northern Hemisphere air pollution regulations is evident — though not an actual net decline because of offseing increases in Asia; see Box 2.1 — as is the resultant flaening of effective forcing.
90 2.7 Observation and modelling
Figure 2.13: Illustrative time dependence of aerosol optical thickness and effective forcing from the GISS GCM (Fig. 3 panel (c), Hansen et al. 2007b). Reproduced with kind permis- sion from Springer Science & Business Media B.V.
Figure 2.14: Illustrative model-derived temporal evolution of the surface forcing and instantan- eous radiative forcing (panels from Fig. 2.23, Forster et al. 2007). Copyright the Intergovernmental Panel on Climate Change, 2007.
91 Chapter 2 Aerosol effects on climate
Figure 2.14 shows an illustrative model-derived temporal evolution of the surface and instantaneous radiative forcing59 for all main agents over the same period — though less discernible in this graphic, the aerosol direct and cloud albedo radiative forcing are similarly flaened from the 1990s onward, while the LLGHG forcing continues to grow apace. Newer quantitative estimates of individual forcing agents and net forcing discussed in Chapter 3 have modified this picture somewhat, but the principal aspects of forcing over time remain.
CCSP (2009) concludes their review of aerosol modelling by returning to the persist- ent large differences in model calculations of aerosol forcing distributions, the large uncertainties in modelling atmospheric processes and emission profiles, and the fact that agreement among models on total AOD oen obfuscates substantial variation in component species and their consequent contribution to (direct) forcing. Similarly re- inforced is the warning that a degree of agreement on global averages among models, and between models and observations, masks much larger discrepancies at the regional level. Because the spatial and temporal heterogeneity of aerosols means their effects are far more significant regionally, ‘it is insufficient or even misleading to just get the global average right’ (CCSP 2009, 81).
Crucially, modelling of indirect effects remains a great challenge. Efforts to reduce un- certainties are confounded by substantial intermodel discrepancies in the complex input parameters described above (particle size, composition, mixing state, microphysical and atmospheric processes), and confronted by a lack of necessary global-scale observational data for comparison. Again, the majority of models available for IPCC AR4 did not even include aerosol indirect effects. e final CCSP statement is compellingly blunt: ‘Im- provements must be made to at least the degree that the aerosol indirect forcing can no longer be used to mask the deficiencies in estimating the climate response to green-
59 is is the ‘true’ forcing of a radiative perturbation applied instantaneously, with no temperature or atmospheric state adjustments permied; Fi in Figure 2.2.
92 2.7 Observation and modelling house gas and aerosol direct RF’ (2009, p. 81). Such strong caution is echoed by Stevens and Feingold (2009), who warn of the great difficulties remaining in modelling of cloud microphysics and the continued use of parameterisations; global model estimates of the cloud lifetime effect in particular may well underestimate the true uncertainty.
Chapter 3 next examines alternative approaches to estimate net aerosol forcing that are able to avoid many of these persistent modelling difficulties.
93
3 Climate response and associated risk
e range of aerosol species, their properties, and array of effects on climate are laid out in Chapter 2. is chapter now expands from that foundation to site aerosols within the full context of anthropogenic climate change. Recent studies of planetary energy balance uncover important new insights into climate response, and provide alternat- ive, empirically-driven means to infer net aerosol forcing beyond the primarily model- derived estimates discussed in section 2.6. e potential ultimate consequences of a re- duced aerosol loading under abatement scenarios are then investigated, particularly in relation to risks of dangerous climate change. Finally, available quantitative metrics are reviewed for their capacity to provide meaningful, policy-relevant assessment of both LLGHG and aerosol contributions to anthropogenic climate change.
3.1 Climate response to forcing
e actual response of the climate system to any forcing is the proper concern for policy makers. antifying the forcing exerted by particular agents is policy-relevant only to the extent that they can meaningfully indicate this response over time. e principal measure of ultimate climate response is the change in global mean surface temperature,
95 Chapter 3 Climate response and associated risk
TS, at radiative equilibrium. Per eq. 2.7 in section 2.3, this is global mean radiative forcing multiplied by climate sensitivity. However, RF alone does not readily indicate how much of that perturbation is yet to be responded to, or, conversely, how great a fraction of the perturbation has already changed climate. Policy makers need information on the transient response, not only the final result. In that light, it is worth now re-examining RF in more detail.
3.1.1 Radiative forcing revisited
Some of the difficulties surrounding radiative forcing and the range of methodologies available for calculating it were discussed in section 2.3. Further issues also exist, par- ticularly relating to comparing forcing agents using their RF values. RFs are usually calculated as averages for the Earth as a whole, which obscures potentially significant variation in regional forcing (Forster et al. 2007). Net RF may be zero if substantial re- gional asymmetries cancel out, even though global mean temperature is still affected. At the same time, similar paerns of climatic response can be caused by RFs exhibit- ing very different spatial paerns and hence appearing quite dissimilar. e spatial and temporal heterogeneity of aerosol emissions and their effects are not adequately rep- resented if global-scale RF alone is the metric used for assessing climate change (CCSP 2009). Comparison of particular forcing agents can also be misleading if dissimilar RF methodologies are employed: forcing values can vary substantially between methods, and express qualitatively distinct physical processes. Similarly, summation of calculated per-agent RF values is subject to a range of uncertainties and methodological difficulties. In all, while RF is clearly a useful and important indicator of perturbations to the climate system, Forster et al. (2007, 133) warn that it ‘provides a limited measure of climate change as it does not aempt to represent the overall climate response’.
Even within a single methodology, not all radiative forcings are ‘interchangeable’. Con-
96 3.1 Climate response to forcing sider RFs derived for greenhouse gases and for aerosols. GHGs perturb climate by ab- sorbing and then re-radiating OLR, so that an increase in GHG concentrations causes more energy to be absorbed by the planetary surface — GHGs produce a longwave radiat- ive forcing. Aerosols affect climate via qualitatively different and much more heterogen- eous mechanisms. e most significant radiative effects of anthropogenic aerosols are in modifying incoming shortwave radiation; they both change planetary albedo through reflection, and further modify surface irradiance through atmospheric absorption (sec- tion 2.5.11). e distinction with regard to GHGs is highly salient: absorbed shortwave radiation is the originating energy input to the Earth-atmosphere system, and temper- ature responds rapidly to significant changes in insolation. Furthermore, unlike GHGs, the forcing exerted by aerosols at TOA can differ substantially from that at the surface. Aerosol absorption in particular may produce small or even nil observable change in TOA RF while producing a large negative surface forcing. As explained in section 2.5.4 and 2.5.11, this is because aerosol absorption extinguishes surface shortwave irradiance without necessarily inducing a corresponding change in TOA shortwave flux.
Radiative forcings for particular agents acting on longwave versus shortwave radiation are then most meaningfully compared with reference to final planetary radiative equi- librium. Direct comparison of RF values during ongoing transient perturbations may obscure large differences in temperature response and other important climate variables (eg, Forster et al. 2007, 195-9). e radiative forcing concept is hence important but alone is insufficient to capture the range of climate responses generated by forcing agents of substantially differing character. Forcing efficacy (section 2.3.1) addresses some of these problems, but cannot resolve them all. As we will see in the sections that follow, the transient response is likely to become increasingly critical for mitigation policy, making these difficulties of intercomparison via global-average RF an emerging policy issue.
97 Chapter 3 Climate response and associated risk
3.1.2 Equilibrium climate sensitivity
As explained in section 2.3, equilibrium climate sensitivity defines the ultimate response to a forcing of the Earth-atmosphere system induced by a change in boundary conditions, commonly given as the equilibrium ∆TS induced by a doubling of atmospheric CO2 concentrations. A brief further explanation of this factor is warranted, in part because of the interdependence of derived climate sensitivity and inferred net aerosol cooling inherent in inverse modelling studies (as noted in section 2.7.2; see also Tanaka and Raddatz (2011)). Climate sensitivity derived from model studies is therefore subject to structural uncertainties, and as noted in section 2.3, these have proved challenging to reduce.
Hansen and Sato (2011) and Hansen et al. (2008) instead derive equilibrium climate sens- itivity from the paleoclimatic record. eir central estimate remains 3 ℃ for doubled CO2 as in many previous studies, but the uncertainty is narrowed slightly to give 3 ±0.5 ℃ rather than the ±1 ℃ commonly found previously. Stated more technically, their central −2 estimate is 0.75 ±0.125 K/W m for 2×CO2.
In both studies, the authors emphasise the need to clearly differentiate between the fast- feedback or ‘Charney’ sensitivity,1 and the full slow-feedback sensitivity of final equi- librium. Fast-feedbacks refer to near-immediate changes in response to a forcing by water vapour, clouds, natural aerosols, snow cover and sea ice; slow-feedbacks involve changes to surface albedo, including vegetation and ice sheets, and LLGHGs, which lag by decades to millennia (depending on the forcing itsel). Most discussions of climate sensitivity actually describe fast-feedbacks only — and this is the value given here — but slow-feedback sensitivity, or ‘Earth system sensitivity’ may be as much as double this, particularly due to CO2. e full Earth system sensitivity should therefore be an important consideration for policy decisions around ‘acceptable’ climate change and cor-
1 Named in reference to the lead author of the first report examining climate sensitivity in 1979.
98 3.2 Planetary energy imbalance and climate response responding atmospheric LLGHG concentrations. However, this thesis is concerned with the extent to which anthropogenic aerosols represent a threat to GHG mitigation efforts themselves, and hence climate response in the short term is the relevant factor. Fast- feedback sensitivity is therefore the applicable form and should be assumed henceforth.
3.2 Planetary energy imbalance and climate response
e planetary energy budget is ‘a fundamental characterization of the state of the cli- mate’ (Hansen et al. 2011, 13441); it provides a powerful tool for evaluating the climate response to forcing and the extent of perturbation to radiative equilibrium (Murphy et al. 2009; Hansen et al. 2005a; Church et al. 2011; Trenberth et al. 2009; Huber and Knui 2012). Imbalance over several years between the energy absorbed by the Earth (absorbed insolation) and that emied back to space (OLR) is the ‘primary symptom’ of planetary thermal inertia, thereby providing ‘an invaluable measure of the net climate forcing’ (Hansen et al. 2005a, 1431). A positive planetary energy imbalance means that more en- ergy is being absorbed than emied — the climate system has not yet fully responded to a net positive forcing. at extra energy will hence continue to heat the planet, as over the long term emied energy must match absorbed energy to restore radiative equilib- rium: further warming is ‘in the pipeline’. e geophysical processes that constitute the climate response are complex, but the overall picture is well illustrated by Figure 3.1.
Figure 3.1 shows an idealised step change in GHG concentration inducing a positive radiative forcing (top panel). By raising atmospheric longwave opacity, elevated GHG levels cause outgoing longwave radiation to space (OLR) to fall immediately (boom panel), inducing a positive radiative forcing. e planet now has a positive energy im- balance (the shaded area). e climate responds at two time scales (mid panel): a rapid partial surface temperature increase over about a decade, with consequent increase in OLR, reducing much of the initial energy imbalance; followed by the long response of
99 Chapter 3 Climate response and associated risk
Figure 3.1: Illustrative sketch of the response to positive energy imbalance (Fig. 1, Murphy et al. 2009). Reproduced with kind permission from John Wiley & Sons, Inc. the oceans over centuries, slowing as radiative equilibrium is approached. ese scales reflect the small heat capacity of the planetary surface and atmosphere relative to the vast heat store of the oceans.
antification of thermal inertia and associated climate response is highly policy-relevant. e delay between imposition of a forcing and realisation of its full effect means that a degree of GHG stabilisation-target overshoot may be manageable: if the net forcing can be reduced in time, the ultimate temperature response will be lessened. e question is, how much time? Accurate assessment of planetary energy imbalance helps to answer that question; it provides a key global diagnostic of anthropogenic interference, defining ‘how much current climate forcings must be altered to stabilize climate’ (Hansen et al. 2011, 13441).
100 3.2 Planetary energy imbalance and climate response
Ocean heat content is the predominant component of the planetary energy budget and the principal source of thermal inertia — around 90 % of additional energy goes to heat- ing the oceans (Murphy et al. 2009; Trenberth and Fasullo 2010). erefore, while the ultimate response to any forcing is the change in global mean surface temperature at equilibrium, TS, a function of climate sensitivity, the pace of that change depends largely on the rate of heat exchange between the surface mixed layer and deep ocean, and, to a much smaller degree, rates of ice sheet disintegration (Hansen et al. 2005b; Wigley and Schlesinger 1985; Hansen et al. 2008, 2011). Calculating the planetary energy budget over time requires combining ocean heat content data with the observational temperature re- cord (land and sea surfaces) and all forcings. ese forcings include the well-constrained LLGHGs, for which atmospheric concentrations are directly measured; other GHG spe- cies; anthropogenic aerosols, which are primarily tropospheric; and significant vari- ations in natural boundary conditions, principally the solar radiation cycle and volcanic aerosol emissions, which are reasonably well-quantified as they reside in the strato- sphere and hence survive for a few years. e change in ocean heat content (known as ‘uptake’) roughly equals the energy imbalance: while the imbalance continues — surface warming is not yet sufficient to restore OLR to radiative equilibrium — that energy is heating the oceans as the planet’s major heat reservoir.
Accurate knowledge of the full profile of ocean heat content — and its rate of change — both at surface level (to ~700 m) and at depth (to 2000 m), is consequently vital for quantifying that imbalance. e recently-established international Argo float network,2 continuously measuring ocean temperature throughout the vertical water column to 2000 m, is now making possible ‘dramatic improvements’ in this knowledge (Hansen et al. 2011).
Planetary energy imbalance can shed light on the actual rate of change in TS during
2 Argo has continually expanded coverage and data quality since its initiation in 2000. Measurements became available from 2003, with full deployment achieved in 2007. See http://www.argo.ucsd.edu/.
101 Chapter 3 Climate response and associated risk the transient response to forcing, where climate sensitivity by itself can tell us lile. is rate of change of the climate response is highly important for mitigation strategy and for determining how much ‘room to move’ exists under conditions of stabilisation target overshoot; reliable estimation of the temporal paern of temperature response is enormously valuable. e climate response function determines that rate of change, defined as the fraction of equilibrium change to TS realised as a function of time, sub- sequent to imposition of a forcing (Hansen et al. 2011). Climate sensitivity tells us how much, the climate response function tells us how fast.
antifying the planetary energy budget affords one final insight: an empirically con- strained derivation of total negative aerosol forcing (Murphy et al. 2009; Church et al. 2011; Hansen et al. 2011). Positive forcings are the GHGs and minor changes to solar output over the ~11 year cycle; this is offset by any stratospheric aerosol cooling. Total energy absorbed at the planetary surface is balanced by increased OLR as a function of rising TS, and the storage of energy as heat in the oceans and melting of ice. e difference between this total energy absorbed and the remaining positive forcing — the energy residual — is the net negative forcing from the sum of aerosol direct and indirect effects.3
e energy budget since 1950 determined by Murphy et al. (2009) is illustrated in Fig- ure 3.2, where panel (a) shows the total cumulative positive forcing and panel (b) shows shows the cumulative negative forcing, including ocean heat storage and the land and atmosphere energy absorption leading to increase in TS. To reiterate, ocean heat content is a proxy for planetary energy imbalance.
3 As well as any unknown mechanism; though none is suggested by the evidence.
102 3.2 Planetary energy imbalance and climate response
Figure 3.2: Cumulative planetary energy budget since 1950 (Fig. 6, Murphy et al. 2009). Repro- duced with kind permission from John Wiley & Sons, Inc.
3.2.1 antitative estimates of energy imbalance and aerosol forcing
Studies providing specific estimates of planetary energy imbalance, both current and his- torical, are difficult to compare — they may use differing methodologies, assess possibly overlapping but discrete time periods, and do not necessarily give values for a consist- ent set of indicators. eir estimates of the absolute and relative forcing contribution of aerosols may more profitably be compared, however. ese estimates from relevant available studies are discussed below, with a summary given in Table 3.1.
An early quantitative estimate is that of Hansen et al. (2005a). Imbalance did not exceed a few tenths of a wa per square metre until the 1960s, but increased GHG forcings gives an imbalance in 2003 of 0.86 ±0.15 W m−2, with aerosol forcing estimated at –1.39 W m−2 (approximately 50 % uncertainty). Commied warming with then-current atmospheric composition was a further ~1 ℃ above 2000 mean temperature. ough an important study, this work has effectively been superseded by Hansen et al. (2011), discussed below.
103 Chapter 3 Climate response and associated risk
Murphy et al. (2009) performed an energy balance GCM study strictly constrained by observations of surface temperature, GHG forcing, solar radiation, volcanic emissions, and ocean heat content. Estimated planetary energy imbalance is given graphically but not stated explicitly in the text. ey infer from energy residuals a total anthropogenic aerosol forcing best estimate4 of –1.06 ±0.4 W m−2 for the period 1970–2000 at the 1σ confidence level. e authors note that their derived residual forcing agrees with the direct and cloud albedo effect forcings for 2005 in Forster et al. (2007), and fits the total aerosol forcing estimate of Denman et al. (2007). However, the range of –1.9 W m−2 to –0.3 W m−2 at the 2σ confidence level excludes large negative aerosol forcing beyond –1.9 W m−2, hence serving to narrow the AR4 uncertainty range shown in Figure 2.12. Murphy et al. (2009) question the validity of an apparent jump in energy residuals around 1995, greater than –1.5 W m−2 in 2000, concluding it is likely due to errors in ocean heat uptake data or temporary heat transport to the deep ocean.5 ey ‘largely exclude’ an actual increase in aerosol forcing itself as, in part, no increase in albedo or AOD was evident in satellite observations since that time.
Murphy et al. (2009) underscore the ‘striking result’ that since 1950 only about 10 % of the positive forcing of GHGs and solar radiation have actually gone into heating the Earth: 20 % has been balanced by increased OLR; a further 20 % has been offset by volcanic aerosol emissions;6 and the remaining 50 % has been offset by the combined direct and indirect effects of anthropogenic aerosols.
In a short perspective article, Trenberth and Fasullo (2010) state a post-2000 planetary energy imbalance of 0.9 ±0.5 W m−2. However, ocean heat content data then available to the authors suggested a discrepancy in observed energy absorption: it appeared less than expected, thus preventing closure of the energy budget. Hansen et al. (2011) address
4 Other estimates are given (some larger) depending on specific ocean heat content and GHG forcing datasets; the one cited here is their best estimate overall. 5 Interestingly, the higher rate of heat transport to the deep ocean noted by Hansen et al. (2011) below would seem to support this. 6 Particularly El Chichón in 1982 and Mt Pinatubo in 1991.
104 3.2 Planetary energy imbalance and climate response this point directly, arguing that their calculations using revised and expanded data show no such discrepancy, a smaller imbalance, and hence a closed budget (pp. 13436–7).
Church et al. (2011) used sea level and land water reservoir data in concert with updated ocean heat content data to examine the energy budget since 1961. Applying the ap- proach of Murphy et al. (2009), they infer aerosol forcing from energy balance residuals, finding a central estimate of –0.8 ±0.4 W m−2 for the 1980s and early 1990s. is value, though lower for the best estimate, has an uncertainty range consistent with the results of Murphy et al. (2009) and Hansen et al. (2005a). (It is however outside the bounds of Hansen et al. (2011) below.) e authors also discuss a post-1995 jump in energy resid- uals to about –1.5 W m−2, similar to that noted (but rejected) by Murphy et al. (2009). However, here Church et al. (2011) argue that the updated ocean data do not readily support a conclusion that this is an artefact of measurement error. In fact they conclude that this jump and inferred increase in negative aerosol forcing is ‘robust’ (p. 6), and necessary to close the planetary energy budget, given the slowdown in TS increase over the previous decade. e authors suggest a range of possible reasons for an increased negative aerosol forcing:
• A substantial surge in developing country sulphur and other aerosol species emis- sions, such as that postulated by Kaufmann et al. (2011) (see section 3.6).
• A small decline in stratospheric water vapour on the order of 0.1 W m−2 to 0.2 W m−2, which is not taken into account in their analysis.
• Moderate volcanic activity (a natural contribution to enhanced cooling).
• Emerging evidence from lidar observations at Mauna Loa Observatory and Boulder, Colorado reported by Hofmann et al. (2009). ese show increased aerosol backs- caer in the stratosphere at altitudes of 20 km to 30 km of around 4 % to 7 % per year since 2000, indicating partial transport of tropospheric aerosol to the strato- sphere, due in part to a large increase in Chinese coal burning since 2002.
105 Chapter 3 Climate response and associated risk
e comprehensive study of Hansen et al. (2011) calculates energy imbalance of 0.58 ±0.15 W m−2 for 2005–2010, a period coincident with the most prolonged solar minimum of the satellite era (ie, a small negative solar forcing). is period is also marked by a trans- ition to decreased upper ocean heat storage and increased heat uptake at depth. For 1993–2008, upper ocean heat storage dominated, but calculated energy imbalance varies substantially with the particular ocean heat content analysis, ranging from 0.59 W m−2 to 0.80 W m−2. Figure 3.3, taken from this work, shows an indicative history of planetary energy imbalance over recent decades; a useful summary is given below it.
Net aerosol forcing is inferred to be –1.6 ±0.3 W m−2 in 2010, the largest of any to date, implying substantial forcing from indirect effects. Hansen et al. (2011) remark that their findings agree well with the ‘insightful’ work of Murphy et al. (2009) discussed above. ey find aerosol forcing of –1.2 ±0.3 W m−2 for the 1970–2000 period Murphy et al. used, and estimate from Murphy et al.’s Fig. 4c a forcing of about –1.5 W m−2 in 2000, consistent with their own. Hansen et al. (2011) further note that though their findings do not exceed the a priori estimates of AR4 including all indirect forcings (discussed in section 2.6), they do exceed the range used in most model simulations undertaken for TAR and AR4.
Inference of the total aerosol cooling effect from energy imbalance in this way is a ro- bust approach, but it cannot determine the contribution of individual species or precisely quantify the relative role of direct and indirect effects. Hansen et al. (2011) declare that ‘continued failure to quantify the specific origins of this large forcing is untenable, as knowledge of changing aerosol effects is needed to understand future climate change’ (p. 13421) and that ‘until aerosol forcing is measured, its magnitude will continue to be crudely inferred, implicitly or explicitly, via observations of climate change and know- ledge of climate sensitivity’ (p. 13422).
Huber and Knui (2012) employed an alternative method for anthropogenic aribution of climate warming based on ensemble modelling of energy balance driven by ‘boom-
106 3.2 Planetary energy imbalance and climate response up’ estimates of historical forcing, constrained by observations of temperature since 1850 and ocean heat uptake since the 1950s. ey found an average net forcing between 1850 and 2010 of 0.54 W m−2, ranging between 0.36 W m−2 and 0.76 W m−2, and a most likely current net forcing of 1.6 W m−2 (no uncertainty stated). Aerosol direct and indirect effects (including volcanic emissions to the stratosphere) have offset ‘about hal’ of the
7 warming effect of GHGs, giving a total change in TS since the 1950s of about 0.55 ℃.
Figure 3.3: Indicative planetary energy imbalance over time (Fig. 15b, Hansen et al. 2011). Imbalance is the top of the red area, which shows energy uptake by ice and land and air warming; ocean heat uptake is in blue. Real-world interannual climatic variability (noise) is removed in this representation. Measured ocean denotes Argo data era.
[A] precipitous decline in the growth rate of GHG forcing about 25 [years] ago caused a de- crease in the rate of growth of the total climate forcing and thus a flaening of the planetary energy imbalance over the past two decades. at flaening allows the small forcing due to the solar cycle minimum, a delayed bounceback effect from Pinatubo [volcanic aerosol] cooling, and recent small volcanoes to cause a decrease of the planetary energy imbalance over the past decade. (Hansen et al. 2011, 13439)
7 Somewhat confusingly, the authors also state that ‘About 83% of the accumulated energy from carbon dioxide forcing alone … is offset by the combined negative direct and indirect effect of aerosols’ (p. 1), and later that simulated temperature increase from GHGs of 1.31 ℃ is countered by direct and indirect aerosol cooling of –0.85 ℃, which is approximately 65 %.
107 Chapter 3 Climate response and associated risk , obtained from 2 − content discrepancy based on available OHC data and estimated energy imbalance, as discussed above. Offset is interpreted from data presented in source graphic Fig. 3(a). Offset is derived relative to total LLGHG forcing of slightly more than 3 W m source graphic Fig. 18(b). Aerosol cooling offset ‘about hal’ of LLGHG forcing, though this includes stratospheric aerosol contribution. Notes translation of fossil fuel emissions with effect parameterisation rather than inferred. Offset explicitly stated as discussed above. 46 % 50 % 50 % 50 % Not given Authors posit a possible heat ≈ Not given Aerosol forcing is estimated by ≈ ≈ ≈ Period 2000 2000 ) 2 − (W m –1.6 ±0.3 2010 Inferred aerosol forcing Offset –1.06 ±0..4 1970– –1.5 Period Estimate 2000 2010 ) 2 − Imbalance Estimate (W m 0.86 ±0.15 2003 –1.39 ±0.7 2003 0.9 ±0.5 post 0.58 ±0.15 2005– Method GCM study checked against OHC data Energy balance GCM study, constrained by obs of positive forcing and OHC Unstated. Employs energy budget approach of Murphy et al. 2009, with additional constraint of land and ocean water balance data Energy balance GCM constrained by obs, including updated OHC data Ensemble modelling of energy balance from ’boom up’ historical forcing estimates, constrained by obs of temp and OHC 2009 2005a 2011 2011 2010 Estimates of planetary energy imbalance and inferred aerosol forcing Study Hansen et al. Murphy et al. Trenberth and Fasullo Church et al. Hansen et al. Huber and Knui 2012 Table 3.1:
108 3.3 Commied warming
3.3 Commied warming
e additional temperature response expected without further anthropogenic interfer- ence is known as the committed warming, where commitment refers to the con- sequences of historical emissions. Projection of this commitment is ‘a fundamental met- ric for both science and policy’ (Armour and Roe 2011). Mahews and Weaver (2010) argue that discussion of commied warming — viewed as ‘unavoidable’ due to climate system inertia — is oen misleading, owing to a basis in fixed atmospheric concentra- tions, ignoring atmospheric drawdown by carbon sinks. is is actually a ‘constant- composition commitment’,8 such as atmospheric stabilisation targets expressed as vari- ous CO2 ppm concentrations, which in fact require ongoing GHG emissions to maintain — the IPCC stabilisation ranges in AR4 are a highly influential example (Tab. SPM.5, IPCC 2007c).
Mahews and Weaver (2010) argue the correct definition is for cessation of all emis- sions, the ‘zero-emissions commitment’, where atmospheric concentrations will begin to decline following emission termination, leading to temperatures remaining roughly constant or beginning to fall. It is hence socioeconomic inertia, not geophysical, that will determine long term future warming. However, it must be stressed that Mahews and Weaver (2010) refer to CO2-commitment only, and so do not take into account the long residence times of some non-CO2 LLGHGs, or the cooling offset of anthropogenic aerosols.
Armour and Roe (2011) point out that once these are taken into account the abrupt ter- mination of aerosol emissions induces immediate significant temperature increase (see section 3.4.1). ey argue that this temperature spike should be treated separately. e ultimate commitment at equilibrium is the commied warming due to past GHG emissions, aligning with Mahews and Weaver (2010) and the “‘geophysical” warm-
8 Hare and Meinshausen (2006) describe this as a “‘present forcing” warming commitment’.
109 Chapter 3 Climate response and associated risk ing commitment’ identified by Hare and Meinshausen (2006). e transient commit- ment is defined as the peak temperature increase following emission termination.
3.3.1 Derived climate response function
As outlined in Box 3.1, the energy balance study of Hansen et al. (2011) reached two important conclusions: first, inferred current net negative aerosol forcing is large,9 in turn indicating that indirect effects exceed direct, possibly substantially; second, the climate response function is more rapid than represented in models, due to the study’s finding of less-efficient nature of ocean heat mixing. If correct, aerosol masking is hence greater, and when it is removed the climate will respond faster to the then-strengthened positive GHG forcing. is approach to deriving a climate response function appears to be novel in the literature at time of writing. e authors do emphasise however that it is necessarily provisional, partly based on qualitative reasoning given the data available (but incorporating sensitivity analysis to varying key assumptions), and that substantial effort is being expended on active investigation of the flaws indicated in the GISS ocean model.
Hansen et al. (2011) derive a climate response function producing 40 % of the change in
TS within the first 5 years, followed by a subsequent long, slow ‘recalcitrant’ response phase, producing 75 % of the change in TS within 100 years. ey note that while the slow response tail remains in this assessment as with previous research, it is exaggerated in many climate models (including that of their own GISS model). e rapid response in TS ‘implies that even moderate ongoing changes of the climate forcings can have a noticeable effect, despite the fact that the climate system is still in a mode of trying to come to equilibrium with forcing changes that occurred over the past century’ (p. 13436).
9 For details and comparison to other energy balance studies, see Table 3.1 above.
110 3.3 Commied warming
Box 3.1: Is climate response faster than previously thought?
In late 2011, James Hansen and colleagues published a detailed assessment of the climate re- sponse function (Hansen et al. 2011). eir study determined net climate forcing and the rate of ocean heat mixing, constrained by the detailed observations of ocean heat content now afforded by the Argo network, in concert with observed land and sea surface temperatures. A number of research groups are actively assessing the raw Argo data, and the study carefully examined each group’s output in turn. Hansen et al. conclude that most climate modelsa mix heat into the deep ocean too efficiently. Novel application of ocean mixing data derived from observed downward transport of transient chemical tracers (such as chlorofluorocarbons) further support this finding, as do other empirical indicators such as an unrealistically deep thermocline at lower latitudes. e implications of this likely flaw are enormously significant. slower ocean heat mixing ⇒ surface ocean warms faster ⇒ climate responds more rapidly A relatively less efficient real-world ocean heat mixing means that heat is transported to the depths slower than represented in climate models. Ocean surface waters will therefore be warmer than expected, because they retain more of the heat absorbed due to a persistent positive forcing. Consequently, global mean surface temperature will be higher than for more efficient ocean heat mixing, which means that the climate response to that forcing is in turn faster than models indicate. Yet the major climate models reproduce recent warming with good skill. How? Building on pre- vious work by Knui (2008)b, Hansen et al. (2011) argue that the answer must be that net forcing is actually less than calculated by climate models: an exaggerated net forcing employed by mod- els counters their unrealistically slow climate response function, generating a good match to the observed temperature record. is conclusion is not made arbitrarily; indeed the authors note that a ‘family of solutions’ is consistent with observed warming, involving varying values for the key factors of net forcing, equilibrium climate sensitivity, and the climate response function (which is dependent on ocean heat mixing). Climate sensitivity is known from paleoclimatic evidence, as discussed above. Planetary energy imbalance and observed ocean heat content provides the means to discriminate between the remaining possible solutions. e positive for- cing of LLGHG concentrations is well-constrained by observations, so this term is accurately known. In contrast the forcing of anthropogenic aerosols is ‘practically unmeasured’. Ocean heat content constrains the feasible range of the climate response function. e net forcing de- rived from planetary energy imbalance therefore permits the aerosol forcing term to be inferred, and the result strongly indicates that aerosol negative forcing is underestimated, both their direct effects and indirect influence on clouds. a Possibly because of a common ancestor in the GFDL Bryan-Cox ocean sub-model. b ough here it must be pointed out that Knui (2008) arrives at largely the opposite conclusion, that total aerosol forcing is relatively small.
111 Chapter 3 Climate response and associated risk
As of the completion of this thesis, however, no other study is known to have reached a similar conclusion nor replicated the particular findings regarding modelled ocean heat mixing efficiency — the Hansen et al. (2011) result may hence be regarded as something of an outlier.
A contrasting view is given in Oo et al. (2013), using an intermediate type of energy budget approach. Oo et al. combine data-based estimates of heat uptake from the oceans, land surfaces, ice, and atmosphere, with a current-generation multi-model en- semble estimate of total and aerosol forcing to determine equilibrium climate sensitivity and ‘transient climate response’.10 eir study finds a lower temperature increase for both equilibrium and transient response, based on data from the most recent decade — the implication is that temperature response under conditions of a substantial drop in aerosol loading is not likely to be a serious concern. However, aerosol forcing is here obtained from forward modelling studies and is in fact further reduced to beer accord with recent satellite-constrained estimates (Oo et al. 2013, Supplemental); the value used is substantially lower than those surveyed above and those reviewed in section 2.6. Given the potential systemic underestimates in both models and satellite observations (section 2.7), this input aerosol forcing value may be biased low. e resultant values for transient and equilibrium temperature response may therefore also be underestimates. While the approach of Oo et al. (2013) offers an important contrast to the comparatively strong negative aerosol forcing estimates discussed above, available observational data remains constrained and the limited understanding of complex microphysical aerosol effects on cloud persist. In my view, such model-based studies therefore cannot yet rule out the risk of a large and rapid temperature change consequent to a weakened aerosol mask implied by those larger negative forcing estimates.
10 e laer is defined as the change in TS evident at the point where 2×CO2 is reached under conditions of a 1 %/yr rate of CO2 increase.
112 3.4 Dangerous anthropogenic interference
It is worth highlighting the conclusions of Murphy et al. (2009) here: the relatively small fraction of energy actually retained by the Earth thus far reminds us that in reality energy imbalances can heat the planet quickly, and that ‘it is energetically possible for surface temperature to increase or sea levels to rise much more rapidly than they have in the recent past’ (p. 11).
