The preceding chapters, following the conceptual framework presented in Chapter 2, examined and evaluated evidence about the relationships and mechanisms that could link climate change and climate events over the next decade to outcomes of importance to U.S. national security. This chapter draws upon that analysis to address a core element of the project’s statement of task: identifying “variables that should be monitored and ways that indicators of climate change, impacts, and vulnerabilities might be developed and made useful to the intelligence community in assessing climate-related threats to U.S. national security.” Our premise is that the intelligence community needs a monitoring system capable of (a) supporting a continuing series of assessments of the likelihood and nature of security threats arising as a result of climate events in combination with other conditions, (b) informing timely preventive measures, and (c) supporting emergency reaction. An effective system must integrate information about climate, social stress, and response variables. It must be based on carefully considered priorities regarding the variables to be included, provide sufficiently high-resolution measures in space and time for critical locations and systems, and be actively and continuously managed and improved.
It is important to note that this study deals only with the opportunities provided by monitoring based on open-source materials. We recognize that the intelligence community also has access to classified information, but we do not have a basis for assessing how that relates to open-source materials.
We reemphasize that, as discussed in Chapter 1, there are important factors this study did not address that will also affect climate–security connections. In particular, countries adopt policies that may interact with
climate events in ways that create security threats. For example, U.S. agricultural price interventions, as in use of corn for fuel, can have a direct impact on food prices in fragile societies that can be amplified by climate events that reduce agricultural production. Decisions to protect a country against climate events can also create or accelerate crises; for example, a unilateral decision by Turkey regarding the management of headwaters for the Tigris–Euphrates system could immediately generate crisis conditions. Another example is climate geo-engineering, which is already attracting considerable attention (Royal Society, 2009) and may become a source of conflict. In addition to the monitoring capacity we recommend here, the intelligence community will need the means to monitor, understand, and make forecasts concerning such developments.
The analysis in the previous chapters, developed from our conceptual framework which was informed by an analysis of available knowledge, indicates that from a natural security perspective the climate events of most concern are those that would create the equivalent of a perfect storm: a country or region of importance to U.S. national security that experiences an extreme climate-related event or the effects of a climate-related shock to a global system that meets a critical need, that has significant human and economic assets in harm’s way, where those assets are highly susceptible to harm, where local coping ability is static or decreasing, and where official response systems prove to be ineffective. In order to assess the likelihood that a country or region will experience a conjunction of these factors, all of these dimensions of security risk need to be monitored, that is, data about all these phenomena need to be repeatedly collected and examined.
Conclusion 6.1: Monitoring to anticipate national security risks related to climate events should focus on five key types of phenomena:
- Climate events and related biophysical environment phenomena;
- The exposures of human populations and the systems that provide food, water, health, and other essentials to life and well-being;
- The susceptibilities of people, assets, and resources to harm from climate events;
- The ability to cope with, respond to, and recover from shocks; and
- The potential for outcomes of inadequate coping, response, and recovery to rise to the level of concern for U.S. national security.
In the domain of climate and biophysical environment variables, it is particularly important to monitor and estimate the likelihood of potentially
disruptive climate events occurring in countries and regions of security importance to the United States or affecting global systems that meet critical needs in those places. Climate science provides considerable expertise for identifying, monitoring, and estimating the likelihood of single disruptive physical events occurring in particular places. Other kinds of science are needed in conjunction with climate science to define the monitoring needs for events that are more than just physical, such as climate-driven increases in food prices or outbreaks of infectious disease. These other types of science are also important for defining methods for monitoring and anticipating clusters and sequences of potentially disruptive events that might affect particular regions of interest and for considering the potential for climate events to generate shocks to integrated global systems of potential national security importance.
To monitor exposures to potentially disruptive events requires an understanding of where events are likely to occur as well as of who and what is or will be present in those places. Climate science can tell us what to monitor to foresee where potentially disruptive events may occur; social sciences that focus on population dynamics, economic development, and migration can tell us what to monitor to foresee what will be in harm’s way. Several of the social, economic, and political conditions that contribute to exposure can be projected with some confidence based on available data; of those that cannot, many can be monitored at the country level and below.
Monitoring the degree and nature of susceptibility to harm from climate events should focus especially on places and systems of security concern. It needs to consider different susceptibilities to different kinds of events as well as differences among populations separated by place or differentiated by class, race, ethnicity, religion, or other social cleavages.
