Earth’s climate provides the environment in which humanity has evolved and in which human societies have expanded and thrived. It also periodically generates events that disrupt those societies—in some historic cases, apparently causing the failure of entire civilizations, although in many of those cases considerable dispute exists about the precise cause (Butzer, 2012). Climate events are disruptive when they harm people and the environmental, social, and economic assets that people depend on and when societies are unable to cope, respond, and recover effectively. Disruption thus depends on the conjunction of events with human vulnerability. However, major climate events are not always seriously disruptive. Past civilizations have sometimes faced very serious stress from environmental change without collapsing (Butzer, 2012; Butzer and Endfield, 2012).
When climate is changing, potentially disruptive climate events occur at different rates, with different intensities, and perhaps in different places from what people would expect from past experience. Those responsible for anticipating the risks, including the intelligence and security communities, turn to climate science in the hope of getting useful estimates of changing overall risks of future events of interest and forecasts of when and where such events will occur on time-scales longer than those of normal weather forecasts.
This chapter focuses on climate events, a key element in the conceptual framework presented in Chapter 2 for considering the connections between climate change and outcomes of security concern. It briefly summarizes the scientific bases for providing risk estimates and forecasts for potentially disruptive climate events on time-scales up to a decade or so. We consider
the potential for major, abrupt climate change as well as for single climate events and clusters and sequences of such events that could be sufficiently disruptive as to raise concerns about U.S. national security. The next two chapters examine how such events might disrupt social, political, and economic systems sufficiently to raise national security concerns.
The predictability of weather and climate varies with the time-scale; moreover, predictive efforts face different challenges at different time-scales. Weather forecasting focuses on predictions at time-scales up to about two weeks and is based on the premise that the atmosphere behaves according to a set of deterministic equations such that if the initial state of the atmosphere is known, its evolution can always be determined. Changes or errors in the initial state limit predictability on longer time horizons; two weeks is normally considered the limit of atmospheric predictability.
Climate models are based on the same basic set of equations that predict shorter-term weather variations, but they also include terms that represent a coupling of the atmosphere with the ocean and land surfaces, which inherently have memories of climate longer than the atmosphere does. Climate models have a coarser resolution than weather prediction models, which limits the level of accuracy in their simulations of atmospheric and ocean dynamics and their interactions with climate. Instead of forecasting actual day-to-day changes in weather over a period of a week or more, climate models concentrate on simulating the processes that govern the interannual and longer-term climate variability of the coupled ocean– atmosphere–land system.
To date, climate prediction has focused mostly on two time-scales: seasonal and centennial. Seasonal predictions, like weather forecasts, are dependent on initial values, and thus their ability to make predictions relies on information provided by initial ocean conditions (Latif et al., 2010), particularly sea surface temperatures, which strongly influence atmospheric circulation. The greatest contributor to predictive skill on a seasonal time-scale has been an understanding of the dynamics of the El Niño–Southern Oscillation (ENSO), which influences the yearly variability of rainfall and temperatures over broad sectors of the globe and even global mean temperatures.
On centennial time-scales, the evolution of climate remains chaotic and irregular and depends on external changes in radiative forcing (the influence of a factor, such as solar radiation or anthropogenic changes in atmospheric composition and land cover, on the balance of incoming and outgoing energy in the Earth-atmosphere system). Thus on such time-scales, climate projections are sensitive to assumptions about how future radiative forcing
will change as a function of energy consumption, land cover, and other drivers of change in Earth’s radiation balance. On such time-scales, the system loses its memory of the initial conditions, and the future trajectory of the coupled climate system is strongly determined by the external forcing.
Recently observations and models have suggested that it might be possible to make decadal climate predictions by using knowledge about regularities in the natural climate system on that time-scale, especially conditions involving the state of the ocean. However, climate projections based on emission scenarios indicate that decadal-scale variability is also influenced by the accumulated impacts of anthropogenic radiative forcing. Decadal-scale predictions therefore require information both about radiative forcing (e.g., levels of greenhouse gases and aerosols) and about the current observed state of the atmosphere, oceans, cryosphere, and land surface (Hurrell et al., 2010).
Understanding and making predictions of decadal variability is still very much in its infancy, and evaluating these predictions is a major challenge. It is straightforward to verify daily weather forecasts through statistical and historical data, but verification is more difficult for decadal forecasts. Observational records are simply not consistent enough or long enough to quantify prediction skill. Even the basic characteristics and mechanisms that describe climate on a decadal scale are poorly documented and not well understood. Testing models against observed climate variability provides some means of verification and thus can offer some confidence in using these models to simulate future climate, but even those efforts are hindered by a lack of subsurface ocean observations and satellite data.
As part of the work for the next Intergovernmental Panel on Climate Change (IPCC) report, modeling centers have coordinated decadal hindcast and prediction experiments covering the period from 1960 to 2035 in what is known as the Coupled Model Intercomparison Project Phase 5 (CMIP5). Part of the work involves an atmosphere–ocean general circulation model with a resolution of approximately 50 kilometers (31 miles) being run to make decadal predictions out to 2035. Lower-resolution versions of the same model will be run with coupled carbon cycles and biogeochemical processes to help quantify the magnitude of important feedbacks that will determine the degree of climate change in the second half of the 21st century (Hurrell et al., 2009).