3.4 Dangerous anthropogenic interference
e central objective of climate change mitigation is to prevent dangerous anthropo- genic interference in the climate system, as articulated by Article 2 of the UNFCCC (reproduced on page 8). Dangerous interference is usually understood as the need to keep temperature rise above the preindustrial level to some upper limit, most commonly the 2 ℃ ‘guard rail’, a widely-nominated goal in international mitigation negotiations (Richardson et al. 2009; WBGU 2009; Anderson and Bows 2011).11 e rate of that rise is as important as the magnitude, however, and this facet is oen neglected in policy dis- cussion, despite its inclusion in Article 2 (Ross and Mahews 2009). e rate of change maers most to natural ecosystems: temperature rising too rapidly may overwhelm adaptive capacity, threatening species extinction, ecosystem vulnerability, and loss of biodiversity. Ross and Mahews (2009) noted that no consensus as to a safe rate of change threshold exists in the literature, but rates above 0.2 ℃/decade seem likely to be dangerous for most systems. At localised warming rates above 0.3 ℃/decade, the bulk of ecosystems are unable to adapt (Raes and Seinfeld 2009). Any rapid transient warming triggered by the loss of a significant fraction of the aerosol mask therefore risks wide- spread ecosystem damage, even if the magnitude is tolerable.
Further, dangerous anthropogenic interference may manifest across a spectrum of vul- nerable elements in the climate system, as well as natural ecosystems (Lenton et al. 2008;
11 See also discussion in section 1.2.
113 Chapter 3 Climate response and associated risk
Schellnhuber 2009).12 Nonlinearity is a prominent feature of the Earth-atmosphere sys- tem, and abrupt changes may occur at differing levels of temperature rise. If tipping points are exceeded for long enough, changes may be triggered that are irreversible on any time frame meaningful to humans. Nonlinear shis in some elements also risk sig- nificant positive feedbacks, exacerbating warming further, such as the loss of highly reflective Arctic sea ice; methane release from melting permafrost and undersea marine hydrates; or ice sheet disintegration in Greenland and Antarctica. Recent research indic- ates substantial Arctic methane release is already occurring (Anthony et al. 2012), and that Greenland disintegration may be triggered by lower temperatures than previously thought (Robinson et al. 2012).
Ramanathan and Feng (2008) argue that the reality of aerosol cooling means that the commied warming relevant for mitigation policy is that of the full GHG forcing. Fig- ure 3.4 shows this commitment as a probability distribution, based on a climate sens- itivity probability density function; tipping elements are superimposed, demonstrating the temperature thresholds assessed by Lenton et al. (2008). Aerosol-induced temperat- ure change may therefore play a central role in understanding dangerous anthropogenic interference, so we turn now to quantitative assessments of its possible magnitude and pace.
3.4.1 Aerosol-induced temperature change
Bond et al. (2011, p. 1506) describe the effect of short-lived species as akin to ‘a burst of energy’ applied to the Earth-atmosphere system, the temperature response to which ‘is very different from response to longer-lived species’. e effect of an instant removal of a large fraction of aerosol cooling would be similar to a sudden and sustained spike in solar output: energy imbalance increases by the same amount as the negative aerosol
12 Recall the initial discussion in section 1.2.
114 3.4 Dangerous anthropogenic interference
Figure 3.4: Dangerous anthropogenic interference for a range of tipping elements (Fig. 1, Ramanathan and Feng 2008). Commied warming due to GHG emissions 1750–2005; temperature thresholds as assessed by Lenton et al. (2008), except for Himalayan gla- ciers. Copyright the National Academy of Sciences, 2008. forcing that has just been lost, as more solar radiation now reaches the surface. ant- itative estimates of the likely magnitude and rate of change to TS induced by significant shis in anthropogenic aerosol loading over time are therefore key complements to the knowledge of the climate response function in determining effective mitigation policy. e temperature response to such changes in shortwave radiative flux is hinted at by the anomalous diurnal temperature range observed during the grounding of all commercial aircra13 in the United States following the September 2001 terrorist aacks: over this three-day period, the temperature range rose 1.1 ℃ above the mean, the only such period in 30 years where the increase was greater than two standard deviations (Travis et al. 2002).
13 Aircra contrails are known to, in part, affect transfer of incoming solar radiation.
115 Chapter 3 Climate response and associated risk
Model studies have been undertaken to examine various aspects of the temperature re- sponse to a significant loss of aerosol cooling, either in scenarios of abatement, or where deliberate geoengineering schemes are aborted or fail. Brasseur and Roeckner (2005) modelled a ‘drastic reduction’ in aerosol load, specifically sulphate. ey found that TS could rise by 0.8 ℃ in under a decade if all anthropogenic sulphate were removed, with regional increases of 1 ℃ on most continents and as much as 4 ℃ in the Arctic. Hare and Meinshausen (2006) examined one illustrative scenario where all emissions cease in- stantly, which induced pronounced TS rise of around 0.3 ℃ to 0.4 ℃ in the years following (depending on the year when emissions cease), remaining elevated for some decades.14 Ramanathan and Feng (2009) state that current LLGHG concentrations give commied warming of 2.4 ℃, with negative aerosol forcing of about –1.4 W m−2 offseing that in- crease to date, leading to a rise of 1.3 ℃ if aerosols are removed. Model studies of aerosol and GHG interaction also indicate nonlinearity in their combined effect on temperature response (Ming and Ramaswamy 2009).
Mahews and Caldeira (2007) modelled a complex set of circumstances comparing BAU emissions and deliberate geoengineering, with abrupt failure occurring at particular years this century. With BAU emissions offset by geoengineering, a failure in 2075 in- duced a rate of change as high as 4 ℃/decade in the years immediately following. eir lowest scenario was 2 ℃/decade, still ten times faster than current rates. e authors state that ‘global temperature responded quickly to changes in incoming solar radiation’ in all cases (p. 9951). eir results showed rapid convergence to a non-geoengineered climate forced by remaining LLGHG concentrations when geoengineering was abruptly removed, a clear analogue for a precipitous loss of aerosols.15
Similar work by Ross and Mahews (2009) examined future deliberate geoengineering
14 is study used a simpler GSM than the current state of the art. 15 Mahews outlined preliminary model results for a loss of anthropogenic aerosols in the case of mit- igation, with immediate warming of 0.2 ℃ to 0.3 ℃, followed by convergence with the CO2-only result (per. comm. 2011).
116 3.4 Dangerous anthropogenic interference and the consequences of abrupt removal. ey found that very rapid warming can oc- cur in the year immediately following removal, around 0.4 ℃ to 0.5 ℃ per year (depend- ent on climate sensitivity) and beyond 1.0 ℃/decade under some circumstances. ese conditions are not directly translatable to more common mitigation scenarios, but they indicate a capacity for rapid change, particularly in the years immediately following a significant loss of aerosol forcing.
In a specific form of commied warming study (refer to section 3.3 above), Tanaka and Raddatz (2011) examined two future emissions scenarios: SRES A1B to 2100 and a hypo- thetical total cessation of all anthropogenic emissions in 2020.16 eir approach is based on a deliberate analysis of the interdependent inverse relation between climate sens- itivity and negative aerosol forcing inherent in inverse modelling studies constrained by the temperature record. ey find that an initial abrupt warming under the cessa- tion scenario is the ‘hidden commitment’ represented by aerosol emissions, arguing that this has received lile aention in studies of commied warming (noting the exception of Armour and Roe (2011)). For all ranges of climate sensitivity considered, this hid- den commitment under cessation led to a marked initial TS increase, with the rate of increase strongly dependent on sensitivity. Importantly, even with low sensitivity the rate of increase exceeds 0.2 ℃ per decade.
Other studies examine implications for temperature change from a loss of aerosols due only to imposition of air pollution controls. Raes and Seinfeld (2009) state that the Kloster et al. (2009) scenario of maximum possible air pollution control to 2030 concur- rent with continued increase in GHGs leads to long term increase of 2.2 ℃,17 with a rate of increase between 2000 and 2030 of 0.4 ℃/decade globally, and as high as 0.8 ℃/decade over large areas of the Northern Hemisphere. Any robust imposition of air pollution controls is thereby likely to increase the current warming rate of around 0.2 ℃/decade.
16 Emissions scenarios are examined in section 3.6.3. 17 e Kloster et al. (2009) paper is somewhat ambiguous as to what base this increase is relative to — presumably it is preindustrial climatology.
117 Chapter 3 Climate response and associated risk
ere is no sure way to know in advance the rate of temperature rise induced by aero- sol emission abatement consequent to climate change mitigation, or indeed from strong air pollution controls. e outcome is too heavily dependent on aerosol emission rates which cannot reliably be predicted in advance. Scenario studies examining the rate of change in future emissions then bracket the possible rate of temperature change, but more exact projections are inherently extremely difficult. ese model studies are also subject to the large uncertainties surrounding indirect effects, which may not be fully in- cluded, and to the possible errors in ocean heat mixing discussed above. But the primary risk is that under scenarios of wide-scale loss of aerosol cooling, the rate of temperature change is likely to exceed adaptive capacities; even if the global mean is tolerable, the rate of change over continental areas may be dangerous. On the other hand, Schellnhuber (2008) points out that in reality reductions of aerosol emissions could be more haphaz- ard and gradual, possibly avoiding any significant warming spike. Regardless, the funda- mental implication must be that the full ramifications of air pollution and climate change mitigation policies on aerosol cooling must be understood.
Raes and Seinfeld (2009) provide a useful illustration of these ramifications for temper- ature rise and its rate of change, reproduced in Figure 3.5 — the dashed lines indicate the commonly perceived outcome of mitigation based only on LLGHGs, the solid lines indicate the potential for unintended side effects when aerosols are also considered.
3.5 Toward appropriate metrics
Appropriate measurement of aerosols and their effects is problematic. As discussed pre- viously, metrics such as radiative forcing are not a good fit for these spatially and tem- porally heterogeneous forcing agents, tending to obscure important differences region- ally and between TOA forcing versus that at the surface or in the atmospheric column.
118 3.5 Toward appropriate metrics
Figure 3.5: Illustrative response to air pollution and climate change mitigation policies (Fig. 1, Raes and Seinfeld 2009). Reprinted from Atmospheric Environment, Copyright 2009, with permission from Elsevier.
Aempts to refine RF using forcing efficacy or the related RF index18 do not substan- tially address these concerns. is problem is even more pronounced at the point of mitigation policy instruments, commonly reliant on global warming potentials and CO2- equivalence, which are far from suitable for short-term forcing agents in general, and especially for those with negative forcing effects. Beer quantification of aerosol emis- sions and their radiative effects is clearly necessary.
3.5.1 The problem with GWP and CO2-e
e global warming potential (GWP) is an index expressing the climate change impact of a forcing agent relative to CO2. GWP is defined by the IPCC as the ratio of time- integrated global mean RF induced by a 1 kg pulse emission of some compound relative
18 e ratio of total CO2 + non-CO2 RF to that of CO2 alone, developed to assess sectoral emission profiles (especially transport and aviation) (Unger 2010). Fuglestvedt et al. (2010) highlight that RF index has been misapplied in some cases, incorrectly framing it as a per-emission metric where it is not, as RF depends on previous emissions.
119 Chapter 3 Climate response and associated risk
to a 1 kg pulse emission of a reference gas, CO2 (Forster et al. 2007). e time horizon is usually either 20, 100, or 500 years, with 100 years the most commonly stated and the metric applied by the Kyoto Protocol.19 GWP is employed within multi-gas abatement frameworks by deriving a carbon dioxide equivalent (CO2-e) emission mass that would produce the same time-integrated RF as the non-CO2 species. In this way, all species can be expressed in ‘equivalent’ CO2 amounts and then treated as interchangeable (and tradeable). ere is no actual requirement in the definition of GWP that CO2 be the reference gas, but in practice this is the standard, and the IPCC calculates GWPs on this basis (eg, Forster et al. 2007, 210–3). is equivalence therefore implicitly assumes that all agents induce an exclusively positive forcing, as CO2 does.
GWP has always been subject to strong criticism. Shine (2009) reviews the history of the global warming potential, highlighting the strong caveats in the IPCC First Assess- ment Report in 1990 regarding the lack of agreement on a methodology to combine all relevant factors of a forcing agent within a single metric. In fact, GWP was chosen to illustrate the difficulties in doing so. But once adopted, GWP became vital for the multi- gas approach of the Kyoto Protocol where, inter alia, its transparency is an advantage, despite real concerns such as a lack of specificity as to what climate effect was actually being compared. e Kyoto Protocol may not have been possible without it. Shine (2009) canvasses criticism of the IPCC for failing to fully evaluate alternatives and for failing to integrate the perspectives of all Working Groups,20 in effect treating the issue as a question of physical science alone and thereby neglecting the serious mitigation policy implications. e IPCC has begun to address these concerns, however; alternatives are likely to be beer evaluated for the Fih Assessment Report.
19 Shine (2009) notes there is no real scientific reason for that choice over other time periods, and Fuglestvedt et al. (2010) point out that the decision reflects value judgements as to the importance of future impacts. 20 In particular that of WG III (mitigation), where a range of alternatives have been raised, such as eco- nomic damage functions or price ratios taking into account how close temperatures are to some target limit; though these too may suffer from a range of contentious assumptions regarding abatement cost, discounting, and so on.
120 3.5 Toward appropriate metrics
e problem for application of GWP and especially CO2-e to cooling aerosols is that such agents do not satisfy the positive forcing assumption, and are subject to marked regionality in effect. To be at all meaningful for these species, GWPs must be negative. Studies have calculated negative GWPs for species such as organic maer and sulphate (Bond et al. 2011; Fuglestvedt et al. 2010), but doing so hardly reflects the metric’s original design. Constructing GWP from time-integrated RFs compounds the serious difficulties in intercomparison of RF between GHGs and aerosols, as the stratospherically-adjusted
TOA methodology (Fa in section 2.3) is likely used for both despite its poor representa- tion of aerosol effects. e global RF is only a ‘crude approximation’ for such regionally and temporally heterogeneous species (Smith and Wigley 2000a, pp. 446,448).
In general, GWP has been strongly criticised for methodological flaws that are directly relevant to policy, and this purely for GHGs. GWP is based on integrated RF arising from pulse emissions rather than the sustained emission over time seen in reality, and this integral takes no account of dynamic responses to that forcing or of changes in the relation of emissions to actual atmospheric concentrations, which are oen nonlinear (Smith and Wigley 2000b).
e large disconnect in atmospheric residence time between CO2 and aerosols weakens the approximation further still. ere is lile logical equivalence of aerosol lifetimes
(days) with that of CO2 (decades to centuries) — not at even 20 years, let alone horizons of 100 or 500. Moreover, its applicability for any short-lived agent is tenuous at best when using the conventional 100 year horizon, even for those having unambiguously positive forcing. How can tropospheric ozone, a non-uniformly mixed species with at- mospheric lifetime on the order of months, accurately be quantified as an approximation of CO2 unit forcing integrated over 100 years, despite the climatic forcing of ozone be- ing undoubtedly positive? Further, 100 year GWP is highly problematic for mitigation policy if the goal is to limit warming to some target maximum: GWP treats all pulse emissions equally, regardless of how close the target actually is; a low GWP value for
121 Chapter 3 Climate response and associated risk aerosol species (or potent warming species such as methane) is then dangerously mis- leading if the target is near (Shine et al. 2007). Shindell et al. (2009) also demonstrate that relative forcing is not static, as atmospheric chemical interactions between GHGs and other species such as aerosols affect their GWPs. Taken together, GWP is hence ‘ques- tionable for the short-lived non-CO2 effects and robust values do not currently exist’ (Unger 2010, 5332).
For cooling aerosols the implication is unavoidable: they are not equivalent to carbon dioxide in any meaningful sense. Policy instruments reliant on CO2-e that aempt to include cooling aerosols are thus likely to be seriously flawed and unworkably reduc- tionist.
3.5.2 Alternative methods
Some of the problems with GWP and CO2-e might be avoided by careful use of the ra- diative forcing methodology. But RF is not without its own difficulties, as discussed above and in section 3.1.1, even with the newer definitions, and ambiguity as to which definition is being applied in any given quantitative assessment also remains a concern. RF is also applicable principally at planetary or regional scales, and hence unlikely to be suited to the task of reporting and emission target-seing, in the manner that CO2-e conventionally is. In fact, Ramanathan and Feng (2009, 46) argue that RF is ‘an inad- equate and even an inappropriate metric’ for evaluating the regional effects of aerosol loading. Fuglestvedt et al. (2010) point out that RF is “backward looking”, in that it sig- nifies the forcing exerted by cumulative emissions relative to some baseline state, rather than being concerned with the future impact of emission. Further, mitigation policy really needs metrics that express information on a per-emission (mass) basis, especially if any form of trading is to be involved. Finally, a policy-relevant metric is needed that is capable of properly capturing aerosol effects, most importantly those with negative
122 3.5 Toward appropriate metrics forcing.
No readily deployable alternative metric is apparent today that perfectly meets those broad criteria, but two approaches offer substantial improvements over GWP and CO2- equivalence — each is examined below, and a summary of their characteristics follows in Table 3.5. Global temperature potential is analogous to GWP, but redresses some of its limitations regarding time horizon and lack of clarity as to what effect is being compared. Specific forcing pulse was constructed explicitly for short-lived species, principally aer- osols, and is well-suited to quantifying emission-pulse effects on a regional basis. Both metrics are likely to be important for future mitigation policies as they are further de- veloped, but both also have limitations of their own that will continue to pose challenges for aerosol-integrated policy design. Perhaps this is unavoidable to some degree, as any aempt to devise a metric that can assign exchange values between differing forcing agents by ‘[a]dding together the climate impact of species that have different character- istics is a bit like adding apples and oranges’ (Tol et al. 2008, p. 3, via Fuglestvedt et al. 2010) — determining what should be compared or aggregated, and why, is a subjective decision of policy.
Other approaches that incorporate economic perspectives — damage functions, abate- ment costs, and the like — are not considered here. ese inherently involve subjective assumptions and hence move away from a purely physical concern with the underlying geophysical changes.
3.5.2.1 Global Temperature Potential
e global temperature change potential (GTP) was developed to represent the actual change in global mean surface temperature induced by a forcing agent, rather than the RF (Shine et al. 2005b, 2007). GTP ‘follows the general philosophy’ of GWP, but a major difference is that GTP indicates the change in TS at time t (an end-point) rather
123 Chapter 3 Climate response and associated risk than the integral of changes up to t, which in effect ‘remembers’ the earlier temperature change (Shine et al. 2005b; Shine 2009).21 As for GWP, GTP is calculated as the ratio of change in TS induced by an emission of 1 kg of a gas relative to that of 1 kg of CO2. Distinction is made between GTP for a pulse emission and one that is sustained over time, wrien GTPP and GTPS respectively. GTPS is then calculated for specified time horizons, such as the 20, 100, and 500 years used for GWP. e value of GTPP at a given time may be strongly dependent on a given emission scenario (due to differing residence times), if a target temperature is taken into account.
A limitation of GTP is the difficulty in applying it to aerosol species, which was only briefly examined in the original paper. Shine et al. (2005b) points to the same problems as calculating GWP for such short-lived species, including inhomogeneous changes to atmospheric concentration resulting from an emission, and the significant uncertainties surrounding radiative effects and interaction with other atmospheric species. ey sug- gest that GTPS would be the appropriate metric, because GTPP would be very small for time horizons beyond a year. While that statement is of course true, GTPS would suffer the same flaws as any metric that compares such short-lived species to the LLGHGs. Shine et al. (2007) provide a more extensive account of GTP for use with short-lived species. ey calculate GTPP for black carbon, while reiterating the substantial chal- lenges in quantifying radiative effect: GTPP ranges across three orders of magnitude, depending on emission scenario, climate sensitivity, and proximity to a target temperat- ure threshold; where that threshold is near, GTPP exceeds 20,000, but even at its smallest
GTPP is above 100. For comparison, similar GTPP values for methane and nitrous oxide range between 0 to 100 and 200 to 300, respectively; the 100 year GTP for methane is only 4, compared to a GWP of 25 (Shine 2009; Fuglestvedt et al. 2010).
21 Fuglestvedt et al. (2010) highlight the fact that the original calculations of GTP (Shine et al. 2005b, 2007) were performed with a simplified model, which did not include a deep ocean and hence understates the ‘climate system’s long-term memory to a pulse perturbation in forcing’, leading to errors in estimated GTP for short-lived species at far time horizons. For this reason they prefer the approach of Boucher and Reddy (2008), who employed a GCM, though this adds complexity to the metric.
124 3.5 Toward appropriate metrics
In their detailed study, Fuglestvedt et al. (2010) calculated comparative GWP and GTPP across a range of time horizons for BC, OC, and sulphate aerosol. ey explain that studies examining regional variation in forcing for these species do not in general agree, producing differing results for the highest metric by region. Consequently, Fuglestvedt et al. (2010) conclude it is premature to determine regional values, choosing instead to use only global means; they further include only direct aerosol forcing effects. e results for GWP and GTPP are reproduced in Table 3.2, where GWP is calculated at the usual 20, 100, and 500 year time horizons, while GTP is for 20, 50, and 100 years. Emissions of SO2 are translated to sulphate (SO4), allowing for lifetime and chemical conversion. e difference between GWP and GTP is large in all cases, underscoring the deliberate choice of metric required for policy making.
Table 3.2: Comparative GWP and GTPP at time horizon H years for BC, OC, and SO2 (reproduced from Tab. 5, Fuglestvedt et al. 2010)
BC OC SO2 (as SO4) GWP H=20 1600 –240 –140 H=100 460 –69 –21 H=500 140 –21 –12
GTPP H=20 470 –71 –41 H=50 77 –12 –6.9 H=100 64 –10 –5.7
Shine et al. (2005b, p. 297) suggest the advantage for policy of GTP over GWP is that it represents ‘an actual (if crude) impact’ rather than the more abstract integrated RF — it is closer to the actual impacts that are the cause for concern (Shine et al. 2007).
GTPS may also be preferable because it represents a sustained change in emissions rather than a single pulse (which of course describes reality); Shine et al. (2005a) show that
125 Chapter 3 Climate response and associated risk
GTPS can also readily be modified to incorporate forcing efficacy for a given agent (see section 2.3.1). Conversely, Shine et al. (2007) examine GTP applied to mitigation policy with specific maximum temperature targets, suggesting that GTPP is appropriate and superior to GWP by virtue of its time-dependency and responsiveness to the chosen policy objectives. Fuglestvedt et al. (2010, p. 4657) also regard GTPP as the appropriate choice for ‘target-oriented’ policy making, noting that GTPS is ‘quite closely related’ to GWP and hence not a real improvement.
3.5.2.2 Specific Forcing Pulse
Bond et al. (2011) created a new metric designed to capture the geographically and tem- porally specific effects of short-lived forcing agents. e specific forcing pulse (SFP) is defined as the energy immediately added to or removed from the Earth-atmosphere system at a receptor geographic region, per emission mass of some chemical species in a source region over that species’ full lifetime. SFP is explicitly restricted to those spe- cies with a lifetime less than one year, the meaning of ‘immediate’ — a species with an e-folding lifetime of 4 months should have 95 % of its effect captured by a one year in- −1 tegral. Units are GJ g for a specific species (S) at a stated emission region (Rj) exerting S effect on a receptor location (E), wrien as SFPE(Rj). An SFP with negative sign denotes energy removed from the climate system.
SFP was developed to accurately express the climatic effects of aerosols and other short- lived species. e restriction on lifetime ‘effectively divides species that have impacts only in the very near future and those for which accumulated burdens are important’ (Bond et al. 2011, 1506). SFP captures the energy added to a region — not power, as RF does — because this beer reflects the short atmospheric lifetime of such species. eir perturbation of the climate system is akin to a ‘burst of energy’ (a delta function),22 not
22 ough their lifetimes are non-zero.
126 3.5 Toward appropriate metrics the sustained flow over time appropriate to long-lived forcing agents (an integral). As an expression of energy change, SFP can be used directly in planetary energy-balance ap- proaches, but Bond et al. (2011) warn that forcing is not equivalent to warming and that there is no simple way to translate SFP to a change in temperature, even on a regionally- specified basis.
Important limitations of SFP are that it considers only the direct effect of forcing agents as measured at TOA, including changes to the cryosphere (snow and ice albedo) — in- direct effects on cloud are not quantified, and these ‘could greatly alter the estimate of impact’ (Bond et al. 2011, 1510). e authors point out the ‘dearth’ of model studies that examine cloud response to emissions rather than total impact,23 which is necessary for any measure to include cloud effects. Presumably there is no a priori reason why cloud or other indirect effects could not be added subsequently, as understanding increases.
SFP offers a policy-relevant metric that quantifies the impact of short-lived species per emission mass, but avoids the need for subjective value decisions such as time horizon, as GWP does. However, Bond et al. (2011) demur from employing SFP in mitigation instru- ments as a trading metric, explaining that ‘it captures impacts that cannot be reflected in globally-averaged, integrated metrics’ (p. 1508).
In developing the SFP, Bond et al. (2011) calculated forcings for black carbon (BC) and organic maer (OM). Table 3.3 lists SFP values as averages for all regions,24 and these are hence temporally and spatially emission-dependent. OM uncertainties are asymmetric.
BC BC SFPatmo is all-region direct atmospheric forcing, SFPatmo+cryo is total including effect on cryosphere. Energy is fossil fuel and biofuel combustion; OM includes organic carbon aerosol species. More than an order of magnitude separates the warming effect of BC per emission mass compared to the cooling of OM, as BC is ‘far more effective at interacting with visible radiation’ (Bond et al. 2011, 1511). BC energy-related SFP varied by around 23 Here they cite the study of Chen et al. (2010) (briefly discussed in section 2.4.2.3 on page 42) that found indirect cloud effects actually outweighed BC positive forcing through absorption. 24 See full Tab. 1 in Bond et al. (2011) for regional variation.
127 Chapter 3 Climate response and associated risk
45 % between regions and by a factor of 4 for OM. Seasonality for BC forcing is marked in the Arctic due to availability of sunlight; open burning SFPBC is greater as burning primarily occurs in the NH summer.
Table 3.3: Specific forcing pulse for black carbon and organic maer (reproduced from Tab. 1, Bond et al. 2011)
−1 BC BC OM (GJ g ) SFPatmo SFPatmo+cryo SFP Average energy 0.96 ±0.46 1.11 ±0.47 –0.042 (–0.02, –0.08) Average open burning 1.16 ±0.63 1.23 ±0.63 –0.074 (–0.03, –0.15) Global average 1.03 ±0.52 1.15 ±0.53 –0.064 (–0.02, –0.13)
Bond et al. (2011) then used calculated SFP to derive global mean RF and GWP for both species, comparing these to the values reported by AR4. Table 3.4 shows these derived values for both the anthropogenic emissions specified in AR4, and an updated inventory (it is not clear whether this is anthropogenic only). Updated emissions are 4.8 Tg BC and 15 Tg OM from energy, and 2.6 Tg BC and 30 Tg OM from open burning — estimated forcing is consequently systematically larger. Bond et al. (2011) note that their estimate of total anthropogenic BC direct RF (+0.40 W m−2) is 18 % greater than AR4 for the same emissions (+0.34 W m−2), which is an improved estimate as it includes beer understand- ing of internal mixing effects causing increased absorption. BC on snow (+0.05 W m−2) is less than the AR4 best estimate of +0.10 W m−2,25 incorporating newer physically-based studies.
At time of writing, no SFP calculated for other cooling aerosol species such as sulphate were evident in the literature, but there is no reason why this cannot be done given the OM|OC precedent.
25 See section 2.5.4.
128 3.5 Toward appropriate metrics
Table 3.4: Comparative RF and 20 & 100 year GWPs for BC and OM, derived from SFP (repro- duced from Tab. 2, Bond et al. 2011)
BC TOA direct BC direct+cryo OM TOA direct
Anthropogenic RF, AR4 emissions (W m−2) Energy related 0.25 ±0.12 0.29 ±0.16 –0.03 (–0.01, –0.06) Open burning 0.15 ±0.06 0.16 ±0.13 –0.10 (–0.04, –0.20) Global total 0.40 ±0.18 0.44 ±0.29 –0.13 (–0.05, –0.25) Total RF, updated emissions (W m−2) Energy related 0.28 ±0.19 0.33 ±0.22 –0.04 (–0.02, –0.08) Open burning 0.19 ±0.07 0.20 ±0.08 –0.14 (–0.05, –0.27) Global total 0.47 ±0.26 0.53 ±0.30 –0.17 (–0.07, –0.35) GWP, 20-year Energy related 2400 ±1600 2800 ±1800 –110 (–40, –210) Open burning 2900 ±1100 3100 ±1300 –180 (–70, –360) Global average 2600 ±1300 2900 ±1500 –160 (–60, –320) GWP, 100-year Energy related 690 ±450 790 ±530 –30 (–12, –60) Open burning 830 ±330 880 ±370 –53 (–20, –100) Global average 740 ±370 830 ±440 –46 (–18, –92)
3.5.2.3 Alternative metrics summary
Table 3.5 summarises selected important characteristics for the GTP and SFP metrics based on the preceding discussion. is set of characteristics is intended to be indicative only as such a summary cannot fully convey the metrics’ complexities and nuance. Filled circles indicate that the metric satisfies the characteristic in question; outline circles indicate the characteristic is partial; crossed circles indicate the characteristic is satisfied but is regarded as a disadvantage for accurate quantification of aerosol cooling.
129 Chapter 3 Climate response and associated risk
Table 3.5: Comparison of selected characteristics of alternative metrics
Characteristic GTPP GTPS SFP Refers to particular explicit geophysical effect • • • Refers to a readily understood impact† • • ◦ Designed to be appropriate for short-lived forcing agents ◦‡ • Expresses effect for point time • • Requires selection of time horizon (integral) ⊗
Calculated effect is relative to GHG (CO2) ⊗ ⊗ Evaluation limited to global-scale effect ⊗ ⊗ Captures spatially heterogenous species • Includes aerosol indirect effects – – – Has been calculated for cooling aerosol species • •
† GTP expresses the relative potential to change temperature, whereas SFP expresses energy added or withdrawn from the climate system, an inherently less understandable measure for non-experts.
‡ As noted in section 3.5.2.1, more recent assessments of GTP indicate GTPP can be useful for short-lived forcing agents.
3.6 Aerosol emission rates
ite obviously, recent trends in anthropogenic aerosol emissions and scenario projec- tions for future emissions are central considerations in evaluating the inherent risk of the aerosol mask. As noted in section 2.4, declines in aerosol emissions in North Amer- ica and Europe over recent decades have oen been balanced by significant increases in Asia, particularly China and India. Indeed, Kaufmann et al. (2011) argue that surging Chinese coal combustion in the past decade led to rapid increases in sulphur emissions and a consequent suppression of temperature increase: Chinese coal consumption re- portedly more than doubled in only four years from 2003 to 2007, where the previous doubling took 22 years; in that four year period, China accounted for 77 % of the 26 % overall rise in global coal use.
A number of studies (eg, Skeie et al. 2011; Unger et al. 2010; Shindell and Faluvegi 2010) produce time series of aerosol emissions as calculated radiative forcing; Figure 2.13 (p. 91) showed one example. However, these are dependent on accurate modelling of the RF
130 3.6 Aerosol emission rates induced by a unit emission, which as we have seen tends to understate indirect effects, or exclude them entirely. Inferred aerosol forcing derived from planetary energy imbalance studies (section 3.2) further suggests that calculations of RF in this manner underestimate the net effect; questions regarding BC effect on cloud (Chen et al. 2010; Spracklen et al. 2011) also add uncertainty.26 I therefore present available data on emissions by mass rather than derived RF, or even SFP, as these give the clearest picture of the underlying physical activity.
Important emission inventories are maintained by a number of groups, such as Bond et al. (2007) for BC and OC (with updates given in the discussion of SFP above) or Smith et al. (2011) for sulphate. Aggregate data for total anthropogenic emissions of primary aerosol or secondary precursor species are examined first, followed by a sectoral profile.
3.6.1 Aggregate emissions
Skeie et al. (2011) provide detailed profiles of aerosol emissions from the preindustrial (1750) to 2010, based on a recent comprehensive dataset compilation combined with chemical transport modelling to derive atmospheric burden. Time series evolution of regional aerosol burdens for the period 1900 to 2010, relative to 1850, are reproduced in Figure 3.6 for sulphate (a), fossil fuel and biofuel (FFBF) derived OC (b), biomass burning
27 (BB) OC (c), SOA (d), and fine-mode nitrate (as HNO3) (e). Figure 3.6 panel () plots the evolution of modelled global mean aerosol burden since 1900, relative to 1750, using data provided in Skeie et al. (2011). e additional species FFBF BC and BB BC are included;
fine-mode nitrate is separated as NO3 and NH4.
26 Refer to the discussion in section 2.4.2.3 and 2.5.2. 27 Figure 3.6 (e) shows the original plot from Skeie et al. (2011) where nitrate is wrien ‘NO3’; how- ever, the text explains nitrate’s complex activation pathway from NH3 and NOx precursor emissions (Figure 3.7), including interaction with ambient sulphate and meteorological conditions, describing the plot as showing HNO3 concentration. See section 2.4.2.4 also.
