COPYRIGHTED MATERIAL

0002006766.INDD 1 1/10/2014 7:15:48 PM Vocabulary of Climate

Plate 1.1 Vocabulary of climate. How we discuss climate today is illustrated in this ‘wordle’, 1 created by Neville Nicholls using the text in Chapter 3 of the 2012 Intergovernmental Panel on Climate Change (IPCC) Report on Climate Extremes.2 Most of the work drawn upon is collaborative, so the phrase ‘et al.’ indicating with others dominates. Interestingly, precipitation is more frequently used (larger) than temperature; projected and projection s occur much more often than evidence ; and, relevant for this book, models and confidence are both fairly important.

0002006766.INDD 2 1/10/2014 7:15:51 PM 1 Why Model Climate?

‘All models are wrong, but some are useful.’ (Box and Draper 1987 , p4)

‘The strongest arguments prove nothing so long as the conclusions are not verified by experience. Experimental science is the queen of sciences and the goal of all speculation.’ (Roger Bacon ca. 1214–1294)

LEARNING OBJECTIVES After completing this chapter, you will be able to: • recognise the many reasons for having models • track the history of climate theory becoming fact • list the factors affecting planetary scale climate • explain the concept of climate feedback Plate 1.2 Eclipse 2012 – the climate is driven first and give examples and foremost by solar radiation. • recognise the mechanisms whereby persistent and widespread life affects climate.

The Climate Modelling Primer, Fourth Edition. Kendal McGuffie and Ann Henderson­Sellers. © 2014 John Wiley & Sons, Ltd. Published 2014 by John Wiley & Sons, Ltd. Companion website: www.wiley.com/go/mcguffie/climatemodellingprimer

0002006766.INDD 3 1/10/2014 7:15:54 PM 4 The Climate Modelling Primer

1.1 Introduction Table 1.1 A primer, or ‘A, B, C’, of climate modelling Aspects of This book is entitled The Climate Modelling A, B, C climate modelling Primer, a title that presupposes modelling to be a useful exercise, and that readers are familiar with A: Astronomy Astrophysical the idea of models and the reasons for partici­ attributes – orbit, pating in modelling. We assume you are inter­ atmosphere, radiative ested in building or testing models or in exploiting budget, existence/ their results. This foundation chapter tests these prevalence of water … assumptions by examining the important ques­ B: Biology and Life and climate, tion, ‘Why model climate?’. We try to answer this boundaries surface conditions, question in three ways: first by looking at reasons volcanic activity … for modelling in general; by applying a selection of these reasons to climate modelling; and then C: Comprehension Prediction, testing by taking a very different view of Earth’s climate, theories, raising from a distant galaxy, and using this metaphorical questions, bracketing alien climate scoping to investigate some of the outcomes, directing fundamental ingredients of planetary climates and data collection, thus of climate models. In this opening chapter we disciplining policy … cover a wide variety of topics quite quickly to give a sense of the wonderful breadth of climate mod­ els and their achievements. In doing this we do not define or explain in much detail because these explanations constitute the rest of this book. If you come across a concept you wish to understand better, you can locate a further description of it using the index or checking the summary of boxed material at the end of the Preface. The characteristics of climate and hence those that climate models must try to reproduce can be thought of as a primer – or perhaps an A, B, C – as outlined in Table 1.1. • A is for astronomy: any planet or moon with a climate is constrained by fundamental astro­ physical conditions. Figure 1.1 The Macquarie collector’s chest. Collections • B is for boundary and for biology: climate like these were for display and specifically designed as becomes interesting to model most often attractive and persuasive depictions of unusual places. when it relates to living systems and where it Source: Mitchell Library. State Library of NSW – XR 69. touches boundaries. • C is for comprehension: the reasons for con­ and 19th centuries, such collector’s chests were structing, operating and analysing climate built to hold and attractively display novel collec­ models are ultimately to try to understand cli­ tions of scientific specimens. Many voyages of mate change and variability. discovery included natural scientists who would To encourage personal learning, we are employ­ have carried their rare and curious samples home ing an old technique that may be unexpected in in such sturdy wooden chests. Our example this context. It is a ‘collector’s chest’. In the 18th (Figure 1.1), the Macquarie collector’s chest, was

0002006766.INDD 4 1/10/2014 7:15:54 PM Why Model Climate? 5

Table 1.2 The Primer authors’ climate modelling treasures, following the items in the old collector’s chest shown in Figure 1.1 Type Old chest Authors’ treasure collection Visual Paintings The cartoon by Cathy Wilcox illustrating the CMP authors’ research on Amazonian deforestation that was published on the front page of our local newspaper Personal Butterflies, beetles, etc. Results from the Model Evaluation experience Consortium for Climate Assessment intercomparisons created in 1992. These were probably the first global climate model intercomparisons (e.g. videos on CD in CMP2) Oceans Algae and seaweeds Movie featuring the ocean near where we live – ‘Finding Nemo’ (2003 and in 3D in 2012), especially for its depiction of the East Australian Current – the one that carries the turtles Change Exotic stuffed birds Photos from visits to the melting Mont Blanc behaviour glacier when the authors lived in Geneva Pretty things Arrangements of Art work on the cover of The Future of the sea-shells World’s Climate, a book the CMP authors edited in 2011–12. Both the art itself and the quotation it contains How it works Artefacts An antiquarian water band spectroscope that KMcG bought for AH-S’s birthday that shows water vapour absorption bands (an in-your-hand greenhouse demonstrator)

almost certainly intended as a special presenta­ or virtual links as in our example at the end of this tion piece to celebrate the colony of New South chapter (see Table 1.11). The point of the collec­ Wales once the Governor, to whom it was given, tion is to assist recall of aspects of climate model­ arrived back in the UK. If you are not keen on ling that you may find difficult to understand or stuffed birds and old seaweed, another type of perhaps that you find challenging to explain. Each treasure collection still to be found in some collection is, therefore, rather personal, but not homes is the heritage quilt, and a still more mod­ private, because like Governor Lachlan Macquarie’s ern version is scrapbooking. chest, it will contain amazing illustrations selected Climate Modelling Primer (CMP) readers are for explaining, remembering and sharing. To begin welcome to use whichever analogy they prefer: your great treasure collection, we offer you the collector’s chest, heirloom or heritage quilt or digi­ tangible version of ours (the authors’) in Table 1.2 tal scrapbook. The goal is that, as you read the and later we introduce our e-chest version. Primer, you collect climate-modelling treasures: a At the end of this chapter, we give another of small set of illustrations that you find persuasive, our collection examples and then each Primer pretty and memorable. These can be real objects reader is on their own to collect the best (most such as diagrams, papers, cartoons, printouts, etc. interesting) items for themselves.

0002006766.INDD 5 1/10/2014 7:15:54 PM 6 The Climate Modelling Primer

planet-wide characteristic, undeniably variable 1.2 What is a climate but also subject to change; climate is a topic of huge discussion, if not outright dispute. model? Consequently, what climate modelling involves has changed and will, no doubt, continue to In the broadest sense, models are for learning change. Nonetheless, most people share an about the world (in our case, the climate) and the understanding of what a ‘climate model’ entails. learning takes place in the construction and the Here, we use the analogy of a cooking recipe. manipulation of the model, as anyone who has watched a child build idealised houses or space­ ships with Lego™, or built with it themselves, will know. Climate models are, likewise, idealised 1.2.1 Climate modelling and representations of a complicated and complex cooking: feeding good reality through which our understanding of the Issue: cooking is an interesting analogy for cli- climate has significantly expanded. All models mate modelling. involve some ignoring, distorting and approxi­ Message: the best meals, and models, depend mating, but gradually they allow us to build on many characteristics: fine ingredients, the understanding of the system being modelled. A chef’s skill and the consumer’s attitude, e.g. pal- child’s Lego construction typically contains the ate, hunger/desire and ambience. essential elements of the real object, improves with attention to detail, helps them understand Making a meal and constructing a climate model the real world, but is never confused with the real share, perhaps, three or four essential steps: thing. selecting the ingredients, combining and pro­ In the past few decades, the boundaries of the cessing them, the evaluation (appreciating the climate system that we are modelling have fruits of the kitchen) and, often, considering become much less clear. This evolution, though repeating the recipe. As with any recipe, you can not inhibiting in itself, is exemplified by a quick vary the ingredients of a climate model a little survey of the term ‘climate’ in textbooks a cen­ and create a similar dish or change a lot and cook tury apart – say 1910 and 2010. In the former, up something altogether different. This analogy climate is viewed as constant and stable – the encourages additional comparisons: some ingre­ average weather of a place or region defined in dients are essential, some optional; frequently terms of unchanging seasons, crops, habitability, the order of the steps must be followed rather etc. In the latter, climate is typically viewed as a rigorously; evaluation is a vital part of the process (why cook if no-one eats?) but is poorly quanti­ fied; and, finally, success does not guarantee repeating good outcomes, but is a hopeful sign (Table 1.3). Cooking and modelling share another important feature: it is quite possible to under­ stand how a good meal is constructed and appreciate it without having the detailed culinary skills to replicate it. So it is with models. This Primer is for nourishment and budding food con­ noisseurs but not really for chefs. In this chapter, we intend to take a very quick look at a large selection of climate models. If you are happy to think of this Primer as a recipe book, then this chapter serves as kitchen preparation – we can consider the menu, possible ingredients, tools (even including a dishpan!) and, most

0002006766.INDD 6 1/10/2014 7:15:55 PM Why Model Climate? 7

Speed Date Box 1

Gamers go-for-it: the 2007 Climate Challenge

Meet: http://www.climatemodellingprimer.net/ emissions (in millions of tonnes or l/k101.htm teragrams). A turn consists of selecting up to five policy cards, each of which will use up or Name and date of birth: Climate add certain resources. For example, ‘Import Challenge is a web-based com- Food’ adds food but costs euros and energy

puter game created by the game and adds to the CO 2 emissions. Similarly, development company Red Redemption in ‘Require Energy Efficient Appliances’ costs

2007 and sponsored by the BBC. euros but adds energy and reduces CO 2 emis- sions. Particular policies unlock other cards Fame factor: This game was created in response such as planting large forests. But disasters can to the growing awareness of the role that cli- strike, draining resources unexpectedly and mate models would play in international nego- forcing the player to choose between a very tiation by the worldwide mass media. The BBC expensive, unpopular policy and an expensive, was plugging into public interest in the run-up very unpopular policy. to the famous 15th Conference of the Parties (COP) to the United Nations Framework Coverage: Every policy has an approval rating Convention on Climate Change (UNFCCC) in and, if enough citizens are unhappy with your Copenhagen in December 2007. performance, you will be voted out, which ends the game. Between turns, a newspaper Looks: Climate Challenge is fun: fairly fast but page provides feedback on your progress and also thought-provoking and open-ended. Each public opinion. There was a six-part TV series player works through simulations occupying created at the time the game went live (2007) this century (2000–2100) in which you (the co-produced by One Planet Pictures (UK) and player) become the President of all ‘European dev.tv (Switzerland) http://www.climatemodel- nations’. Your goal is two-fold: radically reduce lingprimer.net/l/k102.htm

your country ’ s CO2 emissions and also manage to remain popular enough to stay in office. On a date: Your date goal is to

The popularity catch is the true reality of the reduce CO2 emissions to the target climate challenge for the world’ s politicians. levels agreed by the global com- The science in Climate Challenge is sound, hav- munity and also to keep your electorate happy. ing been developed at Oxford University using Periodically, you (the player) have to meet the UK Meteorological Office’ s global climate other world leaders at the Climate Change model. Summit and vote on setting new global emis- In the game, each simulation (round) lasts 10 sions limits. This is not unlike the UNFCCC COP turns, each spanning a decade between 1990 meetings. If other leaders feel that you/Europe and 2090. Progress is measured by four world is not doing enough, they will be less inclined resources plus gas emissions: money (in mil- to reduce their own emissions and you will lions of euros); energy (in megawatt hours); have to subsidise them, an expensive way to food stocks (in millions of tonnes); water (in buy votes. There is a fierce sense of reality to

trillions of litres); and carbon dioxide (CO 2 ) this climate model speed date. (Continued)

0002006766.INDD 7 1/10/2014 7:15:57 PM 8 The Climate Modelling Primer

Figure source: www.bbc.co.uk

0002006766.INDD 8 1/10/2014 7:16:02 PM Why Model Climate? 9

Table 1.3 Components of recipes for cooking and for climate modelling Characteristics Meal Climate model Ingredients Some essential, some optional Some essential, some optional Method Ordered and quantified Ordered and quantified Evaluation Does it resemble the photo? Can it simulate present day? How does it taste? How about a different geological era? Did anyone get sick? Are there aspects that are wrong? High nutritional value? Can it predict? Repetition Are changes possible or desirable? Are changes possible or desirable?

importantly, how to combine ingredients to and as the conduit for displaying their results. All create­ a value-delivering climate model. Through­ collections of software (bundles) and hardware out, please take our analogy with a ‘pinch of salt’ (machines) have relationships (human interfaces) (pun intended). Remember that not all dishes win with people: their developers and their users. A favour with all diners and from time to time our neat analogy between climate models and smart­ desire for, and pleasure in, different meals differs. phone apps illustrates the synergies among peo­ This is as true for climate models as for food. ple, platforms and software. Table 1.4 compares Models can be as different as peanut butter the benefits and challenges of nifty phone apps sandwiches and crème caramel; they please and of climate models. ­differently and typically cannot readily substitute The tension between competitiveness and for one another. customer universality of implementation is not A quick review of Figure 1.2 underlines that limited to smartphones and climate models. The just having a menu or list of ingredients (for a same discussion surrounded the development of modern climate model this might comprise CDs, DVDs and, before that, vinyl records and atmosphere, land, ocean, sea-ice, aerosols, car­ 19th-century railway gauges. Users usually want bon cycling, vegetation, chemistry, nitrogen, ice applications to be straightforward and then fre­ sheets and more) does not get the meal ready. quently wish to add on or to mix and match while All modellers also need the recipe for construct­ developers generally regard their system as ‘deli­ ing each dish. The nature of these climate model cate’ or, at the least, worthy of protection. ingredients will be the subject of most of the rest In the case of smartphones and climate mod­ of this book. How to create a climate model will els, some drawbacks – those of infrastructure – depend to a very great measure on what the can be reduced by intentionally creating modeller and user want to predict or understand. applications (phone apps and models) that work Different models demand different methodolo­ on all available systems. National, and even gies and there are a number of ways of illustrat­ international, planning could encourage this. ing this; we have chosen here to highlight the Other problems – especially those arising from strengths and weaknesses of climate models by poor development and testing or from user mis­ examining why people build and use them. application – are less easy to fix. In both cases, the first customer complaints (both (a)s in 1.2.2 Climate models are much Table 1.4) might be fixable by developing an upgrade that solves the problem. However, the more than code second criticisms (the (b)s in Table 1.4) are more Climate models are first and foremost collections to do with the fundamental design: this outcome of software (computer code). As such, they was not intended to be delivered. Of course, require platforms (hardware) on which to operate most systems can be modified to do whatever

0002006766.INDD 9 1/10/2014 7:16:02 PM 10 The Climate Modelling Primer

Mid 1970s Mid 1980s Early 1990s Late 1990s Early 2000s 2010 2013+

Atmosphere Atmosphere Atmosphere Atmosphere Atmosphere Atmosphere Atmosphere

Land surface Land surface Land surface Land surface Land surface Land surface

Ocean and sea ice Ocean and sea ice Ocean and sea ice Ocean and sea ice Ocean and sea ice

Sulfate aerosol Sulfate aerosol Sulfate aerosol Sulfate aerosol

Nonsulfate Nonsulfate Nonsulfate aerosol aerosol aerosol

Carbon Carbon Carbon cycle cycle cycle

Dynamic Dynamic vegetation vegetation

Nonsulfate Atmospheric Atmospheric Ocean and sea ice Sulfur cycle model aerosol chemistry chemistry

Nitrogen cycle Land carbon cycle model Carbon Ice sheets cycle model Ocean carbon cycle model Dynamic Dynamic vegetation vegetation Nitrogen cycle Atmospheric Atmospheric Atmospheric chemistry chemistry chemistry Ice sheets

Figure 1.2 Changing list of components (ingredients) of climate models as it evolved over the past half-century. This diagram is not a recipe because it does not tell how to make the model; it is just the list of ingredients. As such, it comprises only the first step in climate modelling construction. Source: Extended and modified from IPCC 2001.

Table 1.4 Comparison between a climate model and a smartphone navigation ‘app’ (application). Both benefit from users who extend their comfort zone but also suffer from failure of the developers to standardise across platforms and from users’ misapplication Intended platform Code (hardware and its (software) software) Example use Challenges Criticisms Navigation iPhone™ Finding a coffee May not work (a) Doesn’t work in a app shop on Android™ covered mall (b) Doesn’t play music Climate Supercomputer Reforestation May not give (a) Little use for model opportunities the same results sea-ice projection under global on large array (b) Doesn’t include warming of PCs cost-benefit values

their designers wish but some care has to go into phone owners inasmuch as they need to have decisions to add features ‘because we can’ or some understanding of what an app can (and ‘because they were requested’. The analogy cannot) do before setting out to use it for an holds, as climate model users resemble smart­ important task.

0002006766.INDD 10 1/10/2014 7:16:03 PM Why Model Climate? 11

Table 1.5 Top 10 reasons for climate modelling 1.3 Multiple reasons for (in addition to prediction) climate modelling No. Reason In order to identify a set of reasons for conducting 1 Climate models test the robustness climate modelling, we review why people under­ of prevailing theory take modelling of all types for wide-ranging tasks. 3 2 Climate models illuminate salient In a series of lectures in 2008, Joshua Epstein features and core uncertainties described how everyone models all the time but relatively few people recognise their actions as 3 Climate models reveal the model construction and exploitation. Developing apparently simple to be complex the ideas of George Box, Epstein outlines how and vice versa modelling outcomes can be very broad and lists 4 Climate models raise new questions 16 reasons for building or using a model, other and suggest analogies than the obvious one of prediction. To begin to answer the question ‘Why model cli­ 5 Climate models expose prevailing wisdom as compatible or mate?’, we have modified and reduced Epstein’s list incompatible with existing data and to just 10 compelling reasons for being interested hence direct collection of new data in climate modelling. These are listed in Table 1.5. While some readers might have expected 6 Climate models explain this book to focus primarily on climate model 7 Climate models bound (bracket) ­predictions, other strengths and benefits of cli­ outcomes within plausible ranges mate modelling comprise a large proportion of this text. In particular, we examine climate model­ 8 Climate models train practitioners ling from the premise that ‘all models are wrong; and educate the general public the practical question is how wrong do they have 9 Climate models discipline the policy to be to not be useful’.4 That we can and do, in dialogue our daily lives, obtain reliable knowledge from unrealistic models seems paradoxical at first. Dick 10 Climate models encourage sensible Levins argued the case for believing that ‘our thinking and informed discussion truth is the intersection of independent lies’.5 He proposed that, in order to overcome challenges some of the prevalent decadal variability in of modelling complex systems, scientists often ­climate. However, while many large-scale oscilla­ treat the same problem with several alternative tions in the ocean–atmosphere components of independent models. Despite unrealistic aspects the climate system are now recognised and these of their design, if such models are sufficiently oscillations can be reproduced (that is, described) independent and still yield similar results, one can by today’s models, modellers are only just begin­ infer some degree of confirmation. ning to see benefits of the extra complexity. For Throughout The Climate Modelling Primer, we example, during the 2000s, the Interdecadal will refer to the 10 reasons for building, running Pacific Oscillation reversed, cooling the Pacific and exploiting climate models, their strengths and and stalling the human-produced rise of the their weaknesses. For example, the explanatory global average temperature. Climate models can value of climate models is as important as their now reproduce this stall and explore its implications­ use for prediction. There are many ways of illus­ for future climate prediction. Using different trating this, such as the case of simulation of ­initialising conditions affected the simulations of ocean–atmosphere oscillations, the most well the decadal climate changes.6 In other words, as known of these being the El Niño–Southern models become more complete, this complete­ Oscillation (ENSO). Coupling the oceans into ness tends to improve skill of predictions and ­climate models has permitted the examination of increase understanding of climate behaviour.