A better knowledge of the initial conditions needed to initialize the models properly should increase the skill of predictions of the climate system on varying time-scales. An increased understanding of the physical mechanisms that govern climate variability is also critical. In sum, although physical climate science is able to project climate trends based on scenarios for increases in greenhouse gases (GHGs) and to estimate changes in the likelihoods of occurrence for some kinds of climate events in the coming
decade, the ability to anticipate specific climate events on that time-scale is still in its infancy. Improving our understanding of the ENSO cycle and its broader climatic implications currently offers the best route for extending our ability to make skillful forecasts of the risks of climate events from the current limit of a few weeks into the future to a few months or even one or two years. It will also be important to improve understanding of how the GHG-forced changes in climate might change the characteristics of climate variability, such as ENSO.
Despite their inability to predict specific events for the coming decade, climate models can still provide useful information about the range of plausible climate outcomes that could result from the combination of external forcing and variability of the climate system due to internal processes over the next decade. For example, several recent studies have evaluated the distribution of 10-year trends within model simulations that were forced with increasing concentrations of greenhouse gases (Easterling and Wehner, 2009; Santer et al., 2011). The purpose of these studies was to address claims that the lack of a significant positive global temperature trend over the 10-year period following 1998 refutes projections from climate models. And, indeed, the models did show that even under scenarios with relatively high greenhouse gas emissions, decades with a zero or negative temperature trend occur in more than 10 percent of the individual 10-year segments within models.
Figure 3-1 displays the probability distribution functions of 10-year trends from Easterling and Wehner (2009) for global mean temperatures, and it extends their method to look at the probability distributions of 10-year trends for a number of regions for the future (see Appendix C for an explanation of the method used for the regional projections). These functions were calculated using model simulations from the CMIP5 database under a scenario for the first half of the 21st century that assumes a “business as usual” rate of greenhouse gas increase of about 1 percent per year (known as the RCP8.5 scenario). The functions represent differences between projected average annual surface temperature, globally and for various regions, for the coming decade and the average temperature at the start of the decade, expressed in degrees Celsius. The estimates apply to any decadal start date up to 2040. Much as Easterling and Wehner (2009) found for global temperatures, each region has a reasonable chance of cooling over a period of a decade because of natural variability that can act to counteract the warming trend. At the same time, the upper tails of these distributions show that the increase in average global temperature could be as high as 1.8°F (1°C) in a single decade and the increase in average regional temperatures could be as high as 3.6°F (2°C) or more in a single decade— for example, about a 40 percent chance of an increase of at least 1.8°F (1°C) and about a 10 percent chance of an increase of at least 3.6°F (2°C)
in the U.S. corn belt. Although the scenarios of strong and rapid warming over a decade have low probability, the potential implications for society are large enough to deserve much more consideration by climate and security analysts. To our knowledge, similar work has not yet been published for plausible decadal trends at regional scales or for climate variables other than surface temperature.
Although the foregoing analysis suggests very useful predictive skill for changing likelihoods of some kinds of events, even projections that have relatively low predictive skill can be useful to the security community. Low predictive skill is especially likely to be characteristic of projections of climate parameters with low signal-to-noise ratios (e.g., with high interannual variability) and for very low-frequency events, including the most extreme events. Such events deserve special consideration by security analysts when and where the fundamental science suggests that they will increase in likelihood and the data are not yet adequate to rule out this possibility. In such situations, it may be wise for security analysts to consider what-if scenarios involving increased frequencies of occurrence for such events.
Climate models currently used to project climate change over the 21st century generally indicate that there will be a gradual response to increased greenhouse gases and other climate forcing, suggesting quite a low likelihood that major changes will occur within the coming decade. Nevertheless, the simulations do show that abrupt changes are possible in various regions where a particular variable crosses a threshold (for example, if precipitation lowers sufficiently to cause a transition from forest to grassland, or grassland to desert, or if temperature increases enough to cause a transition from snow- or ice-covered surfaces to snow- or ice-free ones). The observed climate record, particularly the paleoclimate record, shows that abrupt or step changes in climate do occur, even within a time span of a decade or less. The possibility of rapid step changes in climate is of great concern in terms of social and political stresses because, with less time to adjust, it is usually harder for societies to cope and respond effectively.
Abrupt climate change is generally defined as occurring when some part of the climate system passes a threshold or tipping point resulting in a rapid change that produces a new state lasting decades or longer (Alley et al., 2003). In this case “rapid” refers to timelines of a few years to decades. Abrupt climate change can occur on a regional, continental, hemispheric, or even global basis. Even a gradual forcing of a system with naturally occurring and chaotic variability can cause some part of the system to cross a threshold, triggering an abrupt change. Therefore, it is likely that gradual or monotonic forcings increase the probability of an abrupt change occurring.
Many cases of rapid change in climate have been observed in the instrumental and paleoclimatic records. The instrumental record is short—only about 130 years in length—and as a consequence does not adequately sample the full spectrum of climate variability. However, even with this limited sampling some examples of rapid regional change can be found in the observed record. For example, strong warming (greater than 4°C) occurred in parts of the Arctic during the 1920s, with most of the warming occurring in the late 1920s (Alley et al., 2003). The Dust Bowl years of the 1930s in the central United States provide another example of a rapid shift, in this case with the reversal also occurring in less than a decade. On a much larger geographic scale, the climatic effects of the abrupt shift in 1976–1977, mainly in sea surface temperatures in the Pacific Ocean, have been documented in a number of global and hemispheric analyses (Alley et al., 2003).