131 Chapter 3 Climate response and associated risk
Figure 3.6: Modelled aerosol burden over time. (a) to (e) are reproduced with permission from right panels of Fig. 9 of Skeie et al. (2011); () plots data provided in Tab. 2 of Skeie et al. (2011). (a) is regional sulphate (SO4) burden since 1900, relative to 1850; (b) to (e) are as (a), for FFBF OC, BB OC, SOA, and fine-mode nitrate, respectively. () is modelled global mean aerosol burden since 1900, relative to 1750, for FFBF BC, FFBF OC, SOA, BB BC, BB OC, and fine-mode nitrate precursors, in Gg (le axis); sulphate is ploed in Tg S (right axis).
132 3.6 Aerosol emission rates
Figure 3.7 plots the underlying emission data provided by Skeie et al. (2011) as total annual anthropogenic and biomass burning primary aerosol or precursors since 1850.28 Figure 3.7 shows that emissions of many aerosol species are at or close to their historical record maxima. Nitrate precursors, BC and OC emissions are substantial, especially from biomass burning for the laer carbonaceous species. Sulphate has declined from its peak in the 1970s and 1980s, but remains elevated; coal combustion in India and China could possibly see a returning upward trend. Importantly, no species shows a significant downward trend over recent decades.
25 70
FFBF BC (Tg) BB BC (Tg) 60 FFBF OC (Tg) 20 BB OC (Tg) 50 NOx (Tg N) 15 NH3 (Tg N) SO2+SO4 (Tg S) 40
30 10
20
5
Carbonaceous aerosol emissions (Tg) 10
0 0 Non−carbonaceous aerosol emissions (Tg N, Tg S) 1850 1900 1950 2000
Figure 3.7: Total annual emissions of major aerosol species since 1850. Data provided by Tab. 1 of Skeie et al. (2011). Carbonaceous species are in Tg annual emissions (le axis), non-carbonaceous species are in Tg N and Tg S (right axis).
Deriving aerosol emission inventories is a difficult task involving complex and disparate data sources and a range of decisions regarding calibration and the like. e various available inventory products from different research groups do not necessarily agree, even within uncertainty bounds (which may be large for some species, where they are
28 e dataset begins at 1750, but as the next time increment is 1850 (ie, a 100 year gap), 1850 has been set as the base year for consistency with the previous plots.
133 Chapter 3 Climate response and associated risk quantified at all; BC is a factor of 2). Granier et al. (2011) note that agreement is best for SO2 and NOx but remains low for BC; agreement in global aggregate is also not ne- cessarily seen in regional contributions. Research into understanding and reducing the differences among inventories is an ongoing challenge. For instance, the total emissions listed by sector in the assessment of Unger et al. (2010, Supplemental) diverge markedly from those shown in Figure 3.7: total OC is some 40 % greater and sulphate is around 30 %
29 greater, while BC is within a few percent. e dedicated anthropogenic SO2 emissions study of Smith et al. (2011) is close to agreement with the data from Skeie et al. (2011) ploed in Figure 3.7, giving a 2005 global total of 115 Tg SO2 (119 Tg SO2 including open burning), or ~58 Tg S (~60 Tg S including open burning).
3.6.2 Emissions by source
Global total aerosol emissions probably maer most for the climate (though with strong regional variation), but proportional emissions by source maer most in assessing the effects of mitigation action. However, inventories by sector are somewhat difficult to obtain. e data provided by Unger et al. (2010, Supplemental) provide emission data by source for a number of key species, but the totals do not agree well with global aggregate emissions, as noted above. An integrated inventory of year 2000 historical emission data by sector is available from the Representative Concentration Pathways Database (v2.0), combining a number of existing inventories and considering differences between them.30 ese data do closely match the totals of Skeie et al. (2011); they are presented as percentages of the per-species total in Table 3.6, giving a broad indicator of relative sectoral importance.
29 Using a factor of ~0.5 to convert Unger et al. (2010) Tg SO2 molecular mass to Tg S. 30 e inventory compilation and assessment of differences among datasets is described by Granier et al. (2011); see van Vuuren et al. (2011c) for a general overview. e dataset used (‘ID - RCPINV’ year 2000 historical data) and future emissions under the RCP scenarios are available online at http://www.iiasa. ac.at/web-apps/tnt/RcpDb (accessed June 2012). Representative Concentration Pathways are discussed in section 3.6.3 below.
134 3.6 Aerosol emission rates
Table 3.6: Proportional anthropogenic emissions of major aerosol species by sector, year 2000 emissions. Major emission sectors are highlighted for each species. Data obtained from the RCP Database (v2.0) ID - RCPINV historical inventory via the online com- parison tool.
Sector SO2 BC OC NOx NH3 Surface transportation 4.0% 17.3% 4.0% 28.6% 1.0% International shipping 10.3% 1.7% 0.4% 14.2% 0.0% Aviation 0.0% 0.1% 0.0% 2.2% 0.0% Power plants, energy conversion, extraction 49.2% 0.7% 1.0% 18.6% 0.1% Solvents 0.0% 0.0% 0.0% 0.0% 0.0% Waste (landfills, waste water, incineration) 0.1% 0.5% 0.1% 0.2% 0.0% Industry (combustion and processing) 25.0% 19.3% 6.3% 12.6% 0.3% Residential and Commercial 7.7% 25.1% 21.6% 7.2% 0.7% Agriculture (waste burning on fields) 0.2% 1.9% 1.9% 0.5% 1.3% Agriculture (animals, rice, soil) 0.0% 0.0% 0.0% 1.7% 73.8% Savannah burning 1.5% 19.1% 30.1% 9.2% 8.5% Land-use change (deforestation) 2.1% 14.5% 34.5% 4.9% 14.3%
Table 3.6 shows that many sectors that produce a significant fraction of anthropogenic greenhouse gas emissions also contribute substantially to the aerosol mask. Energy con- version, particularly electricity generation, factors strongly, as does surface transporta- tion, industry, and the residential and commercial sectors. Further, all of those sectors are (or are likely to be) subject to the first deployments of GHG mitigation policy instru- ments.
Agriculture and biomass burning are also important aerosol sources, but these sectors are more difficult to measure and their inclusion in mitigation instruments such as emissions trading is less likely, at least initially; they are more likely to be covered by specific schemes such as programs that seek to avoid deforestation.
Given the large likely forcing of sulphate, SO2 emissions deserve special aention. Smith et al. (2011) report that the proportion of global SO2 emissions arising from fossil fuel combustion remains about 50 % from coal and 30 % from petroleum. Figure 3.8, repro- duced from Smith et al. (2011), shows the evolution of global SO2 emissions since 1850. ese proportions are in general agreement with Table 3.6, which also highlights the
135 Chapter 3 Climate response and associated risk
Figure 3.8: Global SO2 emissions since 1850 by source and by end-use sector (Fig. 5, Smith et al. 2011) large contribution of industry. However, Ramanathan and Feng (2009) state that coal contributes about 78 % of total SO2 emissions, and oil a further 10 %. From a final-use perspective, SO2 emissions from energy conversion — mainly electricity generation — have risen steadily over the past century to now account for about half of the global total.
For all the finger pointing at rising Chinese emissions, it should be noted that in 2000
Chinese coal-fired electricity generation emied some 5.4 Tg SO2 — less than half of the
US at 12 Tg SO2 (Shindell and Faluvegi 2010). ough as the inventory of Smith et al.
(2011) shows, US SO2 emissions have declined significantly since their peak around the mid ’70s while China’s have risen dramatically in the past two decades, the US remains a major SO2 emier (Ramanathan and Feng 2009).
3.6.3 Future emissions
e range of possible emissions futures in the IPCC Special Report on Emissions Scenarios (SRES) (Nakićenović et al. 2000) played a dominant role in the decade since they were
136 3.6 Aerosol emission rates
Box 3.2: Australian aerosol emissions
As an Australian, it is only right to make brief mention of our national emissions. e National Pollution Inventorya (NPI) data for 2010/11 show that electricity generation is the largest single source category of SO2 emissions at around 44 % of the 1.3 Tg SO2 total, followed closely by non- ferrous metal manufacturing at 43 %, then a large gap to sub-reporting threshold fuel combustion at 3 %. Oxides of nitrogen include aerosol precursor NOx emissions, but also the GHG N2O, making useful comparison difficult at the level of data published. With that caveat, electricity generation accounts for 30 % of the 1.4 Tg total, followed by diffuse motor vehicle emissions on 24 % and diffuse burning on 9 %; coal mining accounts for 5 %. Diffuse burning accounts for about 85 % of the 147 Mg HNO3 emissions to air (as nitric acid), with fertiliser and pesticide manufacturing another 13 %. e NPI does not currently provide data on BC or OC emissions. However, electricity generation accounts for 40 % of non-specific fine particulate maer up to 2.5 µm diameter, coal mining 17 %, and metal ore mining 15 %. For particulate maer diameter 10 µm or less, coal mining contributes 23 %, metal ore mining 21 %, disperse burning 19 %, and disperse windblown dust 15 %. Overall, LLGHG emission sources also account for the bulk of Australian aerosol-related emissions.
a Available at http://www.npi.gov.au.
published in 2000. A pivotal feature of all such scenarios is that they do not incorporate explicit climate change mitigation policies, by design. ey instead represent a range of possible futures where deliberate action to abate GHG emissions is not undertaken; or at least, not undertaken for climate purposes. Further, Cofala et al. (2007) make the important point that in comparison to this extensive scenario analysis of future GHG emissions, similar assessment of future aerosol and precursor emissions is ‘comparably modest’ (p. 8486).
e lack of mitigation in SRES is clearly highly problematic, given the urgent need for abatement if catastrophic climate changes are to be avoided. New scenarios are needed that do explicitly examine mitigation (such as the 2 ℃ maximum temperature increase target), and that also include recent research in areas such as economics, sociology, and technology, as well as the physical science (Moss et al. 2010). In 2007, the IPCC requested new scenarios be developed that met this need (van Vuuren et al. 2011c).
at effort culminated in the Representative Concentration Pathways (RCPs), scenarios
137 Chapter 3 Climate response and associated risk that will be assessed in the IPCC’s Fih Report (van Vuuren et al. 2011b). Each RCP signifies one possible emission scenario that would lead to a specific characterised ra- diative forcing, ‘representative’ of the set of scenarios in the literature that corresponds with that RF. And each should be a ‘plausible and internally consistent description of the future’ (van Vuuren et al. 2011b, 7–8). e scenario is a ‘pathway’, emphasising that the emission trajectory is of interest and not only atmospheric concentrations (Moss et al. 2010); however it is atmospheric LLGHG concentrations not emissions that are the basic input to climate models (van Vuuren et al. 2011b).
Taken together, the RCPs cover the range of year-2100 RFs canvassed by the literat- ure, from 2.6 W m−2 to 8.5 W m−2, and are named according to that RF: RCP8.5, RCP6, RCP4.5, and RCP2.6 (van Vuuren et al. 2011c). RCP2.6 is representative of scenarios that seek to restrain global mean surface temperature rise to 2 ℃ above the preindustrial, re- quiring a 70 % reduction in cumulative LLGHG emissions between 2010 and 2100 relative to the baseline no-mitigation scenario (van Vuuren et al. 2011a).
Tanaka and Raddatz (2011) note that SO2 emissions decline faster in RCP than under SRES. Indeed, the total anthropogenic aerosol burden is projected to ‘strongly decrease’ in this century under all RCPs, and also in the baseline scenario (van Vuuren et al. 2011c, 3). ese declines are in part due to assumed air pollution control policies becoming more stringent over time, as well as abatement stemming from climate change mitiga- tion (van Vuuren et al. 2011b). SO2 emissions drop substantially under RCP2.6 as coal combustion decreases, for example (van Vuuren et al. 2011a).
Figure 3.9 shows the projected decline of aerosol species under all four RCPs.
138 3.6 Aerosol emission rates
Figure 3.9: Projected anthropogenic aerosol emissions 2000 to 2100 under all RCP scenarios for total BC, OC, nitrate precursors, and sulphate (as SO2). RCP8.5 (brown line), RCP6 (red line), RCP4.5 (blue line), RCP2.6 (green line), and year 2000 emissions ID - RCPINV historical inventory (black marker). Images compiled from the RCP Data- base (v2.0) online comparison tool.
139 Chapter 3 Climate response and associated risk
ere are two problems inherent in an RCP-based assessment of the risk of a significant loss of aerosol cooling under climate change mitigation at this time. First, the GCM runs that assess the likely temperature response to a given RCP are, as noted above and in section 2.7, subject to persistent uncertainties in representation of aerosol indirect ef- fects,31 and may be underestimating the resultant net negative forcing. Added to this is the uncertainty in actual current emissions, and the generally unsatisfactory quantific- ation of aerosol burden, optical depth, and so on. ose issues must be redressed, and consensus on an accurate modelling of the full range of aerosol effects established, if nasty surprises stemming from a loss of aerosol cooling are to be ruled out.
Second, given the current state of international climate change mitigation negotiations and continuing record GHG emissions, there is no guarantee that any such pathway will actually be followed. It is surely a very real possibility that emissions continue un- abated for some years to come, only to see a dramatic and precipitous decline driven by a step-change in political response, as discussed in Chapter 4. Moreover, fossil fuel consumption continues to grow, particularly exploitation of unconventional oil in the form of tar sands and surging production of unconventional gas in the form of shale gas or coal seam methane, especially in the US. Added to that is the increasingly apparent scale of fugitive methane emissions in those extraction and processing operations. And Australian coal exports are undergoing enormous expansion, with much more planned. e point here is that the actual shape of future emissions of both GHG and aerosol species is far from certain and cannot be assumed to follow any RCP. Consequently it cannot be assumed that aerosol emissions will follow a path that precludes realisation of a short-term temperature spike when serious abatement action does finally begin. Fi-
31 In the first model results of future GHG concentrations and temperature response presented with the RCP Special Issue of Climatic Change (see van Vuuren et al. (2011c) for an overview), Meinshausen et al. (2011, 235) note that ‘… uncertainties arise as well from second-order effects of tropospheric ozone and aerosols, for example. To the extent that our non-GHG modeling assumptions deviate from current generation chemistry modeling results …, indirect effects via global-mean temperature and gas-cycle feedbacks will impact derived GHG concentrations presented here, although this indirect second-order effect is likely limited.’
140 3.7 Policy implications
nally, even RCP2.6 equates to a peak atmospheric concentration of some 490 ppm CO2-e by 2050 — 440 ppm CO2 alone — and remains above 400 ppm CO2 at 2100, not return- ing below current levels until about 2300 (Meinshausen et al. 2011; van Vuuren et al.
2011b). ese levels are beyond the 350 ppm CO2 long-term maximum suggested by the paleoclimatic evidence as necessary to avoid catastrophic climate change this century (Hansen et al. 2008). erefore, the chapters to follow make no specific reference to projected emission trajectories, but instead tend to assume something of a worst case scenario for mitigation, characterised by a seriously delayed onset of abatement and the resultant risk of triggering dangerous anthropogenic changes without deliberate use of an aerosol-integrated mitigation framework.
3.7 Policy implications
Several important implications for the climate change mitigation policy response may be distilled from this chapter. e conclusions stemming from the planetary energy imbalance studies examined above are threefold:
• Planetary energy budget is highly relevant to mitigation policy development.
• e climate response function derived from energy balance studies may be more rapid than commonly appreciated or indeed modelled; however investigations into associated complexities are ongoing and no clear consensus currently exists.
• e aerosol mask is substantial and ongoing, offseing as much as half of extant anthropogenic GHG forcing.32
Planetary energy imbalance can therefore provide a robust metric for international cli- mate negotiations, indicating as it does both the global magnitude of remaining forcings and framing the consequent time window for mitigative action. ese aspects are not
32 However, GCM studies tend to find an offset around 35 %, as shown in section 2.6.
141 Chapter 3 Climate response and associated risk well communicated by radiative forcing. Planetary energy imbalance also provides fur- ther evidence that the aerosol dilemma cannot be ignored if mitigation is to be effective in avoiding ‘dangerous anthropogenic interference’.
Moreover, the difficulties in quantifying aerosol emissions and adequately representing their cooling effects with the standard metrics of GWP and CO2-e strongly indicate that new alternatives are required for any aerosol-integrated policy framework. e specific forcing pulse appears to meet that need admirably.
Finally, emission inventories, though subject to persistent uncertainties and gaps in cov- erage, show that global anthropogenic aerosol emissions are large and ongoing. How- ever, the relative shis in regional burden — falling in the industrialised countries of Europe and North America, but rising strongly in the developing world — are likely to present significant challenges for international negotiation of any aerosol-integrated UNFCCC Protocol.
142 4 Comparative analysis of mitigation policy
is chapter examines the ontology, form, and scope of mitigation policy approaches applied to the domains of climate change and local air pollution. In particular, it analyses the interrelation and overlap between these two domains, their commonality and conflict in objective and practical implementation. Most importantly, it assesses the degree to which the climate change implications of aerosols are, or might be, integrated within these frameworks.
4.1 Pollution taxonomy
A consistent taxonomy is necessary to effectively examine the inclusion of aerosol emis- sions within climate change mitigation policies, particularly for comparison with GHG and air pollution regulation. Key differentiating characteristics for a given pollutant are rates of assimilation by environmental sinks, the degree of atmospheric mixing, and the types of environmental effects consequent to emission. Whether a particular substance is regarded as a pollutant depends on the combination and interaction of these charac- teristics.
143 Chapter 4 Comparative analysis of mitigation policy
4.1.1 Emission and assimilation
A substance is emied to the atmosphere from some originating physical process or mechanism, termed the source of emission; a particular geographical emission site is commonly termed a point source. Once emied, the substance may1 be affected by some environmental process that absorbs or otherwise aenuates that substance over time, termed a sink. In the troposphere, sink processes are usually chemical reactions, precipitation, or gravitational deposition (Almqvist 1974). rough such biogeochemical sink processes, emissions are assimilated by the environment and, if those processes are not overwhelmed, the substance may be returned to a ‘useable source’ (Daly and Farley 2011, 422).
However, any such environmental sink will have a finite capacity to perform that func- tion over time, likely including both a maximum rate of absorption and a maximum cumulative limit (which itself will be related to the time frame in question). Available sink capacity, along with the environmental impact of an emission (examined below), is therefore an important characteristic in defining whether a given atmospheric emis- sion is regarded as a pollutant. Further, the atmosphere itself — and ambient levels of a pollutant within it — is the site of ‘contamination’ and not an actual sink in this context.2
Pollutants are usefully categorised with respect to their emissions quantity relative to en- vironmental sink capacity. An assimilative or flow pollutant is defined as one whose rate of emission can readily be absorbed (assimilated) by available sink capacity: a given year’s ambient pollution level is independent of the quantity emied at previous times as those past emissions have already been absorbed, hence the pollutant does not accu- mulate in the environment (Tietenberg 2006; Harris 2006). Whether such a substance is
1 Not all substances have naturally available sink processes, such as many of the halocarbons which exist only as products of human industry. 2 Many gas species are permanently resident in the atmosphere in some background concentration (a steady state). ey become pollutants when some threshold is passed, giving rise to environmental impact.
144 4.1 Pollution taxonomy regarded as a pollutant then depends on its environmental effect at this ambient load- ing, despite the lack of accumulation; ie, damage from pollution does not require that the pollutant accumulates.
Expressed symbolically, an assimilative pollutant can be wrien as
∑J At = at + ejt (4.1) j=1
at ≥ 0; ejt ≥ 0 j = 1,...,J (4.2)
Where At is the steady-state ambient pollution loading at time t (usually a given year), at is a random variable representing any natural ambient emission level, and ejt is the emission rate of the jth anthropogenic point source at time t, in the set of all such point sources J. Many traditional air pollutants are assimilative, such as volatile organic com- pounds, and the aerosol species discussed in this thesis.
Conversely, an accumulative or stock pollutant has an emission rate in excess of avail- able sink capacity, or no sinks exist: emissions are greater than can be assimilated and so the pollutant accumulates in the atmosphere. e pollution level at a point in time is the sum of the current emissions rate plus some fraction of all previous emissions, account- ing for the decay rate of natural sink removal processes. Symbolically, an accumulative pollutant can be wrien as the change in atmospheric loading L at time t
∑J ∆Lt = αt + ejt − βt (4.3) j=1
and βt = f (L, s) (4.4)
Where ejt is anthropogenic emissions as before, αt is a random variable representing any natural emission rate at time t, and βt is any natural assimilation of that pollutant during time t. βt represents the pollutant-dependent decay rate which is a function of
145 Chapter 4 Comparative analysis of mitigation policy the loading L, and s, a term representing any dynamic and feedback changes in sink capacity, and hence may also vary over time.
If no assimilative capacity exists, then βt is zero and Lt is simply the sum of all emissions. e total cumulative pollutant loading L at time t is therefore, by definition
Lt = Lt−1 + ∆Lt ∑J Lt = Lt−1 + αt + ejt − βt (4.5) j=1
e enhanced greenhouse effect of carbon dioxide and other greenhouse gases is a cent- ral example of accumulation. In the Earth-atmosphere system at current boundary con- ditions, natural CO2 emissions are large but, in the absence of significant anthropogenic emissions, are balanced by the natural removal processes of the carbon cycle. On av- erage αt is therefore cancelled by βt, such that their expected combined value is zero
(E [αt − βt] = 0), and Lt is thereby in a steady-state on time scales of centuries to tens of millennia.
at steady-state will persist until a shi in boundary conditions occurs, sufficiently modifying natural emissions or sink absorption such that Lt decreases or increases over time (non-anthropogenic perturbations are the Milankovitch orbital forcings, large vol- canic events, changes in continental configuration in geologic timescales, and the like).
Further, the carbon cycle responds under conditions of elevated total CO2 emission, increasing net absorption by sinks, meaning that the βt function strengthens (greater decay) to partially offset those elevated emissions.
To date the airborne fraction of anthropogenic CO2 emissions — that fraction of an emis- sion pulse remaining in the atmosphere in the years following — is holding at about 40 % to 55 % (Le éré et al. 2009; Knorr 2009; Solomon et al. 2009; Bes et al. 2011; Hansen
146 4.1 Pollution taxonomy et al. 2011).3 However, this fraction is unlikely to persist indefinitely under future cli- mate changes, as sinks have finite capacity, are subject to counteracting feedbacks (eg, ocean CO2 outgassing with increasing surface temperature), and hence must eventually weaken.
4.1.2 Atmospheric mixing
Pollutants are further categorised by the extent of mixing within the atmosphere once emied. Mixing is a result of the interaction of factors such as the rate of removal by sink processes (ie, atmospheric residence time for an emission pulse), the physical properties of the pollutant itself, and atmospheric transport through chemical reactions and the physical processes of advection and convection.
A pollutant may be uniformly mixed in the atmosphere within some geophysically meaningful time period, such that the ambient loading is generally equivalent (uniform) at all geographic locations, allowing for transport processes to distribute the initial emis- sion pulse. e ambient pollutant loading is therefore independent of the geographical distribution of the underlying emission sources. As such, a uniformly mixed pollutant can be represented by the previously defined eq. 4.1 and eq. 4.5, for the purposes here. e greenhouse gases are, again, central examples of uniformly mixed pollutants.
A non-uniformly mixed pollutant, in contrast, produces heterogeneous ambient load- ings that vary with geographic location and strength of emissions. e interaction of atmospheric transport and removal processes are such that the ambient pollutant load- ing at any location is dependent on proximity to and emission rate of particular point sources. Ambient loadings are hence highest at point source sites with greatest emission rate, and decline with increasing distance from emission sources. However, transport
3 Of these, Hansen et al. (2011), citing Hansen et al. (2008, Supplemental), is the largest estimate at 55 %; Solomon et al. (2009) gives ~50 %; 40–45 % is the more common.
147 Chapter 4 Comparative analysis of mitigation policy processes and the topography of a particular geographical area may mean that the site of highest ambient loading is some distance from the ultimate emission source. Again, the majority of traditional air pollutants are non-uniformly mixed, such as tropospheric ozone or sulphur dioxide.
Symbolically, a non-uniformly mixed pollutant can be expressed as an extension of the assimilative representation in eq. 4.1 above
∑J Ait = ait + τij ejt (4.6) j=1
ait ≥ 0; 0 ≤ τij ≤ 1 i = 1,...,I (4.7)
Where Ait is the ambient pollution loading occurring at time t at a geographic receptor site i in the set of all considered receptor sites I, ait is a random variable representing any natural ambient pollutant loading incident at that site, ejt is the emission rate of the jth point source as before, and τij is an atmospheric transport coefficient that represents the increase or decrease in pollutant loading at site i due to emissions originating at point source j.
Whether or not any instances of a non-uniformly mixed accumulative pollutant are known, such a pollutant class is not relevant for the purposes here and will not be dis- cussed further.
4.1.3 Environmental impact
A pollutant is cause for concern due to the effect it induces in the environment; the presence of such an effect is a defining characteristic that in fact makes emission of a substance ‘pollution’. Generally speaking, the environmental effect of emission must first be discernible by human societies before it will be regarded as a pollutant and any policy response enacted. However, a failure to identify some substance as a pollutant in
148 4.1 Pollution taxonomy no way precludes the existence of environmental effects, possibly significant ones — the history of the discovery of and policy response to anthropogenic climate change over the past 150 years amply demonstrates this.
Importantly, a pollutant’s effect on the environment is characterised as deleterious in nature, at least as traditionally considered. at is to say, pollution causes negative, damaging, undesired effects, either to humans indirectly through damage to the envir- onment itself (such as smog, acidification, eutrophication) or directly to human health or selements (such as respiratory illness, infrastructure deterioration). In economic terms, these damaging effects make pollutants an environmental bad.4 Further, a pol- lutant may induce more than one single environmental effect, such as CO2 acting both
5 as a greenhouse gas and acidifying the oceans, or SO2 causing acid rain as well as par- ticulate impacts to human respiratory health. ese effects are also commonly known as the environmental impacts of pollution.
e impacts of pollution may not manifest for any and all emission levels: some threshold may first need to be passed beyond which damage begins to occur, or is greater than some socially tolerable level. Consideration of effects that might be apparent but non- etheless regarded as socially acceptable therefore distinguishes impact, as perceived by society, from the more technical distinction between assimilation and accumulation with respect to environmental sinks. is subjective assessment of acceptability forms part of the concept of a ‘socially optimum’ pollution level, where the purported economic costs of mitigation are weighed against the estimated economic cost of environmental damages. A risk management approach would further refine that socially optimum pol- lution level, taking account of uncertainties in both the costs of damage and of mitigation controls. e negative environmental impacts of pollution are well established in policy
4 An economic bad is the opposite of a normal good, in that the object itself has negative value. 5 Ocean acidification demonstrates a higher degree of complexity, in that the ocean is itself an envir- onmental sink for excess (disequilibrium) atmospheric CO2 concentrations while simultaneously being damaged by that very sink function.
149 Chapter 4 Comparative analysis of mitigation policy making and in the regulatory responses that seek to mitigate them.
4.1.4 Applying the taxonomy
Given the codified taxonomy above, we can now compare the properties of aerosols and the long-lived greenhouse gases as pollutants. Table 4.1 lists the characteristics defined above, indicating which is most appropriate for the two aggregated pollutants. at is, LLGHGs and aerosols are shown here as generalised classes of pollutants to demonstrate their dominant common characteristics,6 rather than examining particular species individually. Atmospheric residence time is also restated, to underscore the large divergence in this property between the two classes, as well as its link to the emission– sink function. Finally, impacts are shown at a very high level, simply to differentiate the two broad domains of climate change and local air pollution.
Table 4.1: Comparative dominant pollutant characteristics
Characteristic Aerosols LLGHGs Assimilative • Accumulative • Uniformly mixed • Non-uniformly mixed • Residence time (yrs) 10–2 101 to 104 Impact Climate change damages ◦⋆ • Air pollution damages •
⋆ BC causes positive forcing; aerosols affect the hydrological cycle.
6 Recall from section 2.1.3 that other GHG species such as tropospheric ozone are non-uniformly mixed; moreover, ozone is a significant cause of local air pollution damage.
150 4.1 Pollution taxonomy
4.1.5 A new term for emission sources
One further consideration regarding the emission of pollutants across these two domains is some indication of the common sources for both GHGs and aerosols. at is, in many instances anthropogenic emissions of these two classes stem from the same underlying physical process or industrial activity. Identifying common sources is policy-relevant, because abatement action that affects one pollutant class will necessarily also affect the other if that action targets a single activity that emits both. Policy design should account for such corollary effects. It is hence useful to define a term to indicate instances of common emission source, but none is readily apparent in the literature.7
Given that emissions of the two pollutant classes are tightly coupled to the underlying physical process, I suggest a new term: coupled emissions are the simultaneous emission of both GHG and aerosol (primary or precursors) pollutant species from a single physical activity, a biological or chemical process not readily separable to independent parts, where the quantity of each pollutant class emied is climatically-significant.
Combustion of fossil fuels across a range of technological applications is the archetypal example; though the specific emission profile for a given activity depends on fuel type and the technology employed. Coal-fired electricity generation produces, inter alia, sig- nificant quantities of both LLGHG (CO2) and aerosol (BC, SO2) emissions. Natural gas combustion, in contrast, emits lile to no aerosol species, so coupled emission cannot be read as synonymous with fossil fuels. Indeed, biomass combustion also produces coupled emissions of CO2 along with varying combinations of BC, OC, and other aerosol species (see section 2.4).
7 Caplan and Silva (2005) define a related term, ‘correlated externalities’, but this signifies that a com- mon emission source generates multiple pollutants that have differing externalities. at situation of course describes aerosol and GHG emissions well and is germane to the discussion here, but it refers to the externalities themselves rather than more neutrally pointing specifically to the fact of a common emission source.
151 Chapter 4 Comparative analysis of mitigation policy
4.2 Policy ontology and form
e principal implication of the aerosol cooling dilemma laid out in the preceding chapters is that a specific policy response is required, integrating the full effects of aerosol emis- sions on climate. An obvious question arising from this implication is what should that response look like; what policy instrument should be applied? Clearly, the first place to look for an appropriate instrument is the existing frameworks employed within the climate change and air pollution domains. What may not be so clear is that the form and the philosophical underpinnings of these pollution control policies have a direct bearing on their ability to handle the problem. To appreciate why this is the case requires an exposition of the character of those policies — the ontology and assumptions, both ex- plicit and unstated, underlying their objectives and the regulatory instruments deployed to fulfil them.
4.2.1 Less is beer
e overarching intent of any policy is defined by its stated objective. Pollution con- trol policy-making has traditionally been approached with, at heart, the clear intent to mitigate the environmental damage caused by some offending pollutant: the objective is to mitigate those damages; the means is a reduction in pollutant emissions. Trade- offs will complicate the situation and are likely to result in an actual emission target being non-zero. A prominent example is the economic costs of mitigation versus the economic benefits stemming from the underlying production processes that ultimately generate those emissions (ie, pollution is an archetypal negative production externality). In conventional economic analysis, these benefits would be weighed against mitigation costs using the equimarginal principle.
However the purpose here is to tease out the underlying ontological framing of pollution control policy. So, seing aside complications such as the presence of a minimum effect
152 4.2 Policy ontology and form threshold and real-world considerations of mitigation costs, philosophically there is no a priori reason for a non-zero target emission level, and hence the basic policy objective is to reduce emissions as much as possible. Stated in the negative: the only reason pollutant emissions would not be reduced is precisely those real-world considerations of the costs of abatement technology (which assumes any is in fact available), a loss of benefits from the underlying productive activity, or a minimum geophysical response threshold. Yet even so, should a new abatement technology become available at lower cost, for example, the target emission level is likely to be reduced in consequence. In simplistic language to make this point clear then, the objective of pollution control policy can be crudely characterised as the less emissions the beer.
Particular regulatory instruments are then applied in the service of that objective, spe- cifying various measures such as control technologies or economic regimes that regulate these production processes so as to achieve the emission target. And, of course, these instruments will take into account the real-world considerations set aside in the previ- ous statement. In the simplest case, a single assimilative pollutant generating a single environmental impact might be regulated by a single mitigation measure, such as partic- ular control technologies prescribed for all liable emiers — SO2 abatement via sulphur scrubbers to mitigate acid rain is a seminal example. Increasing regulatory complexity is required in accordance with increasing complexity of the pollution problem: hand- ling non-uniform or accumulative pollutants; multiple individual pollutants, or multiple geophysical damage effects across a spectrum of impact areas (such as eutrophication, acidification); and combinations thereof.
ese more complex forms may be classified as ‘multi-pollutant, multi-effect’ problems (eg, EEA 2004). Taken together, these two dimensions — pollutant types and environ- mental effects — can be interpreted as constituting a general classification schema for pollution control, albeit somewhat informally. at is, a given pollution problem is clas- sified by those dimensions in a rough spectrum, ranging from single-pollutant, single-
153 Chapter 4 Comparative analysis of mitigation policy effect, to multi-pollutant, multi-effect, indicating increasing complexity. e policy ap- proach applied is hence potentially more complex in turn, though that is not necessarily the case.
A fundamental characteristic of even the most complex multi-pollutant, multi-effect policies is that they hold the overarching objective of mitigating negative environmental impacts. Relating this classification schema to the ontological frame, it is therefore axio- matic of pollution control policies that all targeted effects are deleterious in nature, even if this axiom is only ever implicitly assumed. As such, the axiom of deleterious con- sequences is a foundational component of pollution control policy: they are conceived as, and designed to achieve, the reduction of pollutant emissions. is is their raison d’être: less is beer.