0002006766.INDD 11 1/10/2014 7:16:03 PM 12 The Climate Modelling Primer

We approve of, and try to uphold, the idea of modelling as a means of enhancing learning and Reflection on Learning 1.1 understanding, above the desire for prediction. We encourage our readers to consider our 10 climate Recognise the many reasons for modelling reasons, comparing them with Epstein’s having models original 16. Throughout the book, we will point out Virtually all models of importance for the future prediction, explanation and other successes and of Earth and its people exist and operate in a failures of climate models and we encourage read­ complicated, nested framework that also encom­ ers to create their own lists of examples that inter­ passes economics, human development, politics est them, which illustrate how different types of and policies on adaptation to manage exposure models, most of which we construct and use almost to natural and human-induced extremes and unconsciously, underpin our lives and add value. ­disasters. Climate models are no exception (Figure 1.3). Climate models assist in assessments of expo­ sure and vulnerability of human society and ­natural ecosystems to climate. They also allow evaluation of the comparative influence of natu­ ral climate variability and anthropogenic climate disturbance as well as encouraging development of resilience to risks that cannot be eliminated. Outcomes from climate models are today con­ tributing to and influencing demand for policies regarding greenhouse gas emissions and the potential for mitigation of anthropogenic climate change. When thinking about why a model was developed and how its results are used, it is vital to remember this broad context.

Figure 1.3 Schematic of the connections between climate, disaster, development and vulnerability. Many models, including climate models, contribute to our understanding of these interdependencies. Source: IPCC (2012). Reproduced with permission from the IPCC. Disaster

CLIMATE Vulnerability DEVELOPMENT

Natural Disaster risk variability management Weather and climate DISASTER Anthropogenic events RISK Climate climate change change adaptation

Exposure

Greenhouse gas emissions

0002006766.INDD 12 1/10/2014 7:16:04 PM Why Model Climate? 13

Tech Box 1.1

Watch a climate model ‘flower’

Take a glimpse at the behaviour of ‘ready-to- 3. the time characteristics, i.e. what timestep use’ climate models by running a web-based the model uses and what time period it can climate model – for example, simulate? DaisyWorld: http://www.climate- modellingprimer.net/l/k103.htm Compare what you found out about this model Can you identify or discover: with the same attributes for the speed dating model ‘Climate Challenge 2007’ (see Speed 1. any of the equations this model uses? Date Box 1 ). 2. its starting conditions and whether these can be changed by a user (by you)?

Before we can begin to discuss the reasons for 1.3.1 Climate models test the modelling, we need to specify what constitutes a robustness of prevailing model, in our case, a climate model. We shall begin here by asserting that all climate models need theory three things: (i) one or more relationships (equa­ Issue: global warming seen through a Victorian tions) that relate the output to the input; (ii) speci­ rainband spectroscope . fied starting (or initial) conditions; and (iii) some Message: models contribute to examining how time characteristics, usually the time increment the theories work and how observations test them – model uses and the time period covered. comparing gravity and greenhouse . All models are simplifications. Although climate models may appear to be straight­ Most, if not all, models perform roles other than forward, in construction and ease of use, they prediction. Consider the part played by climate produce a surprisingly rich array of simulated cli­ models in confirming7 the ‘theory’ of global mates. The act of simplification is a fundamental warming. The reason for the creation of some of characteristic of modelling. The British physicist the earliest climate models was to be able to and mathematician, Lord Kelvin said in 1889: examine the consequences of the proposal that `.. when you can measure what you are speak­ addition of greenhouse gases to the atmosphere ing about, and express it in numbers, you will result in additional warming. Investigating this know something about it; but when you can­ theory of surface temperature increase as a result not measure it, when you cannot express it in of heat absorption and re­emission by atmos­ numbers, your knowledge is of a meagre and pheric greenhouse gases has involved models for unsatisfactory kind.' (Lord Kelvin aka William centuries. Some of the earliest models, real tools Thomson, 1st Baron, in Popular Lectures and that detected absorption by gases, passed from Addresses, London, 1889, v. I, p73.) the laboratory to everyday shops (Figure 1.4 a). The concept of global (or greenhouse) warm­ However, as Epstein emphasises, apparent ing emerged around 150 years after the idea of simplicity can be deceptive. Coming closer to gravitation.8 It was 1838 when French physicist modern thinking, check out the explanation of Claude Pouillet described how the Earth’ s atmos­ what a model is in Isaac Asimov’s sci­fi story phere increases the surface temperature. The Prelude to Foundation (Asimov, 1988, p 162). theory was also confirmed by observations – in

0002006766.INDD 13 1/10/2014 7:16:05 PM (a)

(b)

Figure 1.4 (a) 1880s’ advertising copy for a spectroscope: a hand-held Victorian scientific toy marketed as a tool for rain prediction, offered for only 3 pounds, 8 shillings and 6 pence. The Grace’s ‘New Direct Vision Spectroscope’ is said to have advantages of being ‘very powerful, portable and efficient’ and able to ‘divide the Sodium lines or the D lines in the Solar Spectrum and show the Rain Band as Separate Lines’. Source: Browning (1883). (b) Possibly the earliest use of remote sensing in meteorology: a so-called ‘rain-band’ spectrum from 6th July 1881. The spectroscope splits daylight into its (rainbow) spectrum. Dark lines, caused by the preferential absorption by water vapour of light at these wavelengths (labelled rainband), appear superimposed on the spectrum as the amount of water vapour increases (grey stippled bands), an indication of imminent rain.

0002006766.INDD 14 1/10/2014 7:16:08 PM Why Model Climate? 15

CSI Box 1.1

Weirdness of Water: Great greenhouse gas

Water vapour is the strongest greenhouse gas warming, it means that + H in our atmosphere. This happens because water when an H2 O vapour mole- molecules (H O) form so that the oxygen atom cule rotates or vibrates there 2 H is pushed to one vertex and the two hydrogen is an interaction with elec- O atoms to others. This configuration causes a tromagnetic radiation, the – dipole charge on the molecule: the oxygen end heat energy emitted from is partially negatively and the hydrogen end the Earth’ s surface. This interaction involves partially positively charged. This charge separa- absorption and emission of radiation in a large tion of the water molecule has many important number of water vapour ‘spectral lines’, mak- consequences. From the point of greenhouse ing water vapour a significant greenhouse gas.

1859, scientist John Tyndall conducted a set of Reflection on Learning 1.2 early spectroscopy experiments demonstrating that water vapour and carbon dioxide absorb Track the history of infrared radiation. The construction of a ‘model’ scientific theory becoming of laboratory spectroscopy applied to the Earth ’ s atmosphere was to become the popular ‘toy’ fact of the late 19th century (Figure 1.4 a). An instru­ Review the history of the two theories of gravita­ ment, termed a rainband spectroscope, capable tion and greenhouse as they pass through almost of ‘fingerprinting’ gaseous absorption became identical steps (Table 1.6 ). General (i.e. wide­ quite a fad around 1870. These pocket spectro­ spread) acceptance proceeds from speculation scopes designed specifically to display the so­ through objections and improvements (usually called ‘rainband’ were created and marketed as scientific), past vindicated prediction to sensitiv­ supposedly useful tools for the prediction of ity analysis pertaining to predictive skill. Along rain.9 They permitted any curious citizen to see a the path, theories collect communities of adher­ colourful ‘rainbow’ spectrum embellished with ents beginning in science, incorporating lay peo­ several dozen dark absorption lines (Figure 1.4 b). ple, encompass the general public and even Danish physicist Niels Bohr concluded that move into portrayal in art and popular culture. the formation of these spectral lines occurs when For example, greenhouse was used to make pre­ elements or compounds interact with radiation, dictions that were later observed: in the 1950s, either absorbing the radiation and increasing their Harvard University professor Richard Goody 10 energy by a discrete amount or lowering their anticipated that the surface temperature of energy and emitting a parcel of radiation. Based Venus would be very high, though this was not on such fundamental foundations, spectroscopy, verified until 1967 by the Soviet Union ’ s probe, the documentation and analysis of these spectra, Venera 4 . is now used in laboratories to determine constitu­ Consider carefully what you think ‘proof’ of a ents of samples in forensic analysis, and by astron­ ‘theory’ entails. What constituency needs to omers to detect chemical species in space. accept such ‘proof’; is this a democratic decision From this established understanding of the or some other process? In this book, we present behaviour of atoms came questions concerning the a series of boxes entitled ‘Climate Model effect of the additional heat from the atmospheric Validation’. What do you think about the concept gases re­radiated back to a planetary surface. of ‘validation’ of a model?

0002006766.INDD 15 1/10/2014 7:16:09 PM 16 The Climate Modelling Primer

Table 1.6 Model involvement in testing the theories of global warming and its history compared to those of gravitation Theory ‘proving’ Gravitation Global (greenhouse) warming Origin 300+ years old (Newton, ~180 years old (Pouillet, 1838) or 1687) older (Mariotte, 1681) Predictions Neptune (1824) from Uranus Industrial warming (Arrhenius, 1896) Venus ‘hot house’ (Goody, 1952) Objections raised Newton disliked ‘action at a Clouds, especially their feedbacks, are distance’ very poor (cf. Lindzen et al. 2001) Prediction conflicts Precession of Mercury’s Different values deduced for climate perihelion (1859) sensitivity: 3 K (IPCC), 3–4 K (Hansen) Improvements Special relativity (1905): Aerosols included in IPCC AR3 E = mc2 (2001d); carbon feedbacks included in IPCC AR4 (2007d) Acceptability: Everyday mechanics is Earth is habitable because of (i) to lay people Newtonian greenhouse gases (ii) to scientists Eddington proves General Human disturbance demonstrated by

Relativity (1919) and space Keeling’s CO2 observations from 1958 agencies use in space and by temperature increases programmes since ~1990 (iii) broad public Stephen Hawking and Star Bill McKibben and the ‘350.org’ Trek movement

Source: Henderson-Sellers (2012). Reproduced with permission of Elsevier.

1.3.2 Climate models illuminate essential for life. The climatically habitable zone salient features and core (or circumstellar habitable zone, CHZ) is the region around any star within which it is theo­ uncertainties retically possible for a planet with sufficient Issue: Goldilocks and rotating dishpans. atmospheric pressure to retain liquid water on its Message: models illustrate and help analyse surface. This habitable zone has to be neither planet-wide complex flows that redistribute ‘too hot’ nor ‘too cold’ but ‘exactly right’ so that energy. it has come to be called the ‘Goldilocks’ zone’.12 Astronomers try to identify planets occurring Climate modelling has two parent groups: first, inside the CHZ of stars and recently estimated the astronomers who probe energy-absorbing that 6% of close stars have planets in the CHZ.13 gases and analyse stellar radiation; and, second, They use this ‘zone’ to sift their observations for the weather forecasters who hope to extend possible life-bearing planets and moons since their predictions beyond days into months and water in its three phase states is a helpful (if years. This mixed parentage has produced odd not essential) component of climatic stability lurches and happy combinations in the under­ (Figure 1.5). standing of climate and how to simulate it.11 Water is very unusual: it is the only natural sub­ One example we can use is water. From the stance on Earth that co-occurs in all three physi­ point of view of astronomy, water is considered cal states (liquid, solid and gas). The properties

0002006766.INDD 16 1/10/2014 7:16:09 PM Why Model Climate? 17

(a) CHZ around Alpha Centauri A and B Proxima (α Centauri C) 11 AU distance at closest a (13,000 astronomical units away from A and B)

pproach

α Centauri A 2 AU limit for stable orbits

Life zone

α Centauri B

The sun and its terrestrial planets (on the same scale)

(b)

Alpha Centauri B from its newly discovered planet Alpha Centauri Bb

Figure 1.5 (a) Schematic of the circumstellar habitable zone (CHZ – in green) around Earth’s two nearest neighbouring stellar systems: Alpha Centauri A and B, together with (b) an artist’s impression of how the latter star might look from its newly discovered planet: Alpha Centauri Bb.14 In other words, if alien life occurs in this constellation it is most likely to be found in one of these green zones and such life probably views its ‘sun’ much as this depiction.15 Source: (a) CHZs of Alpha Centauri A and B from Beech (2012). Reproduced with permission of Oxford University Press. (b) Artist impression by ESO/L. (http://www.eso.org/public/archives/images/screen/eso1241a.jpg). L. Calçada/Nick Risinger (skysurvey.org).

of water that give rise to this include the unusu­ However, there are analogue16 models that ally large latent heats of both condensation and ­illuminate fundamental aspects of the climate solidification; its unusually large specific heat ­system: one is the double pendulum, another is (which means that water bodies absorb heat and the now famous ‘rotating dishpan’. This latter cli­ retain it well); the ‘chaining’ behaviour of mole­ mate model is, as its name suggests, a piece of cules in the vapour phase, which causes surface apparatus comprising a cylindrical pan mounted ‘stickiness’, dimer absorption and many more on a turntable. The aim is to simulate the com­ fascinating characteristics. bined effects of planetary rotation (spinning the Climate models are usually assumed to be dish) and of an imposed equator-to-pole tem­ numerical simulations run on computers. perature gradient (heating the edge and cooling

0002006766.INDD 17 1/10/2014 7:16:11 PM 18 The Climate Modelling Primer

CSI Box 1.2

Weirdness of Water: Water, water, everywhere

‘Earth,’ say observers, especially those lucky enough to have viewed it from outside, would be better named ‘Water’. Almost three-quarters (~71%) of Earth’ s surface is water, the brightest features are frozen water, and the atmosphere is frequently cloudy. The character of our ‘Blue Marble’ is controlled by water and, importantly for climate modelling, by water’ s phase transi- tions: changes of state from ice to liquid, and to vapour and back again. Water’ s very large latent (hidden) heat of vaporisation (2270 kJ kg −1 ), a result of the extensive hydrogen bonding between its molecules, moderates climate because large energies are required to change state. Water ’ s high specific heat (4187 J kg- 1 K- 1 ) means that oceans take a long time to warm and to cool, offering climate change buffering. Water ice is (very unusually) less dense than its liquid state, so lakes and oceans freeze top first, separating atmosphere from water. Photo source: NASA.

the centre) on the motion of the fluid in the cylin­ These patterns of flow, functions of the der. Laboratory experiments of this kind began in pole­to­equator temperature gradient and the the 1950s (see, for example, the work of Raymond speed of rotation, are found in all planetary Hide17 of the UK Meteorological Office and David atmospheres and in oceans, although the latter Fultz of the University of Chicago18 ) and remain an are constrained by continents. The differential important tool for weather and climate analysis. heating (the fact that the equator is heated more In the classic dishpan experiment, the motion than the poles) drives the meridional circulation – of the water in the cylinder is tracked using col­ termed the Hadley Cell in the low latitudes of the oured objects or dye and, as the dish is rotated, Earth’ s atmosphere and Rossby waves in the a variety of patterns is seen when viewed with a mid­latitudes. The ocean circulation possesses camera rotating with the dish. At low speeds, the elements of this equator­to­pole energy move­ flow is around the dish but, as the rotation speed ment complicated by continental boundaries and increases, waves develop, extending almost from the presence of density and salinity differences the centre to the perimeter and incorporating (Figure 1.7 ). smaller, closed circulations. Patterns can be The fundamental characteristics of all atmo­ made clearer by dropping two colours of dye spheres and oceans (fluids) on all astronomical into the water – one close to the cold centre and bodies (planets, moons and even stars) are heat­ another near to the warmed periphery. The dish­ ing as a result of radiation imbalance (incoming pan’ s shift from zonal to wave­like flow is closely absorbed not equal to outgoing emitted) at dif­ analogous to observed shifts in upper tropo­ ferent latitudes and fluid flow carrying energy to spheric dynamics (Figure 1.6 ). resolve this imbalance.

0002006766.INDD 18 1/10/2014 7:16:12 PM (a)

(b) 140° W 160° W 0° E 160°E 140° E

40

120° W 120° E 35

100° W 100° E 30 m/s

25 80° W 80°E

20

60° W 60°E

15

40° W20° W0°E 20°E 40°E

Figure 1.6 Dishpan planets: rotating tank experiments capture the main features of atmospheric circulation. (a) Patterns of tracer (dye) injected as drops into a spinning tank from the warm outside edge (red ) and from the chilled centre (green). Source: Jason Smith, University of Chicago. (b) Weather forecast chart for the Northern Hemisphere for 28 January 2013 showing the isobars of mean sea-level pressure and the 850 hPa wind speed in yellow-green. Source: European Centre for Medium-Range Weather Forecasts (ECMWF): www.ecmwf.int/. © ECMWF.

0002006766.INDD 19 1/10/2014 7:16:16 PM 20 The Climate Modelling Primer

(a) (b) Polar cell

Depression Rossby waves (Jet stream)

Anticyclone Sources Sources Hadley cell AABW NADW

Equator ITCZ

Anticyclone Surface waterDeep water Rossby waves Depression (Jet stream) NASA/NOAA image Polar cell Figure 1.7 Primary features of the atmospheric and oceanic circulation. (a) The atmospheric circulation is determined primarily by the net radiation budgets (excess in the tropics, Inter-tropical Convergence Zone [ITCZ], and deficit near the poles) and the rotation of the Earth (especially the Rossby waves). (b) The thermohaline circulation of the ocean, often referred to as the ‘ocean conveyor’, results in the movement of water throughout the major ocean basins of the world over periods of hundreds to thousands of years. Green dashed lines follow deep ocean water, from their sources off the Greenland and Antarctic coasts, North Atlantic Deep Water (NADW) and Antarctic Bottom Water (AABW) respectively, as it transforms into middle layer and finally surface water ( red ) and closes the ocean circulation system.