The paleoclimatic record indicates that numerous rapid changes occurred in the period after the last deglaciation (i.e., the past 10,000 years). Many of these changes were larger than those recorded in the instrumental record. Examples include abrupt shifts to large-scale drought that were similar to, but more intense and longer lasting than, what was experienced during the 1930s Dust Bowl years. Tree ring records of drought in the western United States show abrupt, long-lasting mega-droughts that came on rapidly and lasted for a decade or more, covering most of the region. It has been argued that large-scale mega-droughts were responsible, at least in part, for the decline and collapse of the Mayan civilization in Mexico (Hodell et al., 1995), the Tang Dynasty in China (Yancheva et al., 2007), and the Anasazi civilization in the southwest United States (U.S. Climate Change Science Program and Subcommittee on Global Change Research, 2008). Such conclusions are hard to support conclusively, however. For example, Butzer and Endfield (2012) judge the Mayan case to be controversial.
What can cause rapid climate change? Some rapid changes are initiated by a cooling process, others by a warming process. Some rapid changes are temporary and reversible, others cause one-way shifts that persist for decades or longer. One recurring cause is massive releases of sulfate aerosols into the atmosphere, which can result in substantial regional or global cooling. Historically the main cause of such releases has been volcanic eruptions. Large volcanic eruptions, such as the 1815 Tambora eruption that led to the world-wide cooling event called the “year without a summer” in 1816, occur on average about once every 1,000 years. Super-eruptions occur on average about once every 10,000 years, with the last two having been in Toba in Indonesia 74,000 years ago and Oruanui in New Zealand 26,000 years ago. It is thought that the Toba super-eruption led to a global “volcanic winter” lasting for several years that was approximately 7.6°F (4°C) colder than today and up to 27°F (15°C) colder in the high latitudes.
Some theories suggest that the Toba eruption and ensuing volcanic winter wiped out most of our human ancestors, leaving only a small population of perhaps a few thousand (Self, 2006).
The above examples of rapid climate change do not fully meet the definition of abrupt climate change given by Alley et al. (2003) because after each of these rapid changes, the climate returned to its previous state. However, it is possible for a monotonic change in certain aspects of the Earth system to cross a tipping point so that climate change not only would be rapid but also would shift the system to a different baseline. For example, as the area covered by Arctic summer sea ice decreases, exposing more ocean surface, the result is an overall darkening and loss of reflectivity for Earth’s surface in that region, which leads to an increase in the absorption of solar radiation; this in turn warms the ocean waters, which tends to increase melting, creating a positive feedback loop. Thus the result of a loss of summer sea ice could be a lasting change in the energy balance of the planet. However, because there are feedbacks that lead ice extent to recover, some analysts argue that a lasting change in the energy balance from this mechanism is unlikely in the coming century (Tietsche et al., 2011).
Could significant and lasting abrupt climate change occur, either globally or regionally, in the next decade or so? The possible scenarios most frequently mentioned include the loss of Arctic summer sea ice, GHG release from melting permafrost, the melting of Greenland glaciers, the breakup of the West Antarctic ice sheet, and a change in the strength of the Indian summer monsoon (e.g., Lenton et al., 2008). At a smaller spatial scale, the loss of specific mountain glaciers or glacier systems, the drying of lakes, and shifts in biomes are likely types of abrupt changes for particular regions.
The likelihood of any of these events occurring by the mid-2020s is difficult to estimate for a number of reasons, including the fact that global climate models do not simulate abrupt climate shifts well, which makes it hard to project when a change in some part of the climate system that might induce a broader abrupt change (e.g., regional or hemispheric temperature or precipitation) will reach a critical value. Lenton et al. (2008) provide estimates of critical values for a number of “tipping elements,” which are defined as features of the Earth system large enough (at least subcontinental in scale) to have pronounced impacts on the climate system if an abrupt change occurs in that feature. Examples would include the Greenland ice sheet and the Amazon rain forest. Even for those tipping elements where critical values are defined (e.g., for rapid melting of the Greenland ice sheet, approximately a 5.4°F [3°C] increase in the local air temperature), estimating a probability that the critical value will be reached in the next couple of decades is difficult if not impossible. Most climate scientists would probably judge the likelihood of an abrupt change that would have large and long-lasting destabilizing impacts on climate and human society by the
mid-2020s to be quite small. However, because of gaps in knowledge about the mechanisms and tipping points for the processes involved, it is difficult to assign a level of confidence to judgments about the preconditions for, or timing of, an instability affecting, for instance, a major section of a large ice sheet. Moreover, there may be other processes in the Earth system, not yet identified, that have tipping points that could lead to abrupt climate change. Because of such gaps in knowledge, the possibility of such events occurring in the next decade or so cannot be totally discounted.