Monitoring the ability to cope with, respond to, and recover from shocks requires measures or assessments of limitations in the capacity of affected people, communities, or sectors to cope on an informal basis as well as limitations to the ability or willingness of governments and other formal assistance organizations to respond after an event occurs. It also requires measures or assessments of the likelihood that responses to disruptive events, particularly by responsible governing authorities, will be (or be perceived to be) inadequate. Past performance in natural disasters may provide useful indicators for most of these factors; indicators of corruption or favoritism in the delivery of public services are particularly relevant for the last. Social science offers a variety of tools and methods for monitoring aspects of susceptibility, coping, response, and recovery. Normal techniques of intelligence analysis are also useful for assessing some of these components of vulnerability, such as the willingness of governments to respond on behalf of only particular segments of their populations in the event of need.
Monitoring the potential that inadequate responses will rise to the level of concern for U.S. national security entails estimating and assessing the ways security conditions in countries and regions of interest could be affected by climate events. A major potential link involves a combination of susceptibility and inadequate response leading to a humanitarian emergency, violence, or political instability. Climate events that disrupt the lives of affected populations are more likely to lead to larger upheavals when the events are serious, when governments underperform expectations in responding, when there is pre-existing dissatisfaction with the government, and when there are organized opposition groups positioned to use dissatisfactions as an opportunity to mobilize confrontations with authorities. The monitoring of many of these security conditions is a standard intelligence function and is related to the monitoring of state fragility.
Monitoring of these five types of phenomena would provide valuable input for national security analysis. Given that security threats arise from combinations of all of these, indicators and monitoring systems should be developed to follow them at various levels from local to national. Monitoring will also need to take into account the fact that some of the above types of phenomena are specific to certain kinds of climate events (e.g., flooding), while others, such as the capacity of emergency-response organizations, have an effect on consequences for many different types of hazards. As a rough generalization, exposures tend to be hazard-specific (e.g., some populations are exposed to coastal flooding and storm surge, others to inland drought), which implies that the monitoring for exposures should be differentiated by hazard type. Coping and response factors (e.g., funding and organizational effectiveness of disaster-response agencies) tend to be much less hazard-specific. Susceptibilities can be either hazard-specific or general. For example, the health status of a population provides an indicator of susceptibility to a variety of stresses (e.g., diseases and food shortages), whereas some attributes of infrastructure reduce susceptibility to only single hazards (e.g., to floods but not wildfires).
It is important to note that although most of the phenomena of all the types we have highlighted normally change on time-scales of months, years, or decades, potentially disruptive climate events often give far less warning. Monitoring of the slower-moving factors makes it possible to use a scenario approach for considering the consequences of rapidly appearing climate events. In this approach, analysts posit the occurrence of a particular potentially disruptive climate event the risk of which is high or increasing and consider how a country, region, or system of interest would likely respond, given what is known or expected from monitoring and assessment of the state of other environmental conditions, exposures, vulnerabilities, and likely responses to inadequacies of coping, response, and recovery. Monitoring and assessing these slower-moving variables will enable analysts to
apply a sort of stress test to target countries, regions, or systems in order to anticipate the social, political, and security consequences that could arise if these events affect them. We return to this idea later in the chapter.
Each of the five types of phenomena highlighted above encompasses many specific elements or variables that could be monitored. Appendix E presents and discusses a wide range of these. A major challenge of monitoring lies in setting priorities: determining which data and kinds of data are most needed. This task involves determining which of the many possible measures of a factor such as exposure to coastal flooding, susceptibility to malaria, or effectiveness of disaster response will be most reliable, valid, and useful as part of a larger monitoring system for assessing security threats.
Setting Data Priorities
Because of the multifaceted nature of the phenomena that might connect climate events to national security concerns and because of the complexity of these connections, setting priorities for monitoring is a significant challenge. Substantial ongoing monitoring activities may prove useful for measuring aspects of the key phenomena. Many are already being carried out by the intelligence community and other parts of the U.S. government as well as by various international organizations. However, much of this work has been organized for other purposes. In the area of climate change, the activities include efforts to forecast climate events and estimate the vulnerabilities to climate change of various aspects of human well-being. Outside of the climate change community, monitoring of environmental conditions and changes is carried out largely by environment agencies, and monitoring relevant to exposure and susceptibility is carried out largely by departments and organizations focused on development or disaster assistance. A number of these monitoring efforts are described in Appendix E.