Climate Model Communication Box 1.1

Modelling hobbyists meeting

This communication exercise is also known as fanatical model train set owner, the ‘CEO in the lift (elevator)’ speech. It com- say, could understand, what climate modellers prises a very quick (2-minute) explanation of do and how their activities are interesting and what you do and why, in your view, it is impor- worthwhile. Use no more than 100 words, or tant. For this example, please imagine you out loud less than 2 minutes of have been invited to a “modelling hobbyists’ speech. An interesting place to convention” – a gathering of all types of mod- check out if you are using jargon ellers. At the opening social, you are asked is at http://www.climatemodel- what you model. Try to explain, in terms that a lingprimer.net/l/k104.htm .

0002006766.INDD 20 1/10/2014 7:16:17 PM Why Model Climate? 21

elegant and extraordinarily pretty demon­strations Reflection on Learning 1.3 of the wonder of planetary-scale circulation in the atmosphere and the oceans (Figure 1.8). List the factors affecting planetary scale climate Climate is controlled by universal laws including 1.3.3 Climate models reveal the conservation and the three laws of thermodynam­ apparently simple to be ics, which C.P. Snow once wittily summarised as complex and vice versa ‘you can’t win, you can’t break even, and you can’t get out of the game’. For planets, these basic Issue: the butterfly effect seen in the swing of a rules include (i) radiation in equals radiation out; double pendulum. and (ii) axial spin adds a Coriolis twist to equator- Message: models capture essential features and to-pole flows. From such basic rules, model­ illustrate unexpected behaviour as seen in cha- lers and experimentalists design beautifully otic attractors.

Figure 1.8 Climate models are designed to represent the major circulation features of the atmosphere and ocean. For example, this atmospheric simulation of the stationary eddy component (the departure from the zonal mean) of the 500 hPa geopotential height (m) in boreal (northern) winter (DJF) uses a coupled climate model (the EC-Earth model). Source: After Hazeleger et al. (2012). Reproduced with permission of Springer Science+Business Media.

80

70

60

50

40

30

20

10

–10

–20

–30

–40

–50

–60

–70

–80

0002006766.INDD 21 1/10/2014 7:16:18 PM 22 The Climate Modelling Primer

Tech Box 1.2

View the Lorenz Attractor unfolding

See : http://www.climatemodelling consists of 5000 spheres, the colour of which primer.net/l/k105.htm changes as the iterations continue. The equations This video allows the viewer to being solved are as identified by Edward Lorenz in witness the dynamic 3D evolution of 1963.19 We will be properly introduced to this extra- the Lorenz Attractor. The simulation ordinary model in Chapter 2, Speed Date Box 2.

2:56

Figure source: http://www.youtube.com/watch?v=iu4RdmBVdps&feature=fvwrel.

Massachusetts Institute of Technology mathema­ was not until Lorenz ’ s seminal work in 1963 21 that tician Edward Lorenz is famous for what has the true nature of behaviour constrained by what become known as the ‘butterfly effect’, although we now know of as ‘fractionally dimensioned he actually used the metaphor of a seagull ’ s geometry’ became clear. Thus, the salient fea­ wing.20 The term describes the chaotic amplifica­ ture of climate (and many other) models is the tion of a tiny disturbance, which means for inherent unpredictability of the systems they climate models that a small change in the initial describe. The core uncertainty is in the final conditions can result in a quite different final outcome of each simulation. As Lorenz himself state (Tech Box 1.2 ). discovered, minute differences in initial condi­ The model Lorenz used, which led him to pub­ tions translate into completely diverse final lish the first description of this now famous effect outcomes. and which has since been found in virtually every It is fairly easy to simulate chaotic behaviour branch of predictive science, consists of just with a real (physical) model – one that you can three equations. Although Henri Poincaré had build yourself. A double pendulum consists of noted the notion of chaotic unpredictability, it two rigid links (sticks) of very little mass with point

0002006766.INDD 22 1/10/2014 7:16:19 PM Why Model Climate? 23

Biography Box 1.1

Meet the modeller: Edward N. Lorenz

Leadership: Lorenz was the first to recognise since Sir Isaac Newton. Speaking soon after his what is now called chaotic behaviour. His discov- death, Kerry Emanuel, also of MIT, said, ‘Ed put ery overthrew the status quo in climate science, the last nail in the coffin of the Cartesian uni- requiring that all climate modellers understand verse and fomented what some have called the the complexity of the system they study. third scientific revolution of the 20th century’.

Popular recognition: Lorenz was famously Read more responsible for naming the ‘butterfly effect’. Gleick , J. ( 1987 ) Chaos: Making a New Science . London : He claimed that this phrase arose as a conse- Vintage . Lewis , J.M. ( 2005 ) Roots of ensemble forecasting . Mon quence of his failure to provide a title for a talk Weather Rev 133 , 1865 – 1885 . in 1972. To plug the title gap, the organiser Lorenz , E. ( 1963 ) Deterministic nonperiodic flow . J Atmos creatively listed, ‘Does the flap of a butterfly ’ s Sci 20 , 130 – 141 . wings in Brazil set off a tornado in Texas?’. Watch Climate modelling connectivity: Lorenz ’ s 1963 http://www.climatemodellingprimer.net work, which laid the foundation for the field /l/k109.htm of chaotic systems, used a truncated form of the atmospheric convection equations of Barry Saltzman, another giant intellect in early cli- mate modelling.

Life and times: Edward Norton Lorenz (23 May 1917 – 16 April 2008) was an American mathe- matician and meteorologist and for many years a professor at MIT. In the early 1960s, Lorenz realised that small differences in a dynamic sys- tem such as the atmosphere – or a model of the atmosphere – could trigger vast and often unsuspected results. It has been said that his profoundly influential work delivered one of Ed Lorenz in 1994. Photo source: © UCAR, photo by Curt the most dramatic changes in our view of nature Zukosky.

weights attached to their ends. These links (see Tech Boxes 1.2 and 1.3 ) that has come to (sticks) are confined to two­dimensional rotational characterise climate dynamics has been discov­ motion about their joint. Initially the motion ered in ecology and, more recently, socio­eco­ appears regular but chaotic movements always nomic systems.22 For example, Figure 1.9 shows occur with the end point of the pendulum even­ discontinuous transitions in the value of supply of tually covering the whole space available to it for money market derivatives (s ) as a function of the a given starting energy (Tech Box 1.3 ). number ( n) of derivatives being traded and the Although first simulated in simple climate mod­ risk (e ) of the various banks operating in a compli­ els in the 1960s, the ‘strange attractor’ behaviour cated international financial system.23

0002006766.INDD 23 1/10/2014 7:16:21 PM 24 The Climate Modelling Primer

Tech Box 1.3

Craft a double pendulum or crochet a Lorenz Manifold

Build a simple device that exhibits chaotic behaviour: an excellent science project or cha- otic climate conversation starter.

1 Build a double pendulum Build it – find out how to create one here: http://www.climate modellingprimer.net/l/k106.htm Get the PDF for this project: http://www.climatemodelling primer.net/l/k107.htm

2 Crochet your own Lorenz Manifold24 Find out how to create this folded surface in wool: http://www.climate modellingprimer.net/l/k108.htm

Consider how these two hobby- ists’ activities relate to one another, Figure source: Osinga and Krauskopf ( 2004 ). Reproduced i.e. what links the behaviour of the double with permission of Springer Science+Business Media. pendulum to the Lorenz manifold surface.

0.9 Figure 1.9 Complex systems exhibit instabilities. e = 0.1 0.8 e = 0.01 Here international trading in financial derivatives is e = –0.01 seen to behave discontinuously. The average supply of 0.7 = –0.1 e any one derivative, s , at competitive equilibrium as a 0.6 function of the number, n , of different derivatives being 0.5 traded, for various values of banks’ risk premium, e . The s 0.4 two ‘branches’ can be viewed as similar to the strange attractors seen in the simulation described in Tech 0.3 Box 1.2 . Source: Haldane and May ( 2011 ). Reproduced 0.2 with permission of Nature Publishing Group. 0.1

0 1 10 n

1.3.4 Climate models raise new Message : social and physical systems with feed- questions and suggest backs grow anomalies and also dampen them so models have to capture this behaviour . analogies Issue : positive and negative feedback seen in a The term ‘feedback’ originates in early electron­ poorly rigged microphone system . ics. In 1909, Nobel Laureate Karl Ferdinand

0002006766.INDD 24 1/10/2014 7:16:24 PM Why Model Climate? 25

Output acts to enhance input sunlight than the ice so the ocean warms still + ­further, making it harder to reform sea-ice and POSITIVE tending to reinforce the first effect. FEEDBACK The importance of the sign of a feedback pro­ cess can be simply illustrated by considering the Process Input Output impact of self-image on diet. Someone slightly overweight who eats for consolation can become depressed by their increased food intake and so Output acts to reduce input eat more and rapidly become enmeshed in a – detrimental, positive feedback effect. On the other hand, perception of a different kind can be NEGATIVE FEEDBACK used to illustrate negative feedback. As a city grows, there is a tendency for immigration but Process the additional influx of industry, cars and people Input Output is often detrimental to the environment, so that it may be balanced by an outflow of wealthier Figure 1.10 Types of feedback. Feedback processes inhabitants, with a potentially negative impact can be classified as positive or negative. In positive on the central city’s economy by reducing invest­ feedback, a portion of the output is fed back to the input ment. Because feedback mechanisms act to fur­ and acts to further stimulate the process. In the case of ther enlarge or suppress the initial disturbance, negative feedback, the portion of the output is their incorporation into models of climate is subtracted from the input and acts to dampen the essential. Early in the story of climate modelling, process the concept of feedback was recognised as important (Feedback Box 1). Braun used the term feed-back to describe unde­ Positive feedback magnifies changes and can sired coupling between elements of an electronic lead to the crossing of thresholds beyond which circuit. In the broadest sense, a feedback occurs a new state is entered from which return is not when a portion of the output from the action of a achievable by removing (or reversing) the original system is added to the input and subsequently disturbance. This idea of positive feedback rein­ alters the output. The result of such a loop sys­ forcing an initial perturbation has been exploited tem can either be an amplification of the process, in many areas such as social systems – for exam­ resulting in the familiar amplifier howl, or a ple, in riot situations where unintentional rein­ dampening, which is used for control in almost forcement (maybe by introducing law enforcement every modern amplifier circuit. These feedbacks personnel) tends to further enrage crowds. are labelled positive and negative respectively. Nobel Prize winner Thomas Schelling exp­ In a climate situation, positive feedbacks act to anded upon the ‘tipping point’ concept in the grow an initial perturbation whereas negative 1970s and a very readable account of such feedbacks reduce the perturbation (Figure 1.10). ­phenomena is found in the book The Tipping Today, ‘feedback’ has become closely associ­ Point: How Little Things Can Make a Big ated with climate; for example, most people rec­ Difference.25 This lists many examples of this ognise its use in the phrase ‘ice-albedo feedback’ threshold-crossing concept and focuses, in par­ even though they might not be able to define ticular, on examples where small changes create the word ‘albedo’ (see Chapter 3). In the mass big effects. The tipping point from which the media, feedback, as the reinforcement of the book takes its title is that point in a system’s impacts of warming on Arctic sea-ice, is in com­ development where a small change leads to a mon usage: as the ice-albedo feedback operates huge effect in a rapid timeframe and spreads as temperatures rise, the sea-ice melts. This leads through the system in a contagious fashion. This to a greater area of dark ocean as the amount of was shown in Figure 1.9 for world markets in bright ice decreases. Dark surfaces absorb more banking derivatives.

0002006766.INDD 25 1/10/2014 7:16:24 PM Feedback Box 1

The ‘CLAW’

Read : Charlson , R.J. , Lovelock , J.E. , Andreae , depiction of the biologically driven climate M.E. , Warren , S.G. ( 1987 ) Oceanic phytoplank- feedback mechanism: dimethylsulfide (DMS), ton, atmospheric sulphur, cloud albedo and produced by oceanic plankton and oxidised climate . Nature 326 , 655 – 661 . in the atmosphere to form a sulfate aerosol that is the main source of CCN over the oceans. This paper is famous for presenting the first The albedo (reflectance) of clouds is altered by testable Gaia Hypothesis26 (that living things changes in CCN density and so biological modi- ‘work’ to modify the climate) and for being fication of the climate ensues. The diagram among the first to discuss the impact of aero- shows how measurable quantities (in rectan- sols (small droplets or particles) on the climate gles) are changed by processes (in ovals) where (the addition of aerosols to cool climate is at the signs show the effect of a positive change the heart of many geoengineering proposals). of the quantity in the preceding rectangle. The paper concludes that counteracting the The CLAW authors note that the least certain

warming due to doubling of atmospheric CO2 of the depicted processes is the effect of cloud requires approximately a doubling of cloud albedo on DMS emission – they show this oval condensation nuclei (CCN). The paper’ s fame (lower right corner of figure) with either plus or has resulted in it being referred to by the acro- minus impact. This must be positive if climate nym built from the first letters of the authors’ regulation is to occur. Thus, if the initial distur- family names: Charlson, Lovelock, Andreae bance is, say, an increase in solar radiation, this and Warren: ‘CLAW’. Here, we focus on its prompts more DMS; DMS is oxidised to sulfate aerosols; more aerosols mean more CCN; this increases the cloud droplet Number concentration Formation of of cloud droplets water-soluble particles number; hence the clouds become (Fixed LWP) brighter and so reflect more solar radiation; and so the initial distur- Cloud nucleation cloud albedo bance is dampened. In Gaian terms, the plankton act to reduce (i.e. dimin- Loss of solar ish) the disturbance to the climate. radiation to space CCN The CLAW hypothesis has its own entry in Wikipedia and there is now Formation of an ‘anti-CLAW’ mechanism created water-soluble particles by Jim Lovelock, one of the authors of this paper and the co-inventor of

2– the Gaia Hypothesis. NSS - SO4 Surface Solar temperature irradiance The way in which the feedback of earth below clouds diagram is drawn here (from the Oxidation CLAW paper) follows a terminology originally proposed by William Kellogg. Questions that the figure DMS (gas) raises include: 1. Can this feedback system be drawn Sea-to-air transport more simply? 2. Does this version show a positive Atmosphere or a negative feedback? Ocean Production of DMS Figure source: Charlson et al. ( 1987 ). Repro- by marine phytoplankton DMS (aq) duced with permission of Nature Publishing ? Group.

0002006766.INDD 26 1/10/2014 7:16:26 PM Why Model Climate? 27

Climate change entered a different regime This time, the character of public perception of in 2007 with the publication of the Intergovern­ anthropogenic climate change was transformed men­tal Panel on Climate Change’s (IPCC’s) Fourth by media coverage of what was an inconsequen­ Assessment Report (AR4) and the joint award of tial error in Working Group Two IPCC AR4 and the Nobel Peace Prize to Al Gore and the IPCC. selected contents of the Climatic Research Unit People no longer asked ‘whether’ human activi­ of the UK’s University of East Anglia stolen emails ties are changing the climate but the more urgent trail.28 The press pushed public perception past a questions of: ‘how fast?’, ‘with what impacts?’ social tipping point in December 2009. The weak and ‘demanding what responses?’. In late 2007, Copenhagen Accord (2009) and the reduced a virtuous cycle reinforced the public’s recogni­ pressure for climate mitigation legislation felt by tion of the need for urgent action to mitigate world leaders are symptoms of the new social change. Positive feedback, including in the state resulting from crossing this irrevocable media, showed climate change to be a risk man­ social threshold. agement problem to be solved by all nations The true integrity question about anthropo­ (Figure 1.11). genic climate change is really whether, and with The mass media reversed (from virtuous to what priority and pressure, urgently required vicious) the direction of their positive feedback actions are agreed and taken. Models are a key on anthropogenic climate change late in 2009.27 component to these decisions.

Figure 1.11 Virtuous and vicious circles created by a positive feedback that operated around two UNFCCC COP meetings (in 2007 and 2009). Public understanding and political will were reinforced through media coverage of the issue of anthropogenic climate change in two strong processes of positive feedback in 2007 and 2009. In the virtuous case, positive feedback strengthens public belief in the reality of climate change and, thus, the political will to respond to the threat at COP13 in Bali in December 2007. In the vicious case, a media clamour of positive feedback strengthens public interest in IPCC errors and in claims (later refuted) of abuse of the processes of IPCC and, thus, greatly reduces the political will to respond to the global warming threat at COP15 in Copenhagen in December 2009. Source: After Henderson-Sellers (2010). Reproduced with permission of Springer Science+Business Media.

If you’re not going to lead, please get out of the way, Bali 12/07

3/10 UNFCCC Secretary quits Virtuous circling 3/10 UK Inst/ Physics Oct 07–Jan 08 11/09 UEA statement email hack 3/10 UK Parliament review 12/09 COP15 Nobel Peace Prize 10/07 sceptics on hack Vicious circling Nov 09 – Jan 10 12/09 WG2 glacier 2/10 WG2 chair melt date error response

1/10 Pachauri on UEA & Penn State glacier error launch reviews

1/10 more errors claimed 1/10 wider UEA emails trail

0002006766.INDD 27 1/10/2014 7:16:26 PM 28 The Climate Modelling Primer

tant in climate modelling: cloudiness changes as Reflection on Learning 1.4 other factors alter. However, correct (valid and robust) parameterisation of clouds in climate Explain the concept of climate models has yet to be achieved. For example, feedback and give examples Figure 1.12 illustrates how cloudiness might Clouds are important: clouds can be gloomy but change: will ‘more cloud’ mean more stratus (low they also create beautiful sunsets. Clouds are level and hence cooler), more cumulus (can be important in climate: they control radiation input towering and very bright) or more cirrus (thin and (because of their high albedos) and they also high)? Each of these changes involves dynamical play a large part in the emitted infrared radiation and thermodynamical feedbacks as illustrated (because of their height in the atmosphere and here for the case of a change in the amount of hence lower temperatures). Clouds are impor­ cumulus convection.

(a) Types of cloud feedback More cumulus

Normal More stratus

More cirrus

(b) Factors associated with cloud feedback

Dynamical Thermodynamical

Greater + + – cloud cover Less moisture convection Less cloud Less surface + convection radiation Less mass circulation + + Less + evaporation

Figure 1.12 Cloud-climate interactions. (a) When climate shifts cause (for example) ‘more cloud’, this can be manifested in many ways: as more low-level layer cloud (stratus), as more towering tall cloud (cumulus) or as more high thin cloud (cirrus). The cloud could either be more extensive vertically or more extensive horizontally. Each of these ‘cloud increases’ creates a very different radiation effect and hence a different feedback onto climate. (b) Examples of dynamical and thermodynamical feedbacks and their directions in the case of a change in the amount of cumulus cloud convection. Source: Modified from Henderson-Sellers and Robinson (1986). Reproduced with permission of Longmans.