Although the likelihood of such abrupt climate change scenarios happening in the next decade or so appears to be quite small, it cannot be estimated accurately. The consequences of some abrupt climate changes, if they occurred, could be quite severe. For example, droughts are a recurring event in the climate system, with a major drought occurring in some part of the world at least yearly. Some drought conditions, such as those that have continued off and on for the past decade in the western United States, could be harbingers of a mega-drought. If that proved to be the case in the western United States, the potential impacts on water supplies as well on as ecosystems, both natural and managed, in that region would be enormous. To the extent that the possible tipping elements leading to such major changes are known, it would be important to monitor those elements and the factors that affect them. It would also be important to monitor changes in the social, economic, and political factors that affect the size of the exposed populations, their susceptibility to harm, the ability of the populations to cope, and the ability of their governments to respond. Where potentially affected areas are important producers of key global commodities such as food grains, it would also be important to assess the effects of climate-induced supply reductions on global markets and vulnerable populations.
The fundamental science of climate change suggests that continued global warming will increase the frequency or intensity (or both) of a great variety of events that could disrupt societies, including heat waves, extreme precipitation events, floods, droughts, sea level rise, wildfires, and the spread of infectious disease. Underpinning many of these extreme events is an acceleration of the global hydrological cycle. For each 1.8°F (1°C) increase in the global mean surface temperature, there is a corresponding 7 percent increase in atmospheric water vapor. Because warm air holds more water vapor than cool air, this leads to more intense precipitation. Essentially, warm air increases evaporation from the ocean and dries out the land surface, providing more moisture to the atmosphere that will rain out downwind. Water vapor is also a powerful naturally occurring green-
house gas. As such it is the source of a very strong positive feedback to the coupled climate system that amplifies any external forcing by a factor of approximately 1.6.
This section discusses trends in some extreme climate events over the past half century, science-based expectations of the futures of these types of events, and the prospects for using the science of climate change to estimate the changing likelihood of such events and to forecast their occurrence.
Trends in Extreme Climate Events
The frequency of extreme high-temperature events driven by global warming is increasing faster than would be the case if only the mean temperature were increasing because the variance of the temperature distribution is increasing as well (see Figure 1-1). Extreme weather and climate events have been responsible for a rapidly increasing loss of lives, well-being, and economic assets in recent decades because of the confluence of the events themselves with increases in the numbers of people and value of property exposed and vulnerable to the events. It has been difficult to determine conclusively whether damaging climate events themselves have yet been increasing in frequency or intensity enough to be detected in trends of damage, normalized for nonclimate factors. We discuss this issue further in Chapter 4.
Climate Change and Extreme Climate Events in the Coming Decade
Effects of Climate Change on Extreme Events
The frequency and intensity of extreme events are particularly hard to project because, among other things, there are by definition few of them, which makes it hard to validate predictive models against experience (see Appendix D for more detailed discussion of statistical issues and methods for assessing the probabilities of occurrence of extreme events). Analyses have, however, converged on a number of expectations for this century, as noted in a recent National Research Council review (2010a). In this century, it stated, “the frequency and intensity of heat waves is projected to continue to increase, both in the United States and around the world,” and “the frequency of cold extremes and the number of frost days will decline in the middle and high latitudes” (p. 223). It is also projected that “the fraction of rainfall falling in the form of heavy precipitation events will increase in many regions” (p. 224). Furthermore, “[r]ecent model projections indicate growing certainty that climate change could lead to increases in the strength of hurricanes, but how their overall frequency of occurrence might change is still an active area of research,” and “projections indicate that
the area affected by drought will probably increase in the decades ahead and that the number of dry days annually will also increase” (p. 265). The models are not highly specific about the timing of these changes, however.
A subsequent IPCC special report on extreme events (Intergovernmental Panel on Climate Change, 2012) summarized available evidence on observed and projected extremes of a variety of important types of climate events globally and at the regional level. Table 3-1 summarizes observed trends in several of these.1 The level of confidence in an observed trend neither implies nor excludes the possibility of changes in an extreme at other geographic scales. The report also offers projections of trends in these events, along with statements about the levels of confidence that the scientific community places in the projections.
Considering the difficulty of validating projections of extreme climate events, it would be a mistake to conclude from the lack of confidence in a projection of change for any type of extreme event that one can prudently act as if there will be no change. When there is good fundamental science behind an expectation of change—for example, in the frequency of extreme high-temperature and high-precipitation events or the likelihood of droughts—combined with noisy data or a small number of events for model validation, there may be sufficient reason for the intelligence community to develop and consider the security implications of scenarios in which the extreme event parameter changes in the direction suggested by the fundamental science.
Effects of Predictable Climate Variation on Extreme Events
Earth’s climate includes various regular cycles that may make it possible to anticipate an increased likelihood of climate events of concern months or longer in advance. Other than the succession of the seasons, the best understood of these is the ENSO (see Box 4-1). ENSO affects weather on most of Earth’s surface in cycles that last two to seven years. The largest and most predictable impacts are in the tropics. The warm phase of ENSO “is usually accompanied by drought in southeastern Asia, India, Australia, southeastern Africa, Amazonia, and northeast Brazil, with fewer than normal tropical cyclones around Australia and in the North Atlantic. Wetter than normal conditions during El Niño episodes are observed along the west
1 The confidence levels used in Table 3-1 are based on three scales: evidence and agreement; confidence; and likelihood (Mastrandrea et al., 2010). The summary terms used to describe the available evidence were limited, medium, or robust, and the degree of agreement was described as low, medium, or high. Levels of confidence were very low, low, medium, high, or very high. Likelihood (probability) was described as virtually certain (99–100%), very likely (90–100%), likely (66–100%), about as likely as not (33–66%), unlikely (0–33%), very unlikely (0–10%), and exceptionally unlikely (0–1%).