There is strong reason to try to identify a small number of reliable and valid indicators to cover a great variety of phenomena. Some potentially useful indicators are already in use or in development in the U.S. intelligence community and elsewhere, and in many cases they are available in the open literature. It is important to emphasize, however, that the basis for constructing such indicators is quite uneven across variable types and across parts of the world. For example, it is possible to develop a map of the western United States with fairly sharp spatial resolution that indicates the risk for forest fires and related events as a function of projected increases in average temperature (see Figure 3-2), but much less confidence can be
accorded to similar maps in parts of the world where there are not such well-developed databases on past fires and on temperatures.
Less well established is the practice of developing indicators for such things as the coping capacity of communities or the ability of governments to mobilize disaster response. For many of these factors, however, sufficient knowledge exists to identify some of the measures that might constitute an indicator and therefore to begin building and testing indicators. For example, a recent analysis of the determinants of political instability features the effects of political institutions and policies for allocating resources (Goldstone et al., 2010). That analysis identified 141 episodes of political instability that occurred between 1955 and 2003 and demonstrated that 80 percent of them could have been predicted two years in advance using an indicator that combined measures of regime type (degree of democracy), infant mortality, incidence of conflict in neighboring countries, and internal favoritism. The greatest likelihood of instability was associated with political regimes intermediate between democracy and autarchy that practiced discriminatory policies.
Despite progress in developing indicators of several of the key phenomena and the connections among them, it is premature to settle on a small number of variables to monitor that will be sufficient to meet the needs of analysis. Research, including experiments and pilot studies, will be required to determine which measures can serve as proxies for which others and thus to develop an efficient and effective monitoring system. In many instances, existing knowledge does not yet support reliance only on quantitative indicators.
Progress can be made by focusing on each class of phenomena separately. Appendix E illustrates the current state of thinking about data needs and discusses examples of current monitoring efforts for climate and biophysical variables, exposures, susceptibility to harm, and coping, response, and recovery. It shows that across these types of phenomena there are substantial differences in the level of consensus within the relevant communities of experts about which are the key variables from which a small and useful set of indicators could be developed.
An example of a high level of expert consensus, and one approach to consensus-building is the work of the climate science community to identify “essential climate variables” (ECVs). In 1998 the Intergovernmental Panel on Climate Change and the United Nations Framework Convention on Climate Change (UNFCCC) established specific requirements for systematic climate observations and a sustained observing system. The Global Climate Observing System, sponsored by the World Meteorological Organization; the United Nations Educational, Scientific and Cultural Organization; the United Nations Environmental Program; and the International Council for Science, is charged with advising the community on global climate ob-
servations and overseeing implementation based on UNFCCC standards. In 2010 the organization developed a list of 50 ECVs that are possible to monitor globally and whose observation could yield significant progress toward meeting the UNFCCC requirements (Global Climate Observing System, 2010; see Figure 6-1). As discussed further in Appendix E, a number of efforts are under way to use these variables to create a more limited set of indicators that could be more relevant to policy making. This indicates a good level of progress in priority setting. However, to our knowledge there has as yet been no serious effort at priority setting among climate measurements, or among environmental measurements more broadly, for the purpose of informing security analysis.
In other domains in which monitoring is needed, there is variation in the degree to which the relevant scientific communities have defined priorities for monitoring and in the degree to which the monitoring systems are relevant to the climate–security nexus. Two other examples of active monitoring efforts that reflect substantial consensus but also suggest the degree to which such systems learn and adapt by gathering and assessing data come from food security and public health.