0002006766.INDD 28 1/10/2014 7:16:31 PM Why Model Climate? 29

Tech Box 1.4

Planetary climate model – effective temperature and surface temperature

The simplest global climate model can be writ- If there are greenhouse gases in the planet ’ s ten as follows: atmosphere then its surface temperature is the sum of this effective temperature and S 4 (1−=ασ) Te (1.1) 4 the greenhouse warming caused by these gases’ absorption and re-emission of infrared Using values of S = 1370 W m- 2 , a = 0.3 and s = -8 -2 -4 radiation. 5.67 × 10 W m K gives for T e (the planetary- wide effective temperature) a value of 255 K, TTse= +∆ T (1.2) or –18°C, in good agreement with the Earth’ s Today ’ s greenhouse effect, D T , of around +34°C average effective (not surface) radiative tem- delivers a global mean surface temperature of perature today. 289 K.

1.3.5 Climate models expose there is some dispute about the geographical prevailing wisdom as extent of the glacial deposits, the climate model­ based topic is fairly independent of these. compatible or incompatible The question posed by widespread geological with existing data and evidence of early global glaciation is, can a fully hence direct collection of glaciated Earth ‘recover’, i.e. warm up, so that new data large­scale oceans are once again unfrozen? It has been noted above that the ice­albedo feedback Issue: geological evidence of massive climate is positive or reinforcing. Once a fully glaciated change (e.g. Snowball Earth) . Earth occurs, almost any climate model will strug­ Message : models test theories about early Earth gle to deglaciate. glaciations – evaluating the impact of strong Using a very simple climate model (Tech positive feedbacks and climate recovery . Box 1.4 ), it is straightforward to pinpoint the aspects of climate that must be modified to alter An effective positive feedback mechanism was a trajectory. This very simple model shows that discussed in the previous section. Climate mod­ globally averaged mean surface temperature els can be shown to be instrumental in probing depends on only two factors susceptible to data by reference to the phenomenon of speedy (near real­time) modification: the albedo ‘Snowball Earth’. At its simplest, the Snowball and the greenhouse effect. Earth idea refers to occasions in the geological Changing the global albedo from today ’ s value record when glacial formations appear to have of 0.3 to a much larger value, commensurate covered the whole Earth, or at least are found with a global glaciation of, say, 0.8, gives an in locations at or near the equator, as well as effective temperature for the planet of 186 K. at mid­ and high latitudes. These periods are Adding today’ s greenhouse increment lifts this to not fully agreed, but seem to begin with a Pre­ only 220 K (or –53°C). This is a ‘best case’ since Cambrian glaciation (the Huronian) that occurred the albedo might have been higher and the solar between 2.1 and 2.4 billion years ago, followed radiation (S ) was certainly lower in the past. These by three Neo­Proterozoic glaciations dated climate model calculations underline how easy it between 650 and 800 million years ago. Although is to encourage persistent glaciation. However,

0002006766.INDD 29 1/10/2014 7:16:32 PM 30 The Climate Modelling Primer

CSI Box 1.3

Weirdness of Water: Buddy, can you spare a dime-r

Understanding water means picturing its name ‘universal sol- + H molecular construction: four hydrogen bonds vent’. The smallest produces an open three-dimensional network – water molecule clus- H H a tetrahedron; the dipole charge on each ter contains only O water molecule means that they are attracted two water mole- H H O to one another. This molecular form and cules: it is the dimer. mutual cohesion are the source of many of the Water vapour dimer H O weird properties of water of great significance absorption contrib- – to Earth, to life and to climate simulation. For utes to greenhouse example, water ’ s high melting and boiling warming and, altho- point temperatures; and its large specific and ugh it has been known to experimentalists for latent heat capacities. Generally, water prefers many years, it remains a challenge to properly to ‘hang together’ but it also dissolves a very code into radiative transfer modules of climate wide range of substances, giving rise to its models.

the geological record also shows that these gla­ ical situation has been termed ‘Dustball’ or ciated periods stopped. Can this simple model ‘Mudball Earth’ while the latter has the name be used to suggest how this could have occurred? ‘Slushball’.30 Both have been extensively exam­ A greenhouse temperature boost of around ined using all types of climate models from very 100 K is required to defrost Snowball Earth. Since simple to fully coupled.31 The climate modelling we know that the greenhouse effect on Venus is barrier to freeze­thaw shifts has also prompted roughly +500 K,29 this heating is by no means additional geological hypotheses – for exam­ impossible, but it does raise interesting ques­ ple, involving changed chemistry of the Earth’ s tions about the geological mechanisms that atmosphere.

cycled CO2 (or other greenhouse gases) in and New geochemical data show two significant

out of the atmosphere on three or four occasions increases in atmospheric O 2 levels at around during Earth ’ s history. 2.4–2.3 and 0.8–0.6 billion years ago, times that This geological puzzle persists and tempts cli­ coincide with the Snowball Earth glacial evidence. mate modellers to probe its possible causes The geochemical story tells of oceanic sulfate and effects. Large mixing ratios of carbon diox­ concentrations increasing in accord with greater

ide can be related to sedimentary rocks later O2 levels, while levels of methane, a strong deposited and the stark facets of the simple cal­ greenhouse gas, appear to be the mirror inverse

culations above can be ameliorated by, for of atmospheric O 2 levels. These so­called ‘oxic example, postulating a lower albedo on the transitions’ are characterised by significant snowball – say because of dust deposited fol­ disturbances in the carbon cycle. A plausible lowing large volcanic eruptions or by arguing case can be made linking the delay between that world oceans would not fully freeze and the appearance of oxygenic photosynthesis might maintain near zero (°C) water tempera­ and oxygenation of the atmosphere to active

tures overlain by wet (and hence grey rather scavenging of newly created O2 out of the early than bright white) sea­ice. The former hypothet­ atmosphere.

0002006766.INDD 30 1/10/2014 7:16:33 PM Why Model Climate? 31

equinoxes. These changes occur in such a way 1.3.6 Climate models explain that two main periodicities are apparent: 23,000 Issue: radiation controls climate. years and 18,800 years. This change, like that Message: models calculate the impact of insola- of obliquity, does not alter the total radiation tion changes – glacial/interglacial cycles follow received but does affect its temporal and spatial solar input. distribution. For example, perihelion is currently on 5 January, in the middle of the Northern The orbit of the Earth is an ellipse around the Hemisphere winter, but 11,000–15,000 years Sun, which lies at one of the foci. There are sev­ from now it will occur in July. At the present-day eral different ways in which the orbital configura­ value of eccentricity, there is a range of ~6% in tion can change to affect the received radiation the solar radiation incident at the top of the and thus the climate. These ‘Milankovitch32 vari­ atmosphere between perihelion and aphelion ations’ describe the changing parameters of the (i.e. ~1411 to 1329 W m-2). Earth’s orbit around the Sun. They are: (i) changes Spectral analysis of long-term temperature data, in eccentricity; (ii) changes in obliquity; and (iii) such as the records in Figure 1.13b, has shown changes in orbital precession (Figure 1.13). The the existence of cycles with periods of ~20,000, Earth’s orbit becomes more eccentric (elliptical) ~40,000 and ~100,000 years (Figure 1.13c). These and then more circular in a pseudo-cyclic way, correspond closely with the Milankovitch cycles. completing the cycle in about 110,000 years. The strongest signal in the observational data, The mean annual incident flux varies as a func­ however, is the 100,000-year cycle. This cycle tion of the eccentricity of the orbit, E. For a ­corresponds to that of eccentricity variations in larger value of E, there is a smaller incident the Earth’s orbit but eccentricity variations pro­ annual flux. The current value of E is 0.017. In duce the smallest insolation changes resulting in a the last 5 million years, it has varied from challenge for climate models. Modelling results 0.000,483 to 0.060,791, resulting in changes in have suggested that the present configuration of the incident flux of +0.014% to -0.170% the landmasses in the Northern Hemisphere (~0.19 W m-2 and ~2.3 W m-2 respectively) from may favour rapid development of ice caps when the current value. ­conditions favour cool Northern Hemisphere The obliquity, the tilt of the Earth’s axis of rota­ ­summers. Almost certainly, these external changes tion, is the angle between the Earth’s axis and in insolation trigger feedback effects in the climate the plane of the ecliptic (the plane in which the system that demand correct incorporation into Earth and other bodies of the solar system orbit models. the Sun). This tilt varies from about 22° to 24.5°, with a period of about 40,000 years. The current value is 23.5°. Seasonal variations depend upon the obliquity: if the obliquity is large, so is the range of seasonality. Although the total received radiation is not altered, a greater seasonal varia­ tion in received flux is accompanied by a smaller meridional gradient in the annual radiation. (Recall that the meridional net radiation imbal­ ance drives atmospheric and ocean circulations – see Figure 1.7.) Owing to gravitational interaction with the other planets, primarily Jupiter, the perihelion (the point of the Earth’s elliptical orbit closest to the Sun) moves in space so that the ellipse is moved around in space. This orbital precession will cause a progressive change in the time of the

0002006766.INDD 31 1/10/2014 7:16:34 PM 32 The Climate Modelling Primer

(a)

A1 Earth

a Sun Aphelion Perihelion b Obliquity P1 22° 23.5°

24.5° (a2–b2) Eccentricity E = √ a 24.5°

(b) (c) Obliquity 100 000 years 24.75 23.25 Larger 43 000 years

Axial tilt ° 21.75 Orbital eccentricity 0.06 24 000 years 0.04 19 000 years E 0.02 0.00 Irradiance 560 at 38°S)

–2 500 irradiance December 440 (W m Amplitude of climatic cycle Vostok ice core –420 –440 –460 δ D (‰)

–480 Smalle r –500 0 100 200 300 400 500 100 30 15 10 7.5 6 Thousands of years B.P. Cycle length (thousands of years)

Figure 1.13 Milankovitch variations in incident solar radiation control the climate. (a) Schematic diagram showing the variations in the three orbital components: obliquity (axial tilt), orbital eccentricity and precession of the perihelion. (b) Variations in these three components over the last 500,000 years together with dD proxy temperature record from the Vostok Ice Core. (c) A spectrum of climatic variations over the last 500,000 years. The graph shows the importance of the climatic cycles of 100,000 years (eccentricity), 43,000 years (obliquity) and 24,000 and 19,000 years (precession of the location of perihelion). The curve is constructed from an isotopic record of two Indian Ocean cores. Source: Imbrie and Imbrie (1979). Reproduced with permission.

0002006766.INDD 32 1/10/2014 7:16:34 PM Why Model Climate? 33

Spotlight on Climate Models Box 1

Palaeoclimate modelling challenge

Read : Braconnot , P. , Harrison , S.P. , Kageyama , tropics). There is a great deal of M. , et al. ( 2012 ) Evaluation of climate models detail given by Braconnot et al. in the supple- using palaeoclimatic data . Nature Clim Change mentary information linked from the publica- 11 , 1 – 8 . tion itself. Their analyses might prompt questions such as: how hard is it to reconstruct The Palaeoclimate Modelling Inter comparison precipitation from palaeo data and how do sci- Project (PMIP), by applying palaeo-evaluation entists reconstruct the temperature of the to simulations made by climate models used coldest month from palaeo data? The figure for future predictions, provides quasi-inde- shows how precipitation anomalies (as MAPs) pendent assessments of model performance. If compare for the five monsoon regions and adequate confidence can be placed in the pal- among the three phases of PMIP and the pal- aeo data, it may be possible to determine aeoclimate reconstructions. An AGCM is an whether a model is sufficiently sensitive to atmospheric (only) global climate model; changes in atmospheric composition or how OAGCM is an ocean-atmospheric model; and well it computes the strength of feedbacks OAVGCM is a global climate model with ocean, that modify the model response to distur- atmosphere and interactive vegetation. bances. This is the basis for the investigation by Braconnot et al. conclude that evaluation of Braconnot et al. spotlighted here. climate simulations against palaeo data shows These authors review the evidence for two that models reproduce the direction and large- aspects of palaeoclimate: the change in the scale patterns of past changes in climate, but mean annual precipitation (MAP) for the mid- tend to underestimate the magnitude of Holocene and the change in mean tempera- regional changes. However, thoughtful inspec- ture of the coldest month (MTCO) for the last tion of this figure may raise questions such as: glacial maximum (LGM). These are assessed for Why is the range (shown by the whiskers of the different regions: for MAP over five monsoon box plots) so much greater for the reconstruc- regions (North Africa, India, East Asia, North tions than for the models? And, since the range America and South America) and for MTCO of observations (truth) is so large, does it mat- over five regions (western North America, ter that the three PMIP simulation sets differ a eastern North America, Europe, Asia and the little?

1,500 3,150

) Reconstructions

–1 PMIP1 AGCMs 1,000 PMIP2 OAGCMs PMIP2 OAVGCMs 500

0 MAP anomaly (mm d –500 North India East North South Africa Asia America Africa

Figure source: Braconnot et al. ( 2012 ). Reproduced with permission of Nature Publishing Group.

(Continued)

0002006766.INDD 33 1/10/2014 7:16:36 PM 34 The Climate Modelling Primer

Discussion question more than a year); and by hind-casting past cli- Climate modellers tackle the need to ‘validate’ mates and assessing how well the model pre- their models in three main ways: by examining dictions compare with reconstructions of these how climate models perform when simulating palaeoclimates. Each of these techniques has a few recent decades to no more than a cen- its own problems and they share the primary tury; by evaluating components of the full challenge of defining ‘how good is good models (e.g. rainfall or radiation) and testing enough?’. Which of these validation tech- the predictions of these components against niques do you believe to be most valuable and observations for a short period (usually no why?

150 Temperature leads CO2 Temperature lags CO2 insolation change creates the largest climate swings and how are latitudinal changes in solar Global 100 460 ± 340 energy transmitted to the whole planet? Analysis SH in the 1980s of CO in air trapped in Antarctic ice –620 ± 660 NH 2 Count 720 ± 330 revealed that greenhouse gas concentrations 50 also increased and decreased over the last glacial cycle. However, neither observations nor models 0 could tease out the exact relationship among –2,000 –1,500 –1,000 –5000 5001,000 1,5002,000 greenhouse gases, solar insolation and climate. Lag (yr) Specifically, proxy data were variously inter­

Figure 1.14 Frequency of occurrence (counts) of preted to suggest that CO 2 was the primary

leading and lagging (years) of atmospheric CO2 driver of the ice ages, a more modest feedback concentration and temperature for the global (grey), on warming and a consequence rather than Northern Hemisphere (NH; blue) and Southern cause of past climate change. Hemisphere (SH; red) during the Pleistocene ice ages Climate models contributed to this confusion

(20–10 kyr ago). CO2 concentration leads the global partly because they had to use snapshot simula­ temperature for the vast majority of the time (90% of the tions rather than through time re­creations and simulations) and lags it in only 6% of the 1000 Monte so could not distinguish the timing of changes in Carlo experiments. Differences between the respective various forcings relative to responses. In 2012, a temperature changes of the Northern Hemisphere and combined analysis of proxy data and new climate Southern Hemisphere are linked to the strength of the model simulations suggested that the local Atlantic Meridional Overturning Circulation (AMOC) Antarctic temperature was strongly correlated recorded in marine sediments. The overall picture is of with and seems to have slightly led changes 33 global temperatures being led by CO 2 concentrations in CO2 concentration (Figure 1.14 ). Whether with an antiphased hemispheric temperature response to this conclusion fully explains the causes of the ocean circulation changes superimposed on this globally Pleistocene ice ages remains to be seen. in-phase warming. Source: Shakun et al. (2012 ). Reproduced with permission of Nature Publishing Group. 1.3.7 Climate models bound The identification of orbital frequencies in (bracket) outcomes within changes in global ice volume (measured using the marine oxygen isotope (18 O/16 O) record as a plausible ranges proxy) demonstrated in the 1970s that climate Issue : land use matters for climate . responds to the Earth ’ s astronomical attitude to Message : models pinpoint how tropical deforest- the Sun. It also posed interesting challenges ation and desertification can alter climate but also for climate modellers: how is it that the smallest in which locations the atmosphere is dominant .

0002006766.INDD 34 1/10/2014 7:16:37 PM (a) 120°W 60°W 0° 60°E 120°E 180°E 90°N

60°N

30°N Sahel

0° ) 8 –1 6 4 30°S 2 Albedos prescribed 0 Permanent desert (0.35) precip (mm d 1324 weeks Idealised marginal (0.14 or 0.35) 60°S Ice cover (0.7) Humid (0.14) 90°S

(b) 90°W 60°W 30°W 90°W 60°W 30°W

0 0° 0 0° +2.5 –125

+2.5 –300 +75 20°S –3.5 20°S

0 40°S 40°S °C mm 0

Jan Jan 60°S 60°S Increase Decrease

Figure 1.15 Historical modelling of the impact of land-use change on climate. (a) Areas for which albedo changes were made in a set of 1970s GCM experiments, by Charney, designed to examine desertification. The inset graph shows the rainfall resulting from increasing the surface albedo from 0.14 to 0.35 in the Sahel region when free evaporation was permitted. (b) Simulated temperature (left) and precipitation (right) changes following replacement of the Amazon tropical moist forest by scrub grassland in a 1980s GCM. These are 5-year means from the end of a 6-year deforestation experiment. Areas of significant increase or decrease (using Student’s t) are shown. Source: (a) Henderson-Sellers and Wilson (1983). Reproduced with permission of the American Geophysical Union.

–120° –60° 0° 60° 120°

20° 10° 0° –10°

Latitude north –20°

Figure 1.16 Climate model investigation of the transport of moisture into tropical regions of interest (four boxes) shown as 10-day back-trajectories of air arriving daily during one example year (2001) calculated using the ECMWF model. Source: Spracklen et al. (2012). Reproduced with permission of Nature Publishing Group.