|Event||Global Change||Confidence||Regional Changes||Confidence|
|Cold days and nights||Decrease||Very likely||Decrease at the continental scale in North America, Europe, and Australia||Likely|
|Warm days and nights||Increase||Very likely||Increase at the continental scale in North America, Europe, and Australia||Likely|
|Warming trend in daily temperature extremes in Asia||Medium|
|Warming trend in daily temperature extremes in Africa and South America||Low to medium, due to insufficient evidence|
|Length or number of heat waves||Increase in many regions with sufficient data||Medium|
|Heavy precipitation events||More regions have experienced increases than decreases, although there are strong regional and subregional variations||Likely|
|Tropical cyclone activity (intensity, frequency, duration)||Unclear||Low confidence after accounting for past changes in observing capabilities|
|Drought||Intense and longer droughts in southern Europe, west Africa||Medium|
|Extreme coastal high water||Increase related to an increase in mean sea level||Likely, at global scale|
SOURCE: Adapted from Tables 3-1 and 3-2 in Intergovernmental Panel on Climate Change (2012).
The El Niño–Southern Oscillation Phenomenon
The El Niño–Southern Oscillation (ENSO) phenomenon is a multi-year cycle marked by changes in relative atmospheric pressures at sea level across the tropical Pacific Ocean and changes in the strength of the Pacific trade winds and the temperature of the ocean’s surface in the central and eastern Pacific. This mode of climate fluctuation has been linked globally to devastating droughts, extreme rainfall events, and the suppression of hurricanes in the Atlantic Ocean, among other interannual climatic changes, and therefore has a significant impact on global society. Recent studies have also linked ENSO to the natural variability of the carbon cycle (Jones et al., 2001). An El Niño event (the warm phase of the cycle, as indicated by sea surface temperature) is characterized by a weakening of the Pacific trade winds, a warming of the ocean’s surface in the central and eastern Pacific, and smaller differences in tropical sea level pressures between the eastern and western tropical Pacific Ocean. Opposite conditions occur in the cold phase, or La Niña. The warm events often last 12 to 18 months and historically occur every 2 to 7 years, although recent anthropogenic climate changes may possibly contribute to more frequent and intense El Niño events. The term ENSO is used to describe the full range of coupled ocean–atmosphere climate variability in the tropical Pacific Ocean.
Over the course of many decades of research, great strides have been made in understanding, monitoring, and predicting ENSO and its effects on climate. During an El Niño event, the warmest waters of the ocean, usually west of the international dateline, spread eastward into the eastern tropical Pacific, a distance one-third the circumference of the planet, and significantly affect the global atmospheric circulation. Via a process known as teleconnection, ENSO is often linked to global extreme precipitation events, either drought or flooding, in many far-flung places. ENSO also appears to play a large role in yearly and multi-year variations in sea level rise, while larger ENSO events can affect decadal trends in sea level by shifting the distribution of rainfall between land and the ocean. ENSO also affects marine and terrestrial ecosystems. During a warm ENSO phase the primary productivity in the eastern tropical Pacific is severely reduced because of weaker upwelling off the coast of Peru. The reduction in productivity affects the population and location of marine mammals, sea birds, and commercial fishing. Year-to-year variability in global atmospheric carbon concentrations is dominated by the ENSO cycle (Rayner et al., 1999). During El Niño, equatorial upwelling decreases in the eastern and central Pacific, significantly reducing the supply of carbon dioxide to the surface (Feely et al., 2006). As a result, the global increase in atmospheric carbon dioxide noticeably slows down during the early stages of an El Niño. Coastal regions are particularly affected by ENSO variability.
coast of tropical South America, subtropical latitudes of western North America, and southeastern America” (Intergovernmental Panel on Climate Change, 2012:155). The cold phase, known as La Niña, generally shows opposite anomalies. The effects of ENSO can also be felt in the other ocean basins and on all continents (Dai et al., 1998). Recent research has shown that different phases of ENSO are associated with different frequencies of short-term weather events such as heavy rainfall and extreme temperatures in the affected regions (Intergovernmental Panel on Climate Change, 2012).
These regularities in the effects of ENSO on climate events allow for skillful short-term climate predictions on time-scales from a season to a year by coupling atmospheric general circulation models with ocean general circulation models initialized with observations of the state of atmosphere and ocean. Forecast skill decreases away from the equator, however. This seasonal climate prediction has now become operational at many of the world’s major weather prediction centers. (NOAA, for example, offers seasonal outlooks online at http://www.cpc.ncep.noaa.gov/products/predictions/90day/ [accessed November 13, 2012].) Prediction on time-scales from years to decades is still very much in the research realm, so it is not yet possible to estimate the eventual skill that might be achieved in forecasting at these time-scales.
The fact that ENSO drives unusual weather patterns on several continents at the same time demonstrates that extreme weather patterns in different regions may not be independent in a statistical sense, suggesting the possibility that extreme climate events may cluster in time, a topic discussed more fully in the next section.