But what if less is not beer? What if less is in fact worse because a key environmental effect is actually beneficial in some vital aspect, such that reducing pollution may ex- acerbate, not mitigate against, some other negative environmental impact?
4.2.2 Less is worse
In the aerosol cooling dilemma, less may indeed be worse; that is the crux of the problem.
Can aerosol effects on climate in the contemporary context be properly understood within the ontological frame that less is beer? I contend that they cannot, because aerosol masking violates that frame’s implicit axiom of deleterious consequences. at is, despite the clear and serious damaging effects of aerosol emission to ecosystems, hu- man selements, human health, and indeed, the climate8 — all of which are negative impacts to human welfare and environmental bads — aerosols simultaneously gener- ate beneficial effects in their masking of the enhanced greenhouse effect, dampening
8 Such as changes to precipitation paerns; see Box 2.2.
154 4.2 Policy ontology and form the global warming that would have occurred in their absence — a contingent environ- mental good and positive welfare impact, relative to that enhanced greenhouse effect. Table 4.2 aempts to illustrate this benefit by extending Table 4.1 above to contrast the actual environmental effects of the two pollutant classes:9 alongside all their undoubted and serious damaging effects, aerosols also generate a large net negative climate forcing.
Table 4.2: Comparative pollutant effects
Effects Aerosols LLGHGs Positive radiative forcing ◦† • Negative radiative forcing • Ocean acidification •‡ Air pollution damages •
† Black carbon has positive atmospheric radiative forcing and perturbs snow and ice albedo.
‡ From CO2.
In the contemporary situation confronting humanity, anthropogenic aerosol emissions cannot correctly be characterised as causing only negative impacts, as pollutants histor- ically have been. e benefit of aerosol masking exists in a very real sense; it cannot be dismissed, and the mitigation policy response must be capable of fully incorporating it. But the contingent nature of this benefit must be heavily emphasised: first, the value of the aerosol mask is entirely predicated on the prior existence of significant anthro- pogenic climate change being well underway; second, the confounding implications of that mask for mitigation actions affecting coupled emission sources are certainly not positive, as we shall see.
For if current atmospheric aerosol loading masks as much as half of the warming effect of current atmospheric GHG concentrations, as the evidence canvassed in Chapter 2 and Chapter 3 strongly indicates, is it prudent to continue the pursuit of air pollution or climate change mitigation action that assumes removing those aerosols will increase
9 e same qualifying caveat of fn. 6 on page 150 also applies here.
155 Chapter 4 Comparative analysis of mitigation policy net human welfare? Should the risk of significant temperature increase inherent in sub- stantive declines in aerosol emissions not be carefully considered for mitigation policy design? Fundamental changes in mitigation instruments may well be required in light of that consideration. And so policy making must adopt an ontological frame that at least permits the case where reductions in pollution do not add to net human welfare but rather harm it — less pollution can under certain circumstances in fact be worse.
4.2.3 The policy challenge
A full accounting of aerosols’ role within anthropogenic climate change thus presents a serious ontological challenge for policy making: policy objectives and the instruments designed to achieve them must be constructed from the ground up to allow the case where less is worse; and indeed, where less is both beer and worse simultaneously, reflecting the conflicted multi-dimensional nature of aerosols’ place within the climate and air pollution domains. Policy making must be reconceived to incorporate the im- plications of the simultaneous existence of both damaging (air pollution) and beneficial (climatic) effects, and their dynamic co-dependencies such as the reality of coupled emis- sion sources. Above all, climate change mitigation policy making must recognise that reducing aerosol emissions is no longer necessarily prudent nor desirable without fur- ther deliberate and coordinated compensatory measures; or at the very least, the option of doing so should the geophysical evidence require it.
It is difficult to identify another instance of policy making under such conditions. is challenge is distinct, and markedly harder, from more familiar trade-offs between what appear to be opposing objectives, such as short-term economic profit versus the preven- tion of environmental damage. Here the objectives themselves are essentially congruent, not opposed, yet seem to demand contradictory actions to meet them despite their com- mon purpose in seeking to avoid environmental damage and raise human welfare. is
156 4.2 Policy ontology and form contradiction operates on two levels: first, the conflict between air pollution objectives that call for reductions in aerosol emissions, and climate change objectives that now may have to prevent that from occurring; and second, the same conflict arising from the existence of coupled emissions, as even reducing GHG emissions themselves will in key instances also reduce aerosols. erefore, the mandating of aerosol emission abatement in the service of an objective of mitigating air pollution damages may now confront a parallel need to prevent a totally unmanaged loss of those same pollutants in serving the objective of averting dangerous climate change.
e full array of environmental effects at the nexus of air pollution and climate change — the complex interactions and interrelations that are revealed when the pollutants caus- ing them are considered holistically — have not been comprehensively captured by the policy processes in those two domains. Regulatory measures that have always been entirely appropriate when applied to pollution control in traditional form, including cli- mate change, may now be rendered counter-productive, even dangerous, when these complexities are incorporated. Legras (2011) terms this vexed problem ‘incomplete en- vironmental model specification’, arguing that ‘policy approaches to atmospheric pollu- tion have typically been based on incomplete environmental models, in the sense that they have not captured all the contributors to the issue, or if they have, that the correl- ations between the contributors have not been properly described’ (p. 528). Similarly, Caplan and Silva (2005) define a new term, ‘correlated externalities’, denoting the same general issue, noting that ‘previous mechanisms designed to control single externalities are generally incomplete…. [and their] adoption would typically create inefficiencies, since correlated effects are neglected’ (p. 69).
Yet even the necessary complete environmental model specification may not be sufficient to overcome these flaws in policy response. In their ontological framing that less is beer, existing policy instruments — and the ways in which pollution control policies are conceptualised as a whole — may be irredeemable in their current form. at is
157 Chapter 4 Comparative analysis of mitigation policy to say, traditional pollution control policies, spanning the complete range of pollution taxes, tradeable property rights, or direct regulatory stipulations concerning production technology and permissible emissions, may at their foundation be unable to handle the confounding case where less is worse.
A further complication for instrument design in particular is the influence of time hori- zon. Where progressive reductions in pollutant emissions serve the underlying object- ive, as in the traditional case, the policy instrument has performed its function once a target emission level is reached. at instrument will have no ongoing regulatory bur- den aer achieving that target, other than maintaining the restriction on excess emis- sions. It can also safely be presumed that maintenance of that restriction requires a relatively small degree of regulatory oversight, because existing activities do not need to be amended further.
In the case of aerosol cooling, conversely, the policy instrument may be required to main- tain the deliberate and coordinated compensatory measures suggested above. Provision of any such compensatory mechanism is then a regulatory burden that must persist for an uncertain length of time that is unlikely to be well-known in advance, perhaps many decades — it is an active requirement in contrast to the largely passive traditional case of maintaining an already-achieved emission target. In the context of the aerosol cooling dilemma, the need to maintain the compensatory mechanism has important ramifica- tions for the design of the overall policy response, especially regarding ongoing liability and funding; these aspects are examined further in Chapter 5.
158 4.3 A revised ontology
4.3 A revised ontology
A revised ontology for the formation of policy objectives is the necessary starting point for the reconception argued above. e task is, in essence, to expand the multi-pollutant, multi-effect classification schema to formally recognise the existence of contingent be- neficial consequences of some pollutants in some circumstances, and to incorporate the factor of time. However, in order to do so, we must first revisit the ambiguous usage of the terms effect and impact.
4.3.1 Clarifying effect and impact
In the policy discourse, it seems fair to say that the terms effect and impact are oen poorly delineated or used largely interchangeably. But are they the same? Most import- antly, should they have the same connotation of damage? is ambiguity in definition in the discussion above may now be more visible in hindsight (such as the initial statement of the term in section 4.1.3), aer considering the discussion of ontology in section 4.2. To be specific, the central concern is the question of whether effect and impact should both be assumed to refer to the consequences of pollution and, simultaneously, that those consequences are assumed to be deleterious. Such assumptions confuse and con- flate the nature of the consequences of pollution with their underlying mechanism of action. A robust definition should clearly separate the mechanism(s) of action, which is an objective statement of physics and chemistry, from the more contingent — and argu- ably, subjective — assessment of the normative consequences of that mechanism. And the term employed to signify the laer should permit consequences to be assessed as beneficial, given carefully constrained prerequisite circumstances.
e words effect and impact could serve those functions, if appropriately specific defin- itions are given. However, the word ‘impact’ is quite likely to connote damage in and
159 Chapter 4 Comparative analysis of mitigation policy of itself; certainly in the context of pollution control policy that is the sense in which it is used. Redefinition of terms is hence unlikely to remove such connotations in wider discourse, even if it may within a given analysis. It therefore appears necessary to define a new term, both free from any ambiguous conflation with ‘effect’, and without obvious negative connotations. I hence propose refined, specific definitions as follows: effect is the particular objective geophysical mechanism of action in the environment resultant from emission of a given pollutant; any pollutant may produce multiple effects. outcome is the assessed consequences of a given environmental effect with reference to an affected entity (eg, humans, physical infrastructure, other species, the eco- sphere or constituent elements); outcome should include an explicit qualifier that signifies the nature of those consequences as damaging or beneficial (or perhaps, benign).
e term multi-effect is thereby properly constrained to specifically refer only to effect in the sense of mechanism of action. at is, a given pollution case involves multiple geo- physical effects (such as smog, acidification of waterways, acid rain deposition) without inherently prefiguring what the consequences to those effects are, or their nature, par- ticularly regarding human welfare. Accordingly, use of ‘impact’ would be discouraged; where it is used, it might be read as roughly synonymous with effect, or with the subset of possible outcomes that contains only deleterious consequences.
For example, emission of SO2 has the effect of inducing acid rain deposition, the outcome of which is acid damage to natural ecosystems and to human physical infrastructure. However, those same emissions also have the effect of direct reflection of incoming solar radiation and indirect effects on cloud, the outcome of which is a benefit to human and natural systems by dampening the positive radiative forcing of elevated atmospheric
GHG concentrations. SO2 therefore produces multiple effects and multiple outcomes,
160 4.3 A revised ontology and some of those outcomes add to human welfare given the prerequisite circumstance of significant anthropogenic climate change.
4.3.2 Expanded classification schema
With these revised definitions, we can now return to the task of expanding the classi- fication schema. I propose two additional dimensions for the statement of a pollution case: outcome is as per the definition above, codified with explicit cardinal qualifiers indicat- ing the nature of the consequence as either damaging or beneficial for an affected entity; benign outcomes are not considered. Note that outcome should not con- sider any benefit stemming from the underlying activity that produces a pollut- ant, only to the nature of the environmental effects themselves. Outcome is also applied in the aggregate, where a set of particular effects can then be classed as damaging, beneficial, or counter, the last where at least one effect has a damaging outcome and at least one other effect has a beneficial outcome (ie, counteracting or counterposing). e scope of effects considered therefore has a direct bearing on the assessment of aggregate outcome. Clearly the vast majority of pollutants produce effects that are damaging-outcome only. Further, the majority of tra- ditionally considered multi-pollutant, multi-effect problems have scopes that are classed as damaging-outcome in aggregate, such as the consequences of particu- late emissions for local air quality, or, indeed, the consequences of greenhouse gas emissions for climate change. time horizon similarly indicates the period over which the majority of an outcome is manifested. An outcome can then be classed as near, meaning occurring imme- diately, on the order of days to a year; medium, meaning unfolding over a period of some years to a decade; or far, meaning more than a decade is required. At
161 Chapter 4 Comparative analysis of mitigation policy
the aggregate level, a set of particular outcomes may be classed as near, medium, far, or split, where the last signifies that different outcomes operate across differ- ent time horizons. Pollution that damages local air quality is hence near-horizon, while GHG-induced climate change is broadly far-horizon. More importantly, we can now beer classify the full consequences to climate change from the combin- ation of aerosol and GHG emissions: climate change now becomes split-horizon, because the outcome of fluctuating aerosol emissions is near-horizon once GHG- induced climate change is already underway. Note that time horizon is a con- ceptual tool and should not be interpreted as an accurate description of complex realities, wherein outcomes are rarely so neatly delineated. Emissions of methane, for example, produce substantial climatic effects in their first few years in the at- mosphere; indeed, so does carbon dioxide. e point here, to emphasise, is how long is required to produce the majority of the outcome in question.
Employing these additional dimensions in the full schema, we can classify the major pollution cases as follows:
Case 1 Aerosol effects on local air quality, with the scope limited to the traditional air pollution effects, is multi-pollutant, multi-effect, damaging-outcome, near-horizon.
Case 2 Climate change as usually conceived, with the scope limited to GHG effects alone through the enhanced greenhouse effect, is multi-pollutant, single-effect, damaging-outcome,10 far-horizon.
Case 3 ough the enhanced greenhouse effect and its consequences is the major focus, ocean acidification must not be ignored,11 so the classification of climate change is expanded to multi-pollutant, multi-effect, damaging-outcome, far-horizon.
10 Notwithstanding the potential for beneficial impacts in some locations during the initial onset of cli- matic changes (such as decreased winter deaths or increased agricultural yields). 11 In fact, Archer and Rahmstorf (2010) state that in their view, ‘this direct threat to the health of our oceans would be enough reason to halt the further rise of atmospheric CO2 concentration, even if CO2 did not cause climatic changes’ (p. 148).
162 4.3 A revised ontology
Case 4 Aerosol effects on climate change, with the scope limited specifically to climate forcing effects (diminishing a pre-existing positive planetary energy imbalance), is multi-pollutant, multi-effect, beneficial-outcome, near-horizon.
Case 5 Similar to ocean acidification in Case 3, aerosol effects on the hydrological cycle (damaging-outcome) must also be recognised, so the classification is expanded to multi-pollutant, multi-effect, counter-outcome, near-horizon.
Case 6 Climate change when considering all forcing agents is multi-pollutant, multi- effect, counter-outcome, split-horizon.
Any analytical merit provided by this proposed schema risks being obscured by express- ing the four dimensions in word-form only. I therefore offer these concepts in illustrative form below. e diagrams are loosely based on the ‘parallel coordinates’ approach to visualising multi-variate data: each of the four dimensions are represented in a column panel, where the range of possible values is arrayed vertically: pollutant and effect can be one or many, and outcome and time horizon take the cardinal values identified above. A simple marker is then placed in each dimension panel to indicate the classification ap- propriate to the pollution case in question. Multiple markers within the outcome or time horizon dimensions signify a case of either or both of the counter-outcome, split-horizon classification.
Pollution Case 1 in the list above is illustrated in Figure 4.1. e initial ‘minor’ form classification of climate change as the primary enhanced greenhouse effect, Case 2, is illustrated in Figure 4.2. Figure 4.3 shows the more complete Case 3 form, using a dashed marker to represent the inclusion of damaging-outcome ocean acidification. Figure 4.4 illustrates Case 4, where the orange marker is used simply to underscore the beneficial- outcome of reduced net forcing.
163 Chapter 4 Comparative analysis of mitigation policy
Pollutants Effects Outcome Horizon One One NearBeneficial
Many Many Damaging Far
Figure 4.1: Illustrative pollution classification: aerosols as local air pollutants (Case 1)
e additional complexity of damaging-outcome aerosol effects on the hydrological cycle in Case 5 is shown in Figure 4.5. And finally, the full implications of aerosol effects within climate change, Case 6, are illustrated in Figure 4.6, where the shaded effect marker is used to underscore the parallel existence of positive and negative forcing influences on planetary energy balance, and, again, the presence of two markers signifies counter- outcome, split-horizon.
Pollutants Effects Outcome Horizon One One NearBeneficial
Many Many Damaging Far
Figure 4.2: Illustrative pollution classification: GHG effects on climate (Case 2)
164 4.3 A revised ontology
Pollutants Effects Outcome Horizon One One NearBeneficial
Many Many Damaging Far
Figure 4.3: Illustrative pollution classification: GHG effects on climate and oceans (Case 3)
Pollutants Effects Outcome Horizon One One NearBeneficial
Many Many Damaging Far
Figure 4.4: Illustrative pollution classification: aerosol effects on forcing (Case 4)
Pollutants Effects Outcome Horizon One One NearBeneficial
Many Many Damaging Far
Figure 4.5: Illustrative pollution classification: aerosol effects on forcing and hydrological cycle (Case 5)
165 Chapter 4 Comparative analysis of mitigation policy
Pollutants Effects Outcome Horizon One One NearBeneficial
Many Many Damaging Far
Figure 4.6: Illustrative pollution classification: all forcing agent effects on climate (Case 6)
With this expanded classification schema as a basis, we can now assess whether exist- ing policy responses on the ground account for the counter-outcome and split-horizon nature of the aerosol cooling dilemma, and the potential they have for doing so.
4.4 Accounting for aerosols in climate change policy
e analysis of pollution control policy given in general form above can now be ap- plied to investigate actual policy instruments in the climate change and air pollution domains, as they relate to the aerosol cooling dilemma. at is, an assessment of the degree to which aerosol cooling is, or might be, accounted for in current or proposed climate change mitigation policy — or perhaps, climate change concerns incorporated within air pollution policy — and the potential advantages or disadvantages to be con- sidered in doing so.
4.4.1 Linkage synergy and co-benefits
Climate change and air pollution policy objectives are both implemented through reg- ulatory abatement of pollutant emissions that arise, commonly, from industrial produc-
166 4.4 Accounting for aerosols in climate change policy tion activities, where those activities have not historically been subject to cost imposts that reflect the damaging outcome of those emissions. at is, damages arising from such pollutants are negative production externalities (a net loss of welfare). Accordingly, an obvious question is to what extent their objectives might beer be served by harmon- ising their individual policy instruments; what opportunities for synergy exist. EEA (2004, p. 7) characterises this as ‘linkage’, defined as ‘the connection between air pollu- tion and climate change … [considered] in the broadest sense, ranging from atmospheric linkages to linkages in emission control and policies’; in climate change policy this may be referred to as ‘ancillary benefits’ stemming from an abatement action. Indeed, given the close relationship of climate change and air pollution — the well-recognised exist- ence of coupled emissions for many regulated activities — it is ‘surprising’ that these domains have largely been separated in policy efforts (Swart et al. 2004).
Several authors have investigated the potential for such synergy and co-benefits in link- ing climate change and air pollution mitigation frameworks (eg, Swart et al. 2004; EEA 2004; Rypdal et al. 2005; ApSimon et al. 2009; Tollefsen et al. 2009; Rypdal et al. 2009; Nemet et al. 2010). Adapting the terminology of Caplan and Silva (2005) somewhat, realised co-benefits can be characterised as correlated positive externalities, in the sense that they are positive (welcome) spillover effects. A summary of potential co-benefits derived from the literature is presented in Table 4.3.
4.4.2 Linkage barriers and conflict
ere are, inter alia, clear local and regional human health and environmental advant- ages from GHG abatement that also reduces aerosol emissions and vice versa; ie, coupled emission abatement. Co-benefits undoubtedly exist in linking climate change and air pollution mitigation policy frameworks, and it is not my intention to downplay or dis- count the important synergies these offer. However, negative unintended consequences
167 Chapter 4 Comparative analysis of mitigation policy
Table 4.3: Co-benefits in climate and air quality policy linkage
Co-benefit and synergies Common multi-pollutant emission sources (energy, agriculture) mean air pollution policies will also reduce GHGs, and vice versa. Abatement of such multi-pollutant sources by policies in one domain can lead to cost reductions in the other. For example, studies indicated up to 40 % savings for European air pollution control costs if changes in energy systems implied by Kyoto Protocol compliance were implemented (Swart et al. 2004). Economic efficiency gains may be possible through linked multi-pollutant, multi-effect policies, lowering effective abatement costs in both domains. Possible economies of scale from integration of emission assessment and reporting requirements.† Coordination of policy response engenders a more holistic, integrated approach to pollution challenges.
† ough section 4.4.2 notes that increased administrative and transaction costs are also possible. also exist in parallel to those co-benefits under certain circumstances. ese consequences of abatement actions have ramifications that are in conflict with and run counter to the objective of either or both domains.12 Such conflicting consequences are hence a barrier to effective abatement once the full interactions of these pollutant classes are understood — and they have received comparatively less-robust aention in policy making. Employ- ing the revised policy ontology, these consequences are archetypal counter-outcomes.
ese conflicts inherent in climate change and air pollution mitigation policy linkage are investigated in relatively more detail here than the discussion of co-benefits in sec- tion 4.4.1 above. A summary of conflicts identified in the literature is given in Table 4.4.
12 EEA (2004) refers to these complications as ‘trade-offs’, a term that I do not feel adequately commu- nicates the full nature of the conflict.
168 4.4 Accounting for aerosols in climate change policy
Table 4.4: Conflicts in climate and air quality policy linkage
Conflicts and barriers A shi from fossil fuel to biomass energy sources induced by climate change policies may damage human health due to increased particulate emissions profile of biomass combustion.
Some air pollution control measures can increase energy use and hence fossil fuel CO2 emissions for the same output; eg, SO2 scrubbers in coal-fired electricity generators reduce thermodynamic energy efficiency, requiring greater coal combustion and therefore CO2 emission for the same electricity production. (Legras (2011) argues this is a seminal example of unintended consequences stemming from incomplete environmental model specification.)
Controls on many air pollutants (such as SO2) may have the unintended side effect of increased climate forcing, despite benefits to air quality from reduced emissions. ere is potential for loss of momentum for further air quality structural changes in industrialised and developing countries if air pollutant regulation becomes subsumed within climate policy, given the political challenges surrounding the laer.
Existing domain-specific metrics and accounting methodologies (such as GWP and CO2-e in the climate change policies) may be at least ambiguous, at worst inapplicable outside of their current domain. Distinct differences exist in the spatial and temporal character of air pollution versus climate change effects; accordingly, regional and local responses are most appropriate for air quality, versus the need for global coordination in addressing climate change. Human capital and knowledge is generally domain-specific and somewhat isolated: ‘the number of people with knowledge of both areas is very small and appropriate joint institutional arrangements are largely absent’ (EEA 2004, 42). Existing detailed domain-specific policy frameworks are reasonably entrenched in their respective domains and may not easily be adapted to incorporate the other.
Rypdal et al. (2005) examined the inclusion of short-lived species (aerosols and tropo- spheric ozone) within climate change mitigation frameworks, identifying particular dif- ficulties in doing so:
• Many such species are created in secondary formation processes rather than dir- ectly emied, adding complexity.
• Large uncertainties in radiative forcing exist.
• Determination of appropriate geographical extent and the presence of important regional differences.
169 Chapter 4 Comparative analysis of mitigation policy
• e nonlinear relation of atmospheric concentration to source emissions.
eir greater concern, however, is the risk of creating incentives to increase air pollution if negative aerosol forcing is brought into the total emission budget by subtracting it from the positive component of GHGs. Rypdal et al. (2009) subsequently label this the risk of creating perverse incentives to emit unquestionably damaging air pollutants. is serious impediment is examined in Box 4.1.
Beyond these technical aspects, Rypdal et al. (2005) also identify a series of political and policy issues concerning inclusion of short-lived species within mitigation frame- works, all of which tend to increase policy complexity and administrative burden. Cli- mate change mitigation schemes would need to be linked to those regulating air qual- ity. Linkage in turn introduces the problem of regulating the same substance multiple times for different purposes, potentially increasing administrative and transaction costs; though some studies do suggest efficiency advantages from economies of scale. Report- ing requirements would necessarily expand to incorporate these species’ climatic effects, and new metrics and relative weightings may be needed (see discussion in section 3.5 and 4.5.1).
4.4.3 Are they included currently?
In a word, no. Current climate change mitigation policies and international negotiations in effect completely neglect any recognition of the role of aerosol cooling. A signal ex- ample of this is that the word ‘aerosol’ does not appear in the text of the Kyoto Protocol,13 nor is any match found for a keyword search of all UNFCCC official documents, which covers the fourteen years since.14 Shine et al. (2005a) note that all short-lived species are
13 Aerosols are listed in the general definitions of ‘sink’ and ‘source’ in the UNFCCC (United Nations 1992), and these two lines are the only instances. 14 Available at http://unfccc.int/documentation/documents/advanced_search/items/3594.php, as of Janu- ary 2012.
170 4.4 Accounting for aerosols in climate change policy excluded by the Protocol, all aerosols among them. At the same time, climate change effects remain excluded from air quality regulation — the two policy domains continue to act largely in separation (Rypdal et al. 2009; Swart et al. 2004).
at separation is not without justification, of course, and quite apart from the risk of perverse incentives to pollute (Box 4.1). Swart et al. (2004) point out the example of tropospheric ozone’s absence from the Kyoto Protocol despite its significant posit- ive forcing, both because of its formation through secondary processes and because its precursors are already covered by the Convention on Long-Range Transboundary Air Pollution (CLRTAP).
antification of aerosol climatic effects has been and remains a major uncertainty in empirical climate research and modelling efforts, along with the role of clouds. In fact, though I am entirely convinced by the inferences of Hansen et al. (2011) described in Chapter 3, and believe their work will become a landmark study in the field, they are yet to be replicated, and hence sit outside the consensus at time of writing. And finally, in more prosaic, institutional terms, aerosols and their precursors have traditionally been understood as local air pollutants — indeed, the prime original motivation for air quality regulation, which significantly predates any for climate change mitigation. Aerosols’ role in climate, while recognised decades ago, has quite understandably been a distinctly subordinate concern of climate policy-makers, if seriously considered in that context at all.
171 Chapter 4 Comparative analysis of mitigation policy
Box 4.1: Perverse incentives as an impediment to discourse?
I surmise that the risk of creating a perverse incentive to pollute is an important reason behind aerosol cooling’s lack of inclusion within climate change mitigation policy efforts, and indeed an apparent lack even in the policy discourse. is risk was cited to me verbally in person by an eminent researcher in the field of mitigation pathways, in part explanation as to why no policy response was required. Unpacking the risk of perverse incentives further suggests at least three prominent facets of the problem: • At the international level, a country could in effect offset and thereby diminish their ap- propriate GHG mitigation obligation by subtracting their assessed negative aerosol for- cing from reported GHG emissions. Indeed, the dirtier their air the beer, giving incentive to at least reduce the rate of decline in air pollution, if not reverse those declines outright. • Domestically, polluting firms with coupled emissions would perceive similar incentives in the context of a carbon price signal or other abatement instrument if their aerosol emissions could offset their GHG liabilities. e relative merit order of inter– or intra- sector abatement is likely to be significantly altered by including aerosol cooling. • Further, domestic aerosol-only polluting firms could be awarded — or agitate for — a range of benefits for their emissions, such as offset credits within GHG abatement schemes, diminution of air quality regulatory obligations, and so on; and again, the same potential for delayed or reversed abatement would apply. It is hence easy to see why aerosol cooling might be neglected in climate change policy making to date. Consciously or otherwise, the extraordinary challenge of perverse incentives is reason enough to maintain the separation of climate change mitigation and air quality regimes — and this is only compounded by the sheer complexity of aerosol climatic effects and their large un- certainties. Until relatively recently, integrating aerosol cooling within climate policy was on the whole simply not justified. Rypdal et al. (2005) argue that negative aerosol forcing alone renders their incorporation within international mitigation frameworks ‘too complex to be politically feasible’, and moreover that ‘a regional climate agreement with links to both a global climate agreement and various air quality concerns may be too complex to negotiate’ (p. 41). In later work, Rypdal et al. (2009) cite perverse incentives, along with the uncertain state of current knowledge regarding NOx and SO2, as reason to explicitly suggest ignoring short lived species in abatement policies based on integrated radiative forcing and GWP (pp. 855, 867). ough this is an understandable position, I contend that the risk of these perverse incentives is no longer sufficient reason to demur from a frank discussion of the aerosol cooling dilemma and possible solutions to it. e metrics must change, not the problem ignored. As I have shown, the balance of evidence — particularly that emerging since AR4 — now points to a clear danger from an unmanaged withdrawal of aerosol negative forcing, despite the truly perverse nature of the policy implications. e unrelenting pace of GHG emissions means that we will soon arrive at the historical moment where political ‘feasibility’ must be dictated by geophysical reality, and we abandon the delusion that it might be the other way around.
172 4.4 Accounting for aerosols in climate change policy
Climate change mitigation policy to date, internationally and within particular coun- tries, remains overwhelmingly concerned with abatement of the greenhouse gases, primar- ily the carbon-based species. Indeed, even short-lived positive forcing agents such as tro- pospheric ozone or black carbon have yet to be included in international agreements.15 However, a substantial number of studies and policy-option reports have been published in recent years, advocating or identifying BC mitigation pathways as an important ‘win- win’ opportunity for climate and air quality, especially in the developing world (eg, UNEP 2011; UNEP and WMO 2011; Shindell et al. 2012). Future negotiations might hope- fully begin to pay aention to these opportunities.
Regardless, the Kyoto Protocol gives effect to the principle of ‘comprehensiveness’ of abatement action through a basket approach to the six LLGHG species proscribed in its Annex A:16 mitigation obligations can be satisfied by abatement of any of those spe- cies, as best serves a particular country’s circumstances (United Nations 1998; Rypdal et al. 2005; Tollefsen et al. 2009). is basket approach is enabled by expressing mitig- ation obligations in mass unit CO2-e, using 100 year GWPs (see section 3.5, and 4.5.1 below). e precedent has informed mitigation framework design since: national or regional mitigation schemes such as the landmark European Union Emissions Trading Scheme or the Australian Clean Energy Act continue to regulate the same six species.
e predominant mitigation policy instruments actively proposed or implemented around the world are carbon price mechanisms, either in the form of an emissions tax or emis- sions trading scheme (ETS), where the only emissions considered are LLGHGs. ese instruments sit within the traditional policy ontology discussed in section 4.2: the less pollution the beer,17 as GHG emission abatement (multi-pollutant) serves the mitiga- tion objective of reducing the expected rise in TS and arresting ocean acidification (multi- 15 Perhaps in large part for the reasons outlined in section 4.4.2. 16 e individual gases carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), and sulphur hexaflu- oride (SF6), and the hydrofluorocarbon (HFC) and perfluorocarbon (PFC) classes. 17 At least as regards the beneficial effect of emission reductions on future climate change; rising eco- nomic costs act as a dampener for the rate of abatement deemed to be plausible.
173 Chapter 4 Comparative analysis of mitigation policy effect, damaging-outcome). e details of policy instrument design of course add a great deal of complexity and nuance — third-party abatement or sink-enhancement offsets; tax and liability exemptions or cross-subsidies; price ceilings, price floors; intertemporal provisions such as permit borrowing and banking; free permit issuance and other distor- tions (against all economic advice); cap over-allocations; incomplete sectoral coverage; and so on — but the goal remains a reduction of atmospheric GHG concentrations.
Aerosols are not represented in any meaningful way whatever.
4.4.4 Could they be?
Two plausible avenues for an accounting of aerosol cooling are evident in the preceding discussion: linkage of existing air quality and climate change mitigation instruments, or direct inclusion of aerosol species within the laer.
Some authors assessing policy linkage have reached the conclusion that doing so is pos- sible and synergies are aainable, at least in some sectors. For instance, ApSimon et al. (2009) assert that measures that abate both GHG and aerosol emissions within the trans-
18 port sector constitute a ‘win-win’ outcome, including reduction of NOx, a necessary precursor to tropospheric nitrate aerosol formation. But, curiously, this is despite reit- erating the masking effect of aerosol species such as sulphate and nitrate in the same concluding section (p. 677). e implication is surely then that any consequential di- minution of that mask (in this case, by reducing nitrate) is of negligible risk to climate; yet the authors do not actually make such a case, nor indeed any at all. is neglect of aerosol cooling therefore somewhat undermines the findings of pollution control cost savings from climate and air quality policy linkage in studies reviewed by ApSimon et al. (2009) — there is no clear accounting of any additional effective cost from lost aerosol
18 In a specific example, a behavioural shi from larger to smaller petrol-engine passenger vehicles in the UK, producing reductions in CO2 and NOx.
174 4.4 Accounting for aerosols in climate change policy cooling. Again, such omissions can perhaps be interpreted as a demonstration of the limits imposed by an unrecognised policy ontology that is blind to the uncomfortable reality that pollution can have contingent beneficial effects.
Other studies can be said to neglect negative aerosol forcing in differing manners: some recognise but explicitly exclude aerosol cooling, as in Tollefsen et al. (2009), who set negative forcing to zero in their model of efficiency gains from policy linkage; or by outright omission, as in the quantitative assessment of GHG abatement costs that factor in air quality co-benefits in Nemet et al. (2010). e former case stems in large part from a deliberate recognition of the serious difficulties faced by any aempt to include aerosol cooling effects in an accounting of linked policy instruments — an investigative position that has obvious merit, yet I argue is nonetheless insufficient. e laer case, however, appears to me to be fallacious reasoning (finding cheaper abatement), though inadvert- ent, due to omission of these crucial confounding effects. e resultant conclusions, such as that ‘the full inclusion of [air quality] co-benefits in the design and evaluation of climate policy would almost certainly enhance social outcomes because these co-benefits are large’ (emphasis added, Nemet et al. 2010, 6), must hence be regarded with scepti- cism. e omission of aerosol cooling effects in Caplan and Silva (2005) is also obvious, despite their useful investigation of integrated global climate and regional air pollution emissions trading schemes for these correlated externalities arising from shared emis- sion sources.