0002006766.INDD 35 1/10/2014 7:16:43 PM 36 The Climate Modelling Primer

Humans are now recognised as major agents in forests are replaced by cropland. One area that regional-scale changes of the character of the is undergoing deforestation is the Amazon Basin Earth’s surface. These include desertification, ­ in South America. The important change in re- and deforestation, urbanisation and engi­ deforestation is in the surface hydrological char­ neering of major rivers, lakes and dams. Climate acteristics, since the evapotranspiration from a modellers have investigated the climatic effect of forested area can be many times greater than such changes in the nature of the Earth’s conti­ from adjacent­ open ground. Most climate model nental surface for over 40 years. The more recent ­simulations of Amazonian deforestation show a evaluations not only demonstrate human distur­ reduction in moisture recycling (because of the bance of climate flowing from land-surface lack of the moist forest canopy), which reduces changes but also allow consideration of pos­sible precipitation markedly (Figure 1.15b). However, mitigation techniques to try to res­tore aspects of the available global model experiments do not regional climate previously disturbed. agree on whether an increase in surface temper­ Desertification is a problem affecting millions ature occurs. The largest impacts are the local of people. The sparse vegetation natural to arid and regional effects on the climate, which could and semi-arid areas can be easily removed as a exacerbate the effects of soil impoverishment result of relatively minor changes in the climate and reduced biodiversity accompanying the or by direct influence of human activity such deforestation. It has proved possible to detect as overgrazing or poor agricultural practices. impacts resulting from tropical deforestation Removal of vegetation and exposure of bare soil propagating to the global scale by increasing increase albedo and decrease soil water storage, the length of the integrations and adding ensem­ because of increased run-off. Less moisture ble members to the suite of simulations ­available at the surface means decreased latent studied.34 heat flux, leading to an increase in surface From the early 1970s, when Jule Charney cal­ ­temperature. On the other hand, the increased culated the possible impact of marginal desert albedo produces a net radiative loss. In climate land use, past the first deforestation simulation in model calculations, the latter effect appears to the 1980s to the present day, climate models dominate and the radiation deficit causes large- have been used in combination with observa­ scale subsidence. In this descending air, cloud tions to obtain an increasingly persuasive case and precipitation formation tend to be sup­ for reduction of human impact in vulnerable pressed and aridity increases (Figure 1.15a). This regions.35 For example, Figure 1.16 illustrates a global simulation involves a surface albedo modelling result (here back trajectories from four change for a group of semi-arid areas. It can be tropical forests (boxes)) obtained by combining seen that an increase in surface albedo does satellite observations of tropical precipitation seem to decrease rainfall. Use of a global model and vegetation density with model simulations of emphasises that all parts of the climate system atmospheric transport to evaluate the effect of are interlinked. Although this particular model forests on tropical rainfall.36 This model analysis includes many simplifications, the results are established that in more than 60% of the tropical illustrative of the types of surface-induced cli­ land surface, air that has passed over dense and matic effects that are captured by models. areally extensive vegetation in the preceding few At present, around 30% of the land surface of days produces at least twice as much rain as air the Earth is forested and about 10% is cultivated. that has not crossed forests. However, the amount of forest, particularly in the Water occurs in different isotopologue forms tropics, is rapidly being reduced while reforesta­ (see CSI Box 1.4) that can be traced in both tion is prevalent in mid-latitudes. As a conse­ ­models and in the real world. Figure 1.17 quence, the surface characteristics of large areas ­illustrates the likelihood (bracketed by observa­ are being greatly modified. Modellers have tions in this case) that water in the Alamo River attempted to examine the climatic effects of (white circle) is derived from different locations in ­forest planting and clearance. The change in sur­ western North America. Warm colours indicate face character can be especially noticeable when locations at which the modelled precipitation

0002006766.INDD 36 1/10/2014 7:16:43 PM Why Model Climate? 37

CSI Box 1.4

Weirdness of Water: Isotopes and isotopologues

Water ’ s isotopologues (not isotopes because Isotopic depletion, now in proxy analyses, aids water is a compound, not an element) arise climate modelling through determining the from bonding among the various isotopes of global temperature fluctuations during ice hydrogen and oxygen. Because these isotopo- ages to measuring the biospheric recycling of logues can be separately identified, they pro- water in tropical forests. (D = 2H). vide climate modellers with tracking and process measurement capabilities.37 For exam- H D ple, the ‘heavy’ water isotope ( 1 H2 H16 O) binds more strongly to other water molecules and so 16 16 requires more kinetic energy than its common O H O H H cousin ( 1 H1 H16 O) to evaporate. As a conse- HDO: 0.1557 × 10–3 quence, water vapour above an open water 18 O surface, such as an ocean, will contain rela- H D 18 –3 tively fewer ‘heavy’ water molecules than the H2 O: 2.0047 × 10 ocean itself. As the moist air mass moves across 16 a continent, the ‘heavy’ water molecules will O D tend to precipitate out more readily, further D O: 5 10–8 [Heavy water] depleting the water vapour of ‘heavy’ water. 2 ×

0.00 6.80e-3

Figure 1.17 Combining model simulation and data analysis can provide greater insight than either technique applied separately. Output from the Isoscapes Modeling, Analysis and Prediction (IsoMAP) showing the relative likelihood (unit-less ranging from red (very likely) to deep blue (unlikely)) that water in the Alamo River (white circle ) is derived from different locations in western North America. Source: Bowen et al. ( 2012). Reproduced with permission of the American Geophysical Union.

isotope ratios are more similar to the isotopic the water in the Alamo River is derived from the character of the river water, than the blue col­ distant headwaters of the Colorado River drain­ ours. This analysis shows that a large fraction of age basin.

0002006766.INDD 37 1/10/2014 7:16:46 PM 38 The Climate Modelling Primer

1.3.8 Climate models train 1 week 1 month 4 months1 year

practitioners and educate 100 the general public Nuclear war: Baseline, 5000 MT Issue: when is a model ready to inform, e.g. Low-yield airbursts, 5000 MT 10000 MT full exchange nuclear winter? 3000 MT exchange 3000 MT counterforce Message: models illuminate policy choices and 1000 MT exchange also become embroiled in policy disputes. 10 300 MT SH 100 MT city attack It is frequently noted that weather plays an important role in warfare: there are many stories and anecdotes from the Second World War, the most famous being the weather forecast for the 1.0 D-Day landings. The impact of climate forecast

on warfare, or rather the possibility of war, is less Vertical optical depth well known but at least as important. This case of ­climate modelling informing military options 0.1 El Chicón relates to the issue of Mutually Assured simulation Destruction (MAD) that, it was believed, stopped the Soviet Union and the United States of America from embarking on nuclear conflict ­during the Cold War because, once begun, the 0.01 obliteration would be complete. The premise of 105 106 107 108 MAD naturally tempted military advisors to pro­ Time after detonation (sec) pose more limited war strategies until a climate model demonstrated unequivocally that even Figure 1.18 Climate model results can become this was doomed. iconic. This famous simulation is from a paper known as The basic theory of what has come to be ‘TTAPS’ and shows the time evolution of the disturbing termed ‘nuclear winter’ is that nuclear explosions effect of nuclear war on the atmospheric climate. Optical themselves, and to a much greater extent the depths (i.e. scattering plus absorption) calculated by ensuing fires, would cause massive injection of a one-dimensional radiative-convective model for soot (black aerosols) into the atmosphere.38 This wavelength of 550 nm are compared for a natural effect was well known, and had been observed disturbance (the El Chichón volcanic eruption) and a during earlier conflicts. However, climate models range of nuclear war scenarios. Optical depths of ≤0.1 are further improved understanding by permitting negligible; those ~1.0 are significant and when >2 there consideration of the effect of the height of the are serious consequences. Source: Turco et al. (1983). injection (up to 10–15 km above the surface) and then allowing calculation of the resulting impacts of these dark upper tropospheric aerosols. The Although there were earlier accounts of this models showed that the dark particles tended to process, the public and policy recognition mostly absorb solar radiation, warming the air and push­ arose when a famous climate modelling paper ing the smoke still higher. Once in the strato­ known now by the initials of its authors as TTAPS sphere, this black aerosol would persist for many (Richard P. Turco, Owen Toon, Thomas P. years, as the tropospheric removal process of Ackerman, James B. Pollack and Carl Sagan) was rain washout would not be available. The smoke published in the journal Science in 1983.39 The stays high and absorbs sunlight, so that the sur­ model employed was a one-dimensional face temperature drops and sets up a positive ­radiative–convective model extending (unusu­ feedback loop, encouraging these conditions to ally) to the mesopause. As such, it permitted the continue (Figure 1.18). prediction of the vertical characteristics of the

0002006766.INDD 38 1/10/2014 7:16:46 PM Why Model Climate? 39

Model Validation Box 1

Regionalising climate with Thornthwaite

Read : Elguindi , N. , Grundstein , A. ( 2012 ) An How to test the veracity of predic- integrated approach to assessing 21st century tions made by climate models has been a chal- climate change over the contiguous U.S. using lenge since the first models were used. the NARCCAP RCM output . Clim Change 117 , Side-by-side comparison of pictures (usually 809 – 827 . maps) of output variables was the earliest

1) CCSM a) CRCM-CCSM b) MM5-CCSM

2) HADCM3 c) HRM3-HADCM3 d) WRF-CCSM

3) CGCM3 e) RegCM-CGCM3 f) CRCM-CGCM3

4) GFDL g) GFDL hires h) ECP2-GFDL

Saturated Wet Moist Dry Semiarid Arid 12345612 3 45612 345612 345612 3 45612 3456 TorridHot WarmCoolColdFrigid

Figure source: Elguindi and Grundstein (2013). Reproduced with permission of Springer Science+Business Media.

(Continued)

0002006766.INDD 39 1/10/2014 7:16:54 PM 40 The Climate Modelling Primer

method employed. This simple ‘by eye’ com- the ‘current’ climate (1971–2000) simulated by parison was slightly improved by mapping not four AOGCMs (1–4 in left column) and the just a single parameter but a compound varia- RegCMs (a–h) in the centre and right columns. ble composed of more than one element of It is common to find that higher resolution in the predic ted climate. The earliest of these models gives rise to better climate simulation in evaluations exploited ‘climate regimes’ most areas of rapidly changing topography (e.g. commonly due to three researchers: Holdridge, mountains). Does this expected ‘improvement’ Köppen and Thornthwaite. These climate clas- show in the figure? With reference to the future sifications date back to the middle of the 20th climate predictions, Elguindi and Grundstein century and have been quite widely applied. find that ‘the U.S. will become drier, particularly In the 2012 paper by Elguindi and Grundstein, across the Midwest as the moisture boundary the Thornthwaite classification is employed shifts eastward, and in the Appalachian region’. to facilitate a comparison among predicted There are a number of ways that prediction of climates for the continental USA using six the ‘current’ climate may be evaluated. Thinking regional climate models (RegCMs) forced by about these, can you suggest what hazards coupled global climate models (GCMs). The might affect such attempts to ‘validate’ the pre- authors select the models they investigate from dictions shown in this figure? RegCMs participating in NARCCAP (North American Regional Climate Change Assessment Program). Each model was run for 30 years for Discussion preparation questions the current climate (1971–2000), and 30 years 1. How easy do you find it to compare model simulating the future climate (2041–2070) predictions ‘by eye’? using the A2 scenario from the Special Report 2. Research climate classifications and their on Emissions Scenarios of the Intergovernmen- use for evaluating climate model tal Panel on Climate Change.41 The figure shows predictions.

global climate following a large­scale nuclear war As with the science on nuclear fallout, the sci­ but not regional impacts. While this geographi­ ence about climate change is often misconstrued cal limitation was serious, and has since been as being controversial among experts whereas resolved by using general circulation models, the there is widespread consensus on the overarch­ main point – that such a conflict is equally unwin­ ing points of social concern. nable because the surface cooling is global and That the original TTAPS results altered military profound – carried into the mass media as well as practitioners’ views is undeniable. The interest of affecting military and policy strategies. the general public is also clear in the media trail There arose a muddled controversy when in from science paper, via policy documents to 1986 two different modellers, Starley Thompson more generally accessible policy journals (e.g. and Stephen Schneider, wrote a paper entitled Foreign Affairs), to the mass media. Many ‘Nuclear Winter Reappraised’. 40 Review of this subsequent climate simulations have shown that paper shows that the goal was not to discredit the original TTAPS theory is, by and large, cor­ the TTAPS conclusions but to slightly ameliorate rect although the details differ as a function of them – neatly dubbed a ‘nuclear autumn’ (rather such aspects as the regional input of aerosol and than winter). The media and policy fallout (pun the time of year investigated. The main point, intended) from the perceived dispute among cli­ frequently underlined by climate scientists, is mate models was a modest presage for the later that, in addition to the large number of deaths sceptic­driven discussion about global warming. and massive infrastructure disruption resulting

0002006766.INDD 40 1/10/2014 7:16:54 PM Why Model Climate? 41

from a nuclear exchange, there is a major climate ClO+→+ O Cl O2 disturbance that is most likely to persist for years if not decades. Carl Sagan underscored the cli­ These reactions can repeat and continue to matic disturbance of large-scale fires arising from destroy ozone. bombs or incendiary devices in relation to the There are a number of famous names involved Iran-Iraq war in the early 1990s. More recently, with the story of ozone depletion above both global and regional model simulations have poles and the investigation of chemistry that determined that oil-well fires are unlikely to cre­ makes this occur and the international treaty (the ate extensive enough smoke plumes to result in Montréal Protocol) that caused the discontinua­ their lofting above the tropopause, but extensive tion of industrial production of the chlorofluoro­ burning of towns and cities could produce such carbons (CFCs). Three well-known atmospheric plumes. To date, development of large enough scientists, Paul Crutzen, Mario Molina and smoke plumes that are self-heating and thus self- Sherwood (Sheri) Rowland, were awarded the lofting through the troposphere to the strato­ 1995 Nobel Prize for Chemistry for their work sphere, where they become stuck, remains a on this problem. Widespread alarm about how modelled result observationally unsupported.42 CFCs might affect the Earth’s atmosphere arose following the publication of a landmark obser­ vationally based paper by Farman, Gardiner 1.3.9 Climate models discipline and Shanklin in Nature in 1985.43 Other important ­climate scientists involved in disentangling this the policy dialogue chemistry and observing the Antarctic ­depletion Issue: global climate governance, e.g. the include James Lovelock and Susan Solomon. Montréal Protocol. Climate models have been used to show that Message: models simulate the chemical disrup- the Montréal Protocol has not only averted fur­ tion of a natural atmospheric balance and also ther damage to the ozone layer44 but has helped reveal how much a good international treaty has prevent significant regional climate change benefitted Earth. (Figure 1.19). Chemistry-climate models have been used to The successful Montréal Protocol is often offered analyse the effects of the Montréal Protocol, as a template for future limit-setting measures to i.e. the effectiveness of the ozone recovery that curb greenhouse gas increases. Even though it is this treaty delivered.45 Figure 1.19 shows the not at all clear that this is a good analogy for car­ paired comparison predictions from one such bon dioxide and other greenhouse gas emission ­climate model. In one simulation, the emission reductions, the way in which stratospheric ozone of ­ozone-depleting substances was prescribed depletion came to be understood and its subse­ according to the restrictions of the Montréal quent ­protection by means of international treaty Protocol as compared with a second model run in are a vindication of climate modelling. The which the ozone-depleting substances grew by amount of ozone in the stratosphere is the result 3% annually. This model predicts that the Montréal of a dynamic balance between photochemical Protocol will have saved up to 80% of the global production and loss. The chemistry is catalysed annual total ozone by the end of the 21st century. by chlorine ions (Cl) and/or bromine ions (Br). Further analysis of the simulations concludes that The shorthand form of this stratospheric chemi­ without this Protocol, by 2100, the mesosphere cal cycle is a chlorine atom changes an ozone and stratosphere cool down by 40°C and 20°C, molecule to oxygen: respectively, as a consequence of dramatic ozone depletion. Finally, without the Montréal Protocol,

Cl+→ O32 ClO + O ultraviolet (UV) radiation undergoes a five-fold increase in populated areas in the 21st century, and then this ClO reacts with nascent oxygen (O) resulting in much greater incidence of skin can­ to release the original chlorine atom: cers in many high-latitude regions.

0002006766.INDD 41 1/10/2014 7:16:55 PM 42 The Climate Modelling Primer

Biography Box 1.2

Meet the modeller: Stephen H. Schneider

Leadership: a pioneer in modelling, Schneider Life and times: Stephen H. Schneider (11 possessed the rare gift of being able to explain February 1945 – 19 July 2010) was a professor the complexities of climate science and, rarer at Stanford University from where he led still, the willingness to use it to benefit everyone. assessment initiatives for the IPCC, founded He was the greatest populariser of climate. and edited the journal Climatic Change , and wrote hundreds of books and papers on cli- Popular recognition: a frequent contributor to mate. Schneider was withering about those the media, Schneider coined the term ‘mediar- exploiting uncertainty in order to undermine ology’ to describe the challenges of success- public belief in all types of scientific consensus. fully communicating science to the public. He In one of his last books, Science as a Contact spoke fast, unerringly and with unbounded Sport , he describes developing a positive, prac- enthusiasm – always! tical policy that will bring climate change back under our control, help the economy with a Climate modelling connectivity: Schneider was new generation of green energy jobs and pro- advisor, mentor, co-worker and friend to a very ductivity, and reduce the dependence on fossil large number of climate modellers, from Jim fuels and, ultimately, ensure a future for our- Hansen in the 1970s to Bob Dickinson in the selves and our children. 1980s, and most recently with his wife, Terry Root, on the impacts of human-caused climate Read more change on the distribution and abundance of Mastrandrea, M.D. , Schneider , S.H. (2010 ) Preparing for many species. Climate Change. Cambridge, MA: MIT Press. Schneider , S.H. ( 2009 ) Science as a Contact Sport: Inside the Battle to Save the Earth ’ s Climate . Washington, DC : National Geographic Society . http://www.climatemodelling primer.net/l/k110.htm http://www.climatemodelling primer.net/l/k111.htm

Watch Schneider argues for immedi- ate action on climate: http://www.climatemodelling primer.net/l/k112.htm http://www.climatemodelling primer.net/l/k113.htm

IPCC Synthesis Report Scoping Meeting held in Liège, Belgium, August 2010, dedicated to Steve Schneider’ s memory. Photo source: Jean-Pascal van Ypersele.