An example of an active monitoring effort where experience over time has led to a new understanding of key variables and relationships is the Famine Early Warning Systems Network (FEWS NET). Created in 1985 and funded by the U.S. Agency for International Development, the project is a collaboration among national, regional, and international partners, including expert field personnel on the ground that monitor and analyze relevant data and information in terms of its impacts on livelihoods and markets to identify potential threats to food security. A range of products provide alerts, monthly status reports, outlooks, and in-depth studies based on FEWS NET’s on-the-ground coverage of 23 countries, mostly in Africa, but including Central America, Haiti, and Afghanistan. Less extensive coverage is provided for 15 more countries through partner-based monitoring. FEWS NET, which relies on a combination of physical and socioeconomic indicators to estimate and predict the degree of, and changes in, the food security conditions of vulnerable countries, was one of the earliest users of satellite imagery to monitor rainfall and crop conditions in the developing world. It now also looks at longer-term global climate variability to help assess future threats to food security and inform priorities for climate adaptation activities. One result of its ground-breaking use of food prices and “commodity market networks, market integration, the geographic and economic distribution of food commodities, and cross border trade” (Famine Early Warning Systems Network, 2008:1) was the development of new ways of seeing the nature of socioeconomic threats to food security in some countries—significantly different patterns of food pricing within countries, strikingly divergent global food price influences in different re-
gions, and local prices that track closely with global prices in some places but not in others.
Monitoring and evaluation efforts for public health began more than 150 years ago. Emerging infectious diseases are receiving increasing attention because there is a long history of such diseases surprising societies and causing high morbidity and mortality, sometimes accompanied by social disruption (Lindgren et al., 2012). In response to a presidential directive in 1996, the U.S. Department of Defense established a Global Emerging Infections Surveillance and Response System, with the mission to monitor newly emerging and re-emerging infectious diseases among U.S. service members and dependent populations (Clinton, 1996). The U.S. Armed Forces Health Surveillance Center, Division of Global Emerging Infections Surveillance and Response System Operations coordinates a multidisciplinary program to support the International Health Regulations (IHR).1 The goal is to link datasets and information into a predictive surveillance program that generates advisories and alerts on emerging infectious disease outbreaks (Witt et al., 2011). Datasets and information are derived from eco-climatic remote sensing activities; ecologic niche modeling; and arthropod vector, animal– disease host/reservoir, and human disease surveillance for febrile illnesses. The program includes 39 funded partners working in 92 countries (Russell et al., 2011). Other organizations monitoring emerging infectious diseases include the U.S. Centers for Disease Control and Prevention, the European Centre for Disease Control, the WHO Global Alert and Response network (including the Global Outbreak Alert and Response Network), and the PROMED reporting network at the International Society for Infectious Diseases. These activities could be enhanced to consider how climate variability and change could alter the risks of outbreaks in geographic regions of interest.
Recommendation 6.1: The intelligence community should participate in a whole-of-government effort to inform choices about adapting to and reducing vulnerability to climate change. One of the objectives of this effort should be to build the scientific basis for indicators in this domain.
This effort would support activities by the research communities involved in assessing exposures and vulnerabilities to environmental change to identify a relatively small number of key variables relevant to the social and political consequences of climate events. The effort of the climate science community
1 The IHR, adopted in 2007, establish a mandatory reporting system by all 194 members of the World Health Organization for events that may constitute a public health emergency of international concern.
to identify a small number of “essential climate variables” suggests the kind of process that could be used. A recent effort by the National Research Council (2010b) took an initial step in this direction. The recommended effort might identify sets of variables to monitor that could become elements of indicators of exposures to such events, susceptibilities to harm, and of the likely effectiveness of coping, response, and recovery efforts at the levels of communities, countries, and systems that support critical human needs. It would also support research to develop and validate indicators of key phenomena linking climate and security, as has been done with research on the phenomena of political instability.
The Role of Quantitative Data
Quantitative indicators that combine multiple datasets are often highly useful for giving decision makers a broad picture of a phenomenon of concern. For example, composite indicators are sometimes created to summarize knowledge about various phenomena of interest, such as drought, susceptibility to damage from flooding, health status of a population, emergency coping capacity, or political instability. Developing such indicators for the full range of security threats related to climate change presents a major challenge for several reasons.
Integrating Data Types
Observations relevant to climate-security linkages may come from a great diversity of sources: scientific instruments such as space satellites or ground-based sensors, censuses or surveys conducted by governments or nongovernmental organizations, scanning of communications on mass media or the Internet, and on-site expert observation of qualitative aspects of social and political systems, among others. Some of these data sources are quantitative and others qualitative. Of the quantitative indicators, some are already well calibrated and validated, others much less so. For example, data on some socioeconomic factors, such as demographics and gross domestic product, are routinely collected by well-developed methods. Other data, such as on the ability of a region or country to cope with an extreme event of a particular magnitude or on the condition of resource stockpiles, are often collected through surveys. Because the design, conduct, analysis, and archiving of survey data can be time-consuming and costly, it would be prudent to determine in advance what types of data are likely to be needed and how often the survey should be repeated for critical information, taking into consideration how data might be relevant. This not only would be good planning but also could offer opportunities to identify surveys already being conducted to which additional components could be added.