In my assessment, therefore, linkage studies have been distinctly inadequate in their accounting for the loss of aerosol cooling in theory, and hence do not provide a sufficient basis for integration of these instruments in practice.
What then of expanding existing climate change mitigation instruments to include aer- osols directly? At first blush, aerosols may appear a reasonable fit to the pre-existing notion of offset credits within emission trading schemes, which are commonly based on sink enhancements or, in particular, avoided deforestation. Aerosol offset credits could
175 Chapter 4 Comparative analysis of mitigation policy in principle be applied to emission tax regimes also, though offsets in that context are uncommon. ere are at least three major flaws with such an approach, however: two technical and the other political.
• First, there is a crucial technical difference between credits arising from activ- ities such as improved land management practices, biosequestration, or avoided deforestation,19 and potential credits from aerosol emissions: the former func- tion either as a reduction in total LLGHG emission, or effect a reduction of atmo- spheric LLGHG concentrations by enhancing sink capacity. Even should credit- generating actions later fail (such as loss of biosequestered carbon through fire), it is at least possible to compensate for the then-additional emissions by abating more strongly at another GHG source. A unit aerosol emission does not affect at- mospheric LLGHG concentrations at all, but rather, as explained in Chapter 2, af- fects absorption and reflectance of shortwave radiation. e positive GHG forcing is offset only while that aerosol emission is maintained at a constant rate because of the residence time asymmetry; should the unit aerosol emission represented by a credit cease, a substantial portion of the previous GHG emission remains in the atmosphere to exert its warming influence. Even without any unintended failure of aerosol emissions, then, they cannot be regarded as an offset to continued GHG emission unless they are to persist for all practical purposes indefinitely: as long as is required for natural processes to remove the GHG emission slug in question, which is many decades, far in excess of any analogous precedent policy mechan-
ism. What is more, as atmospheric CO2 concentration itself is totally unaffected by any aerosol loading, ocean acidification will continue to intensify as a direct
consequence of the unaltered CO2 emission rate.
• e second technical flaw is the inapplicability of conventional climate change mitigation metrics to aerosols; this is examined in section 4.5.1.
19 Puing aside serious concerns with the integrity and reliability of these approaches.
176 4.4 Accounting for aerosols in climate change policy
• e political flaw is the high risk of realising the perverse incentive to pollute, as discussed in section 4.4.2 and Box 4.1. at risk cannot be ignored, nor can it be neutralised by climate change mitigation policies in their current form.
Conversely, could climate change effects be included in air pollution regimes? Ryp- dal et al. (2005) touch briefly on the possible inclusion of climate effects within CLRTAP, given its multi-pollutant, multi-effect framework (ozone formation, eutrophication, acid- ification). My analysis suggests this will not be sufficient. First, the multi-effect coverage is for consequent environmental impacts that remain damaging-outcome, and so the ef- fects considered would only be the additional climate impacts of pollutants already regu- lated, such as ozone precursors, not the complex interplay arising from aerosol negative forcing. Alternatively, GHG emissions could be included within air pollution instru- ments, but this does not address the problem.
Second, the framework is unlikely to be readily amendable to incorporate any neces- sary managed loss of negative forcing effects consequent to aerosol abatement. And I speculate that suggestions to do so are likely to be met with strong opposition, as a fundamental corruption of those air pollution regulations.
At a higher level, there is really lile difference between climate change and air pollu- tion mitigation policies in their relevance to the aerosol cooling dilemma. Both seek to reduce pollutant emissions, by definition, so neither is equipped to handle any emer- ging requirement to avoid doing so. If compensatory measures are to be introduced as a means to manage the reduction of aerosol emissions while maintaining some degree of their cooling effect, it is hard to see how air pollution regimes could offer any advantage over climate change frameworks, given the problem is at heart one of climate.
177 Chapter 4 Comparative analysis of mitigation policy
4.5 Confounding complexities
e above investigation of policy objectives and the form of their implementation sur- veys many of the confounding complexities inherent in any aempt to link air pollution and climate change mitigation policy, or to include aerosols within climate policy as currently conceived. As I have argued, there is no sound basis in existing policy instru- ments, in either domain, for the inclusion of aerosol species if the full ramifications of their cooling effects are to be properly represented. is section analyses these com- plexities further as they relate specifically to the problem of coupled emissions and the loss of aerosol cooling resultant from abatement of them.
4.5.1 The problem of CO2 non-equivalence
As highlighted in section 4.4.3 (p. 173), climate mitigation policy instruments have usu- ally been constructed using the basket approach to abatement defined by the GWP and
CO2-e methodologies. e technical difficulties and limitations of applying these metrics to negative forcing agents was discussed in section 3.5.1. In light of these problems, the validity of using negative GWP for short-lived species such as aerosols is surely highly questionable on a physical science basis.
e important concern is the application of negative GWP within climate change mitiga- tion instruments. Even if carefully qualified negative GWPs are defensible on a scientific basis, will those qualifications be understood and properly reflected within mitigation instruments, remembering that these largely rely on the basket approach? Simply put, is not the most likely scenario a linear addition of a country’s GHG emissions (positive term) with that of their net aerosol emissions (negative term), thereby reducing their overall assessed CO2-e liability? Rypdal et al. (2005) describe this as ‘politically rather problematic’, given the obvious risk of perverse incentives, concluding that it is ‘not
178 4.5 Confounding complexities evident’ how negative forcing agents can be handled by the basket approach (pp. 34, 41).
Moreover, the validity of GWP within mitigation instruments has been questioned even for GHGs. Smith and Wigley (2000b) highlighted that GWP is defined in relation to emis- sion pulses, rather than the finite-length emission changes over time which are relevant to policy. Moslener and Requate (2009, 1522) argue that the substitutability assumption inherent in the application of GWP and CO2-e within climate change mitigation instru- ments is ‘not generally valid’. Expressing the set of diverse climatically-active species indexed to a unit of CO2 obscures complex issues such as lifetime, dynamic interaction, or non-monotonic behaviour at different atmospheric loadings. Moslener and Requate (2009, 1522) claim that GWP ‘tells us hardly anything about the reduction priorities for greenhouse gases’. ey further suggest the GWP indexes should rightly be periodically adjusted to take such variations into account, hence modifying ‘exchange rates’ between
CO2 and other GHGs when they are regulated by a single market mechanism.
Expanding mitigation instruments to cover aerosol species that survive in the tropo- sphere for at best a couple of weeks via static 100 year GWPs is hence suspect indeed. I contend, more strongly, that it is in effect a category error to measure negative forcing agents in CO2-e terms. New metrics that beer represent the relative effects of forcing agents are required; GTP and SFP, discussed in section 3.5.2, provide options in that re- gard. And new approaches to mitigation policy that avoid the reductionism inherent in
CO2-e also require investigation, at the very least as applied to negative forcing. ese questions are examined in Chapter 5.
4.5.2 Correlated externality feedbacks
Climate change and air quality damage are negative production externalities arising from emission of GHG and aerosol pollution respectively, with the notable exception
179 Chapter 4 Comparative analysis of mitigation policy of tropospheric ozone that affects both. Where that pollution is generated by a single underlying physical activity such as fossil fuel combustion, these coupled emissions con- stitute correlated externalities, as Caplan and Silva (2005) describe. However, because LLGHGs are accumulative pollutants (section 4.1.1) and climate change is subject to large thermal inertia, they operate on a far time horizon (section 4.3.2), though significant ef- fects occur within a decade (see section 3.2). Aerosols conversely are assimilative pol- lutants and operate on a near time horizon, much less than a year. ese properties are captured in economic terms by classifying climate change as a stock externality, and air pollution as a flow externality (eg, Sandal et al. 1998). at is, externalities that are differentiated by the accumulative (stock) versus assimilative (flow) nature of the relevant pollutants. Abatement of coupled emissions as a result of policy instruments applied to either domain produces positive spillover effects in the other domain, as a co-benefit, by mitigating the correlated externalities — abatement of one pollutant will abate the other,20 as we have seen (section 4.4.1).
e problem is that aerosol emissions act in both domains: they cause the air pollution negative flow externality, while simultaneously distorting the climate change negative stock externality by temporarily masking the warming effect of LLGHGs. Accordingly, abatement of aerosols has a second spillover effect on climate change that is not posit- ive, exacerbating rather than ameliorating the damage of LLGHG emissions in the short term. Where that abatement is of coupled emission sources, the loss of aerosol cooling therefore confounds the aempt to mitigate climate change.
I suggest this set of circumstances can be termed a negative abatement feedback, defined as the spillover effect of some mitigation action that serves to increase rather than decrease the targeted negative externality on a stated time horizon. Moreover, be- cause of the different time horizons, this negative abatement feedback on climate change
20 ough not always, as in the important case where SO2 abatement can increase CO2 emissions from coal-fired electricity generation.
180 4.5 Confounding complexities mitigation overwhelms the marginal benefit of LLGHG emission reduction in the short term, and, due to rapid temperature increase, potentially even in the long term.
e existence of negative abatement feedbacks does not appear to be adequately re- cognised in the mitigation policy literature, despite notable recent warnings from some physical scientists: Ramanathan and Feng (2009, p. 47) suggest caution regarding re- ducing sulphur emissions, noting the ‘strong coupling and feedback effects of air pol- lution mitigation efforts and global warming commitment’; Raes and Seinfeld (2009, p. 5132) warn that while air pollution and GHG mitigation policies are both clearly neces- sary, ‘the possibility that simultaneous implementation of both policies would accelerate warming in the short term …, before stabilizing climate in the long term …, needs to be understood’; Shindell and Faluvegi (2010, p. 3257) note that the building “debt” of con- tinued CO2 emissions from coal burning currently part masked by air pollutants must ‘rapidly come due’ once society finally, inevitably, demands clean air. In their examin- ation of dynamic interactions and optimal abatement strategies, Moslener and Requate (2009) note that environmental economics has a tendency to focus on single-pollutant externalities, treating them separately or as aggregates and thereby missing important ramifications arising from ‘multiple pollutants with different dynamic properties’ (p. 1522). e authors are referring to the interplay between different GHG species (see above), but the observation applies just as strongly to the confounding role of aerosols.
4.5.2.1 The folly of natural gas substitution
Natural gas has oen been touted as a ‘transitional fuel’ in the energy supply sector, providing a bridge between CO2-intensive coal combustion and low-carbon or renew- able energy alternatives. e lower GHG emissions intensity per unit of energy from natural gas — around half that of black coal in combined cycle electricity generation — supposedly forms part of a structural transition away from fossil fuels to renewable
181 Chapter 4 Comparative analysis of mitigation policy energy sources in the long term. us conceived, natural gas can be exploited as an en- ergy source to maintain industrial society while also meeting climate change mitigation goals. Indeed, fuel switching to gas is an expected outcome of an effective carbon price in the short to medium term.21
is position is seriously misguided, at best, for three key reasons.
First, it presupposes that a gradual decline in GHG emissions intensity over coming decades is a suitable realisation of the mitigation effort required. at ignores the per- petuation of carbon ‘lock-in’ (Unruh 2000, 2002) by further entrenching dependence on a finite fossil fuel, making the necessary transition more difficult. Perhaps more im- portantly, much of that substituted natural gas consumption will require investment in new physical infrastructure as existing coal-fired facilities are decommissioned or scaled back, many of these at (or beyond) end of life. is infrastructure investment will be undertaken with an expectation of an operational lifetime that allows cost recovery and a normal economic profit. Yet as long-term net GHG emissions must almost certainly be reduced to zero,22 it is dubious that natural gas combustion can be permied to continue for the decades needed. Expansion of natural gas is then a serious mistake, a form of the ‘carbon bubble’ in financial markets’ asset valuation of fossil fuel reserves (see Leaton 2011).
Second, recent research on the full lifecycle emissions of natural gas production, trans- port, and especially fugitive emissions also seriously challenge the apparent benefit over coal combustion. Wigley (2011) concluded that fugitive methane over the supply chain largely cancels out any benefit from lower CO2 emissions at the point of gas combus- tion. And the current rush to exploit shale formations and coal seam methane may well involve such high fugitive methane emissions that ‘unconventional’ gas is actually
21 ough the reality of the ‘merit order effect’ on wholesale electricity price and dispatchable generation consequent to increased renewable energy is changing these expectations; see Parkinson (2012b) for example. 22 And in RCP2.6 they become negative; see section 3.6.3.
182 4.5 Confounding complexities dirtier than coal: ‘the footprint of shale gas is at least 20% greater and perhaps more than twice as great on the 20-year horizon and is comparable when compared over 100 years’ (Howarth et al. 2011, 679). In GHG terms alone, substitution of other fossil fuels to natural gas may turn out to be lile more than a caricature of meaningful abatement.
Finally, positioning natural gas as a transitionary fuel entirely overlooks aerosol cool- ing. e negligible generation of aerosol and other air pollutants (and their gaseous precursors) during natural gas combustion is an important aspect of its characterisa- tion as a ‘clean fuel’. Yet by burning cleaner than oil and particularly coal, natural gas may produce a greater net positive forcing because its (purportedly) lower relative GHG emissions are not offset by any aerosol contribution. An important caveat here is that this analysis only applies to coal combustion where air pollution controls are not wide- spread or have not led to significant emission reductions. As Box 2.1 notes, control regulations in North America and western Europe have been implemented and emis- sion of key aerosol species has been reduced; however, as section 3.6 shows, non-trivial emissions continue in those regions.
From a global perspective, Ramanathan and Feng (2008) suggest that gas is ‘likely the strongest global warming fossil fuel’ for this reason — coal and oil account for some 55 %
23 and 25 % of global SO2 emissions, where gas is less than 1 %. In the central scenario of fuel switching to natural gas for electricity generation, future GHG emissions may therefore be reduced in comparison to continued coal combustion for the same electric output, but at the cost of a weakened aerosol mask now due to the negative abatement feedback (section 4.5.2). Unger (2010, 5332) expresses these confounding consequences succinctly in the observation that ‘For many key emission sectors, the non-CO2 climate impacts outweigh those of CO2 on short time scales (20-30 years), and therefore radically alter the impacts of emission reduction versus the greenhouse gas only impact currently
23 As noted in section 3.6.2, the recent assessment of Smith et al. (2011) broadly agrees with these figures, giving 50 % and 30 % for coal and petroleum.
183 Chapter 4 Comparative analysis of mitigation policy used in climate policy’.
e notion that natural gas can be a pivotal transitional fuel within a long term mitig- ation strategy is wholly invalid. At global scale, the loss of cooling aerosols inherent in fuel switching from other fossil fuels will increase net positive forcing, even if natural gas had reasonably low lifecycle GHG emissions. It does not. Again, the full emissions profile of industrial activities must be taken into account in assessing their real capa- city to achieve mitigation — climate policies focussed only on LLGHGs are dangerously oblivious to this potential for unintended consequences.
4.6 Unintended consequences: the abatement paradox
To reiterate Table 4.1: aerosols are assimilative, non-uniformly mixed, short-lived pol- lutants, while LLGHGs are in general accumulative, uniformly mixed, and long-lived.24 Avoiding a unit of LLGHG emission therefore has no significant effect on net radiative forcing in the short term (on the order of years). Even a hypothetical instant total cessa- tion of all LLGHG emissions globally will not immediately change atmospheric LLGHG concentrations, due to their atmospheric residence times — the positive radiative for- cing term is consequently unchanged in the short term. e carbon cycle continues to remove CO2 over time, but even aer a millennia some 40 % of the peak concentration over preindustrial levels remains (Solomon et al. 2009); most non-CO2 LLGHGs per- sist for decades to centuries, but some trace species last for many thousands of years.25 Conversely, avoiding a unit of aerosol emission has immediate climatic impact: a sim- ultaneous cessation of all aerosol emissions would remove the aerosol mask in effect instantly, due to their short residence time and lack of atmospheric accumulation. As
24 Methane is the shortest lived of the LLGHGs, with an atmospheric lifetime of 12 years (Forster et al. 2007), though it does partly oxidise to carbon dioxide. 25 AR4 estimates lifetime of SF6 at 3,200 years, PFC-14 is the largest at 50,000 years (Tab. 2.14, Forster et al. 2007, 212-3).
184 4.6 Unintended consequences: the abatement paradox shown in Chapter 3, a marked and rapid rise in global mean surface temperature will result as the full GHG forcing is unmasked in this scenario, a higher temperature which will persist for a decade or more until LLGHG concentrations begin to fall.
As categories of climate forcing agents, LLGHGs and aerosols differ fundamentally in their mechanism of action, as reflected by their categorisation according to the pollutant taxonomy defined in section 4.1. Most importantly, their emission rates have system- atically differing ramifications for commied warming and the rate of temperature re- sponse (see section 3.4.1). Fluctuations in aerosol emissions are therefore of potentially greater concern than those of LLGHGs in the short term: the current perturbed configur- ation of the Earth-atmosphere system is in effect more immediately sensitive to aerosol emissions than to GHGs.
e closer that cumulative LLGHG emissions and atmospheric concentrations come to a safe limit26 and climatic tipping points (section 3.4), the larger the risk of inducing damaging impacts from decreased aerosol emissions over small time periods: fluctuat- ing aerosol emissions will reduce aerosol masking, exposing those higher LLGHG con- centrations; the greater and more prolonged any drop in aerosol emissions, the greater the fraction of those LLGHG concentrations unmasked. Moreover, if a safe limit has already been surpassed, the current aerosol mask is actively protecting against the dan- gerous climate changes that would otherwise result from those existing concentrations. Maintaining that mask until such time as atmospheric LLGHG concentrations return be- low the safe threshold may then be a critical requirement for environmentally effective mitigation policy. at is, while cumulative LLGHG emissions and atmospheric con- centrations are the ultimate determinant of climate change, in the current configuration aerosols act as the short-term controller of temperature.
While abatement of GHG emissions therefore can — and must — proceed with requisite urgency, consequent reduction of coupled aerosol emissions risks unintended serious
26 e concept of a safe limit is developed further in section 5.1.
185 Chapter 4 Comparative analysis of mitigation policy negative consequences. e aerosol dilemma is that their masking comes at the expense of local air pollution impacts and disruption to the hydrological cycle; this trade-off is surely difficult enough.
Yet the threat is greater and far more insidious.
Abatement of activities producing coupled emissions is also confronted by a paradox: those GHG emissions must be reduced, but doing so simultaneously weakens the pro- tective aerosol mask, and this negative abatement feedback may thereby hasten, not mitigate against, truly dangerous climate change in the short term. If the loss of aerosol cooling is more gradual such that a spike in temperature does not result, it could still be- devil even ‘aggressive mitigation scenarios’ — the warming rate already underway may be largely unaffected by LLGHG abatement, as declining aerosols offset any reduction in positive forcing (Strassmann et al. 2008).
Dangerous, even catastrophic, climate change may be realised in a most bier irony. To employ the idiom used by James Hansen, aerosol cooling is humanity’s Faustian bargain: their mask has allowed us to reap the undeniable benefits of fossil fuel combustion since the industrial revolution, delaying the onset of the full warming effects inherent in those greenhouse gases; yet payment comes due when we aempt to transform our societies away from these fuels, only to find the abatement task more complex and more fraught with confounding threats than we have imagined. We cannot now simply ‘turn o’ fossil fuel use, lest we trigger the dire consequences we are seeking to avoid. Such is the fundamental challenge to mitigation policy.
4.7 Dangerous omissions
Given the strong evidence of the aerosol mask, is it prudent to persist with mitigation policy frameworks focussed only on the six gases of the Kyoto Protocol; that assume sub-
186 4.7 Dangerous omissions stitutability between forcing agents; and that take no account of aerosol cooling what- soever? I argue strongly that the answer must be no. Our aempts to avoid catastrophic climate change are fundamentally incomplete in their neglect of a class of pollutants that may mask as much as half of the warming effects of greenhouse gases. e abatement paradox I have described must surely demand a considered policy response that is above all capable of addressing the full perverse complexities of anthropogenic interference in the climate system manifest in aerosol emissions. To persist with current GHG-only policy frameworks is to persist with truly dangerous, yet now known, omissions.
Yet this begs the question, how can such a threat exist but remain apparently unknown in the policy response, even in the policy and political discourse surrounding mitiga- tion? Why is nobody talking about aerosol cooling and the disturbingly real chance of inducing large and rapid warming precisely because of actions aimed at avoiding dangerous climate change? Elements of an explanation can be found in the passages above: the sheer confounding technical complexity of the dilemma, difficult for non- expert policy makers to understand; the serious risk of perverse incentives examined in Box 4.1; the barriers and conflicts listed in section 4.4.2. e physical climate science has also progressed significantly in recent years (see overview in section 1.2), and most climate change mitigation efforts fall dramatically short of what that science says is re- quired purely for GHG reductions. Arguably only in the last decade has the aerosol pic- ture become clear enough that the implications of mitigation policy could be recognised, and policy responses have their own inertia, especially at international level. Indeed, the SRES scenarios discussed in section 3.6.3 describe possible futures where increasing GHG emissions dominate those of aerosols, themselves gradually declining. Given the extent to which SRES has framed the evaluation of likely future climate change and the policy responses to it, the gradual fading out of aerosol cooling effectively rendered it unimportant. Taken together, it is easy to see that the policy discourse evolved to ad- dress a mitigation problem, as generally conceived, in which aerosol cooling is simply
187 Chapter 4 Comparative analysis of mitigation policy not a factor. And finally, underlaying it all is the fundamental policy ontology that less pollution is always to be welcomed.
Or it may be that there really is no such problem and no policy response is warranted, even in light of the recent history of GHG emissions growth and rising concern over tipping points. But if that conclusion is possible today, aer full consideration of the evidence, I cannot find an explicit and unambiguous statement to that effect in the lit- erature. ose that come close tend to assume rapid reductions of CO2 and non-CO2 forcings alike, as well as strong abatement of black carbon.27 BC remains outside of in- ternational climate negotiations, and some recent studies have found that BC’s indirect effect on clouds may make it a net negative forcing, rather than the large positive forcing it was thought to be (see discussion in section 2.4.2.3 and 2.5.2).28 So if the aerosol mask is not a serious threat to mitigation with current scientific knowledge, that may only be possible with some fairly heroic coordinated international action, which shows no signs whatsoever of materialising.
Of course, the physical climate science is evolving rapidly. Research may lead to new understanding of aerosols’ role or the climate response, lessening the threat enough to require GHG abatement only. Most important will be further research into the inter- linked issues of aerosol emission measurement and forcing quantification; detailed un- derstanding of energy imbalance and the climate response function; beer knowledge of the vulnerability of key natural systems to rate and magnitude of further temperat- ure change; and the risk of breaching climatic tipping points. Further, the character of mitigation actions undertaken may affect the degree of need for an explicitly aerosol- integrated response. Activities with negligible associated aerosol emissions (such as nat- ural gas combustion) might be preferentially abated over those with a substantial aerosol
27 Hansen et al. (2000) is a prominent example, referred to by recent work such as Hansen et al. (2011). Ramanathan and Feng (2009) draw similar conclusions regarding the ability to offset the loss of cooling aerosol with simultaneous reductions in BC. 28 Here I must reiterate that the major recent BC study of Bond et al. (2013) does not uphold these earlier findings. Nonetheless, substantial uncertainties remain, so the possibility cannot yet be ruled out.
188 4.7 Dangerous omissions component (such as coal combustion), with the resultant decline in GHGs without loss
29 of aerosols obviating deliberate compensatory measures. Atmospheric CO2 drawdown techniques such as biomass combustion with CO2 capture and sequestration could be- come sufficiently viable that remaining positive GHG forcing can be reduced quickly. A best-case scenario would be that, in sum, these factors mean that suitably robust and timely abatement of GHG emissions is sufficient to avoid dangerous anthropogenic in- terference, and the enormously difficult task of directly incorporating aerosol cooling is simply not required.
What if the reverse is true? As I framed the conclusion of Chapter 3,30 should we not be preparing for the worst case scenarios? What if aerosol masking is indeed the hidden and severe threat to mitigation laid out here and in previous chapters? Given the sclerotic, parlous state of current UNFCCC negotiations, the inexorable, even record-seing recent rise in GHG emissions, and the truly ‘alarming’ nature of the newest climate science, that threat appears to be growing ever larger. e sordid role of vested interests working, for decades, to sow confusion, doubt, and disinformation about climate change and to prevent or diminish responsible action is well documented (Oreskes and Conway 2011; Pearse 2007, 2009; Hamilton 2007, 2010; Schneider 2009). e grim sociopolitical reality is that mitigation action is not being taken with the ambition or scale the science demands, and barely at all.31
29 e potential for this is evaluated in detail in section 5.4. 30 Specifically, section 3.6.3. 31 ough individual countries and regions are taking steps, including resolutions passed by their na- tional legislatures in South Korea and Mexio in early 2012. Overall, this is nowhere near enough.
189 Chapter 4 Comparative analysis of mitigation policy
In e Great Disruption, Paul Gilding (2011) presents a convincing argument that no meaningful climate change mitigation will occur within the existing sociopolitical frame. Gilding argues that a step change in collective human political consciousness will not occur until the fundamental systemic risk of collapse is ‘in our faces’ (p. 103); mass recognition of the ‘new world’ we now actually live in. Societies must finally compre- hend climate change and other related systemic disruptions as directly and immediately threatening their daily lives and families — day in, day out. Most importantly, the key impacts will be economic, not simply environmental. Only understanding of the crisis at that dramatic scale will generate the political will for the action necessary. At that mo- ment, Gilding predicts the change will be sudden, widely supported, and transformative of our socioeconomic structures. Crucially, that moment may yet be some years from now,32 and during the intervening period GHG emissions seem destined to continue un- diminished, with an arrest of their rate of increase perhaps the most optimistic outcome for the short term. Hence, when anthropogenic GHG emissions do finally peak and mit- igation begins in earnest, their total positive forcing could be substantially higher than today. e abatement paradox could take humanity and natural systems to the brink if the aerosol mask remains unrecognised in such a scenario.
I contend that a discussion of these concerns between both climate scientists and policy analysts now can only be of benefit. Recognition of aerosol cooling is a vital component of a comprehensive and prudent risk management approach to climate change. A policy response to the abatement paradox is surely best devised in some form in advance, rather than under conditions of extreme duress and the nastiest of surprises — that is to say,
32 However, in an article published a short time ago, Gilding (2013) examines the remarkable shis in position of many iconic and highly influential international institutions such as the International Energy Agency (IEA) and the World Bank. eir public statements in the last two to three years are increasingly blunt and strident in warning that large-scale changes are urgently necessary. Crucially, the warnings are couched in terms of the threat to economies. is is perhaps best exemplified by the IEA, once regarded as the principal advocacy body of fossil fuel interests, who now clearly state that known fossil fuel reserves cannot be burnt if we are to preserve a stable climate and economy. Gilding suggests that the sociopolitical tipping point may be closer than previously thought, despite the lack of real progress in international mitigation efforts.
190 4.7 Dangerous omissions a Black Swan (Box 4.2). Such a policy response should at the very least be carefully thought out and instruments designed ready for use — far beer to expend that effort and find it thankfully unneeded rather than discover to our collective horror that the aerosol mask stood between us and catastrophe.
What form, then, must a fully aerosol-integrated climate change mitigation policy frame- work take if we are to be so prepared? In Chapter 5, I investigate the necessary design criteria requisite to an adequate and effective policy response.
191 Chapter 4 Comparative analysis of mitigation policy
Box 4.2: Beware Black Swans
e erudite insights of Nassim Nicholas Taleb (2010) are surely applicable here. Taleb instructs us on the virtue of not being a ‘sucker’ in the face of uncertainty — to beware of fragility in our knowledge, the Black Swans. A Black Swan event is one with three critical properties: it is an outlier, an occurrence outside of normal expectations and cannot be predicted from past exper- ience; it carries extreme consequences; and human fallibility leads us to construct retrospective explanations that appear (oen fallaciously) to make it predictable aer the fact. A Black Swan corresponds with an incomplete cognitive map of the world (p. 343). Most importantly, a Black Swan is subjective: events are only an unexpected shock to those who could not see them coming, or at least posit their potential existence. Few events exemplify the concept beer than the near collapse of US and international finance between 2007 and 2010 (if, indeed, disruption has actually run its course). e potential for systemic collapse was largely excluded by the dominant economic frame of equilibrium mod- elling analysis and related delusions such as ‘e Great Moderation’ (iggin 2010; Bezemer 2009; Keen 2011). To the vast majority, these events were an enormous Black Swan — indeed, even aer the fact many influential former and current bankers stated in various forms that “no one saw this coming” (Bezemer 2009). Taleb himself presaged the potential collapse of a now much centralised banking sector in the original 2007 edition of his work, noting that failures may be fewer but the crises more severe (pp. 255-6 in the 2010 revised edition). But it was the non-orthodox economists using ‘accounting’ monetary models such as Nouriel Roubini, Robert Shiller, and Steve Keen who did see it coming — in fact, they had predicted it in writing (Keen 2011; Bezemer 2009). e lesson here is that, as Taleb counsels, we should prepare for both the positive and the negat- ive consequences of uncertainty. We should recognise and permit the possibility of the extremely disruptive (p. 272). I raise these themes because I believe that the abatement paradox can be cat- egorised as a potential Black Swan embedded in the systemic uncertainties of our understanding of climate. Models cannot yet fully quantify aerosol indirect effects, nor have they been able to adequately predict serious nonlinearities such as ice sheet dynamics affecting Greenland melt or the staggering loss of Arctic sea ice, or reliably identify tipping points in the Earth-atmosphere system. We simply do not yet know where these traps lay — we must therefore act under con- ditions of significant, asymmetric, and to a non-trivial degree, unquantifiable uncertainty. We should therefore heed Taleb’s advice to consider the consequences, the payoffs, of decisions in that light (p. 349): if no abatement paradox exists, then we may have wasted money and effort on unneeded interventions to maintain the aerosol mask. at is not to downplay the import of doing so, but is not the converse dramatically worse? If we largely ignore the risks of a weaken- ing aerosol mask, the consequences might be catastrophic. In my view, that eventuality cannot be ruled out. Under conditions of complexity and uncertainty such as these, it is the severity of the possible consequences of our decisions that maer most.
192 5 Aerosol-integrated mitigation policy criteria
e preceding chapters have argued from the evidence that the avoidance of dangerous anthropogenic interference in the climate system now requires explicit consideration of the aerosol mask. at consideration means that mitigation policy must account for the emission of all forcing agents and for their interactions. Abatement of GHGs alone is no longer sufficient, though it is unquestionably necessary. Abatement of coupled emissions (section 4.1.5) poses unique and confounding challenges to mitigation policy- making in the form of the negative abatement feedback (section 4.5.2). Meeting those challenges will require that aerosol emissions are carefully accounted for and integrated within an overall climate change mitigation framework. Deliberate and likely deeply counterintuitive actions will be necessary in consequence, with real potential to radic- ally alter the relative merit of particular abatement pathways in the transformation of socioeconomic systems toward a low-carbon future.
is chapter elucidates the principal design criteria that an aerosol-integrated mitigation policy framework must achieve if it is to succeed.
193 Chapter 5 Aerosol-integrated mitigation policy criteria
5.1 Policy objective
e overarching objective for global climate policy should be progressive reduction of anthropogenic forcing so as to restrict climate change to some meaningfully defined safe limit. Where to date this objective has been understood as requiring cuts to GHG emis- sions, I contend that a full accounting of all anthropogenic forcing agents is necessary. Precise definition of a safe limit is beyond the scope of this work, but key aspects are by now well known. An unvarnished assessment of the evolving science requires: (i) the rise in TS be restrained to an upper bound of 2 ℃, and quite likely less; (ii) a markedly constrained LLGHG emission budget to the middle of this century; and (iii) a long term atmospheric CO2 concentration substantially below current levels. Further salient con- siderations were examined in section 3.4 — most importantly, that the rate of change in
TS must be given equal weight in defining a safe limit. antification of the overarch- ing objective will hence require some combination of target ∆TS maximum and rate of change,1 cumulative LLGHG emission budget, and atmospheric LLGHG concentrations, informed by insights from the RCP effort (section 3.6.3). Planetary energy imbalance provides a vital measure of progress and the commied warming at a point in time (sec- tion 3.2), and is an essential complement to forward modelling studies, especially with regard to reducing remaining uncertainties in accurate representation of aerosol and cloud effects.
e pivotal point of distinction in comparison to existing mitigation policy, such as that examined in Chapter 4, is the explicit recognition of the net negative forcing contribution of current anthropogenic tropospheric aerosol emissions. Further, that this net negative forcing constitutes a contingent benefit in counteracting as much as half of the positive GHG forcing — the aerosol mask. And finally, that the coupled nature of many aerosol and GHG emissions means that this benefit will be eroded in direct consequence of cli-
1 Here careful separation between transient and ultimate commied warming will be necessary (see section 3.4 on page 109).
194 5.1 Policy objective mate change mitigation action, thereby posing a significant risk of breaching safe limit thresholds for temperature rise and rate of change. Accordingly, there are six necessary design criteria that must be satisfied by any policy response designed to meet the mit- igation objective articulated above.2 ese are stated in concise form here as guidance for the detailed discussion to follow.
DC.1 Emissions of all known anthropogenic forcing agents should be included. Prin- cipally this means the greenhouse gases plus tropospheric aerosols and their pre- cursors.