The question of whether and how anthropogenic climate change may be altering the ENSO cycle is being actively examined by climate scientists (Intergovernmental Panel on Climate Change, 2012). Dai et al. (1998) noted a shift in ENSO events in the mid-1970s toward more warm events, which coincided with record high global temperatures and drought anomalies that were greater than expected. There is some evidence linking these changes in ENSO events to intensified droughts in some drier regions since the 1970s, while the extent of wet areas has declined during this period. These changes are qualitatively consistent with the observed increases in greenhouse gases in the atmosphere, which act to enhance the hydrological cycle. The IPCC special report on extreme events (Intergovernmental Panel on Climate Change, 2012) describes systematic changes in ENSO behavior that have been observed over the past 50 to 100 years, but it concludes that it is not clear what role, if any, increased greenhouse gases have played in this phenomenon. All models used in that assessment predict continued ENSO interannual variability in the future no matter what the change in average background conditions. However, the changes in ENSO interannual variability differ from model to model based on subtle changes
in the physical parameterization schemes of the models (Collins, 2000). The models are not consistent in their projections of ENSO amplitude or frequency in the 21st century, and it is not yet possible to tell which models’ results are most credible. The IPCC study concluded that “it is not possible at this time to confidently predict whether ENSO activity will be enhanced or damped due to anthropogenic climate change” (Intergovernmental Panel on Climate Change, 2012:157). Climate scientists continue to explore links between increased greenhouse gas concentrations and ENSO because the fundamental principles of atmospheric science indicate that greenhouse warming will drive changes in ENSO.
It seems reasonable to infer from this research that ENSO is likely to change in coming decades, even though we cannot at this point determine which parameters will change or to what extent. In our judgment it would be worthwhile for the intelligence community to consider the security implications of a few of the scientifically plausible scenarios for how ENSO might change because changes in the amplitude and nature of ENSO constitute one of the few components of the coupled climate system capable of having a global synchronized impact. However ENSO changes, there will be simultaneous consequences in many places. For example, if the future will bring stronger El Niño events, as some models project, the results will likely include drought in Australia and excess precipitation across parts of the southern tier of the United States occurring in the same year. Such a confluence of events could affect global grain production in ways that could be disruptive, depending on other conditions affecting food supply and demand.
Over the next decade the most likely scenario for single extreme events is a continuation of the trends of recent decades—probably a slow rate of change for now, but with the possibility of a faster one later. Many of these trends involve the continuation of trends that are already being observed, such as those summarized in Table 3-1, such as a warming of days and nights and a trend toward heavier precipitation, with wet areas tending to get wetter and dry areas drier, and with more of the precipitation concentrated in heavy events, particularly in certain regions. The effects of the ENSO cycle, superimposed on these longer-term climate trends, will exacerbate some extremes and dampen others in complex ways. In addition to these events, there remains the possibility of unprecedented extreme events that might occur as a result of abrupt climate change or other climatic phenomena discussed later in the chapter.
By a cluster of extreme events we mean several extreme climate events appearing close in time but not necessarily in space. Clusters of extreme events are a concern from a national security perspective because U.S. government resources and those of other international actors deployed to deal with a security or humanitarian concern related to the first event in a cluster might be unavailable or less available to deal with a second or subsequent extreme event.
Clusters of extreme events may occur as a result of a random cooccurrence of extreme events in different places that have different causes. They may also result from large-scale climate processes that serve as common causes of events in disparate places that seem superficially to be unrelated. An example of such an event cluster was the conjunction of a heat wave and drought in Russia and floods in Pakistan in 2010. In July and August of that year, western Russia suffered from a major drought and heat wave that resulted in a major loss of life and crops as well as large forest fires. Simultaneously Pakistan was experiencing major flooding that also resulted in a major loss of life and property. These two events were linked by more than just their proximity in time. The meteorological pattern that led to the Russian heat wave, in which the large-scale upper-level wind flow developed a strong and persistent ridge, also contributed to the development of the meteorological pattern that resulted in the Pakistani floods—a downstream, leading trough (Lau and Kim, 2012). The fact that these two extreme events corresponded in time with each other and with a single larger meteorological pattern was unusual but not totally unexpected. Circulation events like this one, which cause some event clusters, are known to occur but are not well resolved in current climate models.
A key question for security analysis is whether such event clusters are non-random, indicating a teleconnection phenomenon in which the events in the cluster are intrinsically linked, making their joint probability of occurrence greater than the individual probabilities multiplied together. If so, when one of these events occurs, the likelihood of the other is increased. Knowledge of systematic connections of this type could be valuable for security analysis because it could aid in the development of security risk scenarios.