0002006766.INDD 42 1/10/2014 7:16:56 PM Why Model Climate? 43

350 extreme, example of ­climate models informing dis­ cussion. However, the larger majority of modellers 300 accept (albeit in many cases very reluctantly) that 250 examining the costs and consequences of trying to engineer the planet to reduce the impacts of global Total ozone (D.U.) 200 warming may become necessary. 150 Using a very simple climate model, it is straight­ forward to pinpoint the aspects of climate that 100 must be modified to alter the most likely global warming trajectory. The very simple climate 50 model introduced in Tech Box 1.4 shows that Saved ozone (%) 0 globally averaged mean surface temperature 1960 1980 2000 2020 2040 2060 2080 2100 depends on only two factors susceptible to Year speedy (near real-time) modification: the albedo Figure 1.19 Climate model simulations can and the greenhouse effect. Thus, there are only contribute to improved global governance by illustrating two ways to try to engineer away the greenhouse the effect of changed laws or proposed legislation warming: either reduce the absorbed solar radia­ changes. Here, the positive impact of the Montréal tion or remove some of the polluting greenhouse Protocol is shown in the time evolution of the global, gases. These two techniques are frequently annual mean total ozone with (black) and without abbreviated as solar radiation management (red) the Protocol. Total ozone saved by the Montréal (SRM) and carbon dioxide removal (CDR). The Protocol limitations (%) is represented by the blue line. radiation terms upon which the techniques have Source: After Egorova et al. (2012). to work are shown in Figure 1.20. As the width of the arrows indicates magnitude, it can be seen that operating on the solar radiation near the top 1.3.10 Climate models encourage of the atmosphere or the heat absorbed by the -2 sensible thinking and ground from the atmosphere (both ~324 W m ) is similarly challenging, at least in energy terms. informed discussion Carbon dioxide removal techniques encompass Issue: Geoengineering as an economic problem. activities already under way such as protecting Message: models allow public discussion of the vibrant forests as effective carbon sinks, moving to possibility and timing of intervention to reduce non-fossil fuel energy sources, enriching agricul­ global warming. tural land so that more carbon is retained in soils,

and by scrubbing CO2 out of the air. CDR also Today’s society discusses a host of ideas around involves much less wholesome ­proposals such as

the issue of global warming, ranging from scep­ increasing oceanic uptake of CO2, for example by tics’ disruption of informed discussion46 through fertilisation of the oceans with naturally scarce to setting carbon taxes at a level that promises nutrients, or by forcing a more energetic oceanic

reduction in CO2 emissions and the challenge upwelling. SRM also ranges from apparently of prioritising adaptation investments to try to benign to much more intrusive techniques: col­ defend future generations while not upsetting ouring roofs white or encouraging crops with the current generation’s lifestyle. higher albedos through to inserting large amounts The idea of large-scale, planetary modification of sulfate aerosols into the stratosphere to scatter to relieve, or even resolve, a pollution problem is sunlight back to space or even erecting solar anathema to some people. Indeed, there is a group shields beyond the atmosphere. of climate modellers who still refuse to work on Over the past decade there has been a series the implications of geoengineering because they of assessments of both of these technique types. argue that this research has the potential to reduce In all cases, climate models have been used to society’s willingness to act to reduce greenhouse evaluate the degree of change possible and, gas emissions. This refusal is a stark, perhaps even importantly, the likely consequences of the

0002006766.INDD 43 1/10/2014 7:16:57 PM 44 The Climate Modelling Primer

Biography Box 1.3

Meet the modeller: Susan Solomon

Leadership: Solomon ’ s climate leadership was 1980s and 1990s she led expeditions to most clearly demonstrated in 2007 as co-chair Antarctica and her popular book The Coldest (with Dr Qin Dahe) of Working Group 1 of the March was in the 2001 Books of the Year lists of UN ’ s Intergovernmental Panel on Climate The New York Times, The Economist and The Change during its production of the report Independent . In Antarctica, her team overcame that concluded ‘unequivocally’ that the world the challenges of scientific study during the is warming. polar winter, confirming the ozone depletion’ s cause: the chemicals known as chlorofluorocar- Popular recognition: In Time Magazine ’ s 2008 bons (CFCs). Solomon developed a new method Top 100 list of the world’ s most influential peo- for evaluating the ozone depletion potentials ple. Susan has been honoured by having two used as a scale for regulating compounds that geological features in Antarctica – the Solomon damage the ozone layer. These conclusions Glacier (78º23’ S, 162º30’ E) and Solomon Saddle helped lead to a global ban on CFCs. (78º23’ S, 162º39’ E) – named in recognition of her achievements as a scientist. Read more Solomon S. ( 2002 ) The Coldest March: Scott’ s Fatal Antarctic Climate modelling connectivity: Solomon shares Expedition . New Haven, CT : Yale University Press . www. a place in the National Oceanic and Atmospheric coldestmarch.com Administration ’ s Top 10 History Makers with Time Magazine ’ s ‘most influential people’: Joseph Smagorinksy. http://www.climatemodellingprimer. net/l/k114.htm Life and times: Solomon is internationally rec- ognised as a leader in atmospheric science. She Watch won acclaim for her perceptive explanation of 2010 Darsh T. Wasan Lecture: http://www. the cause of the Antarctic ozone hole. In the climatemodellingprimer.net/l/k115.htm

Susan Solomon (left) and colleague at McMurdo Station. Photo source: Alexandra Weaver.

0002006766.INDD 44 1/10/2014 7:16:59 PM Why Model Climate? 45

Reflected from atmosphere and surface to space Emitted to space –2 107 W m 235 W m–2

Incoming solar irradiance 342 W m–2

Emitted from Transmitted from atmosphere the surface to space to space 40 W m–2 195 W m–2

Absorbed by Greenhouse atmosphere gases Absorbed by 350 W m–2 atmosphere 67 W m–2

Reflected by Transferred surface from surface 30 W m–2 to atmosphere by convection 102 W m–2 Emitted by Absorbed at surface surface 390 W m–2 Emitted from atmosphere 168 W m–2 and absorbed by surface 324 W m–2

Figure 1.20 Globally averaged energy budget of the Earth: solar in yellow; thermal (heat) in red and moisture (evaporation) in green. Climate models are frequently evaluated in terms of their ability to represent these fluxes. Climatic disturbances that alter components of the energy budget are also simulated using climate models; for example, the impact of geoengineering climate control proposals such as solar radiation management (SRM) and carbon dioxide removal (CDR) techniques. Source: Shepherd et al. (2009) and Trenberth et al. (2009). Reproduced with permission of the American Meteorological Society.

­particular proposed proactive engineering.47 techniques. Evaluation includes how much energy Generally, CDR techniques have more champions reduction is delivered, how much each W m-2 because, if successful, they result in removal of costs, the likely side-effects and the risk associ­

the problem – too much CO2 in the atmosphere. ated with these. Selecting the ‘best’ tool depends However, their outcomes are more difficult­ to on the confidence placed in such estimates. anticipate and, at present, are also believed likely There is one other aspect of geoengineering as to take longer to undertake and to achieve reduc­ probed and revealed by climate models: the pos­ tion of greenhouse gas loading. SRM techniques sibility that such techniques might need to be are usually quicker and easier to undertake but applied very quickly if a so-called ‘tipping point’ carry with them two disadvantages: they do not seems very close (see Section 1.3.4). This is illus­ solve the problem and by further (albeit differ­ trated in Figure 1.21 showing the planetary bound­ ently) altering the climate, they may themselves aries for the Earth.48 The 2009 planetary boundaries cause additional climate disturbances. Table 1.7 concept is based on the notion that transgressing is one example of the application of climate mod­ one or more planetary boundaries is deleterious els to the evaluation of aspects of possible SRM and may be catastrophic for life on Earth because

0002006766.INDD 45 1/10/2014 7:17:03 PM Table 1.7 Comparison of conventional mitigation costs, risks and extent of control offered versus a variety of SRM proposals. Control methods are ranked (roughly) in order of preference taking into account the risks and costs as well as the ability of the method to deliver full removal of the impact of anthropogenic warming Global warming Annual cost Max. control Possible side- ($billions forcing mechanism effects and risk per W m-2) (W m-2) Notes and references

Conventional Reduction in crop 200 –2 to –5 Stabilise CO2 at 450– mitigation yields (low) 550 ppm. Stern (2007) estimates 1% of global GDP per year Space-based Control failure 5 Any Launch costs of $5000 kg-1 reflectors (high) assumed, and replacing Regional climate reflectors every 30 years change (medium) (launch mass of 100,000 Reduction in crop tons) (Keith 2000) yields (low)

Stratospheric Control failure 0.2 Any Injection of 1 Tg H2S pa aerosols (high) by aircraft (Robock et al. Regional climate 2009, Lenton change (medium) and Vaughan 2009) Changes in 1.5 to 5 TgS yr-1 to

stratospheric offset 2× CO2 chemistry (medium) Desert surface Regional climate 1000 –3 Maintenance and albedo change (high) ecological issues likely Ecosystem impacts render this impracticable (high) (Gaskill 2004) Cloud albedo Control failure 0.2 –4 Operating costs of (high) 300–400 autonomous Regional climate vessels pa seeding clouds change (high) with seawater droplets dispersed from ocean Grassland and Regional climate Not known –1 Incentives for growing crop albedo change (medium) high-albedo varieties Reduction in crop and cultural effects not yields (low) known (Lenton and Vaughan 2009) Human Regional climate 2000 –0.2 Painting urban surfaces settlement change (low) white every decade albedo (Lenton and Vaughan 2009)

Source: Shepherd et al. (2009). Reproduced with permission of The Royal Society. Notes: 1 Costs in $109 (billions) given per year and per unit of radiative forcing, i.e. $109 yr-1 per W m-2. 2 ‘Control failure’ relates to the failure of the geoengineering control. As the aim of SRM techniques is to reduce absorbed solar radiation, failure could lead to a rapid warming much more difficult to adapt to than the climate change in the absence of geoengineering. Control methods that produce the largest negative radiative forcing and which rely on advanced technology carry the largest risks of failure.

0002006766.INDD 46 1/10/2014 7:17:03 PM Why Model Climate? 47

Nitr oge c n ycle ospheric P trat (b ho S epletion iog s ne d eo c ph ozo flow ch y o b em cl ru ou i e s nd ca n ar l tio y) ca ifi id c a n a fr e e c s G h l O w o b a a t e l r

u

s

e e

g

n

a

h

c

e

t

a

C

m h

i

l

a

n C

g

e

i

n

l

a

n

)

d

d

e u

n

s

fi

i o

e

t i

t

n

u

a

l

l

u

o

q

p t

l

e B

y

a

i

o t c

i

d o

n i m v

( e

e

r

h

s

i

C t

y

)

l d

o

e

s

fi s

i

t

n

a

u

q ( t

n o e t

y

g

n

i

d

a

o

l

l

o s e o a r

c

i

r

e

h p s o m t A

Figure 1.21 Sectoral view of the Earth’s climate in terms of (inner green) humanity’s ‘safe operating space’ and the red wedges indicating already-crossed boundaries of three critical systems. Of the nine planetary sectors, the boundaries already transgressed (red) are of climate change, biodiversity loss and human interference with the nitrogen cycle. Atmospheric aerosol loading and chemical pollution boundaries are not yet quantified while ocean acidification, stratospheric ozone depletion, global freshwater use and changes in land use remain ‘safe’, as does the biogeochemical flow boundary associated with phosphorus cycling. Source: After Rockström et al. (2009a). Reproduced with permission of Nature Publishing Group.

of the risk that crossing thresholds will trigger ­non-linear, abrupt environmental change within 1.4 Climate models: continental- to planetary-scale systems. The nine sound components in planetary boundaries together define a ‘safe space’ for climate on Earth: (i) climate change; (ii) ocean careful combination acidification; (iii) stratospheric ozone depletion; (iv) biogeochemical nitrogen and phosphorus Issue: climate modelling has many aspects cycles; (v) global freshwater use; (vi) land-use including clever coding, fast/large platforms, change; (vii) biodiversity loss; (viii) chemical pollu­ modellers’ funding, skill and motivation and user tion; and (ix) atmospheric aerosol loading. specifications.

0002006766.INDD 47 1/10/2014 7:17:07 PM 48 The Climate Modelling Primer

Message: community understanding of the stren­ seem like bizarre climate outcomes, caution is gths and weaknesses of climate models is increas- necessary when analysing results of even appar­ ing and demands continued encouragement. ently simple models. We take our kitchen anal­ ogy just a little further to examine another The overview in Section 1.3 of the many ways in important feature of climate systems and hence which climate models are used can help to refine of climate models: the concept of equilibration the ‘recipe’ view of climate modelling. Figure 1.2 time. A pot of hot water removed from a stove showed a long list of ingredients incorporated into will re-equilibrate with the room environment in a climate models ranging from the atmosphere characteristic time that depends upon the differ­ through to ice sheets and nitrogen cycling. On the ence in temperature of the pot contents and the other hand, in this chapter we have been intro­ room, as well as the size and shape of the pot duced to very simple climate models containing as and the contents of the pot. Such characteristic few as only three components: the simple radia­ times are a vital component of climate and thus tive budget model (see Tech Box 1.4) uses albedo, of climate models. solar radiation and a greenhouse increment to It is common to express equilibration times in compute surface temperature and Lorenz’s model terms of the time it would take a system or sub­ (see Tech Box 1.2) computes just the three direc­ system to reduce an imposed disturbance to a tional components of wind. This apparent conflict fraction 1/e ≈0.37 of the disturbed value, termed between complexity and use of climate models the e-folding time. A smaller temperature differ­ pervades today’s literature and even the media ence, a smaller pot or a larger surface-to-volume and the arts. There is no rule that gives the amount ratio of the container will result in relatively of detail required in any particular model. shorter e-folding times. Large e-folding times characterise subsystems that respond only very 1.4.1 Ingredients and method slowly. Table 1.8 lists equilibration times for a range of subsystems of the climate system. The The real challenge of turning good cooking into longest times are those for the deep ocean, the great cuisine is how ingredients are combined – glaciers and ice sheets (hundreds to thousands in fact, the chemistry of cookery. In this chapter, of years), while the remaining elements of the we have glimpsed the extraordinarily complex ­climate system have equilibration times ranging behaviour that can arise from apparently straight­ from days to years. forward combination of model ingredients. Since Often the importance of feedback effects as few as three ingredients can produce what depends upon the timescale of behaviour of the subsystems they affect and so the concept of timescale of response is crucially important to all aspects of climate modelling. This timescale is variously referred to as the equilibration time, the response time, the relaxation time or the adjustment time. It is a measure of the time the subsystem takes to re-equilibrate following a small perturbation to it. A short equilibration timescale indicates that the subsystem responds very quickly to perturbations and can therefore be viewed as being quasi-instantaneously ­equi­librated with an adjacent subsystem with a much longer equilibration time. The very long equilibration times of the deep ocean and ice sheets pose a particularly difficult problem for ­climate modellers. The methods by which the short response time of the atmospheric features

0002006766.INDD 48 1/10/2014 7:17:08 PM Why Model Climate? 49

Wiring the World Box 1

Real wires and the real world

Climate models have to embrace very many before a climate model is con- components and relationships. These have to structed and checked repeatedly during its be clearly identified and the links defined evolution and use. Connections among model

(a)

(b) Fleet management Parks and open Heat Sea level rise Storm surge Rainfall Tree spaces Events Property management management management

Waste education Disease Asset Business management Pest control engagement Human health Stormwater Communications planning Flooding Customer service Population Infrastructure Insurance Operations Development planning

Public liability Heat island Transport Pollution Water supply Staff training impacts Fire Property damage Parks and trees Capital works Emergency management Ground water Procurement Economic access Political Tourism impacts Planning controls Compliance Infrastructure Strategic Social plan planning Sydney 2030

Figure source: (a) CERN. © 2013 CERN. (b) Sydney Coastal Councils Group. Reproduced with permission of the Sydney Coastal Councils Group. (Continued)

0002006766.INDD 49 1/10/2014 7:17:11 PM 50 The Climate Modelling Primer

components’ organisation can be described in a wiring diagram and a feedback diagram (the a variety of different ways: a common form is latter will also feature in boxes). the so-called ‘wiring diagram’. Any electrical The local government fact sheet from which system has a wiring diagram; the more the picture (b) is taken claims that it represents complicated the electrical connections, the key climate change drivers, the impacts of cli- more frightening the wiring, and the diagram. mate change, likely management responses We were pretty amazed when we witnessed and the relationships between these.50 This is a the mass of electrical wires inside the CERN49 wiring diagram but it may not quite satisfy the ATLAS experiment in 2007, just before it was needs of climate modelling. Most of our sealed for experimentation. ATLAS is 45 metres ‘Wiring the World’ diagrams are those that long, over 25 metres high and weighs about underpin construction of models or are used to 7000 tonnes (the same as the Eiffel Tower or a analyse their results. hundred empty 747 aeroplanes). Its wiring (pictured in (a)) is pretty hairy! Wiring the World: two minute thought Descriptions, including models, of compli- prompts cated systems need (1) very many connections • If you had to check the wiring of one of CERN ’ s (wires) and (2) very careful checks of connec- experiments, how would you go about it? tions (wiring). In ‘Wiring the World’ boxes in • Suppose the second picture (b) was to be each chapter, we illustrate and examine this made the basis of a climate model. Are there type of ‘components and links’ diagram. It is aspects you would change, items to add or important to recognise the difference between remove and why?

Table 1.8 Equilibration times for several subsystems of the climate system Climatic domain Seconds Equivalent

Atmosphere Free 106 10 days Boundary layer 105 24 hours

Ocean Mixed layer 106 –107 Months–years Deep 1010 300 years Sea-ice 106 –1010 Days–100s of years

Continents Snow and surface ice layer 105 24 hours Lakes and rivers 106 10 days Soil/vegetation 106 –1010 Days–100s of years Mountain glaciers 1010 300 years Ice sheets 1011 3000 years Earth ’ s mantle 1015 30 million years

0002006766.INDD 50 1/10/2014 7:17:12 PM (a) ESCAPE (Rotmans et al., 1995)

ECONOMICS

POLITICS Root & Schneider (1995)

ECOLOGY INCREASING COMPLEXITY CLIMATE DEMOGRAPHY ASSESSMENT Nordhaus (1991) CETA (Peck & Teisberg, 1992)

Bellouin et al. (2011) Fully coupled OAGCM HadGEM2-ES DYNAMICS AND RESOLUTION with high spatial OAGCM at R15 resolution (Washington & Meehl, 1989)

CLIMATE 2-D SD (Potter et al., 1979) MODELLING

CHEMISTRY EBM SURFACE AND OCEANS (Budyko, 1969) RADIATION

1D RC (Manabe & Wetherald, 1967)

(b)

Integrated assessment models

Energy Economy

Health Agriculture and forestry Climate models Atmospheric Human processes settlement and Terrestrial infrastructure carbon Water cycle Ecosystems & Vulnerability Impacts, Adaptation Sea level rise Oceans

Cryosphere

Figure 1.22 Climate model type schematics showing ‘categories’ of climate models and underlining the recognition that assessment of consequences dominates many modelling activities. (a) Balancing of climate assessment models on the pinnacle of climate models with examples. (b) Categorisation by model application following the IPCC Working Group structure of science (models), impacts (vulnerability) and responses (assessment). Source: (b) Moss et al. (2010). Reproduced with permission of Nature Publishing Group.

0002006766.INDD 51 1/10/2014 7:17:13 PM 52 The Climate Modelling Primer

can be linked to the much slower response time and how they enlarge or dampen disturbance; of the ocean and ice sheets are discussed later in and in terms of their intended application (see this book. Understanding feedbacks can only Figure 1.22a). come through careful examination of the action of likely perturbations and the relative equilibra­ tion times of various parts of the climate system. 1.4.2 Climate model prediction: The very wide range of timescales in the climate system is reflected in the wide range of climate getting the right result for model types described in this book. the correct reason Our last use of the cooking and climate model­ Throughout the Primer we will be examining ling analogy is to illustrate the mixed, changing ­predictions made by climate models and review­ and sometimes confusing terminology. Recipe ing their skill. Climate models come in an almost instructions frequently differ according to culture bewilderingly wide variety of types: written des­ and over time: for example, ‘browning’, crisping’, criptive (hot dry summer and warm wet winter); ‘toasting’ and ‘flashing under the grill’. In the dis­ in laboratories (such as the rotating dishpan cussion of Table 1.8, the rich terminology around ­analogue model); and of course many types equilibration was noted. This unsettled naming is of numerical climate models. Demonstration of also a feature of the last decade’s climate models. some aspects of climate modelling is most use­ For example, Figure 1.22 illustrates two ways of fully accomplished by visualisation developed representing the breadth of climate models that out of analogue representations as well as the operate today: (a) balancing of climate assessment simulation-based videos which abound on the models on the pinnacle of climate models and (b) web. Making a real model work in your office, categorisation by model application. There is an laboratory or home is a persuasive means of old-fashioned convention that discusses ‘physical’ demonstrating skill (Table 1.9). or ‘physically based’ models. This means climate The omission and inclusion of parameters, models that are mostly based on physical laws ­processes and timescales in climate models are (radiation, fluid dynamics, etc.), even though, as vexed. The ‘gotcha’ rule applies strongly to Figure 1.22a illustrates, these models have for ­climate modelling: the overlooked can be critical. some time included aspects of chemistry and biol­ We have highlighted the way in which climate ogy. The ‘assessment’ area of climate modelling is models incorporate many of the very strange and almost as confusing, with integrated assessment extreme properties of water in this chapter – the models (IAMs), impacts models and earth system harder you look at water, the weirder it becomes. models of intermediate complexity (EMICs). Sometimes, the latter are solely physical models, which incorporate climate elements with long response times, say parts of the cryosphere, not 1.4.3 Climate models pushing the yet included in the parameterisations of global envelope ­climate models (GCMs) (discussed in Chapters 3 As this chapter has illustrated, climate models, and 4). Our conclusion is that, as with recipes, while modellers and modelling take many forms and it would be less confusing for novices if everyone deliver to wide-ranging audiences. Since it is used the same terms, this ‘one size fits all’ does not very hard to try to define what is, and therefore exist now and, frankly, probably never should. is not, a ‘climate model’, we demonstrate the The climate system, and therefore any model breadth of the discipline today by posing two designed to represent it, can be described in a possibly outrageous questions of the family of wide variety of ways: in terms of subsystems (see climate models. Figure 1.22b) and their directions and types of interactions; in terms of characteristics such as • Can climate models tell us about extrater­ equilibration times (such as Table 1.8); to capture restrial life or them about us? the feedbacks (i.e. the processes and interactions)­ • Can climate models inform global governance?