DC.2 Aerosol emissions must be tracked specifically using metrics able to adequately quantify their cooling effects (section 3.5); ie, quantify the aerosol mask.
DC.3 Weakening of the aerosol mask by withdrawal of negative forcing consequent to aerosol emission reductions must be deliberately managed within the mitigation framework, employing suitable intervention so as to ensure that the trajectory of net forcing does not breach safe limit thresholds.
DC.4 e residence time asymmetry (section 2.1.3) of tropospheric aerosols further re- quires that management of their withdrawal must be subject to tight temporal constraints so as to ensure that interannual fluctuation of net forcing does not breach safe limit thresholds.
DC.5 Managed withdrawal of aerosol masking must not diminish either the scale or the pace of GHG abatement.
DC.6 Perverse incentives to pollute (Box 4.1) should be avoided to the greatest extent possible, while ensuring all preceding criteria are satisfied.
ese criteria encapsulate a full expression of anthropogenic climate change as a multi- pollutant, multi-effect, counter-outcome, split-horizon pollution control problem (sec- tion 4.3.2).
2 A seventh operational criterion is added on page 228 to strengthen the list here.
195 Chapter 5 Aerosol-integrated mitigation policy criteria
A policy response that satisfies these criteria will undertake a managed withdrawal of anthropogenic aerosols as a fundamental task, in parallel with necessary abatement of GHG emissions within an integrated mitigation framework. As per criterion DC.3, this managed withdrawal refers to deliberate acts of climate intervention that compensate for the progressive weakening of the aerosol mask as their emissions fall. Categoric- ally and unequivocally, that compensation must be in addition to, separate from, and not dependent on parallel GHG abatement, as criterion DC.5 demands. For the reas- ons detailed throughout this thesis, the inescapable need is to maintain a substantial proportion of the masking effects thus far provided by anthropogenic aerosols, thereby preventing dangerous transient warming during the decades-long structural transform- ation to decarbonised societies. is deliberate masking will continue until such time as the residual positive forcing of remaining atmospheric LLGHG concentrations can be considered safe and the need for ongoing intervention consequently relaxed. e task is in effect a balancing act between urgent GHG abatement and compensating for the consequences of that abatement on aerosol masking of remaining atmospheric LLGHG concentrations.
5.2 Discussion and scope
e policy response necessary to enact the managed withdrawal is discussed in detail below. Preliminary terms are first defined and required technological capabilities out- lined. From this foundation, the remainder of this section investigates the implications and challenges for development of effective solutions. e discussion ends by seing the boundary for analysis of specific implementation models presented in section 5.3.
196 5.2 Discussion and scope
5.2.1 Definition of terms
Key terminology is now defined as a preliminary basis for discussion. Some terms will be expanded subsequently. positive forcing agent (PFA) e greenhouse gases and those aerosol tropospheric species with an overall warming influence through exertion of a net positive cli- mate forcing. negative forcing agent (NFA) ose tropospheric aerosol species with an overall cool- ing influence through exertion of a net negative climate forcing.3 Unfortunately, this term is necessarily difficult to codify due to the limitations of applying radi- ative forcing methodologies to aerosols (section 2.3 and 3.1.1),4 the importance of surface forcing (section 2.5.11), and the highly-contingent nature of absorbing aerosols such as BC (whether or not they induce net warming through direct and semi-direct effects, or net cooling through cloud indirect effects). Species unequi- vocally covered by this term are sulphate, OC, SOA, nitrate, and mineral dust. carbon price mechanism For the purposes here, a domestic GHG abatement instru- ment can be taken as a given, because it need not markedly differ from those already suggested or in practice. ese are all likely to be market incentive based (emission tax or emission trading scheme), so the GHG abatement instrument can be referred to as a carbon price mechanism. is term then serves both as short- hand for that instrument, as well as the understanding that the instrument employs a price signal. compensative masking e deliberate intervention to replace some fraction of neg-
3 Explicitly restricting the definition of NFA to aerosols has the corollary that all NFAs are also short lived. e term ‘short-lived climate forcers’ (SLCF) is now used in the literature (eg, Bond et al. 2011; UNEP 2011), however that term captures both aerosol species with strong positive forcing (BC) and short-lived GHGs (tropospheric ozone). As the purpose here is to refer only to negative forcers, NFA is used rather than SLCF. 4 And those limitations also apply to warming aerosols that should be captured by the PFA definition.
197 Chapter 5 Aerosol-integrated mitigation policy criteria
ative forcing now lost as a direct result of changes to anthropogenic aerosol emis- sions. at is, to compensate for a weakening of the aerosol mask. Compensative masking may refer to the aggregate total anthropogenic effort, or to an individual country’s contribution. compensative measure A particular practice or technology employed to achieve the desired compensative masking; expanded in section 5.2.2. compensative masking period Analogous to and likely aligned with the compliance period of the carbon price mechanism, the compensative masking period refers to the operational temporal interval within the mitigation framework that (i) defines the reporting interval and (ii) where compensative masking subsequent to declines in NFA emissions is to occur.
NFA emission baseline e assessed representative historical per-facility NFA emis- sions prior to the introduction of the aerosol-integrated mitigation framework. Depending on measurement capability this baseline value may be an average emis- sion rate over a number of years, or some other representative statistic. It should be explicitly retrospective to avoid any manipulation of emission rates prior to introduction of the implementing mechanisms.
5.2.2 Plausible compensative measures
Compensative measures are the central enabling technologies of the managed with- drawal. While specific masking technologies need not be prescribed, it is useful to clas- sify the broad categories of plausible measures. ree such categories are outlined in Table 5.1. Research into these technologies is underway by a range of groups such as the European IMPLICC initiative examining aerosol injection.5 Further development
5 See http://implicc.zmaw.de/Home.551.0.html.
198 5.2 Discussion and scope and careful implementation of these capabilities will be an important function of the policy response.
At this point it is necessary to acknowledge that ‘compensative measure’ sounds all too much like ‘geoengineering’, yet I have thus far given no direct examination of the term and its connotations.6 Box 5.1 explains this conspicuous absence. e measures catalogued here are indeed forms of geoengineering — but the context of their use is as a last resort as an integral component of genuine mitigation that unequivocally is founded upon GHG abatement, as I have stressed above. A gulf separates what I propose from those who position geoengineering as a means to continue business as usual, the ultimate techno-fix.
Table 5.1: Categories of plausible compensative measures
Category Description Continued particulate emissions Continued deliberate emission of tropospheric aerosol species, though without co-emission of GHGs. Geoengineering: solar radiation Geoengineering techniques classed as solar management (SRM) radiation management, such as injection of sulphate or other aerosol to the stratosphere or troposphere; artificial enhancement of marine sea salt aerosol loading, or other processes to enhance marine cloud fraction or brightness; or manufactured reflective particulates.
Geoengineering: CO2 drawdown Also known as carbon dioxide removal, CDR. Geoengineering techniques classed as drawdown of atmospheric LLGHG concentrations (most likely CO2) by biological or chemical sequestration, beyond and in addition to current natural rates of removal.
6 e only prior mention is indirectly in discussing the likely rate of aerosol-induced temperature change in section 3.4.1, specifically the findings of Mahews and Caldeira (2007) and Ross and Mahews (2009).
199 Chapter 5 Aerosol-integrated mitigation policy criteria
Box 5.1: Geoengineering can never substitute for real GHG abatement
Is this thesis a call for geoengineering by subterfuge? Should the reader interpret the omission of geoengineering as tacit endorsement of it? Does this work seek to insinuate geoengineering as a policy option because, in effect, we’re already doing it? Emphatically not. Geoengineer- ing commonly connotes deliberate intervention in the climate system to supposedly counteract anthropogenic climate change without and instead of serious reductions in actual GHG emissions — ‘end-of-the-chimney fixes’, as Schellnhuber calls them (2011, p. 20277). Commonly geoen- gineering refers to ‘solar radiation management’ (SRM): techniques that essentially mimic the function of scaering aerosols, reducing absorbed solar radiation. In their review, e Royal Society (2009) distinguishes SRM from a second class of techniques that remove CO2 from the atmosphere, ‘carbon dioxide removal’ (CDR), but these are not usually the focus of geoengin- eering proposals. Hamilton (2012) distinguishes between those who embrace such intervention, and those for whom humanity’s ‘penchant for self-delusion’ means that we have no choice. e outcome is the same no maer the intent: geoengineering is no solution to anthropogenic climate change if it permits the slightest extension of GHG emissions. GHG abatement is urgently necessary and an inviolate, immutable, inescapable requirement for genuine mitigation. Solar radiation man- agement in particular does nothing whatsoever to avoid further damage through ocean acidi- fication. Any deployment of geoengineering in ways that allow GHG emissions to continue is simply abdicating responsibility for the manifestly necessary transformation of our socioeco- nomic systems away from reliance on fossil fuels and unsustainable exploitation of natural re- sources and the ecological systems on which we depend. Worse, geoengineering could lead to a form of climate arms race reminiscent of the Cold War and its doctrine of ‘mutually assured destruction’ (Schellnhuber 2011). Or individual countries — even individual groups or persons — may decide to interfere with the climate for their own benefits; some may even wish to in- crease GHG emissions to overcompensate for geoengineered cooling if they perceive warming is in their interests. Ultimately mitigation must arrest the root causes of climate change if success is to be long lived — surely it courts disaster to imagine we might re-engineer the climate by interfering with sun- light or even draw carbon dioxide out of the atmosphere, yet continue to emit the same gases, unabated. What I argue, in contrast, is that geoengineering will be necessary in concert with GHG abatement, possibly for decades. Geoengineering is re-framed as a complement, not a substitute. Only within that explicitly limited tactical context is geoengineering a ‘reasonable’ action, and only because the real task of mitigation has been — and continues to be — so long avoided. In the poignant words of Schellnhuber (2011, p. 20277), echoed in the thoughtful work of Schneider (2008), geoengineering may be ‘the last best hope’ in the event of unrestrained fossil fuel con- sumption this century.
200 5.2 Discussion and scope
Options within and between each category involve differing and largely unknown costs, significant geophysical uncertainties, and the potential for unintended damaging side ef- fects. Costs can be expected to further vary across capital and operational expenditures, with SRM and drawdown likely to have distinctly different cost profiles over time. It is reasonable to expect that SRM will be weighted toward ongoing operational costs that are relatively constant, with lile to no further expenditure once operations cease. Drawdown, in contrast, will have high up-front costs to establish facilities, lower oper- ational costs (largely maintenance and assurance), but with a long tail of monitoring to ensure sequestration.7
Continued deliberate emission of particulate air pollutants is far from ideal, given the established damages to human health, infrastructure, and natural systems. ough it could be argued that the consequences of deliberate particulate emissions are at least beer known than for the alternatives.
SRM via an array of postulated techniques is the closest approximation to the cooling ef- fects of aerosols. In fact, some methods use aerosol species directly, chiefly sulphate. A salient example is injection of SO2 to the stratosphere to generate sulphate aerosols, max- imising atmospheric lifetime by replicating the natural role of large volcanic eruptions. Crutzen (2006)’s essay discussed this possibility, raising interest in SRM markedly. But there is great concern that deliberate human intervention in the stratosphere may have serious negative consequences, such as possible damage to the ozone layer. Schellnhuber (2011, p. 20277) emphasises that Crutzen ‘ has consistently argued then and ever since that such a climate engineering scheme would be implemented out of despair only’. More radical proposals involve installations of mirrors or other reflective structures in space to intercept sunlight before it reaches the Earth, thus effectively increasing planetary
7 e Royal Society (2009) note that geoengineering cost estimates remain largely speculative in the absence of empirical data and experience.
201 Chapter 5 Aerosol-integrated mitigation policy criteria albedo.8
If atmospheric drawdown can be achieved at scale and with acceptable risk of sequest- ration failure, it appears to be the beer option, for three key reasons. First, it tar- gets the root cause of anthropogenic climate change, elevated LLGHG concentrations themselves. Second and related to that, lowering atmospheric concentrations lessens the burden now placed on the oceans to increase CO2 sequestration, arresting the dan- gerous consequent rise in ocean acidity. ird, dependent on the specific mechanisms, sequestration may require lile ongoing action (and funding), beyond monitoring and maintenance of the facilities. Contrast this to the need for constant active intervention in the case of SRM. A major impediment may be exorbitant cost in the case of artificial mechanisms, even if viable technology can be engineered — in the vicinity of US$1000 per tonne CO2 in the case of industrial air capture (Schellnhuber 2011, citing House et al. 2011). However, if used to complement not substitute for wholesale GHG abatement, perhaps those costs will be tolerable.
Improved land management and forestry are also possible sequestration options. In fact, negative CO2 emissions in this manner form part of the RCP2.6 strong mitigation scen- ario (Schellnhuber 2011; Meinshausen et al. 2011); though Meinshausen et al. (2011) note doubts as to whether biosequestration could be sustained for extended time periods.
Another alternative may be to capture and sequester CO2 from biomass combustion, es- timated to cost far less than air capture at around $150 to $400 per tonne (Schellnhuber 2011); though this requires available and reliable sequestration facilities.
8 Lunt et al. (2008) examined this ‘sunshade’ geoengineering in a GCM study and found that it would work, but they reiterate that continued ocean acidification means it is not an alternative to GHG abate- ment.
202 5.2 Discussion and scope
5.2.3 Which emissions, which sources?
Comprehensiveness demands that emissions of all climatically-significant forcing agents should be included, as per criteria DC.1 and DC.2. e primary intent is to introduce deliberate mitigation coverage of anthropogenic tropospheric aerosols that contribute to the radiative mask (NFAs).9 Further, other GHGs beyond the six classes captured by the Kyoto Protocol should arguably also be included in some form, tropospheric ozone precursors being a prominent example (PFAs).10
e qualifier ‘climatically-significant’ obviously allows for the exclusion of some agents if their forcing contribution is a small fraction of the total, though how specifically sig- nificance might be defined is not codified here. Practical operational constraints in areas such as measurement capability or the geographical location of emission point sources are also potential limiting factors in determining coverage thresholds — some less com- mon species or emissions from certain point sources may be too difficult to track effect- ively (with current technology). While unavoidable to some extent, policy makers must be vigilant in striving to minimise such limitations. If coverage is too narrow and too many species or point sources excluded, the framework’s environmental effectiveness will quickly degrade, compromising the policy objective. A targeted scientific review to provide explicit advice on these questions is therefore firmly in the public interest.
e mitigation framework must cover the principal negative forcing agents: the sec- ondary aerosol species of sulphate and nitrate, and hence their precursor gaseous emis- sions; and the primary particulate OC and biomass burning aerosols.11 SOA, mineral dust, and other less prevalent forms of particulate maer may need to be included also, but beer quantification of their emissions and forcing contribution is required for an accurate assessment. It must be highlighted that coverage of BC is quite problematic
9 Refer to section 2.4 for the main species of concern. 10 e questionable comprehensiveness of the Kyoto Protocol was discussed in Chapter 4. 11 ough the last is largely made up of non-fossil BC and OC.
203 Chapter 5 Aerosol-integrated mitigation policy criteria as the only significant aerosol PFA. Effective inclusion may not be practicable until the serious limitations of existing metrics examined in section 3.5 are resolved. And there is the question as to whether BC in fact causes a net negative forcing through currently poorly-quantified indirect effects (section 2.4.2.3 and 2.5.2), hence is not readily classi- fied as either PFA nor NFA in any absolute sense. If, for the interim, BC is categorised as a PFA it is probably most suitable to cover its emissions either under the carbon price mechanism directly, or as a regulated air pollutant.
Which economic sectors and activities are to be covered? ose that generate coupled emissions are the focus, as per criterion DC.3. GHG abatement obligations applying to these activities will drive reductions in NFA emissions, giving rise to the negative abate- ment feedback (section 4.5.2). If we assume that aerosol-integrated mitigation is imple- mented additional to an existing carbon price mechanism, then those coupled emission activities should already be covered. But what of activities that generate non-coupled NFAs only; ie, no GHGs? Should they now also fall under the framework, their emissions tracked and perhaps subject to compensative masking obligations if they decline? As air pollutants, it is likely that these sources will be covered by any pre-existing air quality regulations, hence raising the difficulty of scheme linkage, the issue of double-regulation and double-coverage,12 and potential conflicts between respective policy objectives and legal instruments.13 A reasonable test for inclusion is those activities’ proportional con- tribution to total NFA emissions and the extent to which they are expected to decline over time — if the contribution is small or declines are likely to be slow, such activities could be exempted, but subject to periodic review as experience is gained and measure- ment capability evolves.
What of coupled emissions not commonly covered fully by carbon price mechanisms? Example sectors include agriculture, transboundary activities (aviation, shipping), some
12 See the Oikonomou and Jepma (2008) policy evaluation framework. 13 e same is also likely to apply for coupled emissions to some degree, SO2 regulation of coal combus- tion being an obvious example.
204 5.2 Discussion and scope direct household emissions not captured by upstream liabilities, and disperse activities like land vehicular transport (eg, Perdan and Azapagic 2011; Flachsland et al. 2011); al- though the EU ETS now includes aviation (European Commission 2013). ese areas highlight the trade-offs in policy design between maximal coverage in the service of environmental effectiveness, the related cost, and administrative difficulties of imple- mentation. But that tension already exists with carbon price mechanisms in various jurisdictions14 — transboundary activities and agriculture are key instances of poor cov- erage, for political reasons and measurement constraints — so while inclusion of aerosols increases the difficulty, these issues of too-limited coverage ultimately must be redressed more generally in any case.
A further major concern is biomass burning emissions, a significant source of BC and OC. Here it is important to distinguish open burning from biofuel combustion. Emis- sions from open burning of forests and savannah are generally uncontrolled and are likely impractical to capture within aerosol-specific regulatory frameworks. Burning of agricultural waste and especially biofuel is generally deliberate (though still diffuse) and existing or planned regulations could perhaps be expanded in future to incorporate aerosol emissions in some way. However, developing countries are the largest source of biofuel emissions, raising pressing questions of equity and institutional capacity. Should such countries be obliged to expend scarce funds on compensative measures, given the serious health damages caused by domestic biofuel use and BC’s significant positive for- cing contribution? Open burning emissions are more globally distributed, with signific- ant contributions from industrialised countries in Europe, North America, and Australia, alongside those of developing countries in Africa and Asia (van der Werf et al. 2006). Yet developing countries are the least responsible for the cumulative LLGHG burden that threatens dangerous climate change — why should they be expected to pay for the loss
14 e Australian mechanism, for example, excludes the majority of land transport in its initial years by exempting transport fuels; although a previous proposed legislative scheme was to include them.
205 Chapter 5 Aerosol-integrated mitigation policy criteria of aerosol emissions that have actually helped suppress it as a result of mitigation ef- forts that render improvements in forest protection or land management? Moreover, is there a realistic capacity to do so, either financially or as regards monitoring and other technologies?
In the near term the most reasonable approach may be effort to expand aerosol meas- urement capabilities for both categories of biomass burning to beer assess the future need for any compensative masking. Importantly, the REDD and REDD+ mitigation programs15 seeking to arrest GHG emissions from deforestation and degradation by providing financial incentives to developing countries may offer a means to include ad- ditional compensative masking payments if aerosol emissions show significant declines over time.
5.2.4 Measurement and reporting
If we are to manage the withdrawal of aerosol emissions we must be able to usefully measure them. New units of account are necessary to do so. In fact, section 3.5 strongly suggests a general need to review the metrics and accounting methodologies used to track climate change. Such a review might identify a metric capable of accurately rep- resenting both GHG and aerosol emissions; or more correctly in the context here, PFA and NFA classes. But the need is clearly for an effective means to quantify the specific effects of NFA emissions. On that basis, there is no requirement for the chosen metric to facilitate trading (inter-species substitutability within the policy instrument), in the manner of GHG emissions under the ‘basket approach’. In effect, PFA and NFA emis- sions are to be treated separately, hence two distinct metrics may be employed with no requirement for ‘translatability’ between the two.
15 For details about these programs see the UN-REDD site at http://www.un-redd.org/.
206 5.2 Discussion and scope
Explicitly then, a suitable metric for NFAs is one that:
• quantifies climatic forcing by unit mass emission;
• does not assume ‘equivalence’ with CO2 or any other GHG;
• adequately captures the short atmospheric residence time of aerosol species (sec- tion 2.1.3) as an assimilative pollutant (section 4.1.1);
• is not reliant on (long term) integrals of radiative forcing;
• does not require the selection of time horizon;
• is capable of quantifying regional variation; and
• captures indirect effects on cloud and other complex processes (section 2.5).
Of the metrics currently available, the specific forcing pulse (SFP) examined in sec- tion 3.5.2.2 appears to be a good fit for tracking NFA emissions. While no metric yet fully represents aerosol indirect forcing,16 SFP can readily be updated to redress this limitation as modelling capability improves. Note, however, I assume here that SFP can be developed for sulphate and other relevant species beyond OC.
If we assume that SFP or some other suitable metric is defined, the next issue is the particular requirements for reporting. Non-coupled NFA emiers will presumably be liable reporting entities under any pre-existing air quality regulations; however, as typ- ically only aggregate particular maer data is reported — ie, not species-specific — these regulations are likely to be inadequate in their current form. It may be that coupled emiers also already report in-scope NFA emissions, but that will vary by jurisdiction and cannot be guaranteed. Irrespective of the existence of air quality regulations, NFA emissions per species should be reported explicitly for climate change mitigation pur- poses in a consistent and integrated manner. Logically that in turn suggests that any air quality reporting regulations be unified and harmonised with those for climate change
16 To do so inherently requires that full representation in climate models.
207 Chapter 5 Aerosol-integrated mitigation policy criteria
— the set of all species relevant for these two domains are, aer all, pollutants. Harmon- isation may also lower net administrative and compliance costs for all reporting entities and for the regulator(s).
Minimum emission thresholds for reporting and any consequent liability must also be carefully considered. If liability thresholds are too high, too few firms will be covered if total emissions stem from a large number of point sources with sub-threshold emis- sion rates. Conversely, lower thresholds that capture larger numbers of low-emiing firms may impose cost and administrative burdens beyond what is strictly necessary for environmental effectiveness. One possible solution for the problem of a large num- ber of small point sources may be to use sectoral aggregates in establishing obligations under the scheme. A central agency could collate aggregate sectoral emissions, and in turn satisfy any liabilities for compensative masking that may result from declines over time rather than the individual firms.17 However, measurement capability may not be sufficient to facilitate this at acceptable cost, so some estimation or averaged emission factor could be applied for sub-threshold point sources (akin to economic input/output data), informed by periodic measurement and inspection of those sources by the agency. Other trade-offs of cost against practicality are probable in seing reporting and liability thresholds, so a level of inaccuracy ought to be assumed at the national level. Under- estimation is of greater concern that overestimation, so in turn the total compensative masking obligation may have a ‘buffer’ applied to reduce risk, at least for the early years.
5.2.5 How much and for how long?
How great a proportion of a weakening aerosol mask must be replaced by compensative masking — and for how long — will be determined by a complex and interrelated set of factors. Prominent among them are the trajectory of LLGHG emissions and rising
17 ese questions are discussed further in section 5.3.
208 5.2 Discussion and scope atmospheric concentrations, in relation to targets commensurate with restraining tem- perature rise to some maximum limit; how close temperature rise is to breaching that limit; how close identified climatic tipping points are to being surpassed; indications from planetary energy imbalance, as a measure of evolving net forcing; and continual revisions of risk and the consequent definition of dangerous anthropogenic interference, tempered by unfolding climate damages, which may alter these agreed targets.
It seems likely that individual countries will adopt differentiated responsibilities — as they have done for GHGs, ostensibly, under the Kyoto Protocol — taking account of equity concerns, technical capability, and ability to pay, as well as underlying aero- sol emission profiles. However, it does not necessarily follow that per-country bur- dens for the provision of compensative masking will mirror those for GHG abatement. For example, the wealthier industrialised countries could be expected to compensate for the bulk of their cooling aerosol emissions, even 100 % for some initial period, so as to counterbalance markedly less compensation in the developing world, in line with his- torical responsibility for causing the climate change problem. Again, biomass burning is a prominent challenge here, as aerosol emissions from biofuel combustion are much higher in developing countries, especially BC (Ramanathan and Carmichael 2008).18 e regional effects of aerosol emissions themselves may also require that compensative masking be deployed in specific geographic locations in ways that diverge markedly from current approaches under Kyoto.
Adopting a cautious stance, it is prudent to assume that compensative masking will be required for many decades until safe atmospheric boundaries are firmly emplaced, climate damages are controlled, and human industrial structures are reconfigured on a sustainable basis. e policy response should therefore be designed with long-term operation in mind.
18 Recall also the important distinction between open burning and biofuel combustion noted in sec- tion 5.2.3.
209 Chapter 5 Aerosol-integrated mitigation policy criteria
5.2.6 Temporal and spatial considerations
Short aerosol atmospheric residence time has an important bearing on policy design. GHG abatement instruments can in general allow significant temporal emission vari- ation as emissions in a given year are not critical for temperature targets, providing total emissions over a number of years meet caps on average and stay within cumulat- ive budgets. e same does not apply for the short-lived NFA species, because changes in emissions translate directly and in effect immediately to changes in atmospheric loading and hence net forcing (section 2.1.3). e requirement to ensure that net forcing does not breach specific thresholds (criterion DC.3) therefore necessitates that variability in compensative masking is subject to rigorous temporal constraints.
ose constraints will be set by expert scientific advice, but in general terms variation in net compensative masking must be maintained within a narrow tolerance range. A sustained breach of that range could lead to modified net forcing sufficient to induce damaging temperature changes, given the emerging evidence of a faster response func- tion (section 3.3.1) and approaching tipping points (section 3.4). Recall that the rate of increase in temperature is itself a crucial factor. Fluctuations in compensative masking may be possible within, perhaps, a single year, but must return to the target masking level within the following year, before substantial changes in climate response occur.19 e compensative masking period must therefore be defined to reflect these require- ments. e period should align with the necessary constraints but may possibly have finer temporal resolution (ie, if the constraint is yearly the period could perhaps be set at six month intervals, but not biennially).
A suitable (though imperfect) conceptual analogy is the requirement for electricity net- works to be operated within the strict bounds of the ‘technical envelope’ at all times — divergence beyond these limits in key parameters such as voltage stability and fre-
19 ese time frames are illustrative so as to make the principle clear.
210 5.2 Discussion and scope quency will result in failure of the grid and loss of supply; hence, generation (supply) and load (demand) must be closely matched continuously. We might label the analogous construct a multi-parameter forcing envelope, which includes both the absolute value of net forcing and its rate of change.
Accordingly, controls must be in place to manage the risk that a private compensat- ive masking provider may fail, resulting in a significant undersupply and consequent breach of the forcing envelope tolerance. is risk relates primarily to SRM measures, given they must be continually replenished. Drawdown measures entail a different risk profile in their requirement for long term monitoring and maintenance of sequestration facilities — in all likelihood far longer than any human experience. Risk of failure here is then also a concern, if less acute, given revenues for the service may expire while op- erational costs continue. ese risks of undersupply must be carefully considered in the design of the policy response.
e need to maintain compensative masking within a strict tolerance range also pre- cludes the use of intertemporal flexibility mechanisms for NFA emissions within the mitigation scheme. Equivalent flexibility in carbon price mechanisms in the form of per- mit banking and borrowing (assuming emissions trading) cannot be extended to cover either NFA emissions or compensative measures. at is, excess NFA emission or com- pensative masking in one period does not cancel out a matching quantitative shortfall in subsequent periods, due to the temporal sensitivity detailed above — the aerosol mask must be maintained within the established tolerance range at all times. However, draw- down compensative measures could potentially allow a limited form of intertemporal flexibility because of their reduction in actual LLGHG positive forcing. Temporary cred- its for excess drawdown may provide a solution, similar to the general form discussed by Maréchal and Hecq (2006, and references therein), but assessment of this potential is not provided here.
Spatial and temporal considerations are also a factor for the actual aerosol effects be-
211 Chapter 5 Aerosol-integrated mitigation policy criteria ing replaced — and therefore also for SRM compensative measures. A given NFA unit mass emission in geographic location A may produce a substantially different effect were it to be emied in location B. e assessed obligation for compensative masking may consequently differ between the two locations for nominally equivalent emissions. Conversely, spatial effect variation could also modify the efficacy of particular compens- ative measures, changing the effort and expenditure required to achieve a given target masking (easier and cheaper, or harder and more costly). Further, emission in a source geographic location A may produce forcing effects in a separate receptor site (location B), and efficacies there may differ from those of the source region. Moreover, tempor- ality influences all of these complexities further still, as effects may vary substantially depending on when emission occurs, including not only seasonally but even the time of day.
Taken together these factors suggest that SRM compensative measures may need to be licensed with specific spatial constraints that limit, or mandate, where and when they may be deployed. However, the additional complexity and likely cost involved would need to be weighed against these differences in forcing effect.
5.2.7 Compensative masking services
Developing compensative masking capability will require the creation of service pro- viders able to deploy particular approved compensative measures. An obvious question is whether that service can be provided by private firms, or restricted to a function of the state. Assuming appropriate regulation and oversight, there is no a priori reason why private firms cannot be licensed to provide compensative masking. at is, compensat- ive masking can be established as a competitive market. Licensed private firms would compete on cost for the provision of approved compensative measures, along with their aendant operational requirements, selling those services within suitable geographical
212 5.2 Discussion and scope constraints. Economic efficiency suggests such a market will be the least cost approach and would foster technological innovation over time. With careful regulatory oversight within strict boundaries, private firms should be able to deliver the necessary compens- ative masking at an acceptable level of risk.
A compensative masking market could be instituted in a number of forms, such as bilat- eral contracts, contract auctions, tradeable certificates,20 potentially even a spot market. e appropriate market form depends on two key factors that determine the manner in which demand for these services is created: the definition of liable entities bearing oblig- ation to procure compensative masking; and the actual purchaser of masking services in the market, which may be an actor separate to those liable entities (such as a regulator).
5.2.8 The question of accounting
e managed withdrawal of anthropogenic aerosol emissions requires that they are ad- equately measured, as per criteria DC.2.21 Once quantified, the question of how to account for them within existing mitigation frameworks inevitably arises. In examining this question of accounting rules, it is useful to distinguish between the international and domestic contexts.
Proper consideration of criterion DC.5 excludes approaches within international agree- ments that treat NFA emissions as a negative national budget term, subtracting them from the positive GHG emissions term (as previously discussed in section 4.4.2) and thereby diluting per-country GHG abatement obligations. Domestically, criterion DC.5 should similarly prohibit the use of NFA emissions as offset or credit instruments within existing national carbon price mechanisms;22 additional arguments for such prohibition were given in section 4.4.4. e underlying reason is the same in both contexts: by al-
20 Analogous to green certificate schemes for renewable energy. 21 See section 5.2.4 for detail on measurement. 22 Detailed analysis of offset and credit is given in section 5.3.5.
213 Chapter 5 Aerosol-integrated mitigation policy criteria lowing NFA emissions to be counted as, in effect, ‘negative GHG emissions’, the quantity or rate of real GHG abatement is reduced. is is so because an abatement obligation under a domestic carbon price mechanism or international agreement can now be sat- isfied by credits derived from the masking effects of existing NFA emissions, displacing genuine reductions in net GHG emissions.23
Nominal GHG abatement targets are thereby likely to be achieved at lower cost by virtue of the diminished requirement for actual structural adjustment of the national economy. at is, the requirement for and progress in structural adjustment away from GHG- generating activities is lessened by the quantitative reduction in real GHG abatement target because of the NFA accounting offset. Perverse mitigation outcomes are possible here: countries with substantial NFA emissions but few air pollution controls are able to claim significant progress toward their GHG abatement target; yet, as many aerosols are generated by coupled emission activities, these same countries are also likely to be substantial GHG emiers in the first place. Aerosol cooling would hence begin to mask perceived GHG emissions, as well as their climatic effects.
Such an outcome is in violation of the fundamental requirement to prevent exactly that distortion — the express purpose of criterion DC.5. National and global GHG abate- ment targets simply must be based on actual declines in underlying GHG emissions, not ‘cooking the books’. Here it is important to unambiguously recognise the equitable and just principle of ‘common but differentiated responsibilities’ under the UNFCCC: while GHG abatement targets themselves must be set in undistorted GHG abatement terms, in no way should that statement be interpreted to undermine the agreement that the in- dustrialised countries ‘should take the lead in combating climate change and the adverse effects thereo’ (Principle 1, Article 3, United Nations 1992). Developing countries may have GHG abatement targets that actually allow emission increases, in accordance with that principle.
23 Or from existing non-GHG-abatement credits, suspect though these oen are.
214 5.2 Discussion and scope
It follows from the non-distortion criterion that the facilitation of tradability among
GHG species via the basket approach (in CO2-e terms) cannot be extended to aerosol species because, as explained in section 4.5.1, aerosols are not meaningfully equival- ent to carbon dioxide nor any other GHG. at is, GHG species are not logically inter- changeable with aerosol species in general and NFA aerosol species in particular — legal instruments created to represent the two classes (GHGs and aerosols) therefore cannot be inter-traded and are not fungible within or between international or domestic climate change mitigation schemes.24
Given these constraints, it is difficult to identify any method by which accounting rules could be revised to directly include NFA emissions within existing mitigation frame- works, domestic or international. A viable solution to this problem will instead re- quire the different categories of forcing agent — positive and negative — be tracked separately, prohibiting tradability between them. Such an approach might involve an aerosol-specific instrument operating in parallel to GHG abatement instruments, with individual targets set for each.25 Rather than artificially combining the two categories of forcing agent in the pursuit of a single target, as would be required for direct inclusion within existing frameworks, this ‘two-track system’ instead links the two by coordinat- ing aerosol– and GHG-specific targets under a unified assessment of temperature rise, planetary energy imbalance, and other relevant factors. A two-track system gives the necessary visibility to the role of aerosol cooling, while the separation between tracks ensures that GHG abatement is not compromised.