Generally this sort of clustering of events can occur when the large-scale atmospheric flow around the mid-latitudes of the northern and southern hemispheres, which contains a series of about four to six meanders, or waves (called Rossby, or planetary, waves), develops specific kinds of stable patterns, resulting in the simultaneous occurrences of drought, drought and flood, or flooding events in different parts of the globe. Herweijer and Seager (2008) show that the simultaneous occurrence of drought in North America, parts of Europe, South America, and Australia has resulted from
the development of Rossby wave patterns, likely a response to specific patterns of cold sea surface temperatures in the tropical eastern Pacific Ocean (La Niña). Much as with the recent Russian drought/Pakistani flooding cluster, when drought has occurred over North America and Europe (including western Russia), unusually wet conditions have occurred over Pakistan and northern India as well as over other parts of the globe (Herweijer and Seager, 2008). The ENSO phenomenon creates systematic connections of climate conditions that affect large parts of the planet more or less simultaneously, leading to predictable clusters of extreme climate events during strong El Niño and La Niña phases (see Box 4-1). Such phenomena suggest that the simultaneous occurrence of drought in parts of the northern and southern hemispheres coincident with flooding in other regions is a distinct possibility in the next decade or so. Other clusters of extreme weather or climate conditions may also become more likely than in the past.
Related to the concept of event clusters is the notion of compound events, which occur when different kinds of climate events are linked in the same place. An obvious example is the conjunction of drought with wildfires or crop failures. The first event drives the second, but they are different in that different people or groups are susceptible to damage from different parts of the compound event and different organizations may be involved in response. In Chapter 5 we discuss an example of the conjunction of drought, heat wave, and forest fire in Australia in 2010.
If climate events and extremes were independent in a statistical sense, the likelihood of a cluster or a compound event of any size could easily be estimated mathematically. But as the above example makes clear, extreme events in different parts of the world can be driven by common underlying forces and thus have an intrinsic relationship such that when one such event occurs, the likelihood increases that other extreme events linked to them by common causes will also occur. In statistical language, such events are called dependent.
We conducted two exploratory analyses in which we used data from 1901–2002 to examine statistical relationships between pairs of conditions that climate science indicates may be dependent on underlying large-scale climate patterns (see Appendix D for details). In one case, the correlation between heat in Russia and heavy precipitation in Pakistan, we found no general association between the two phenomena in the century-long record, suggesting that the association in 2010 was without recent precedent. In the other, an association between droughts in regions of the southwestern United States and a region in Argentina and Uruguay, we estimated that the probability of drought exceeding a 10-year return level in one of the regions given a similarly serious drought in the other was almost three times what it would be if the events were statistically independent and that for more serious droughts, the dependence effect was even stronger.
These results are preliminary, and conclusions should be tempered by questions about the accuracy of the data, the limited number of data points (102 annual observations), and the small number of previous examples that were extreme in both variables. The results suggest, however, that event clusters of this sort could be increasing in likelihood more rapidly than the underlying single events and that by combining climatological and statistical analysis, it may be possible to develop better estimates of the likelihoods of occurrence of extreme event clusters of interest. However, the scientific analysis on this matter is still in its infancy.
A climate condition or event can sometimes precipitate or facilitate a series of other physical and biological events, each linked to the others through deterministic processes, that together promote conditions of potential security concern. We can illustrate the concept of climate event sequences, although not the potential national security implications, with a sequence of events that took place recently in western North America.
The recent pine bark beetle outbreaks in western North America have their roots in multiple mechanisms, but climate change is believed to be a factor driving at least some of them (Bentz, 2008). Elevated temperatures, particularly consecutive warm years and elevated minimum temperatures, can speed up reproductive cycles and reduce cold-induced beetle mortality, both of which increase beetle populations. Moreover, shifts in precipitation patterns and associated drought can promote bark beetle outbreaks by weakening trees and making them more susceptible to beetle attacks. Droughts, combined with forest die-off from beetle outbreaks, in turn increase the amount of fuel in forests, increasing the susceptibility of forests to fire, especially during heat waves, which are likely to be longer and more intense with climate change. The ongoing climate change, because of its warming, also increases evaporation and thus intensifies the risk of fire (see Figure 3-2). Dead forests provide an opportunity for fires to spread more widely. The extreme forest fires in Colorado in the summer of 2012 occurred in one of the highest-risk regions in the western United States. Severely burned forest lands are also more prone to erosion in storms (e.g., Benavides-Solorio and MacDonald, 2005), indicating that forest fires increase the risks of soil degradation and of mudslides. Climate change may thus be playing at least four different roles in this dynamic: It promotes bark beetle infestations, weakens trees, dries the environment, and creates weather conditions conducive to fire outbreak. These conditions, connected in sequence, increase the risks of major forest fires and their hydrological and human consequences.
The kinds of linkages involved in this example and in other plausible
event sequences are in general poorly understood and therefore not very predictable. Nevertheless, the example suggests that there may be a very large number of plausible ways by which climate change could set in motion a sequence of events in ecological, hydrological, or other deterministic physical and biological systems that could become seriously disruptive and that might create security concerns.
Climate events occurring in one part of the world have the potential to affect other parts of the world through important, globally integrated systems other than climate itself. One example is the potential influence of climate events on the world supply—and therefore the prices—of international traded commodities, such as grains. By this mechanism an event such as the 2012 drought in the central United States, still developing as this is
being written, could affect world corn or wheat prices in ways that make essential foods unaffordable for populations in Africa or Asia. Another example of a global system shock would be constraints on the availability of humanitarian aid for a country because aid providers are responding to situations elsewhere in the world. Yet another would be a climate event that altered the distribution of a major pathogen affecting people or staple crops. These examples, which are discussed in greater detail in Chapter 4, indicate that there are numerous ways in which climate events could create shocks to integrated global social, economic, health, or technological systems and thus have effects far removed geographically from where the events occur.