0002006766.INDD 52 1/10/2014 7:17:13 PM Why Model Climate? 53

Table 1.9 Summary of models used in this chapter as examples of the top 10 reasons for undertaking climate modelling Climate model type: laboratory Reason for (L) or numerical (N) Section modelling Example in this chapter Rainband spectroscope (L) 1.3.1 Test theory Global warming Rotating dishpan (L) 1.3.2 Illuminate features Goldilocks zone and and uncertainties planetary waves Double pendulum (L) 1.3.3 Show complex can Butterfly effect be simple and vice versa Microphone circuit (L) 1.3.4 Raise questions, Feedbacks suggest analogies 0D Energy Balance Model (N) 1.3.5 Agree (disagree) Early Earth Snowball with data

Monte Carlo Simulation (N) 1.3.6 Explain Milankovitch and CO2 lead/lag temperatures Global Circulation Model (N) 1.3.7 Bracket ranges of Deforestation/ outcomes desertification Single Column Radiative 1.3.8 Train professionals Nuclear winter Convective Model (N) and public Chemistry-Climate Model (N) 1.3.9 Discipline policy Montréal Protocol Economic models (N) 1.3.10 Encourage sensible Geoengineering thinking Multiple models in assessment (N) 1.4.1 Prediction Risk assessment

Alien climate modelling ­conjectured ‘alien’ process is not very far-fetched. In May 2012, NASA scientists determined that Right now, we (Earthlings) are already examining our galaxy will collide with the Andromeda planets that orbit distant stars to evaluate their Nebula51 in about 4 billion years, about as far atmospheric make-up and hence the likelihood into the future as the period of a water-based cli­ that they host life. For example, the June 2012 mate on Earth but less than the age of our solar transit of Venus offered an important opportunity system (around 5 billion years). This forthcoming for the orbiting Hubble telescope to be trained collision is not a cause for alarm, or even much on an object in our solar system specifically to interest, on Earth, but imagine that in Andromeda, aid in correcting and confirming our ability to a galaxy far, far away….. an alien life form has evaluate the atmospheric constituents of very been asked to apply climate modelling to assess distant exoplanets. A little closer to home, as the likelihood of sentient beings inhabiting a Curiosity (the NASA explorer) tours the Martian smallish planetary system discovered by the surface, experiments with climate models are astronomers on her world to be orbiting a rather being used to interpret its findings in terms of ordinary star – the one we call the Sun. This the likelihood of widespread life there.52

0002006766.INDD 53 1/10/2014 7:17:13 PM 54 The Climate Modelling Primer

Our imaginary alien researcher knows how to these worlds have massive day/night climate detect the possibility of intelligent life because contrasts she studied this in high school. The evaluation on • the eccentricity of the orbit determines how Andromeda happens in two parts: A – the astro­ large the seasonal changes are – the more physics and B – the beings. In (A) a well-tuned nearly circular, the smaller the seasonal con­ software package automatically tests the physics trast; extremely eccentric orbits may give rise of the stellar system and the chemistry of the to less habitable climates atmospheres of the planets and larger satellites • the axial obliquity (or tilt of the planetary (moons) in that stellar system. If this delivers likely axis to the plane in which planets and star living planets, our alien researcher begins (B) exist) also contributes to the seasonality. in which she seeks for evidence of persistence of Combinations of large obliquity and large climatically stable conditions that can be con­ eccentricity can lead to extreme and persis­ strued to be imposed (controlled by sentient tent ‘ice’ ages beings). This search is in accord with the • the existence of a magnetic field around the Andromedan view that the clearest demonstra­ planet protects its atmosphere and surface tion of intelligent life’s emergence is integration from destructive stellar emanations and is with planetary systems. therefore believed to be a useful contributor Any search for intelligent life assesses the star to overall habitability. type, which determines its radiance characteris­ tics and development. It is believed in Andromeda The astrophysical and chemical assessments that single stars (i.e. not binaries or trinary star above are well known to us but how aliens might systems) are more likely to host planets and evaluate our activities can only be conjecture. moons in climatically stable zones (see Figure 1.5). The radiative signature of the biosphere may Planets are known to form from the debris have two components: non-sentient and sen­ remaining in the proto-stellar disc, following tient. Chemical disequilibrium, a sign of living ­stellar condensation. Close to most stars, sili­ systems,53 can be detected in absorption/emis­ cates and minerals accrete, building planets that sion lines in spectra (see Figure 1.4b). Lovelock54 are smaller and denser (rocky planets) while, at deduced as long ago as 1965 that a planet bear­ larger distances from the star, accretion is of ice ing life can be easily distinguished from a sterile cores that then capture surrounding gas, creat­ one and he concluded: ing gas giants. All of these may have moons (sat­ ellites) that are large enough to retain an ’… atmospheric analysis, is simple and prac­ atmosphere over astronomical timescales, i.e. tical as well as important in the general many billions of years. problem of detection of life. A detailed and Assessments are undertaken of planets and accurate knowledge of the composition of the moons large enough to have retained an atmos­ planetary atmosphere can directly indicate the phere for most (preferably all) of the lifetime of presence of life in terms of chemical disequi­ their star. Once a planet (or moon) is discovered librium.’ (p569) to have an atmosphere, its physical properties are checked, especially rotation rate, eccentricity Emissions directly attributable to sentient life of its orbit around the star, axial obliquity, and can also be sought. Radio wavelength radiation the presence of a magnetic field (see Figure 1.13). (TV and radio programmes) have first to be These are important because: ­separated from naturally occurring radio waves and then, perhaps, such a data stream might • the rotation rate determines the length of reveal the existence, even behaviour, of sentient days/nights – no preferred value is sought beings. Andromedans incorporate social science but phase-locked planets (where the rotation in their climate evaluation and, indeed, some rate is equal to the stellar orbital period, i.e. aspects of sociology and economics are being day length equals year) are avoided because incorporated into our climate models.

0002006766.INDD 54 1/10/2014 7:17:13 PM Biography Box 1.4

Meet the modeller: Carl Sagan

Leadership: Sagan, while popularly recognised stability. He was a renowned communicator for astronomy, led climate-modelling science about science and an advocate for both the in a number of ways. Perhaps the most impor- environment and for humanity. Sagan’ s inter- tant was his role in the famous ‘Nuclear Winter’ est in the evolution of life began during his assessment. The simulated climatic effect of a undergraduate studies when he worked with nuclear war is that large amounts of smoke the geneticist H.J. Muller and wrote a thesis on and soot ejected into the upper atmosphere the origins of life with physical chemist H.C. reduce sunlight for many months or even Urey. Sagan was linked to many searches for years, plunging the whole world into very low extraterrestrial life, including the famous temperatures and, it was argued, making any ‘Arecibo message’ that he co-wrote with Frank such nuclear detonation a global catastrophe. Drake. This radio message was transmitted to space from the Arecibo radio telescope on 16 Popular recognition: The very popular PBS November 1974, aimed at informing potential series Cosmos ; and his novel Contact is the extraterrestrials about Earth. basis for the 1997 film of the same name star- ring Jodi Foster that ends with the dedication Read more ‘For Carl’. TTAPS paper: Turco, R., Toon, O., Ackerman, T., Pollack, J., Sagan, C. (1983) Nuclear winter: global consequences of Climate modelling connectivity: Carl Sagan multiple nuclear explosions. Science 222, 4630, 1283–1292. and Lyn Margulis, the co-inventor (with James www.jstor.org/stable/1691639 Lovelock) of the Gaia Hypothesis of planetary climatic stability, were co-workers and once Watch married. Carl Sagan, Stephen Hawking and Arthur C. Clarke. God, the universe and everything else (1988). http://www.climatemodel- Life and times: Carl E. Sagan (9 November 1934 lingprimer.net/l/k116.htm – 20 December 1996), the American astrono- Carl Sagan ’ s last (1996) interview: http:// mer and science communicator, wrote seminal www.climatemodellingprimer.net/l/k117. papers on planetary habitability and climate htm LINK 1.17

Carl Sagan. Photo source: © Druyan-Sagan Ass, Inc.

0002006766.INDD 55 1/10/2014 7:17:16 PM 56 The Climate Modelling Primer

Climate Model Communication Box 1.2

Climate modelling shared with non-professionals

Bringing about changes in global governance • What level of climate, environ- demands widespread public support. Such sup- mental and modelling knowledge do you need port depends on how effectively issues of envi- to understand this magazine ’ s articles (high ronmental degradation, including climate school, university, postgraduate or profession- changes, are understood. In this activity, you ally trained)? Give reasons and examples to read articles with which you are unfamiliar. illustrate your answer. (Less than 200 words) Choose a well-known ‘professional’ journal or magazine (e.g. Economist, National Geo- 2. Characterisation of climate modelling graphic, New Yorker, Rolling Stone, Foreign in this magazine Affairs) that has an audience very different Find at least one (preferably two) articles from from your knowledge base – something you the past 5 years in this magazine that describe would not normally read but professional peo- climate models or predictions made using ple access. There are two parts to this task: them. Give the full references of these. familiarising yourself with the ‘aliens’ who Write a review of one article (500 words read this magazine and evaluation of their maximum) that explains: characterisation of climate modelling. • the aim of the article • how the author describes the climate models 1. Evaluating the magazine generally used From one recent issue, carefully answer the • the audience for the article following questions. • how the author makes use of layout features • Who reads this journal? Give reasons for such as diagrams and tables your answer using examples and quotes • your personal opinion (with reasons) on the from the publication. (Less than 200 words) worth of the article.

Global governance outcomes from public are confident enough about climate model climate modelling results to pursue discussions on additions to or Another aspect of climate modelling that warrants changes in international law. This, in turn, depends thought is the use of model results to encourage, on climate modellers informing policy – a chal­ or even mandate, global changes to governance lenge that the IPCC has faced for many years. structures and laws. In Section 1.3.9, the example In addition to the Earth system boundaries of the Montréal Protocol was raised as both a argument, the issue of geoengineering of the cli­ successful result of scientific understanding (and mate (see Section 1.3.10) has already prompted modelling) of the atmospheric changes and their calls for much improved global governance consequences, and as a possible template for before any experiments are undertaken on plan­ etary­scale climate engineering. global structures to limit CO2 emissions. There are sound arguments in favour of governance changes In the lead­up to the UN Conference on in the transgression of three of the boundaries 55 Sustainable Development in June 2012 (known identified in Figure 1.21 : the rate of biodiversity as Rio + 20), and the High­Level Summit on the loss, the rate of transformation of the nitrogen Millennium Development Goals in 2013, there cycle and climate change. has been a growing debate on how to draw up Hope of global governance improvement renewed and expanded global development depends on whether members of the general goals that bring together the twin objectives of

0002006766.INDD 56 1/10/2014 7:17:17 PM Why Model Climate? 57

Climate change

Fre ge NMENTAL C sh n IRO EILIN w ha NV G at c E er e u s s u st sp e d ju ac d e n n f a a o L e r f h a u s FOUNDA AL T m I IO C e O N a p S h Water h N n o T i Food i s t t r s y p o Income h s g

o o e l r n u y

t s Health a i

s n c r

d y e Education c

v

i l

e

d Gender

s

o

i

equality B

Resilience

Social

O

T

N c equity

e

Voice E

a

n

IN M n

o

i P

Energy a

t C Jobs

c

e O

l L i

L d

p U

E i f

e S

i

c

d I V

V E a

e

E D t i n

o

o A

IC n

z N

M

D

O O

S N

U O

C

S

E T

A E

INA L B

C

h

e

m

i

c

l

a

l o

p

s

g

o

n

o

i l

l

r

u

d

t

a

e

i

o

o

l

n

a

c

i

r

e

h

p

s

o m o t A

Figure 1.23 Diagrammatic combining of the environmental sectors (outside) of the ‘safe operating space’ of Rockström et al. (see Figure 1.21) with societal sectors (inside) likely to be involved in managing the Earth’s systems, i.e. provision of a human systems’ underpinning of the planetary boundaries. Source: Raworth (2012).

poverty eradication and environmental sustain­ an area, shaped like a doughnut, that represents ability. The case that members of the general an environmentally­ safe and socially just space in public are confident enough about climate which humanity can thrive (Figure 1.23). model results to pursue discussions on the This last section has strongly stretched the idea most appropriate way of dealing with the of what can flow from climate modelling: into the consequences­ of large-scale climate change is Andromeda Nebula and into human systems,­ presumed in a recent Oxfam Discussion Paper: especially those that inform inter­national law. ‘Living in the Donut’ by Kate Raworth.56 The Maybe this is going too far but models of all sorts establishment of biogeochemical and physical already inform international negotiations and we boundaries depends­ on climate models. In the saw (Section 1.3.8) that climate­ model simulations Rio + 20 paper, Oxfam extends this concept, have affected warfare. adding human factors to the planetary ones to It may be that rules of civilisation develop­ create a single framework. The social foundation ment57 may come to be included in climate or forms an inner boundary, below which are many ‘Earth system’ models in the future. Around the dimensions of human deprivation. The environ­ world, some early adopters among climate mod­ mental ceiling forms an outer boundary, beyond ellers are currently examining the relationships which are many dimensions of environmental between the number of agents and the complex­ degradation. Between the two boundaries lies ity of their interactions.

0002006766.INDD 57 1/10/2014 7:17:17 PM 58 The Climate Modelling Primer

Reflection on Learning 1.5 1.5 Climate modelling: Recognise the mechanisms about this book whereby persistent and widespread life affects climate Issue: climate models are being, and will con- tinue to be, used and abused. 58 As long ago as 1965, James Lovelock pointed Message: this Primer is designed to develop out that any planet bearing life must differ very better-informed climate model users. substantially from a sterile one in at least two ways: an ‘extreme departure from an inorganic In this first chapter, we have tried to identify why steady-state equilibrium of chemical potential’ people model, how these motivations relate to the and a planet-wide orderliness of structures and skill of climate modellers and to the challenge faced of events which are ‘utterly improbable on a basis by users of their model output. It is also important of thermodynamic equilibrium’. Thus, before any to mention here what this book is not about. This is human probe alighted on any foreign planet, not a book on global (or greenhouse) warming, Lovelock described a simple life detection although the core lessons can help one better scheme: ‘… a detailed and accurate knowledge understand the role of models in studies about of the composition of the planetary atmosphere climate change. In this last section, we explain our can directly indicate the presence of life in terms goals for readers and indicate our own views. of chemical disequilibrium’. This life detection system underlines how critical life is to the cli­ mate while the very need to detect distant non- 1.5.1 Climate modelling: read the Earth life is suggestive of how important climate is for life. Recent Earth system model develop­ label and exercise care ments are functionally dependent upon clear The most important reason for modelling is to representation of climate-life interactions and understand and then explain. This chapter interdependencies (Figure 1.24). ­considered the reasons for climate modelling,

HUMAN COUPLED SYSTEM ENVIRONMENT

INSTITUTIONS (Banking, judicial, education) BioGeochemistry POLICY DECISION MAKING (Soil fertility, OM levels, fluxes) (Incentives, conservation, land tenure) LAND USE BioDiversity CULTURE (Species, community, landscape, (Perceptions, values, ethics) genetic, functional)

ECONOMY UTILIZATION BioPhysics (Valuation, cost-profit, discounting) (Albedo, energy exchange, structure)

DEMOGRAPHICS (Number, gender, age)

Figure 1.24 This simplified ‘interacting boxes’ diagram shows some two-way interactions between the human (left) and environmental (right) subsystems. This representation of the relationships between environmental and human systems can be compared with the sectoral structure in Figure 1.23. The question for climate modellers is how much of this diagram must be incorporated into a climate model. Source: Nobre et al. (2010). Reproduced with permission of the American Meteorological Society.