24 e corollary of basket approach-like tradability among individual aerosol species is more of an open question, but that does not affect the current discussion. 25 e use of a parallel targets framework is briefly canvassed by Rypdal et al. (2005, citing Fuglestvedt et al. 2000), though they referred to short versus long lived climatically active species; ie, tropospheric ozone would be included in the short-lived species target, along with cooling aerosols.
215 Chapter 5 Aerosol-integrated mitigation policy criteria
5.2.9 Domestic implementation
e overarching policy objective stated in section 5.1 applies at global scale, but imple- mentation will predominantly occur within national jurisdictions. Aspects of the link from global to national have been examined in the preceding discussion, but a number of important questions remain. What then are the requirements for individual coun- tries? Should all countries have the same obligations, met in largely the same ways, or should differentiated responsibilities be applied here also? Indeed, if obligations are set with respect to current NFA emission rates, does that not advantage industrialised countries who have reduced their emissions, achieving much improved air quality, while developing countries now contribute a large fraction of the aerosol mask that has global benefits (see section 3.6 and 3.7)? Will each individual country manage or commission their own compensative measures, or might a new international agency be created to undertake those efforts, or at least oversee their coordination among participating na- tional schemes? How are the equity and capability issues raised in section 5.2.3 to be handled? More broadly, how might an international regime agreed in good faith come to be manipulated or ‘gamed’ to secure unfair advantage; what safeguards should or could be put in place to prevent that? Will any serious aempt to discuss integration of aer- osols in climate change mitigation negotiations end up being lile more than a stalking horse for geoengineering, bolted on to business as usual as in Box 5.1?
ese are extraordinarily difficult questions with no ready answers; doubtless, there are many more. Future international negotiations that grapple with them are certain to be as fraught and complex as anything seen to date.26 And I cannot hope to solve them here.
Rather, my goal is to contribute to the beginnings of a dialogue that might discover workable solutions. Seeking to identify and account for all possible permutations of
26 Indeed, recall from p. 172 in Box 4.1 that Rypdal et al. (2005) concluded such a task too complex to be feasible.
216 5.2 Discussion and scope international negotiation overwhelms that intent. But by limiting the task to an incom- plete yet meaningful subset of the full problem, possible solutions can be investigated and useful insights developed.27 In that light I will make a set of assumptions that narrow the array of variables to be considered, to limit uncertainty.
Accordingly, let us consider the case of an aerosol-integrated domestic mitigation instru- ment applicable to a country in Kyoto Protocol Annex I. e boundary is drawn to rep- resent an industrialised country (such as Australia) subject to a successor international mitigation Protocol under the UNFCCC (or equivalent). An economy-wide GHG abate- ment scheme is assumed to be already in place (the carbon price mechanism defined in section 5.2.1), and domestic aerosol emissions are now to be included within an extended overall mitigation framework. Specifically, assume that:
I e country participates in an aerosol-integrated UNFCCC mitigation Protocol; where
II Each country is obliged by that Protocol to abate greenhouse gas emissions, and accordingly to deploy endorsed compensative measures consequent to reductions in domestic coupled negative forcing agent emissions; and
III e equivalent compensative masking quantity to be maintained is defined exo- genously by the Protocol as a proportion of assessed baseline negative forcing contribution, which is expected to decline over time as per section 5.2.5; and fur- ther
IV Solar radiative management compensative measures should be implemented do- mestically where possible, but allowing transboundary flexibility through joint im- plementation projects (or similar) in recognition of the practical difficulties faced by countries of small geographic size; verified drawdown may be obtained from any other certified participating country, including non-Annex I.
27 Althaus et al. (2007) counsel that effective policy analysis first requires the problem be defined with sufficient structure and boundary, so that there exists a realistic prospect that a solution can be found.
217 Chapter 5 Aerosol-integrated mitigation policy criteria
With these assumptions in place we can now investigate the design models for imple- menting a domestic mitigation framework that satisfies these stated international obliga- tions and the defined policy objective. For clarity, that domestic framework is henceforth referred to as the Scheme, and the international agreement as the Protocol. Importantly, the masking proportionality factor of assumption III may conceivably exceed 1. is is plausible if it is determined that industrialised countries should shoulder additional compensative masking burden so as to avoid shortfalls; section 5.2.3 examined possible reasons why that may occur. Let us label this the ‘Protocol factor’, to be referenced by the implementing Scheme model.
5.3 Implementation model evaluation
A range of possible Scheme implementation models were considered in the process of developing a workable policy instrument able to fully satisfy the design criteria laid out in section 5.1. Four of those models are evaluated in the sections to follow, as well as related analysis of their political economy with respect to the interdependent questions of who pays, when, and for what. All four models were found to be flawed. ey are presented here, however, because by analysing their features, implications, and their flaws, valuable insights are obtained that lead to and support the fih model ultimately adopted — the subject of Chapter 6. To that end, each model is presented with an over- view of the proposal stating its key features, analysis of its function describing the flaws identified, and concludes with the lessons learned. For simplicity, the models consider effects applying to a hypothetical firm operating a single coupled emission facility.
218 5.3 Implementation model evaluation
Box 5.2: e polluter pays principle
A prominent consideration in seeking to design an aerosol-integrated mitigation framework is the legal principle that requires those that produce pollution to pay for the damages caused. e polluter pays principle is a normative doctrine of environmental law. Although its pre- cise legal definition remains elusive, the core of this principle stems from the fundamental, logical, and fair proposition that those who generate pollution, not the government, should bear pollution costs. e principle underlies much of modern environmental law, and in re- cent years, has become increasingly important in guiding environmental policy, especially at the international level. (Nash 2000, 466)
5.3.1 Loss-liability model
Proposal
An obvious starting point is to impose an additional obligation on coupled emiers that requires them to fund the provision of compensative masking as the carbon price mech- anism induces their NFA emissions to decline. As each liable firms’ NFA emissions de- crease relative to their established historical baseline, they are replaced by compensat- ive measures in the compensative masking period following withdrawal. e obligation could take a number of specific forms, such as liable firms purchasing compensative measures directly, or via stipulated payments to a regulatory authority which manages that procurement. Critically, the compensative masking obligation is cumulative: new obligations arising from NFA emission decreases in a given period are summed with all previous periods’ obligations, which are ongoing.28 In this way the cooling offset is maintained but perverse incentives to pollute are avoided and the impetus for GHG abatement is not compromised. We can label this the loss-liability model.
28 ough drawdown measures might provide masking for a number of periods, up front.
219 Chapter 5 Aerosol-integrated mitigation policy criteria
Analysis
Because the compensative masking obligation is additional to liabilities existing under the carbon price mechanism, it is an additional cost impost to liable firms. At first blush this appears entirely appropriate. As an extension of existing mitigation frameworks that internalise negative production externalities, firms would be required to pay for their pollutant emissions in line with established legal norms — the polluter pays principle defined in Box 5.2. ere is a flaw in that logic, however, for implementation of compens- ative measures is qualitatively distinct from traditional pollution abatement. Polluters are, as usual, obliged to reduce their GHG emissions, incurring the overall abatement costs for doing so, or pay for their continued emission under the carbon price mechan- ism. Yet because those emissions are NFA-coupled, polluters are now further required to pay for compensative measures to replace these now-removed NFA emissions.
Firms can therefore rightfully claim to be subject to double-regulation (by two climate- related instruments) and to a form of double-coverage (a single activity covered twice) involving two separate costs.29 Worse, this double coverage is contradictory: they are first forced to pay the cost of abatement, and then pay a second cost to compensate for the NFA pollutants thus removed. Moreover, when viewed as providing a contingent benefit, these NFA emissions are no longer accurately characterised as exclusively causing mar- ginal external damages, so a designation of negative externality is far from clear-cut.30 is implementation difficulty is a manifestation of the ontological challenges evinced by Chapter 4. In such a conflicted policy regime, the appropriateness of additional cost impost is certainly questionable.
Such ‘philosophical’ difficulties might nonetheless be dismissed if the necessary object- ives are met, but there are two further consequences that cast stronger doubt on the
29 See Oikonomou and Jepma (2008) for an overview of these issues. 30 ough a counterargument might be the inverse: NFA pollutants have even greater external costs than previously understood, because of their need to be replaced owing to the large damages arising from coupled GHG emissions.
220 5.3 Implementation model evaluation feasibility of doing so. First, the additional obligation necessarily raises the cost of pro- duction for liable firms. If abatement occurs, total cost is then the sum of compensative masking expenditure and the underlying marginal abatement cost. Alternatively, a li- able firm may defer abatement until this total cost is less than the prevailing carbon price, hence GHG emissions are continued and incur larger carbon price liability. Both possibilities result in production costs greater than would occur in the absence of the aerosol-integrated framework. is burden may affect the underlying economic viabil- ity of those firms over time to such an extent that they become insolvent or bankrupt, ceasing operations altogether; ie, more rapidly than would have been induced by a car- bon price signal alone. ose firms’ residual NFA emissions consequently cease imme- diately, yet they are no longer funding the provision of compensative masking for their previously-withdrawn NFA emissions. eir total masking contribution will therefore be lost without government intervention — the net loss in masking effect is a quantity equivalent to the firm’s NFA emission baseline.31
Second, imposition of such an additional cost poses a serious political obstacle, to say the least. Given the oen tortured history of mitigation policy-making to date, it is certainly possible that political and legal challenge from affected coupled emier interests would be so determined and protracted that Scheme implementation is greatly delayed. ough sound public policy should not be held hostage to such narrow sectional interests, this is a very real practical constraint on effective legislative action that cannot be ignored. Under the assumption that a carbon price mechanism is already in effect, significant quantities of coupled NFA emissions may be lost during such delays due to the negative abatement feedback.
A variation of the loss-liability model would substitute ex ante environmental bonds in
31 ough the reduced net forcing resulting from previously-purchased drawdown measures would not be affected in the short term.
221 Chapter 5 Aerosol-integrated mitigation policy criteria place of the ex post obligations above.32 Here liable firms would be required to post a bond determined at Scheme commencement, nominally sufficient to cover their com- pensative masking obligations in accordance with baseline NFA emissions. Several prob- lems exist. First, the issue of additional cost burden remains; in fact, it may be greatly compounded as the bond must cover future prospective NFA emissions for a signific- ant number of years. Second and related to this, estimation of those costs ex ante is extremely unlikely to be accurate in any meaningful way. ird, by definition a bond is intended to ensure a firm’s future compliance with regulatory requirements known in advance: when the mechanism functions as designed, the firm ought to satisfy those requirements in exchange for return of the bond. Yet in this context there is lile reason to expect the firm to be in operation at this future date, and so in practice the bond is in effect a form of insurance — as Gerard and Wilson (2009) note, the long time horizons involved mean that the necessary surety providers may not be prepared to underwrite such bonds in the first place.
Lessons
• Assigning direct liability for provision of compensative masking to coupled emit- ters introduces a risk of unexpected loss of NFA emissions and sudden loss of funding for that masking.
• Additional cost impost on major GHG emission sources carries significant political risk for implementation.
• Funds for continuation of compensative masking are unreliable and potentially volatile in the loss-liability model.
I conclude that the efficacy of imposing additional cost on coupled emission sources is doubtful, both ex ante and ex post, and aempting to do so represents a significant risk 32 Gerard and Wilson (2009) provide a useful review of these ‘financial assurance mechanisms’, examin- ing their potential application for long-term carbon sequestration.
222 5.3 Implementation model evaluation of policy failure. But what if firms did not actually have to pay for that compensative masking?
5.3.2 Cost-neutrality model
Proposal
e main elements of the loss-liability model can be reconfigured to deliver the same compensative masking under the same conditions and with the same liabilities, but at no effective additional cost to liable firms. is cost-neutrality model can be implemen- ted by allowing coupled emier firms to deduct their compensative masking expenses from financial obligations to the state.33 ere are two principal options: diminished carbon price financial liability, or deductions against company tax liability (or the like).
Analysis
Consider first diminished liabilities under the carbon price mechanism. As per criterion DC.5, this option must not suppress GHG abatement — thus the key word, financial liability. If the mechanism is an emissions tax, implementation is straightforward: in- curred compensative masking expenditure is fully deductible against carbon tax liab- ilities payable. e firm then faces the same net cost over the compensative masking period, as the additional compensative measure costs are rendered neutral by deducting that amount against the firm’s carbon tax liabilities for their continuing GHG emissions in that period.34 Any such ex post adjustment on a per-firm basis has no effect on the carbon price signal applying to all other liable entities, which remains the regulated emission tax rate. As the adjustment is cost-neutral for each firm, by definition, there is
33 Note that deductibility is used here in the sense of full cost recovery rather than reduced assessable taxation liability only. 34 at is, that fraction of original baseline GHG emissions that have not yet been abated and are thus subject to the carbon tax.
223 Chapter 5 Aerosol-integrated mitigation policy criteria also no per-firm suppression of effective carbon price signal for continuing GHG emis- sions — the abatement incentive for future periods is unaffected (the full tax rate), as deductions are only permissible aer expenses have been incurred.
If the mechanism is an emissions trading scheme (ETS), implementation is more com- plex. Deductibility cannot be achieved in a way that diminishes the quantity of GHG emission permits that the firm must surrender in the compensative masking period, analogous to deductions against a carbon tax, as this would reduce overall abatement quantity in the Scheme — real CO2-e emissions would be greater than the number of permits, exceeding the emission cap.35 Lowering firms’ permit demand will reduce per- mit scarcity by freeing up displaced permits for other emiers, hence depressing the carbon price signal; although, it should be noted that this price depression may not be a concern in and of itself, only the uncorrected cap-exceedance.
e price adjustment necessary for cost neutrality could instead be delivered by granting firm-specific discount rights for emission permits bought at auction in future compens- ative masking periods. In broad terms, a firm would bid at auction for the quantity of permits needed to cover expected GHG emissions in that period, in the same manner as under an unmodified ETS. However, when paying the government for receipt of those permits, the firm exercises their discount right and payment is reduced by the amount deemed to cover their compensative masking expenditure in the previous period.36 In this way the number of GHG emission permits required by the firm is undistorted, as is their bid price at auction, because the discount right is only exercised to recover com- pensative measure costs already incurred, thereby dampening any incentive to underbid.
e second cost neutrality option is for a deductibility mechanism outside of Scheme financial flows. Designating compensative masking expenditure an allowable deduction against company tax liability is one such mechanism. e details will depend on the 35 Offset and credits already allowed under the carbon price mechanism do not affect this outcome. 36 To avoid possible distortionary effects on the carbon price signal in future periods as the cap is pro- gressively tightened, the lifetime of these discount rights could be strictly limited.
224 5.3 Implementation model evaluation particular taxation framework, but in general terms cost-neutral deductibility should be relatively simple to implement. As the deduction stands outside of the mitigation frame- work, there is no distortion of GHG abatement incentive or carbon price signal. Indeed, by using the broader taxation framework, this option may have lower administration costs and ease of implementation.
e fall in net state revenue will be equivalent under the cost neutrality model, regardless of deduction option (and for each type of carbon price mechanism). at is, equivalent to each firm’s compensative measure costs in the compensative masking period (or in the previous period in the case of ETS permit auction discount rights). However, despite the possibility for lower administrative costs in the company tax option, the principal advantage of price adjustments under the carbon price mechanism is that this ensures revenue changes are contained within the Scheme itself. At the very least, containment has symbolic and political benefits; it may prove the wisest option for legal and regulat- ory purposes also.
One possible reason is that over time actual compensative masking expenditure for in- dividual liable coupled emier firms may come to exceed their carbon price liability. Even though the GHG emission cap is progressively tightened, driving the carbon price upward, liable firms may eventually have small enough residual GHG emissions that their resultant carbon price costs (unit emissions multiplied by price) are less than their ongoing compensative measure costs. is inverse relation is a possible general char- acteristic of aerosol-integrated mitigation over time: as the carbon price rises, GHG emissions fall; but compensative masking must increase to replace the cumulative lost coupled NFA emissions, hence compensative measure costs rise. We cannot meaning- fully predict when or even if this cost inflection point will occur for any given firm, but the Scheme must be designed to handle the possibility. erefore, we must examine the potential for compensative masking to become an increasing financial burden of the state.
225 Chapter 5 Aerosol-integrated mitigation policy criteria
Lessons
• Significant constraints may arise for implementation dependant on the form of carbon price mechanism, tax or ETS.
• Containment of compensative masking costs within the Scheme is a desirable design goal.
• Cost-neutrality may not be sustainable over longer periods as firm outlays for compensative masking may come to exceed ongoing carbon price liabilities.
• A general cost-inflection point where compensative masking costs exceed carbon price revenue is likely to exist.
• e Scheme must be explicitly designed to withstand variations in revenues with respect to ongoing funding of necessary compensative masking.
5.3.3 Funding and ultimate liability
We have thus far considered models where coupled emier sources are designated liable entities under the Scheme, obligated to procure compensative masking as their NFA emissions decline, with and without additional cost impost. Recall from the loss-liability model in section 5.3.1 that a major concern is that additional cost impost is likely to hasten the closure of liable firms relative to what would occur under a carbon price alone. e cost-neutrality model of section 5.3.2 is designed to alleviate this closure risk, yet because compensative masking expenditure is still incurred firms may eventually exhaust their deductibility potential against a falling carbon price liability. at is, their obligation to pay for compensative masking will no longer be cost neutral.
ese risks therefore point to questions of ultimate liability for ongoing compensat- ive masking, and the availability of necessary funding during the managed withdrawal. In fact, these are critical design considerations owing to the fundamental distinction
226 5.3 Implementation model evaluation between GHG emission abatement and maintenance of the compensative mask. To ex- plain: GHG emissions continue not as a deliberate action but only as a byproduct of the underlying productive economic activity. Should that activity cease, GHG emis- sions are entirely abated; they do not continue. In that regard, the overarching climate change mitigation objective of GHG emission abatement is achieved — though the intent of policy is to foster technological change of the underlying activity so as to decouple and remove those GHG emissions, rather than for the activity to cease per se. But the requirement for compensative masking to be maintained subject to strict operational constraints, as detailed in section 5.2.6, is the inverse: the activities that produce that masking (compensative measures) are themselves deliberate and must continue to op- erate while compensative masking is required, regardless of the continued existence of the original liable firm.
Cessation of compensative measure activities results in a failure of the system to meet the objective of a managed withdrawal. Consequently, unambiguous responsibility for and funding of continued procurement of compensative masking must be guaranteed in the event that the original liable entity ceases to exist for any reason — only the state can function as that ultimate guarantor.
If the state is to assume ultimate responsibility for procurement of compensative mask- ing, is it sensible to implement any model that first places direct liability on coupled emiers? Only a single failure of a significant liable emier is necessary to trigger the assumption of responsibility by the state. To underscore this point: the Scheme must be robust to the failure of any liable entity at any time. Moreover, it is an expected out- come of a carbon price signal strengthening over time that liable firms with large GHG emissions or large abatement costs will close — some of these are likely to be coupled emiers.37 Further, individual firms or industry sectors may fail in unexpected ways
37 Note that this does not contradict the potential for a deductibility cost inflection point, as a diverse range of firms are subject to the carbon price mechanism with heterogenous emission and operational cost profiles.
227 Chapter 5 Aerosol-integrated mitigation policy criteria due to a myriad of social, economic, or political pressures that cannot be predicted in advance and may have lile relation to the mitigation framework, not the least of which will be the increasing severity of climate change impacts themselves. To restate the in- sights of Taleb (2010): it is simply foolish to make plans that do not recognise and allow for uncertain, unpredictable events that carry large negative consequences.
Lessons
• Ultimate liability for continued funding of compensative masking must be care- fully considered in Scheme design.
• e Scheme must be designed to robustly handle the loss of liable funding entities.
is is a critical implication for Scheme design, so I restate it explicitly as an additional operational criterion of the list in section 5.1:
DC.7 e Scheme must be robust to the failure of any participant entity, ensuring that compensative masking continues to be deployed at the level required for as long as is necessary.
Let us now consider a model that provides that robustness.
5.3.4 Central authority model
Proposal
A central authority model is one possible alternative, characterised as robust to indi- vidual firm failure. Here, liability is lied from the underlying coupled emission source, obligation to procure compensative masking is accordingly removed, and no related costs are incurred by the firm. Rather, the central authority — an independent regu- latory agency of the state — determines the aggregate compensative masking required and procures that quantity from a compensative masking market (see section 5.2.7).
228 5.3 Implementation model evaluation
Analysis
An important advantage of this model is that the authority has economy-wide purview and hence can incorporate demand for compensative masking from a range of sources, not only NFA and GHG emissions of firms subject to reporting obligations under the Scheme.38 In particular, the authority can factor in demand corresponding to the NFA baseline emissions of firms that have since ceased to operate. rough this central au- thority the state assumes total responsibility for compensative masking, thereby address- ing the question of ultimate liability. at leaves the question of funding.
Given an operational carbon price mechanism is assumed, revenue from emission tax receipts or periodic emission permit auction are an obvious possible funding source. Deductibility for compensative masking expenditures39 is a form of revenue redirection, but as source firms are not liable for compensative masking, deductibility is not relevant here. e equivalent under a central authority model is to hypothecate a portion of carbon price revenue to directly fund compensative masking. Hypothecation is logically consistent with the principle of cost-containment within the Scheme (see section 5.3.2, p. 225).
However, redirection of funds arising from carbon price revenue necessarily reduces the amount remaining for other important purposes. Addressing climate change will require the state to fund an expanding array of programs such as industry structural adjustment, household compensation, increasingly vital adaptation measures, interna- tional assistance, and so forth. As climate damages manifest and the abatement task be- comes harder, all these claims on limited funds will rise. Moreover, the revenue stream from GHG emiers can be expected to eventually peak and begin to fall, perhaps precip- itously, as net GHG emissions decline to zero, despite a continually rising carbon price
38 For example, the authority could enact the buffer referred to in section 5.2.3. 39 Examined in the cost-neutrality model (section 5.3.2).
229 Chapter 5 Aerosol-integrated mitigation policy criteria applying to those emissions that do remain.40
Viewed in GHG abatement terms alone, a decline to zero is entirely desirable — indeed, the whole point of a sufficiently strong carbon price is to phase out the underlying pol- lution activities; it is not a revenue raising instrument. Carbon price revenue therefore cannot be relied on to fully fund the actions necessary to deal with climate change over the long term. e combined cost of mitigation and adaptations programs will eventually overwhelm that revenue, making broad government funding inevitable. Compensative masking is no exception.
Lessons
• NFA emissions occurring outside of the Scheme or no longer produced are relevant considerations for any central authority.
• Redirecting carbon price revenue to fund compensative masking reduces the quant- ity available for other necessary programs.
• Carbon price revenue must be expected to fall to zero over the Scheme’s opera- tional time frame.
Finally, it is instructive to now consider a counterfactual model implied by the discussion of offsets in section 5.2.8, despite having ruled it out.
5.3.5 Conferred value model
Proposal
Let us revisit the issue of offsets and credits deriving from existing NFA emissions. To proceed, let us first codify these somewhat vague terms from their prior general usage in section 5.2.8 as follows: 40 is is the more general case of the possible cost inflection point outlined in section 5.3.2.
230 5.3 Implementation model evaluation
NFA-offset an accounting practice whereby the existing NFA emissions generated by
coupled emiers are deemed negative CO2-e units, offseing the assessed carbon price liability of the source firm only (ie, NFA-offsets are non-transferable).
NFA-credit the generation of saleable instruments by any NFA-emiing firm which are then purchased by second-party firms subject to a carbon price liability. is use of ‘credit’ is therefore equivalent to the general concept of additionality cred- its within existing mitigation programs, generated by avoided deforestation, bi- osequestration, improved waste management, and the like.
Consider a scenario where the prohibition against fungibility established in section 5.2.8 is relaxed. Assume one-way fungibility is permissible between separate instruments representing NFA and GHG emissions — in the context of an ETS, the NFA instrument is surrendered in place of a GHG emission permit. e particular method by which that fungibility is implemented need not be specified for this discussion, but will require conversion from unit NFA emission to an ‘equivalent’ quantity of GHG-avoided.
Fungibility confers an economic benefit for the emission of NFA pollutants to the emier: they are assigned positive economic value. We can hence label this the conferred value model. Serious distortions of the carbon price mechanism may result, where the form of that distortion depends on the particular type of instrument, NFA-offset or NFA-credit. Each is examined in turn.
Analysis
An NFA-offset for a coupled emier constitutes a firm-specific reduction in compliance costs, effectively dampening the carbon price signal affecting that firm in comparison to their full GHG emission liability. Reducing compliance costs for continuing GHG emissions subject to the carbon price has the corollary of decreasing that firm’s effective abatement costs. Either it is now cheaper for such a firm to continue producing a nominal
231 Chapter 5 Aerosol-integrated mitigation policy criteria
GHG emission level (reduced by virtue of abatement previously undertaken), or their GHG emissions could return to a higher level for the same nominal compliance cost. e changes to underlying compliance and abatement cost curves are not linear, however, because the coupled nature of NFA and GHG emissions means that GHG abatement also necessarily reduces NFA emissions, and vice versa.
It therefore follows that by returning to a higher GHG emission level a firm may actually be able to lower its costs further, by virtue of the consequent increase in NFA emissions and corresponding increased capacity for NFA-offsets.41 Such shis in per-firm com- pliance cost may be sufficient to alter relative inter-firm or inter-industry abatement costs to such an extent that GHG-intensive firms may continue operation at the ex- pense of otherwise less intensive emission sources that lack significant coupled aerosol emissions. Coal-fired electricity generators are again the archetypal example where off- seing is likely to have material impact: NFA-offsets could lead to an outcome where one of the highest GHG emiing activities faces a lower carbon price signal relative to other electricity generation sources, such as natural gas.
Further, the carbon price signal will be depressed economy-wide when implemented as an ETS, even though NFA-offsets are restricted to the source firm. Price depression oc- curs due to the GHG emission permit-displacement effect of NFA-offsets first described in section 5.3.2 — aggregate permit supply rises as the firm’s permits are freed up, thereby decreasing permit scarcity and hence lowering permit cost for all liable firms. Gov- ernment receipts from permit auction in the subsequent compensative masking period may also decline if excess permits are bankable. Where the carbon price signal is im- plemented as an emissions tax, distortion is limited to the individual firm employing NFA-offsets, as these have no influence on the tax rate applying to other liable firms. In both cases, the real GHG emission quantity is greater than in the absence of NFA-
41 is possibility highlights the need to carefully specify the rules governing calculation of NFA emission baseline.
232 5.3 Implementation model evaluation offsets: emission cap exceedance under an ETS will be the quantity of permits freed up by NFA-offsets (though if cheaper permits were used instead of existing credits, the ex- ceedance is reduced); an emissions tax produces equivalent decreased GHG abatement by the coupled emier.
NFA-credits produced from non-coupled sources (ie, no GHGs are generated) risk further distortions. e sale of deemed NFA-credits to second-party firms allows them to offset their carbon price liabilities in the same manner as described above. In this scenario, the positive economic value assigned to the emission of NFA pollution represents an additional revenue source for the generating firm — and as they are non-coupled sources they have zero carbon price liability, by definition.
e NFA-credit price will in general be less than both the prevailing market price of an equivalent quantity of GHG emission permits, and a purchasing firm’s marginal GHG abatement cost. If the purchasing firm can abate emissions or purchase an emission permit at less cost than the NFA-credit price, we can expect them to do so. e only production costs incurred for an NFA-credit by the source firm will be transactional and administrative: the cost of having the NFA emission deemed by the regulatory author- ity.42 Given the low prices of existing offset and credit instruments available — which usually involve some form of actual productive activity, where NFA pollution is to date a zero-value negative production externality — we can expect NFA-credits to be very low cost. Transactional and administrative costs would serve as a price floor.
Without legislated restrictions, polluting firms therefore have an incentive to continue or even expand NFA emissions, if the profit obtainable from NFA-credit sales is non- zero and greater than cost of control liabilities under any air pollution regulations. e economic value of NFA-credits will hence dampen the appetite with which regulated firms seek to abate their NFA emissions under such air pollution regulations, distorting
42 It is a reasonable assumption that such firms will already be subject to some form of emission reporting obligation.
233 Chapter 5 Aerosol-integrated mitigation policy criteria the price signals of any market-based instruments. Similarly, if controls are implemen- ted via prescriptive technology standards,43 the economic value of NFA-credits serves to reduce firms’ effective NFA abatement costs, even if the level of actual abatement is unaffected — incentives to change underlying production processes or the relative cost profiles of competing firms will then be lessened in much the same way as the use of NFA-offsets distorts the carbon price signal for coupled emiers, described above. If NFA-credits are an option for all firms with GHG emission liabilities — that is, not only source coupled emiers as for NFA-offsets — the depression of carbon price and govern- ment receipts may be greater than for the use of offsets alone. A possible advantage to general NFA-credit availability, however, is a lessened distortionary impact on per-firm compliance costs, because that option is now available for any firm to purchase.
e quantity of real domestic GHG abatement decreases by the sum of the deemed ‘equi- valent’ quantity of NFA-derived instruments within the overall mitigation scheme for both fungibility types detailed above. For both types, the effective carbon price signal is suppressed and government receipts from that carbon price mechanism are eroded. And for both types, the positive economic value conferred to NFA pollutant emissions violates the principle of polluter pays (defined in Box 5.2), which may represent a legal barrier as well as an ethical failing of policy. Moreover, allowing the emission of NFA pollutants to now in some way ‘compensate’ for GHG emissions may constitute pay- ment for the production of local air pollution — and the damages those pollutants cause — whether that payment is direct or indirect and even if this payment is not the policy intent. In violating the polluter pays principle, such payment is clearly also a perverse incentive to pollute and therefore fails to satisfy criterion DC.6. alitatively, this per- verse incentive is at its worst in the case of NFA-credits from non-coupled emission sources. ough constraints could be imposed to limit the scope of these NFA-derived
43 Harris (2006) notes that the commonly used term ‘command and control’ is a loaded one, implying inherent inferiority to market mechanisms and summoning ‘images of failed communist economic sys- tems’ (p. 360, fn. 2).
234 5.3 Implementation model evaluation instruments, this failure to uphold a central tenant of environmental law is unavoidable.
Lessons
• Changes to firm compliance costs under the Scheme alter effective abatement costs.
• Shis in abatement costs may affect the order of progressive abatement action from particular GHG emission source types.
• Under an ETS, permit displacement due to NFA-derived offset mechanisms may place downward pressure on the prevailing carbon price for a given GHG emission cap.
• Credits for non-coupled NFA emissions have no actual production costs other than administrative and transactional; credits from such sources constitute a significant perverse incentive.
• Conferring positive economic value to NFA emissions likely violates the polluter pays principle.
On principle then, and in accordance with the conclusions of section 5.2.8, the use of NFA-credits is again rejected as a legitimate policy option. is position is strengthened by observing that while NFA emissions from non-coupled sources are an important com- ponent of the overall aerosol mask, those emissions are not directly affected by GHG abatement policies and so are not subject to the negative GHG abatement feedback. It cannot be assumed that those emissions will not fall, weakening the aerosol mask, but successfully meeting the challenge of managing the withdrawal of coupled emissions will also provide the means to address declines from non-coupled sources.
e use of NFA-offsets by coupled emiers is similarly again rejected, where implemen- ted in the form detailed above. Diminished real GHG abatement fails to satisfy criterion
235 Chapter 5 Aerosol-integrated mitigation policy criteria
DC.5, as originally argued in section 5.2.8. However, proper consideration of all for- cing agents generated by coupled emiers leads to the uncomfortable conclusion that the potential for significant shis in relative abatement costs engendered by offsets has a defensible geophysical basis: NFA emissions already exist, they currently act to sup- press the full climatic effects of past coupled GHG emissions, and it is these two facts that are the very motivation for the managed withdrawal itself. erefore, if an alternat- ive form of implementation can be devised that does satisfy the non-distortion criterion DC.5, then effective shis in relative abatement costs should not be rejected even if they nonetheless represent a form of perverse incentive. In fact, as distasteful as it may be and despite apparent violation of the principle of polluter pays, an objective assessment of the implications of such shis indicates that some form of conferred benefit may be a necessary component of effective climate change mitigation. e ramifications for abatement pathways are now examined in section 5.4.