Many extreme events, such as hurricanes, tornadoes, other severe storms, floods, heat waves, and wild fires, occur on the time-scale of days to about two weeks. A key question for climate research is how such extreme events will change within a warming climate. With such warming the statistics of weather are no longer stationary, and the linkage between weather and climate emerges as a research priority. Today’s climate change models are generally considered to provide an adequate representation of the large-scale secular trend in climate. However, gaining a better insight into how climate change will affect extreme weather requires that high-resolution numerical weather prediction models (and their inherent fast physics, such as cloud–radiation–precipitation interactions) be run in climate mode. This presents a major computational and resource challenge and has been the reason for the call for a seamless approach to weather and climate forecasting (World Climate Research Programme, 2009). Such a unified approach was first implemented by the U.K. Met Office in 1993. In the United States the National Centers for Environmental Prediction of the National Oceanic and Atmospheric Administration developed the Coupled Forecast System (CFS) in 2004 as a lower-resolution climate version of its Global Forecast System, a weather forecast system for short-term seasonal climate prediction. However, this system has yet to be run on decadal to centennial time-scales in response to GHG forcing. Hence the behavior of extreme events in a non-stationary climate cannot be fully described and projected until climate change models are run with the spatial resolution and physical processes of numerical weather prediction models. Until that time, extreme climate surprises should be expected to be more the rule than the exception.
Physical climate science has developed some skill at estimating the changing likelihoods of the occurrence of certain kinds of climate events, such as heat waves and certain precipitation anomalies, at a decadal time-scale and at a global and, in some cases, a continental or subcontinental geographic scale. However, the ability to foresee specific climate events on a decadal time-scale at the level of medium-sized countries is still in its infancy. Predictive skill at such time horizons and levels of resolution depends on having a more extensive observational system than currently exists and on developing an improved understanding of interannual and decadal processes of climate variability that can be incorporated into predictive models. Estimating the risks of extreme climate events—and especially estimating the places they will occur—is particularly challenging because predictive skill is harder to acquire and to validate for infrequent events and because the frequency and character of such events may change as climate trends continue.
The current state of understanding does not suggest that the distributions of single climate event types in particular places will change sharply in the coming decade, but neither can it preclude such a possibility with high confidence. It is safe to say that extreme climate events will change in their frequencies, intensities, and probably also their locations in the coming decade. The most likely scenario is a continuation of the temperature and precipitation trends of recent decades, probably with a slow rate of change for now, but with the possibility of a faster one later. However, the effects of the ENSO cycle superimposed on longer-term climate trends will exacerbate some extremes and dampen others in complex ways. In addition, there remains the possibility of unprecedented extreme events or conjunctions of events that might occur as a result of abrupt climate change or other climatic phenomena (for example, a more rapid rise in sea level if the melting of ice sheets were to accelerate).
Conclusion 3.1: Given the available scientific knowledge of the climate system, it is prudent for security analysts to expect climate surprises in the coming decade, including unexpected and potentially disruptive single events as well as conjunctions of events occurring simultaneously or in sequence, and for them to become progressively more serious and more frequent thereafter, most likely at an accelerating rate. The climate surprises may affect particular regions or globally integrated systems, such as grain markets, that provide for human well-being.
Some events or conjunctions of events may arise from connections within the Earth system, such as aspects of the ENSO cycle and other
continental-level phenomena that are only beginning to be appreciated and understood. Climate change has moved the physical Earth system into conditions that are unprecedented in recorded history, and we do not understand the complexities of the relationships among the elements of the system well enough to make skillful forecasts of specific events. It is therefore prudent to expect that the likelihoods of such events occurring will increase in the coming decade in most places and that the rate of change will continue subsequently to increase. Models are an increasingly powerful tool for examining the likelihood of changes in means, extremes, and variability of climate.
It makes sense for the intelligence community to apply a scenario approach in thinking about potentially disruptive events that are expectable but not truly predictable. For example, available climate models sometimes disagree about the direction of a climate trend even when the fundamental science strongly suggests that change is likely. In such situations it may make sense to consider the security implications of two or more plausible trends as a way to anticipate risks. It will also be valuable for the intelligence community to have improved forecasting ability for subcontinental climate events and for event clusters and sequences. An improved monitoring of factors that might provide early warning of potentially disruptive events would also be valuable. We discuss monitoring issues in Chapter 6.
Recommendation 3.1: The intelligence community should participate in a whole-of-government effort to inform choices about adapting to and reducing vulnerability to climate change. It should, along with appropriate federal science agencies, support research to improve the ability to quantify the likelihoods of potentially disruptive climate events, that is, single extreme climate events, event clusters, and event sequences. A special focus should be on quantifying risks of events and event clusters that could disrupt vital supply chains, such as for food grains or fuels, and thus contribute to global system shocks.
This research should include efforts by climate scientists to improve fundamental understanding of the effects of climate change on the likelihoods of extreme climate events and also efforts to apply the methods of extreme value statistics to these problems, particularly the problem of estimating the likelihoods of clusters of extreme climate events that are dependent on the same underlying climatic processes. Having improved likelihood estimates for single and clustered extreme climate events would help in defining climate event scenarios for countries, regions, and systems that could be used as the basis for the climate stress tests that we discuss in Chapter 6.