Monitoring systems will require the integration of quantitative indicators of both environmental and social phenomena, qualitative data on social systems, and information gathered by traditional security and intelligence analytic methods. Integration will be necessary in part because different methods are necessary to gather different kinds of information. Some important kinds of information, such as how governments are likely to respond to disasters, are difficult to collect because many governments will be responding without elaborate advance planning or training. Also, different methods of validation are appropriate for different kinds of data and information (see, e.g., King et al., 1994; George and Bennett, 2005; Brady and Collier, 2010). All methods should be used to gain insight. Where the same kind of information can be obtained by multiple methods, this situation creates an opportunity to use each method as a check against the others and thus increase overall confidence in assessments. We note that it may be possible to advance the objectives of both analysis and risk reduction if information gathering is done through open dialogue, sometimes with the assistance of the U.S. government.
Developing the needed broad monitoring system for climate-related security threats will also require integration of climate science, various branches of social science, and security analysis. The integration of the social science of natural disasters and disaster response with other forms of analysis will be particularly important for assessing the security consequences of climate change because many disruptive climate events will be perceived and responded to as natural disasters.
Judgment is involved in creating indicators, even when they are built on highly reliable observations. Expert assessment is needed of the accuracy and validity of indicators and of whether other relevant information should be taken into consideration. For example, weather forecasts are only partially based on the vast amounts of data analyzed by highly sophisticated computer models; skilled meteorologists modify the forecasts based on an understanding of the complex weather systems that extends beyond what can be coded into a model and of the performance of a particular computer model for a weather variable in a specific region.
Coverage and Resolution
Because of the various purposes for which data collection has been organized, existing data may or may not have the coverage or offer the degree of spatial or temporal resolution needed to track and analyze the key variable sets in ways needed for security analysis. In many cases, data need to be collected at higher resolution than in the past in order to support improved analytic assessment. High-resolution monitoring will be especially important for highly significant and highly vulnerable locations.
The appropriate level of spatial and temporal resolution for indicators varies, however, with the substantive domain. Some indicators, such as of the capacity of national governments to provide response and recovery support after a disruptive event, are most appropriately measured at the national level. Others, such as the risk of coastal flooding, may need much finer resolution, especially in areas of dense population. It may be necessary to develop indicators of community coping capacity separately for communities defined by geography and for communities defined by business relationships or cultural similarities. Similarly, risk indicators may require more frequent updating for some climate events than for others. The need for temporal resolution is probably greatest during the development of a slow-moving climate event, such as a drought, and in the immediate aftermath of a disruptive climate event.
For many existing and potential indicators the required spatial and temporal resolution is finer than what is currently available. In setting priorities for indicator development and improvement, the intelligence community should take into account the gaps between the existing and the desired resolution and should invest in improved resolution for those indicators judged to be the most needed and the most useful in places of concern. When considering how to invest in the development of indicators focused on a particular country or region, it will be worth considering the extent to which an indicator could be applied elsewhere. It should also be kept in mind that existing datasets may provide—or may be analyzed to provide—useful information. Over time, determining the needed data coverage and resolution tends to proceed from an initial assessment of the main data needs and to evolve as the monitoring system is used.
Validation involves determining the extent to which an indicator actually measures the phenomenon it purports to measure—a task that can be extremely challenging. Validation is a long-term process, especially for measures associated with the likelihood and consequences of disruptive events, which are almost by definition uncommon. A measure of the likelihood of a climate event that occurs only once in several decades may require a century or more for full validation, so validation efforts might involve using one indicator to validate another. For example, a projection of drought risk from climate models might be tested against a drought severity indicator and validated even in the absence of extreme drought events. A measure of response capacity may have to wait for validation until a disruptive event occurs, but if it holds up in two or three disruptive events, confidence in its validity should increase. As already noted, the validation of indicators for many of the factors linking climate change and security will not be a
mechanical exercise but rather will likely require considerable judgment for some time to come.2
It is important to develop and validate monitoring systems now in order to have baseline data for future studies of climate events, for the associated exposures and elements of vulnerability, and for social and political stress analyses. Open data sharing, information regarding source codes, and transparency in the analyses are also essential elements of this process. Validation is particularly an issue with emerging monitoring technologies, such as those involving sophisticated data-mining algorithms (e.g., of Internet postings) or remote observations that are overlaid on geographic information systems. Such techniques may produce outputs that catch the eye and are very impressive on first glance, but they are sometimes closely held by their developers and difficult to validate, especially if they involve infrequent events. Indicators and monitoring results should be interpreted with caution until these techniques develop a record of validation.