5.4 Ramifications for abatement pathways
Abatement (or emission) pathways are possible sequences of abatement activity over time that lead to a target emission reduction goal or, more importantly, satisfy a given cumulative emission budget. e optimal pathway, at least in economic terms, is one that achieves the mitigation objective at least overall cost.44 e job of mitigation policy is then to encourage (or perhaps require) abatement sequences that follow the optimal pathway. Indeed, as Manne and Richels (2004) note, this is a principle made explicit in the UNFCCC.45
e preceding analysis suggests that recognising the offseing effects of coupled NFA emissions within a mitigation framework could lead to significant shis in the relative
44 See, for example, Manne and Richels (2004). 45 Specifically, Principle 3 of Article 3, United Nations (1992).
236 5.4 Ramifications for abatement pathways abatement costs of existing activities. ose shis hence change the relative ranking of particular abatement pathways with respect to optimality: activities with high GHG emission intensity and otherwise high abatement costs have lower costs if their coupled NFA emissions are conferred positive economic value; pathways that include longer life- times for such activities (abatement occurs preferentially elsewhere first) may thereby be less costly than without aerosol-integration. is is a fundamental result of the re- quirement to replace lost NFA emissions with compensative masking — that masking must be paid for, so delaying the loss of NFA emissions delays additional cost.46
e flow-on effects of positioning natural gas as a suitable substitute for coal combustion in the electricity sector, as first discussed in section 4.5.2.1, epitomises such shis in cost. In general terms, the NFA emissions lost when retiring an existing coal-fired electricity generator will have to be replaced by compensative masking — these costs are now added to those of the new gas-fired plant and its carbon price liability. Abatement pathways that involve early removal of coupled emiers and expansion of natural gas substitutes may hence be rendered sub-optimal. In fact, if compensative masking costs are large, early removal of coupled emiers may be sub-optimal in general, even if replaced by low-carbon alternatives. Critically, if compensative masking production capability is missing or insufficient, all such pathways could simply fail to meet their fundamental purpose of avoiding dangerous climate change.
ese ramifications for abatement pathways exemplify the disturbing, even paradoxical, ethical and political challenges inherent in the managed withdrawal. While it is beyond question that further expansion of coal and oil use is a foolish mistake, we must con- front the confounding reality of existing NFA emissions when constructing mitigation policy and the configuration of mitigation instruments, which are a crucial determin- ant of the priority assigned to particular abatement pathways. And so we must ask
46 Alternatively, abatement pathways might be reconstructed to explicitly account for aerosol emissions, giving the same outcome.
237 Chapter 5 Aerosol-integrated mitigation policy criteria ourselves whether it may be wiser to follow a pathway that allows coal combustion47 to continue longer than indicated by GHG-emission profiles alone, instead prioritising GHG abatement from other, low-NFA sources such as natural gas. Is it in fact preferable to continue with existing coal-fired infrastructure rather than substitute for natural gas that in all likelihood must in turn be replaced with genuinely low-carbon alternatives long before that capital stock reaches end of life? Would doing so give the best possible welfare benefit over the long term and at least cost, reluctantly accepting the distressing inherent trade-offs that come from the very real air pollution damages for which coal burning is responsible? Is the large contribution of coal to historic aerosol masking a further indictment of its polluting nature, a hidden cost serving to hinder full under- standing of climate change risk — or is it quite the opposite, a deeply perverse yet vital ‘redeeming’ feature that we are forced to admit has shielded us from the full brunt of its GHG emissions even while simultaneously causing such destruction to human life and natural systems? is is not a question science alone can answer, it is a question of ethics, which I discuss in Box 5.3.
Box 5.3: e ethical dilemma of aerosol cooling
At the heart of aerosol cooling’s challenge to policy is a vexing ethical dilemma: does the abate- ment paradox’s threat of triggering catastrophic climate disruption outweigh the damage being done by aerosols right now, today? Does avoidance of catastrophic climate change — avoidance of future lives lost, of large scale species extinction, of destroying, in the words of Hansen et al. (2008), the planetary conditions ‘on which civilization developed and to which life on Earth is adapted’ (p. 217) — justify intentionally continuing air pollution that takes and diminishes lives today, that still damages human and natural systems? In writing this thesis I have asked my- self this question many times. Given what the climate science points to, given the scale of the threat to the very foundation of life on Earth, I conclude that if such a horrendous decision must be made, then yes, preventing catastrophic climate change must be the highest priority. For that reason alone, if nothing else, it is my earnest hope that the transitionary framework I will shortly propose can be implemented to minimise damaging air pollution but still avoid precipit- ously weakening the aerosol mask; that can usher in far less harmful alternatives quickly, while GHG emissions are arrested.
47 And other important NFA emission sources.
238 5.4 Ramifications for abatement pathways
If we accept the existence of an abatement paradox then the ideal pathway is surely one that sees strong mitigation policies preclude expansion of natural gas and any other fossil fuel, shuing down coal combustion in favour of rapidly deployed renewable energy or other genuinely low-carbon alternatives, with compensative masking implemented in parallel to manage the consequent NFA masking loss. But this may well not happen; certainly nothing of the sort is happening now. e process of establishing a compens- ative masking capability and the regulatory framework to govern it are still to be even considered. Yet evidence is emerging that coal-fired generators are already being dis- placed in some areas by the combined drivers of the merit order effect of rising renew- able energy penetration and the modest carbon pricing so far implemented (Parkinson 2012a). In the US, low gas prices are believed responsible for recent large falls in coal- fired electricity generation’s market share (Lacey 2012).
And so perhaps the ‘least worst’ policy option is to adopt transitionary mitigation frame- works that preference coupled emiers over purportedly lower GHG-intensity substi- tutes. To fully acknowledge the implications: I am suggesting taking deliberate steps to preference continued coal burning over substitution for gas because of the masking ef- fect of coupled NFA emissions, despite the damage they cause as air pollutants. Doing so could buy time before the real push for wholesale abatement finally begins and, crucially, allow compensative masking capability to be established. If we accept this most wicked of bargains, what might be the optimal Scheme implementation model that achieves the policy objective while minimising the damages? Chapter 6 proposes an answer.
239
6 Scheme implementation: the balancing market
Chapter 5 established the policy objective of an aerosol-integrated mitigation framework and set key design criteria for Scheme implementation. A range of possible implement- ation models were evaluated, but each was found to contain serious flaws with respect to both those criteria and the overarching policy objective. is chapter harnesses the lessons gained from analysis of those flaws to propose a domestic Scheme implement- ation model without an inherent unacceptable risk of failure — a model that confronts the confounding, paradoxical reality of the aerosol mask.
6.1 Restricted direct value model
I propose that a workable solution lies in a modified form of conferred economic value model, combined with elements of the central authority model. e pivotal differences are to reconfigure the way in which value is aributed to NFA emissions so as to avoid the distortionary impacts identified in section 5.3.5, and to refocus market forces so as to drive the shi to alternative abatement pathways postulated in section 5.4. is new approach is constructed to deliver the managed withdrawal of coupled aerosol emissions while giving vital time to carefully introduce and build out compensative measures. Im- portantly, it is also resilient to unpredictable changes in private firms.
241 Chapter 6 Scheme implementation: the balancing market
I label this the restricted direct value model. Its essential feature is to pay directly for the contingent benefit of coupled aerosol emissions during a transitionary phase of uncertain duration.
6.1.1 Overview and rationale
e key problem with the conferred value model1 is that NFA-offsets suppress the car- bon price signal and, via emission permit displacement, diminish real GHG abatement.2 at is, they interfere with the carbon price mechanism in ways that undermine its fun- damental purpose. Yet NFA-offsets also serve to reorder abatement pathways such that the GHG abatement of coupled emiers is delayed. In accepting the logic of section 5.4, such reordering is now an intended outcome of the Scheme. e crux of the problem is therefore the interference itself — how can we confer value to affect the relative abate- ment order of GHG emission activities without diminishing GHG abatement?
We can do so by:
i Establishing a market for compensative masking operated by a central controlling authority, as briefly considered in section 5.2.7;
ii Having that authority set demand in the market as a function of baseline NFA emissions, factoring in the extent of masking required under the Protocol, which I emphasise will decline over time (refer to the Protocol definition in section 5.2.9, and see section 6.2.6 below for the specific demand-seing algorithm);
iii Licensing coupled emiers as eligible providers for their NFA emissions under temporary conditions; and
iv Paying them for that ‘service’ as market participants.
1 See the Lessons given on page 235. 2 is assumes an ETS is in operation rather than an emissions tax; see section 5.3.5 for the difference in effect on price signal.
242 6.1 Restricted direct value model
Coupled NFA emissions thereby obtain positive economic value directly in the market: firms receive revenue for the contingent benefit of those emissions.3 By separating the beneficial outcome of aerosols from the damaging outcome of their effects as air pol- lutants in this way, the Scheme is able to avoid conflict with air pollution regulations through the interplay of prices. I must emphasise this point clearly: the Scheme does work to deliberately support continued emission of NFA pollutants, thereby raising the ethical dilemma recognised in Box 5.3. But as section 6.4 shows, air pollution control regulations can still be deployed to target the worst of those damages, and can function effectively alongside the Scheme in doing so. e ethical dilemma is not removed, but damages can be minimised as much as possible.
Further, the license available to coupled emier firms is explicitly restrictive, precluding any use of NFA emissions as credits in the manner examined in section 5.3.5. In concert with the enabling rules that govern the market forces at play, this restriction ensures coupled emier participation is limited to a transitionary phase.
A primary function of the Scheme is to shi the rank order of abatement pathways that preferentially abate non-coupled GHG emission sources (lile or no NFA emissions). GHG abatement must still occur and is undiminished in real terms as the GHG emission cap remains in place under the ETS (see next section), but that abatement is now obtained preferentially from non-coupled sources.
6.1.2 Carbon price mechanism integration
e model and its effects described in this chapter assume that the carbon price mech- anism is implemented as a cap-and-trade ETS. As will become clear, this is principally because of the price interaction effects with the new market mechanism and the con- sequences for how a target GHG abatement level is likely to be achieved. Importantly,
3 e contingent benefit of aerosol masking is examined in Chapter 4.
243 Chapter 6 Scheme implementation: the balancing market these price interaction effects mean that the cap will be met without the need for any modification of the ETS itself. A further assumption is that ETS permits are allocated largely by auction, maximising carbon price revenue for government.
If the carbon price mechanism is an emissions tax, however, modifications are required. is is because the price interaction effects detailed in section 6.3.1 and especially sec- tion 6.3.3, will not occur. e carbon tax rate must be raised in order to accommodate the shis in relative carbon price tolerance of coupled emier firms under the Scheme (detailed in section 6.3.3). If the tax rate is not adjusted, GHG abatement will be less under the Scheme and the policy objectives will not be satisfied. Adjustment of the tax rate is possible, but adds additional administrative complexity, and calculation of the ad- justment factor is unlikely to be straightforward. Such difficulties ultimately reinforce the general position that an ETS is the superior instrument choice for climate change mitigation because it specifies an emission cap, leaving the price to be determined by the market.
6.2 Scheme implementation
e major task required to implement the restricted direct value model is to create the compensative masking market and define its governing rules. e new market will op- erate in parallel with the carbon price mechanism under the unified Scheme. It has the following properties:
• Supply is provided by private firms or competitively-neutral government enter- prises participating in the market as compensative measure providers (CMPs).
• Specific eligible compensative measure technologies are prescribed in regulations and subject to the licensing process defined in section 6.2.1.
• Coupled NFA emissions are designated a special type of compensative measure,
244 6.2 Scheme implementation
subject to the specific restrictions defined in section 6.2.5.
• A central authority regulates and operates the market and oversees or conducts the licensing process. Logically this should be an additional responsibility for the existing authority having jurisdiction over the carbon price mechanism. Let that authority be labelled the Regulator for this discussion.
• Market demand is determined by the Regulator for each period via the process detailed in section 6.2.6.
• A reverse auction mechanism is used to satisfy that demand in each period, as detailed in section 6.2.7.
Stretching the analogy of electricity generation in section 5.2.6 further, compensative masking can be viewed as akin to the balancing and ancillary services market for real- time matching of load with supply. As the term ‘compensative masking market’ is some- what turgid for common use, let us therefore adopt the balancing services analogy and thus label the new Scheme mechanism a balancing market.
Accordingly, the compensative masking period is labelled the balancing period, re- taining its definition as per section 5.2.1. e balancing period is set to an equivalent interval of, and should be coincident with, the ETS compliance period. is period is as- sumed to be one year for the purposes of this discussion. However, as experience with the Scheme develops it may be more efficient — both economically and in relation to for- cing effects — to increase the frequency to biannual or quarterly market auctions, while ensuring that the temporal constraints of the forcing envelope are upheld (section 5.2.6).
e final question is the source of funding for the balancing market. Two principal options are proposed and evaluated in section 6.2.8: payment from government revenue, or an additional charge levied against GHG emiers. Both options have advantages and disadvantages, and the choice between them (or any other) may be based on political circumstances in the enacting jurisdiction.
245 Chapter 6 Scheme implementation: the balancing market
6.2.1 Licensed compensative measures
e categories of compensative measures identified in section 5.2.2 are adopted without further specification or restriction. However, use of these measures in practice is to be regulated through a licensing process with the following general aspects:
• Each prospective compensative measure technology must be submied to the Reg- ulator for evaluation and, if approved, is listed as an accredited compensative measure.
• Licenses may then be issued for the use of a particular accredited compensative measure, subject to appropriate operational criteria (such as spatial or temporal constraints as discussed in section 5.2.6).
• Licenses are granted to specific applicant firms that demonstrate compliance with the governing regulations for deployment of a particular accredited compensative measure.
• Separate licenses are required for each accredited compensative measure a firm wishes to supply; a single firm may acquire an unlimited number of licenses.
• e total number of compensative measure licenses is not restricted in general; any firm that meets the necessary operational and governance criteria for an accredited compensative measure can obtain a license for its use. However, license quantity may be restricted if determined necessary by the Regulator.
Clearly, cautious evaluation of proposed compensative measures will be required before licenses can be granted, and scientific advice will be needed as a basis for accreditation — this will take time. Consequently, CMP firms created specifically to provide services to the market from Scheme commencement will face delays while prospective technologies are certified. Certification of SRM-class measures could be a protracted process indeed.
246 6.2 Scheme implementation
ere is also the distinct possibility that previously-certified technologies could fail and fail badly. is might only become known aer they have already been deployed for some time. Failure may occur simply because they do not function as expected at scale, or, worse — potentially catastrophically so — they could induce such severe unintended side effects that they cannot be allowed to continue. Exhortations to extreme caution in considering geoengineering exist for very good reason (eg, Schellnhuber 2011; Schneider 2009).4
e following clause is therefore added to the licensing process above:
• e Regulator is to conduct periodic re-accreditation reviews of all licensed com- pensative measure technologies; a review may also be initiated by government direction in accordance with relevant enacting legislation. Accreditation may be revoked with immediate effect as a result of these reviews.
6.2.2 Coverage and reporting liabilities
e Scheme is implemented primarily by establishing the balancing market itself. A prerequisite is that legal liability for reporting is expanded to include the emission of aerosols and their precursors. All firms should report emission mass quantities per spe- cies in the balancing period following emission. Following the discussion of section 5.2.3, the most important in-scope species are those designated NFA: the precursor gases of secondary sulphate and nitrate aerosol, and to the extent possible, SOA; the primary particulates OC, mineral dust, and biomass burning.
BC remains a difficult question. On the balance of evidence it should be categorised as a PFA, but ongoing research into the extent to which BC acts as CCN could shi its categorisation. Within the balancing market context, BC is therefore not relevant. As a PFA, it could be incorporated into the carbon price mechanism, or regulated under
4 And refer again to Box 5.1.
247 Chapter 6 Scheme implementation: the balancing market air pollution regulations (interaction with which is examined in section 6.4). e same applies to the precursors of tropospheric ozone. Despite their not being eligible under the balancing market provisions, these species should also be subject to reporting obligations to improve quantification.
Reporting obligations under the Scheme should be extended to all firms that emit the pollutants listed. Specifically, this means that all firms covered by the existing carbon price mechanism now have increased reporting obligations, and those firms not covered by the existing carbon price mechanism are now subject to reporting obligations under the Scheme. Moreover, these expanded reporting requirements again point to the bene- fits of a unified reporting framework for all atmospheric pollutants rather than treating GHG and local air pollutant emissions separately, as highlighted in section 5.2.3.
e Scheme does not directly require any modification of existing sectoral, facility, or firm coverage rules under the carbon price mechanism. ose sectors currently exempt can continue to be excluded until measurement and other barriers are overcome. eir omission is acceptable on the assumption that the absence of a carbon price signal should translate to the absence of significant falls in NFA emissions that are actually caused by GHG abatement; there is no negative abatement feedback. However, it may be the case that the GHG emissions of firms in these sectors are subject to other, possibly indirect, regulations such as vehicle fuel efficiency standards. e increased reporting obligations for otherwise-exempt firms will improve visibility of NFA emissions and allow them to
5 be tracked. e annualised adjustment factor (αt) used in the calculation of balancing market demand in section 6.2.6 therefore provides a means to capture significant changes over time.
If coverage under the carbon price mechanism is extended aer the balancing market has commenced, newly-added coupled emission firms should be treated in the same 5 Disperse non-point source emissions from vehicles (oil combustion) will be challenging to include. e best option is likely to involve estimation using average technological efficiency and aggregate fuel consumption data.
248 6.2 Scheme implementation manner as existing firms. eir NFA emissions should be incorporated under the rules established below regarding eligibility (section 6.2.5) and NFA emission baseline (sec- tion 6.2.4), and included in subsequent market demand (section 6.2.6).
Finally, the extent to which any additional legal liability is established by the Scheme, beyond the reporting obligations above, is dependent on the funding mechanism adop- ted. ese mechanisms are detailed in section 6.2.8.
6.2.3 antification metric
e metric adopted for measurement and reporting purposes must meet the quantific- ation requirements identified in section 5.2.4. Let us assume this is the specific forcing pulse (SFP), where the receptor region is the geographical boundary of the enacting jurisdiction.6 Crucially, let us further assume that (i) SFP has been calculated for each in-scope NFA species, and (ii) that these values are additive. As per SFP notation defined in section 3.5.2.2,7 SFPa denotes the contribution of species a in GJ g−1, SFPb is the contri- bution of species b, and their sum represents the combined specific forcing pulse incident within the Scheme geographical boundary, within appropriate uncertainty bounds. SFP values for each NFA species are to be calculated and set in the Scheme regulations.8
Drawdown-type compensative measures (see section 5.2.2) require explicit treatment as licensed compensative measures. For a unit NFA emission measured in SFP to be removed and then replaced by an ‘equivalent’ reduction in atmospheric CO2 concentra- tion, translation from SFP to GWP is necessary. Although Bond et al. (2011) state that GWP can be derived from SFP, doing so requires selecting an appropriate time horizon,
6 Sub-national receptor regions might be needed in some countries, which could be achieved by match- ing sub-balancing markets, but such a situation is not considered further here. 7 SFP subscript denotes the receptor region, but as that is assumed to be the Scheme geographical bound- ary as a whole, the subscript is omied as unnecessary here. is is consistent with the original use of Bond et al. (2011). 8 Note that the definition of SFP in section 3.5.2.2 is based on forcing pulse over one year. If the balancing period is subsequently made sub-annual, SFP values may need to be re-calculated.
249 Chapter 6 Scheme implementation: the balancing market which then re-introduces many of the serious flaws examined in section 3.5. Even choos- ing a 20-year period in preference to 100 or 500 hardly addresses the fundamental lack of equivalence. is translation problem may be largely insoluble, hence it may be ne- cessary to weight the calculated GWP conservatively (> 1) in order to avoid significant underestimation.
6.2.4 NFA emission baseline
NFA emission baselines for each firm covered by the Scheme must be determined as a prerequisite. More accurately, the NFA emission baselines of each covered facility are required; a given firm may operate multiple facilities, or individual facilities may change ownership over time. Per-facility emissions are calculated by summing the reported unit mass emission of each NFA pollutant species occurring in one balancing period, multiplied by their corresponding SFP value. Expressed symbolically, annual coupled NFA emissions for an individual facility can be wrien as
∑P p εt = ept·SFP (6.1) p=1
ept ≥ 0 p = 1,...,P (6.2)
Where εt is the total forcing pulse for period t in GJ, and ept is the estimated unit mass emission of NFA pollutant species p in period t in grams, for the set of all in-scope species
P. As NFA emissions are coupled to those of GHGs, εt can be regarded as a function of those GHG emissions in the same period, Gt, as per eq. 6.3.
250 6.2 Scheme implementation
Additionally, εjt denotes the total forcing pulse of a particular facility j in the set of all such coupled emission facilities J.9
εt = f (Gt) (6.3)
εjt = f (Gjt) (6.4)
j = 1,...,J (6.5)
e NFA emission baseline is labelled Bj, defined as a representative historical annual total forcing pulse for an individual facility j (in GJ). e baseline is calculated as the mean total forcing pulse for any consecutive two-year period occurring in the 5 years prior to Scheme commencement. is definition is indicative only, following that of the US Clean Air Act;10 alternative definitions that serve the same general function may be used.
Note that the derivation of εjt used herein inherently assumes that it is acceptable to aggregate the forcing pulse of each individual NFA species, as stated in section 6.2.3. Aggregation is necessary to enable the balancing market to function (explained in sec- tion 6.2.6 below). ough a per-species forcing pulse can be recorded for each facility — and is of course transparently derivable via Scheme reporting of per-species emission mass — it is difficult to see a means by which a market could facilitate replacement of each NFA species’ emissions with compensative measures without a common metric. at commonality implies aggregation. In fact, the very notion of compensative meas- ures replacing NFA pollutants requires that the cumulative effects of one be translated to an equivalent quantity of the other. e specific forcing pulse is the means chosen to
9 On notation generally: symbols hereaer commonly include the individual-facility marker, subscript- j; subscript-t indicates a time period, as standard; superscript characters differentiate between related B instances, such as B for balancing market or C for carbon price (eg, Pt for the balancing market total price in a single period, eq. 6.15). 10 Definition of ‘baseline actual emissions’ under Title 40: Protection of Environment, § 51.165 paragraph (a)(1)(xxxv)(A), as of January 2013.
251 Chapter 6 Scheme implementation: the balancing market do so here, and aggregation of forcing pulses in the manner stated is consistent with the intent of that metric.
6.2.5 The residual masking license
e coupled NFA emissions of firms having liability under the carbon price mechan- ism are designated a special form of compensative measure. Firms owning particular coupled emission facilities may apply for a residual masking license authorising the use of NFA emissions for the purpose of supply into the balancing market. e license is named for the masking effect of residual aerosol emissions not yet abated; eligible firms operating coupled emission facilities are consequently labelled residual masking providers (RMPs). Only those firms that are both actively covered by the carbon price mechanism and trigger its compliance threshold are eligible. at is, only those firms that must obtain and surrender GHG emission permits in an ETS compliance period.
e license is fixed-term of short duration (say, one or two years) so as to ensure that no perpetual right to participate as a masking provider is created; firms must reapply aer license expiry, proving their continuing eligibility. RMP firms are also the only legal entities that can be issued a residual masking license, and the residual masking license is the only such license that authorises continued deliberate aerosol emissions11 as a form of compensative measure per the categories identified in section 5.2.2.
e residual masking license allows the firm to offer into the balancing market for supply of a ‘masking’ quantity equivalent to their facility’s forecast SFP-weighted NFA emis- ′ sions in the following period. We label this ex ante expected emission quantity εjt, dif- ferentiated from the actual quantity determined ex post, εjt as per eq. 6.1. Estimation of NFA emissions available in a prospective balancing period is the responsibility of the ′ firm. If a firm underestimates available NFA emissions (εjt < εjt) they forego potential
11 Deliberate in this context refers to the payment RMP firms will receive for their NFA emissions.
252 6.2 Scheme implementation
′ revenue in the balancing market; conversely, overestimation (εjt > εjt) risks penalty for breach of contract (see section 6.3.2). With these constraints in mind, the offer quantity can be wrien as
′ ωjt = f(ε jt) (6.6)
Where ωjt is the quantity offered into the balancing market by an RMP firm for facil- ity j in the coming balancing period t. Importantly, offer quantity in any period is not explicitly limited. e NFA emissions of a coupled emissions facility may consequently grow in response to the additional revenue now available from the balancing market, and other RMP firms operating coupled emission facilities brought online aer Scheme com- mencement may also compete for that revenue. Box 6.1 explains the reasoning behind this lack of restriction.
Finally, aggregate expected eligible NFA emissions and aggregate actual eligible NFA emissions in the Scheme can be wrien as
∑J ′ ′ Et = εjt (6.7) j=1 ∑J Et = εjt (6.8) j=1
′ Where Et is the sum of all RMP facilities’ eligible NFA emissions expected in balancing period t (ex ante), and Et is the actual aggregate NFA emission quantity (ex post).
6.2.6 Market demand
As in the central authority model of section 5.2.7, the Regulator sets market demand for each balancing period in advance. ere are two key differences here. First, rather than being assessed on a per-facility basis as the shortfall between a facility’s current
253 Chapter 6 Scheme implementation: the balancing market
Box 6.1: Ethics and environmental effectiveness
I am compelled to explain the logic behind the decision not to impose limits on the eligible NFA emissions of RMP firms. e original formulation of the restricted direct value model set two explicit limits: (i) only those facilities operating at Scheme commencement would ever be eligible (new facilities would have a baseline of zero; ie, Bj = 0); (ii) no firm was able to offer into the balancing market for a quantity greater than the facility’s NFA emission baseline (ie, ωjt ≤ Bj). A principal reason for these restraints is ethical — a desire that the Scheme in no way supports the emission of local air pollutants beyond the bare minimum necessary (minimise perverse incentives). However, at-times heated debate of the logic underpinning this ethical desire led to the real- isation that the overarching policy objective of a managed withdrawal may best be served by removing these limits. For instance, a newly built facility may be less polluting than existing ones, and its NFA emissions will contribute to the real aerosol loading. Excluding new facilities from the balancing market is therefore difficult to justify. Further, while the intention of the managed withdrawal is ultimately to transition to full compensative masking, these technolo- gies remain largely experimental at best, with poorly quantified costs, uncertain efficacy, and inherent risk of unpredicted damaging side effects. Caution suggests, grimly, we may be beer to stay with the devil we know until understanding and technological capacity improves. e ethical implications stemming from this position (first raised in Box 5.3) are at least some- what addressed by the ability to reduce damage from local air pollution through parallel air pollution regulations. As detailed in section 6.4, the interplay of price effects provides a means to target particularly damaging NFA pollutants, or particularly damaging individual facilities. Further, actual demand in the balancing market eventually decreases over time. On that basis, there is insufficient justification for imposing such limits on the residual masking license itself.
NFA emission and its baseline — the amount lost, now to be compensated for — demand in each period is derived from the sum of the NFA emission baselines of all facilities active at Scheme commencement. Second, for a transitional period in which CMPs are established, that demand is in fact met by the continuing NFA emission of RMP firms operating coupled emission facilities.
Scheme demand in the coming balancing period is then based on the total historical aerosol masking effect to be maintained, adjusted under the Protocol as specified in sec- tion 5.2.9 and in accordance with the overarching objectives of section 5.1.
254 6.2 Scheme implementation
Expressed symbolically, balancing market demand can be wrien as ( ) ∑J∗ Mt = ϕt αt + Bj (6.9) j=1
αt ≥ 0; ϕt > 0 (6.10)
Where Mt is the total demand required in the next balancing period t in GJ, αt is an annualised adjustment to accommodate uncertainties and gaps in coverage as per sec- tion 5.2.4 and 5.2.5, and ϕt is the current exogenously-specified masking proportionality Protocol factor per section 5.2.9.12 J is the set of all coupled emission facilities as in sec- tion 6.2.4, but J* refers to the specific set J at Scheme commencement.
e value of Mt is not dependent on the continued operation of coupled emiers, only the adjusted aggregate historical NFA emission baseline. Consequently, Mt is reasonably predictable in advance, assuming that αt is itself well-quantified (or comes to be as the Scheme evolves) and the Protocol factor is not subject to interannual fluctuation.
6.2.7 Reverse auction mechanism
e balancing market employs an electronic reverse auction mechanism to satisfy de- mand in the next balancing period: firms offer to supply a specific quantity of compens- ative (or residual) masking in blocks, at a stated price per block ($/GJ); the Regulator then awards contracts for supply in ascending price order of these blocks until demand quantity Mt is met. An important feature of supply contracts is that they do not con- stitute a property right in the manner of emission permits in an ETS, and so the reverse auction does not facilitate inter-firm trading. e absence of trading is desirable in this context as it strengthens the prohibition against coupled emiers obtaining any further
12 Recall from section 5.2.9 that the Protocol factor has the potential to exceed 1; ϕt therefore has no upper bound.
255 Chapter 6 Scheme implementation: the balancing market revenue outside the auction process.
A reverse (or ‘procurement’) auction is chosen as an effective and efficient means to obtain supply in the conditions here (Kaufmann and Carter 2006; Kahn et al. 2001): fixed and known demand (at time of auction), with many providers bidding to supply a single buyer (ie, the Regulator acts as a monopsony buyer13). Importantly, the balancing market is the only available market where many of these providers can sell their services.14 ere is therefore no ‘true’ price outside of this market mechanism, especially for NFA pollutants. e auction mechanism itself hence determines real value.
A number of reverse auction systems exist, mirroring their forward auction antecedents (Kaufmann and Carter 2006). Of these, the first-price sealed-bid system in ascending price order is the appropriate choice as there is no requirement for suppliers to see their competitors’ bids, and no requirement for multiple offer rounds for any individual sup- plier. e more important question is whether the sealed-bid auction should be uniform- price, where all bidders receive the offer price of the marginal winning bidder (ie, the most expensive contracted), or pay-as-bid, where each bidder receives exactly their of- fer price. In both cases the marginal winning bidder sets the market-clearing price; the difference is what price inframarginal bidders receive.
Kahn et al. (2001) examined the details of these auction types in the context of a possible shi from uniform to pay-as-bid pricing in the Californian wholesale electricity mar- ket.15 ey conclude strongly and convincingly that uniform price remains the superior option, and that many of the apparent advantages of pay-as-bid evaporate under actual market conditions. Most importantly that includes any expectation that pay-as-bid will be cheaper overall. e principal reason is that suppliers will change their bidding beha-
13 A monopsony buyer is the equivalent on the demand side to a monopoly supplier; ie, the market is characterised by a single buyer exercising price-seing power. 14 e exception is firms offering drawdown services, who are likely able to generate GHG abatement credits. 15 e uniform price reverse auction is commonly used in wholesale electricity markets to determine the dispatch merit order.
256 6.2 Scheme implementation
Table 6.1: Sealed-bid first-price reverse auction types
Type Feature Uniform price All participants are awarded the market-clearing price for their supply. Participants therefore have good reason to offer in at approximately their short run marginal cost (by block), because if competitors are cheaper the firm does not have to sell below cost. For this reason uniform price provides realistic price discovery. Firms have incentive to lower costs so as to maximise revenue. Under normal market conditions should result in procurement at lowest available cost. Competition is discouraged where large suppliers game the market, withholding supply in order to push up the marginal price. Pay-as-bid Participants are awarded their offer price, hence they must take into account costs beyond short run marginal cost, including contribution to fixed cost and profit. Participants must also forecast the expected market-clearing price in determining their offer price (predict their competitors’ behaviour), incurring the costs for obtaining that forecast. Under normal market conditions cannot assure lowest-cost procurement order. Competition is discouraged as small bidders are disadvantaged by the need to forecast (a task subject to economies of scale); by increased opportunities for gaming, strategic bidding, or withholding; and by discouraged capacity expansion. viour away from offering at approximately marginal cost, the authors noting the ‘naïve’ assumption of advocates that bids would be the same in both types. e analysis of Kahn et al. (2001) is summarised in Table 6.1.
Despite these strong arguments in favour of uniform price the balancing market should be constructed as a pay-as-bid system, at least until CMP firms come to dominate supply. e central reason for this decision is that RMP firms have production costs of zero for the generation of NFA pollution — the balancing market is not operating under normal com- petitive conditions. is atypical characteristic largely nullifies many of the advantages of uniform pricing, as detailed in section 6.3.1 below.
Via the pay-as-bid reverse auction, the Regulator awards contracts for supply to bidding
firms for their contribution to Mt, which may be across multiple individual facilities. Each contracted firm is paid at the price it offered into the market for each block, for
257 Chapter 6 Scheme implementation: the balancing market
16 total receipts of Rt per facility. e marginal winning bidder may then be contracted for some or all of their offered masking quantity block(s).
B Net expected supply for the period, St , is the sum of all CMP-contracted compensative m e measures, St , and all RMP-contracted NFA emissions, St . To reiterate, in the transition- ary phase the expectation is that demand will be supplied entirely by the NFA emissions of existing RMP firms using their residual masking licenses.17 Symbolically, net supply is
B m e St = St + St (6.11) e ≤ ′ St Et (6.12)
B Net payment required, Pt , is in turn the sum of the two component contract costs for m e m e St and St , labelled Pt and Pt respectively.
m m Pt = f(St ) (6.13)
e e Pt = f(St ) (6.14)
B m e Pt = Pt + Pt (6.15)
rough this circular process the required aerosol mask is maintained in the transitional phase, buying time for CMP firms to establish themselves but ensuring that GHG abate- m ment actually occurs in real terms. During the transitional phase, St will be small, pos- e ≫ m sibly zero; ie, St St for these initial years. e process flow is illustrated in Figure 6.1 in simplified form using a single RMP firm operating a representative coupled emission facility (coloured orange). Note that the ‘adjustment factors’ input incorporates both αt and ϕt from eq. 6.9 above; blue coloured elements represent market operations; green coloured elements represent CMPs.
16 Or Rjt for facility j in the set of all facilities J, as before. 17 Shortfalls in supply are possible and are addressed in section 6.3.2.
258 6.2 Scheme implementation