0002006766.INDD 58 1/10/2014 7:17:18 PM Why Model Climate? 59

Table 1.10 The Climate Modelling Primer or A, B, C together with questions that models can and do tackle Components Climate model question A: Astronomy How astrophysics Orbit Is the Earth’s astrophysics compatible defines climate with its 4 billion year climatic stability? Atmosphere Can atmospheric species depletion/ increase be explained? Radiative budget How does greenhouse warming interact with astronomical variations in insolation? Water Is water essential for long-term climate stability? B: Biology and Boundaries How life affects climate Life and climate Is pervasive life always revealed by and where to place chemical disequilibrium? climate boundaries Climatic boundary Climate models include and exclude conditions aspects of ‘planet-wide’ systems. How are these boundaries selected? Climate modification Can climate model results make a and awareness of compelling case that people are passive climate control significant climate modifiers? C: Comprehension How climate models Test theories and bracket If models are used to test theories, lead to understanding ranges of outcomes how are models tested? Illuminate features and Do analogue models predict? reveal uncertainties Show that complex can If model results are very variable, how be simple and vice versa can predictions be understood? Raise questions, suggest Models both use data and demand new data requirements data – can the paradox of data ‘double-dipping’ be resolved? Encourage sensible Many global challenges are informed policies by modelling, climate and economic simulations having at least commensurate validity

including viewing planetary climate from far away teristics of climate and thus the features that with the intention of identifying the main (and most climate models endeavour to capture can hopefully most important) characteristics of cli­ be thought of as a primer – or perhaps an A, B, C mate and hence of climate models. The charac­ (Table 1.10). Any planet or moon with a ‘climate’

0002006766.INDD 59 1/10/2014 7:17:18 PM 60 The Climate Modelling Primer

will most likely satisfy some fundamental astro­ ters are concerned with different model types, physical conditions. This climate becomes inter­ their development and applications. Throughout, esting to modellers most often when it relates to we have taken climate models to be predictive living systems. These may either passively (with­ descriptions of regional- to global-scale phe­ out knowledge) or actively (like us) alter the natu­ nomena; hence, empirically based ‘models’ such ral climate. Our reason for constructing, operating as crop prediction equations and water resource and analysing climate models is finally to try to management codes have not been included. understand them and, through them, the It is necessary to introduce the concept of climate. energy balance, especially planetary radiation In our view, all the best reasons for modelling balance, before one-dimensional energy balance climate involve improving explanations and shar­ models (Chapter 3) can be understood. In ing these aspects of climatic characteristics with Chapter 4, models that intentionally consider wider communities. A great example appeared only a few of the important processes of the cli­ in the magazine Rolling Stone in August 2012.59 mate system are examined. These simpler mod­ By Bill McKibben of 350.org fame, this broad els are used to gain deeper understanding of the audience story is powerfully entitled ‘Global nature of feedbacks and forcings within the cli­ Warming’s Terrifying New Math: three simple mate system. These models, which have enjoyed numbers that add up to global catastrophe – and a significant renaissance in the last 15 years, are that make clear who the real enemy is’. now widely known as earth system models of As noted above, this Primer is not primarily intermediate complexity (EMICs). about the effects of anthropogenic greenhouse One way to think about the book is that it begins gas emissions. As scientists, we concur with with simpler models and works towards more the 2012 American Meteorological Society complex ones. This is true but the structure also ­statement.60 In another way, the Primer is very introduces elements of the climate system as they much about human climate change because to are needed. For example, clouds and the cryo­ understand our future climate we (policy makers, sphere (ice, snow and ice clouds) are important for voters, commuters, in fact everyone) must energy balance models (EBMs) and so time is ­understand a little about climate modelling. As invested in glaciers and their modelling in Chapter Kevin Trenberth61 points out, climate scientists 3. The way radiation interacts with the atmosphere are frequently asked about an event ‘Is it caused is explained in Chapter 4, together with an intro­ by climate change?’. But this is the wrong ques­ duction to the components of the oceanic circula­ tion and, being poorly posed, has no satisfactory tion necessary for climate simulation. Chapter 5 answer. Trenberth’s answer, with which we whole­ takes a more ‘how to build it’ approach, looking at heartedly agree, is ‘all weather events are the way in which climate computations are under­ affected by climate change because the environ­ taken in computers. By this point, the reader is, ment in which they occur is warmer and moister hopefully, well prepared to understand the way in than it used to be’ (p283). which energy transfers, ocean and atmosphere dynamics, biological processes and chemical 1.5.2 The Climate Modelling changes are included in coupled three-­dimensional models of the climate system. We also address Primer how these results can be integrated with assess­ In one sense, this book develops the background ments (both of model prediction skills and of likely material required for understanding of the most impacts of climate changes) in the development complex type of climate model, the fully coupled of social and economic policies. climate system model, by illustrating principles in Throughout the book, an effort will be made other, simpler, model types. Chapter 2 contains to underline the importance of simpler models in a history of climate modelling and provides an understanding the complex interactions between introduction to all the types of models to be dis­ various components of the climate system. cussed in subsequent chapters. The other chap­ Complicated three-dimensional models are only

0002006766.INDD 60 1/10/2014 7:17:18 PM Why Model Climate? 61

Climate Model Showcase Box 1

Checking 20 climate models

Read : Bender , F.A-M. , Rodhe , H. , Charlson , R.J. , is undertaken for the recent past: Ekman , A.M.L. , Loeb , N. ( 2006 ) 22 views of the from the mid-1980s to the early 2000s. The global albedo – comparison between 20 GCMs GCM simulations are derived from 20th-century and two satellites . Tellus 58A , 320 – 330 . re-creations using historical forcings that differ somewhat between GCMs, but the effects of A basic premise of all climate modelling is the differences are considered unimportant that, at the top of the atmosphere, there is an for the purposes of this comparison. approximate balance between incoming and The two satellite measurement campaigns

outgoing radiation, given by 0 ≈ C ∂T e /∂t = are the Earth Radiation Budget Experiment 2 2 4 p R S(1 − a ) − 4p R s T e . Here, the first term on (ERBE), giving data from February 1985 to May the right represents the fraction of the inci- 1989, and the Clouds and the Earth’ s Radiant dent solar radiation, after the fraction a (the Energy System (CERES), with data from March planetary albedo) is reflected back to space. 2000 to December 2003. The 20 coupled ocean- The second term represents the outgoing long- atmosphere GCMs performed their simulations wave radiation, with its Stefan–Boltzmann of the 20th century in support of the IPCC dependence on the effective radiative temper- Fourth Assessment Report and archived their

ature, Te . C is the Earth’ s total heat capacity, S model descriptions and the results the solar constant, R the Earth ’s radius and s at the Program for Climate Model the Stefan–Boltzmann constant. Diagnostics and Intercomparison It is known that many of these terms change (PCMDI: http://www.climatemod over time; for example, the solar ‘constant’ ellingprimer.net/l/k125.htm). varies in accordance with solar fluctuations These archives of observations and simulations and as a function of changes in the Earth ’ s allow a comprehensive comparison of the orbit. This paper examines one of these varia- planetary albedo (a ) in data and predictions bles: the Earth’ s planetary (i.e. viewed from made by 20 current GCMs at the time in use for outside the atmosphere) albedo, a . The inves- the IPCC assessments. tigation is interesting because it compares this The climate models do not do well, as the albedo as observed by two different satellite figure shows. GCM-derived albedos are almost sets and as computed by 20 different GCMs. consistently higher than the values observed This comparison of the Earth’ s global albedo by satellites. For the period with global ERBE

CERES ERBE GISS–EH UKMO–HadGEM1 GISS–ER UKMO–HadCM3 MIROC3.2(hires) MRI–CGCM2.3.2 INM–CM3.0 IPSL–CM4 GFDL–CM2.1 PCM ECHO-G CGCM3.1(T47) FGOALS–g1.0 CCSM3 GFDL–CM2.0 CSIRO–Mk3.0 BCCR–BCM2.0 MIROC3.2(medres) ECHAM5/MPT–OM CNRM–CM3 0.27 0.28 0.29 0.3 0.31 0.32 0.33

Figure source: Bender et al. (2006). Reproduced with permission of John Wiley & Sons Ltd. (Continued)

0002006766.INDD 61 1/10/2014 7:17:20 PM 62 The Climate Modelling Primer

data (February 1985–May 1989), the modelled Discussion questions global mean albedo is on average 0.009 above 1. It is recognised that climate model- the measured global mean. This corresponds lers use all the data they can find to set to a difference in radiative flux of almost parameters and correct formulations; 3 W m−2 . The mean level of global mean albedo in other words, to tune the models. according to CERES (using data from the first Can this tuning be removed (or accounted 4 years of this century) is an additional ca. for) in evaluation tests? Can such observa- 0.012 below the ERBE mean, corresponding to tions be used to evaluate the adequacy of an additional flux difference of ca. 4 W m−2 . the models? The authors undertake a detailed regional 2. How much difference do you think is analysis, concluding that the seasonal varia- ‘reasonable’ between the GCMs’ albedo tions of albedo in subtropical areas dominated and the two sources of observations used in by low-level stratus clouds and in dry deserts in this multi-model evaluation? Why do you subtropical areas are the most poorly simu- choose this value? Much greater regional, lated by the models. than global, discrepancies (between models This study gives rise to a range of issues, and data) were found. Why do you think including how useful are efforts to evaluate this happened? Is this finding likely to hold climate model performance and, since essen- for other climate parameters? tially all climate models are tuned to existing 3. This study is about evaluating GCMs in terms observations, what value of planetary albedo of the planetary albedo they calculate. In should be employed in low-resolution cli- other climate models, this planetary albedo mate models such as energy balance models is specified. What do the observations sug- (EBMs). gest about specifying a single unchanging value for this parameter? Is the planetary Review albedo unchanging and, if so, what is the Find other evaluations of climate model ade- cause of such stability? Can this apparent quacy and try to determine if these have global stability be disturbed by human changed over the past two decades. impact or is the system resilient?

one, not necessarily the best, tool for climate models modify only particular components of study. The literature contains many fascinating the models. The size and detail in these models examples of very simple models being used to mean that only through a sharing of effort can demonstrate failures and illustrate processes in progress be made. As the models have become much more complex systems. However, any intro­ increasingly complex, application of the princi­ duction to climate modelling must stress the cru­ ples of software engineering has become an cial role played by computers. Without the recent essential part of the process and has made it growth in computational power and the reduction easier to upgrade and exchange parts of the in computing costs, most of the developments in models. In an uncannily symbiotic lock­step, climate modelling that have taken place over the computers and climate models co­develop. last five decades could not have happened. The exponential growth of the Worldwide A fully coupled climate model (also called an Web and the access to information it enables ocean–atmosphere­biosphere general circula­ have become a fundamental part of life and, tion model – OABGCM) takes about 25–30 thus, also of climate modelling. This has pros person­years to code, and the code requires and cons: it allows global access and has led to continual updating as new ideas are imple­ the widespread use (and sometimes abuse) of mented and as advances in computer science are climate model results. At the beginning of this accommodated. Most modellers who currently chapter, we introduced the idea of a ‘treasure perform experiments with the most complex of chest’ of climate modelling examples and

0002006766.INDD 62 1/10/2014 7:17:21 PM Why Model Climate? 63

­illustrated how each reader might develop such We encourage reflection to lock in learning. At a personal collection with our analogue version the end of the book, we have gathered some (see Table 1.2). The Web encourages an final examples of aspects of climate modelling e-treasure collection, perhaps a little like the we believe this book illuminates. We also hope Macquarie treasure chest (see Figure 1.1). that our examples of personal climate modelling Table 1.11 shows the Primer authors’ climate treasures will encourage or even inspire your col­ modelling e-treasures. lection. Remember, your goal as you read through

Table 1.11 The Primer authors’ climate modelling e-treasure collection (compared with the ‘real-world’ items given in Table 1.2) Type Old chest Authors’ e-treasure collection QR codes Visual Paintings Write a climate change limerick: for inspiration, see http://www. climatemodellingprimer.net/l/k118.htm

My own Butterflies, Skeptical Science experience beetles, etc. http://www.climatemodelling primer.net/l/k119.htm and climate negotiations http: //www.climatemodellingprimer.net/l/ k120.htm LINK

Oceans Algae and Movie featuring the ocean conveyor belt seaweeds ‘The Day After Tomorrow’ (2004). Watch carefully about 6.5 minutes in for 2 minutes (and then enjoy) Change Exotic stuffed Disturbance growth – butterfly effect behaviour birds (BBC) http://www.climatemodelling primer.net/l/k121.htm

Pretty Arrangements Lorenz Attractor explanation things of sea-shells http://www.climatemodelling primer.net/l/k122.htm

How it Artefacts What is a cloud? (NASA) works http://www.climatemodellingprimer. net/l/k123.htm

Testing climate model predictions http://www.climatemodelling primer.net/l/k124.htm

0002006766.INDD 63 1/10/2014 7:17:21 PM 64 The Climate Modelling Primer

this book is to collect great climate model exam­ ­climate models is the International Satel­ ples. Choose ones that you find interesting and lite Cloud Climatology Project (ISCCP).62 that you can work out how to explain to someone Together with other satellite data, this has else. In each chapter, look out for new discover­ been analysed to create a record of cloud ies and add to your collection any that are illus­ amount from 1982 to date (Figure 1.25). trative of points or aspects of climate modelling Consider these measurements in the context you find tricky or amazing (or both). of their use for climate modelling. List as many ways as you can think of how this measure of cloud amount could be exploited by climate modellers. Beside each use, give as many caveats as you can. (You may find it useful to refer back to Figure 1.12.) 4. Most people today understand that some aspects of life affect the climate (for example, plants have an important role in regulating­ the atmospheric burden of carbon dioxide). Oil companies Eco warriors Review any newspaper or magazine (choose one you have not yet studied here) for as many issues as you need to find two articles about how people affect the climate.­ Summarise Elected leaders Developing nations both articles in less than 100 words each.

Part 2 – Discussion questions 1. Human influence on climate is clear (in land- use change effects as well as greenhouse gas warming). On the other hand, climate has changed over the whole lifetime of the Earth, most of which has been without human modification. Construct a quick set of figures (back of envelope numbers) that compare 1.6 Summary: research the effects of people with at least two differ­ ent non-human influences on climate. and review 2. How do you think climate modellers might Part 1 – Review questions better portray themselves to outsiders (aliens)? Research examples of astronomers’ 1. List the five reasons most persuasive, in your efforts to explain our civilisation, e.g. the opinion, for building or using models. You do Pioneer 10 plaque, the Arecibo message not need to restrict your list to climate mod­ and the Voyager ‘Golden Record’, and use elling but, for each of your chosen reasons, these as templates for sharing understanding­ give an example of how a model has contrib­ about climate modelling. Suggest how to uted already and could help in the future. improve the magazine articles you read for 2. Imagine you have to explain the incredible Climate Model Communication Box 1.2. properties of water to a group of keen 11 3. Improving global governance depends upon year olds. What would you say? Allow your­ many issues. To what extent can you make a self less than 500 words in total. What two case for international laws informed by experiments would you show them? results from climate models? Why might 3. One of the longest running data collection some people believe this is hazardous (even efforts designed to assist and evaluate foolhardy) and do you agree or disagree?

0002006766.INDD 64 1/10/2014 7:17:22 PM 0002006766.INDD 65 statistics ( climate modelsshowninFigure 1.22. Source: After Houghtonetal. (1997). ReproducedwithpermissionoftheIPCC. (C-cycle iscarboncycle). This straightforward plotcanbe comparedwiththetwodiagrammaticcategorisationsof Figure 1.26 ( Figure 1.25 4. American MeteorologicalSociety. for ATSR-GRAPE and10.00 dots

Cloud amount anomaly –0.04 –0.02 0.02 0.04 Consider these three sketchesand either the diagram in Figure 1.26 is another. Figure 1.22compares two depictionsand or classifyingclimatemodels. Forexample, There are manydifferent waysofarranging ) andInternationalSatelliteCloudClimatologyProject(ISCCP)anomaliescreatedusingthewholediurnaltime 0 blue line blue

High Low Time seriesofglobalcloudamountanomaliesfromthelong-termmeanmultiplesatellites Comparison ofthecomprehensivenessandcomplexityavariety ofclimatemodeltypes 1985 Comprehensiveness ). For mostofthedata, thelocalobservation timeis1.00 Low models Simple biome

am forMISR). Source: Stubenrauchetal. (2012). Reproducedwithpermissionofthe balance models Energy ATSR-GRAPE PATMOSX ISCCP 1990 C-cycle models assessment Integrated Sea-ice models models Upwelling diffusion/ balance models energy Soil vegetation Complexity AIRS-LMD TOVS-PathB HIRS-NOAA atmosphere 1995 convective schemes Radiative transfer models Year of modelsrelate tooneanother. method ofshowinghowthe different types your choiceorcreate another, stillbetter, select thebest and explainthereasons for atmosphere-ocean Atmospheric general circulation circulation

general 2000 models pm (except3.00 Coupled MISR MODIS-ST MODIS-CE models Why ModelClimate?

pm forISCCP, 10.00 2005 High 1/10/2014 7:17:23 PM

65 am 2010 66 The Climate Modelling Primer

Your Climate Modelling Treasures

Begin to assemble a set of attractive visualisa- interested person – say a friend or tions of aspects of climate modelling and be family member. Use the template sure that you are able to explain these to any in Table 1.11 .

Quick historical literature review

1963 : Lorenz , E.N. ( 1963 ) Deterministic, non periodic flow . J on Climate Change. Cambridge : Cambridge University Atmos Sci 20 , 130 – 141 . Press . 1986 : Thompson , S.L. , Schneider , S.H. ( 1986 ) Nuclear winter 2012 : Oreskes , N. , Conway , E.M. ( 2012 ) Merchants of Doubt: reappraised . Foreign Affairs 64 ( 5 ), 981 – 1005 . How a Handful of Scientists Obscured the Truth on Issues 1987 : Gleick , J. ( 1987 ) Chaos: Making a New Science . from Tobacco Smoke to Global Warming . New York : London : Vintage . Bloomsbury Press . 1999: Feynman , R.P. (1999 ) The value of science. In: The Pleasure 2012 : Shine , K. , Ptashnik , I.V. , Radel , G. ( 2012 ) The water of Finding Things Out. New York: Perseus Publishing. vapour continuum: brief history and recent develop- 1999 : Schellnhuber , H.J. ( 1999 ) ‘Earth system’ analysis and ments . Survey Geophys 33 , 535 – 555 . doi: 10.1007/ the second Copernican revolution . Nature 402 (Supp. 2 ), s10712-011-9170-y . C19 – C23 . 2012 : US National Academy of Science ( 2012 ) A National 2003 : Walker , G. ( 2003 ) Snowball Earth: The Story of the Strategy for Advancing Climate Modeling. Committee on Great Global Catastrophe That Spawned Life as We Know a National Strategy for Advancing Climate Modeling. It . New York : Crown . Board on Atmospheric Studies and Climate; Division on 2009 : Schneider , S.H. ( 2009 ) Science as a Contact Sport: Earth and Life Studies. Washington, DC: National Inside the Battle to Save the Earth’ s Climate . Washington, Academies Press . DC: National Geographic Society . 2012 : Victor , D.G. , Kennel , C.F. , Ramanathan , V. ( 2012 ) The 2012 : Hansen , J.E. , Sato , M. , Ruedy , R. ( 2012 ) Perception of climate threat we can beat . Foreign Affairs 91 ( 3 ), climate change. Proc Natl Acad Sci USA 109 , 14726–14727, 112 – 121 . E2415 – E2423 . doi: 10.1073/pnas.1205276109 . 2013 : Kopparapu , K.R. , Ramirez , R. , Kasting , J.F. , et al. 2012 : Henderson-Sellers , A. , McGuffie , K. (eds) ( 2012 ) The ( 2013 ) Habitable zones around main-sequence stars: Future of the World’ s Climate . Amsterdam : Elsevier new estimates . Astrophys J 765 ( 2 ), 131 . doi: 10.1088/ Scientific . 0004-637X/765/2/131 . 2012 : Intergovernmental Panel on Climate Change ( 2012 ) 2013 : Winton , M. , Griffies , S.M. , Samuels , B.L. , Sarmiento , Managing the Risks of Extreme Events and Disasters to J.L. , Frölicher , T.L. ( 2013 ) Connecting changing ocean cir- Advance Climate Change Adaptation. A Special Report of culation with changing climate . J Climate 26 , 2268 – 2278 . Working Groups I and II of the Intergovernmental Panel doi: 10.1175/JCLI-D-12-00296.1 .

0002006766.INDD 66 1/10/2014 7:17:24 PM 0002006766.INDD 67 1/10/2014 7:17:24 PM