Improvement of Indicators and Analytic Techniques
Each country presents its own unique mix of exposures, vulnerabilities, socioeconomic conditions, and so forth, and different kinds of climate events have different associated patterns of exposure and vulnerability. There is no formula to tell intelligence analysts where to focus monitoring efforts and which variables are most important to monitor.
As climate change proceeds and its human implications continue to be experienced, a clearer understanding is likely to emerge of the mechanisms linking climate events to security concerns and, therefore, of the most important things to monitor. The needed monitoring will be a major undertaking over an extended period, which should avoid duplication of
2 Several persistent and sometimes specialized statistical issues will also need serious attention as part of the development of information that will be useful to the intelligence community. One of particular importance for this study, already discussed in Chapter 3, is dealing with rare events, which introduce a variety of subtle, although well studied, statistical problems. Another is the widely recognized but difficult problem of making estimates from data of widely differing quality when that quality varies systematically rather than randomly, for example, across regions, types of government, and level of economic development. A third is the sophisticated use of statistical controls—both classical control variables and the newer methods for matching cases—to improve estimation. Closely related to this is the application of quasi-experimental design methods and identifying cases where these are appropriate.
More broadly, as the Political Instability Task Force forecasting tournament indicated, there is a need to identify the relative merits of the competing data analysis approaches currently prevalent: frequentist (relying on significance testing); Bayesian (using probability distributions); and machine learning (pattern recognition, broadly defined). Related to this issue is the question of applying an assortment of new, computationally intensive ensemble methods that integrate the results of multiple models, such as Bayesian model averaging.
effort and strive for maximum amounts of synergy and complementarity. Existing databases should be identified and consulted before new ones are developed; feedback should be provided to clarify what indices are most and least useful; and all datasets should be as transparent and accessible as possible so that analysts working on all dimensions of climate change can rely on the work of colleagues and use it to address their own questions of greatest interest—in this case, questions about security issues.
As important as data and monitoring are for assessing the effects of potential climate events on national security concerns, data in the absence of effective analytical techniques to process them and produce useful information are of little help. In some situations, more data could actually make the situation worse in terms of producing useful forecasts. Some of the major challenges are associated with two questions.
What analytical techniques are most appropriate for the growing number of “big data” approaches? Advances in computing power and new developments in data mining have the potential to allow the intelligence community to gather more data about more places more quickly than could be imagined even a few years ago. Perhaps the best example of this challenge is the explosive growth of different forms of social media via mobile technology, which is yielding, essentially in real time, potential new forms of data about political and social trends. The development of sophisticated algorithms is making it possible to perform machine-based analysis of data from social media and other kinds of events data instead of relying on the traditional laborious coding by individuals; meanwhile, advances in translation software enable the monitoring of many more sources of information from all forms of media.
Applying these new technologies will require successfully addressing issues of data quality and reliability, interoperability across databases and different forms of data, risks of false positives, and the sorting out of meaningful signals from large amounts of noise. At present it has to be acknowledged that the capacity to acquire data in many cases exceeds the analytical capacity to translate the data into useful information that can improve understanding of trends in key countries or regions. For example, the development of analytical—as distinct from visualization—methods for social network and geolocated data is in its infancy, compared with techniques for the analysis of older forms of data, such as econometric or survey.
How much data are really needed? Contrary to the assumption that more is better, with statistical analysis, additional variables may simply duplicate what other data already provide or in some cases actually decrease the accuracy of a model because of poor data quality or other statistical quirks. The pursuit of parsimony, or the “reduction of dimensionality,” is important in numerical analysis. The State Failure Project,
the predecessor to the current Political Instability Task Force (PITF), began working with a set of some 700 potential variables and over time reduced its final model to 4. (See Box 5-1; a list of many of the variables examined may be found in Goldstone et al., 2010.) Another example of reducing dimensionality can be found in the work of Cutter and colleagues (Cutter et al., 2003; Cutter and Finch, 2008) to develop an indicator of social vulnerability to environmental hazards. Using spatial and county-level data in the United States on 30 variables relevant to vulnerability (e.g., wealth, employment structure, demographic composition, and other factors known to influence a community’s susceptibility and response capacity), they determined that the variables could be represented by 7 underlying principal components, which were summed to create a single value for each county of a social vulnerability index (SOVI; Hazards and Vulnerability Research Institute, 2012).
Conclusion 6.2: Developing an adequate system for monitoring the conditions that can link climate events to national security concerns will require maintaining critical existing observational systems, programs, and databases; the collection of new data; the analysis of new and existing data; and the improvement of analytic systems, leading to a better understanding of the linkages over time and to improved indicators of key variables where quantitative indicators are appropriate and feasible to produce. It will typically require finer-grained data than are currently available. It will also require improved techniques for integrating quantitative and qualitative information.
We emphasize that improved understanding and monitoring of the various elements of climate vulnerability—a key link between climate events and security concerns—is an objective that the intelligence community shares with the U.S. Global Change Research Program (USGCRP) and many other institutions at federal, state, local, and international levels.
The intelligence community cannot address these challenges alone. Addressing many of the new and enduring methodological problems is largely the province of the academic research community. The intelligence community needs to draw on this knowledge, as efforts like the PITF are doing, to address the interactions of climate events with traditional intelligence community concerns.
The United States, like other countries, lacks a national strategy for sustained, long-term observations for the purpose of informing analysis of relationships between environmental changes, including climate change,
and national security. Multiple environmental monitoring activities and programs exist, organized by both public and private actors; they have diverse purposes and are focused on conditions and processes that range from local to global. Only a few of them are organized to inform security analysis, however, and it is difficult to know how useful the others might be for that purpose. Efforts to develop environmental observation priorities for security analysis should focus on identifying a small number of composite indices designed for specified purposes of analysis or early warning. The same can be said for social, economic, and political observations: Multiple monitoring programs exist, with diverse purposes, and only a few of these are organized to inform security analysis. Organized efforts at indicator development for climate–security analyses remain works in progress. Yet systematic efforts are needed. Progress will require additional work, which should be conducted through collaborations involving climate scientists, environmental scientists, social scientists, and security analysts.
The intelligence community should adopt a risk-based strategy for setting its monitoring priorities. Such a strategy seeks to prioritize the measurement and assessment of the most significant expected security risks that may arise from conjunctions of potentially disruptive climate events; exposures; susceptibilities; limitations of coping, response, and recovery; and the reactions to revealed limitations. A strategy that is risk-based considers the product of the likelihoods of events and the magnitude of their consequences. However, because the likelihoods of key events—and even in some instances the nature of the events—are not well known, monitoring under a risk-based strategy is not an exact science and must be expected to evolve as research and monitoring activities improve understanding of which conditions are most important to monitor and provide increasingly valid estimates of the probabilities and consequences of key events.
Threat Monitoring as a Long-Term Research Activity
Developing a monitoring system for climate-related security threats is a long-term enterprise. As noted above, a considerable amount of effort is already being devoted to monitoring climate events and trends; some aspects of food, water, and health security; risks of natural disasters related to climate change; and certain elements of disaster response capacity by a variety of governmental, nongovernmental, and international organizations for various purposes. Such existing monitoring systems, both open-source and commercial, should be periodically scanned for potential usefulness, but with critical attention paid to indicator selection, data reliability and validity, and cross-case and cross-national comparability.
As we have also noted, the connections between climate events and national security concerns are complex and contingent, with many plau-
sible combinations of climatic events with social, economic, and political conditions that might create risks to U.S. national security. These risks are unlikely to be foreseen by looking only at climate trends and projections or by looking only at political and social trends and projections. To anticipate the risks, analysis needs to integrate three kinds and sources of knowledge: (1) knowledge of political and socioeconomic conditions in countries of interest, (2) knowledge from climate science about the potential exposure of these countries to climate events, and (3) knowledge from social science about the susceptibility of these countries to be harmed by those events and the likelihood of effective coping, response, and recovery at local to national levels. These sources of knowledge come from different communities of experts, which will need to communicate with each other. Making this happen will take time and continued effort.
Indicators based on monitoring efforts can be used even while research and development on them is in an early phase if they are interpreted cautiously as one source of insight among many, including qualitative insight derived from on-the-ground information and experience. Open-source monitoring efforts can help reduce the risks of climate change by helping national and international decision makers anticipate potentially disruptive events and reduce vulnerabilities. Monitoring efforts by the U.S. intelligence community may also have such broader benefits.
Efforts to develop quantitative indicators need to be improved over time to maximize their usefulness for security analysis, and achieving this goal will require a long-term effort with a significant research component. As such indicators are developed and validated, it will become appropriate to assign more weight to the information and predictions they provide. The intelligence community should consider the development of the needed indicators to be a long-range research activity.
A research investment in indicator development is likely to increase in value over time, both because monitoring systems are likely to improve through continued efforts and because potentially disruptive climate events are expected to increase in frequency and intensity in years and decades to come. It is therefore important to begin now to build and test the capability to monitor and anticipate climate-driven security threats. The potential for disruptive events, the elements of vulnerability, and security conditions will all need continued monitoring because they are all changing and can affect each other. For example, responses to recent climate events or other disasters can affect both the future capacity to respond and security-related conditions, such as public support for governments. The research effort needs to integrate monitoring across variable types and methods and should focus on validating indicators, monitoring the appropriate spatial and temporal resolution, and improving analytical techniques, particularly to make effective use of rapidly increasing volumes of data.
The Need for a Whole-Government Approach
Recommendation 6.2: The U.S. government should begin immediately to develop a systematic and enduring whole-of-government strategy for monitoring threats connected to climate change. This strategy should be developed along with the development of priorities and support for research as recommended in Chapters 3, 4, and 5.
The monitoring should include climate phenomena, exposures and vulnerabilities, and factors that might link aspects of climate and vulnerability to important security outcomes, and it should be applicable to climate issues globally. It should also include making and periodically updating priority judgments about when and where high-resolution monitoring is needed.
The recommendation for a whole-of-government approach is consistent with the recommendations of the Defense Science Board (2011) and the strong convergence of the climate change monitoring objectives of the intelligence community as discussed here and those of the USGCRP. As noted in previous chapters, these interagency enterprises have many common needs for monitoring and for the fundamental science that informs monitoring choices, but their efforts are not integrated. As the recent National Research Council (2012a) review of the USGCRP strategic plan noted, “An effective global change research enterprise requires an integrated observational system that connects observations of the physical environment with a wide variety of social and ecological observations. Such a system is a crucial foundation for identifying and tracking global changes; for evaluating the drivers, vulnerabilities, and responses to such changes; and for identifying opportunities to increase the resilience of both human and natural systems” (National Research Council, 2012a:39). Monitoring for the purposes of the intelligence community has the same requirements, although information will be used differently because of the need to focus on threats outside the United States. It makes sense for these different interagency communities to collaborate on the scientific analysis required to design the needed monitoring and assessment systems and, as appropriate, on the development and use of these systems.
Organized international collaboration with potentially affected societies and governments and open sharing of data will be important aspects of developing the needed monitoring systems. A monitoring system capable of anticipating and detecting severe instances of climate-induced social and political stresses in many countries would be of great value not only to the U.S. national security community, but also to the affected countries themselves to guide anticipatory adaptations as well as to international humanitarian assistance agencies and foreign donors for preparing their response capacities and to security analysts in countries other than the United States
and the affected countries as they consider the security implications of climate events. Such a monitoring system with open sharing of data would thus provide a global public good. The U.S. government would also benefit from data-gathering efforts in and by other countries.
Open, international scientific collaborations are also desirable on scientific grounds. The development of compatible concepts, databases, and indicators across countries helps speed scientific progress and improves the ability to learn from experiences in other countries.
International collaboration is likely to be necessary to achieve acceptance of higher-resolution monitoring at critically vulnerable locations, particularly if that monitoring requires an on-site component. Such a system would inherently include elements that could be seen as intrusive in the countries being monitored. Thus its global acceptability would depend on justifying its purposes and legitimizing its rules. In particular, such a system would have to be credibly directed to broad common interests rather than intended to provide some competitive national advantage that might be perceived as hostile. If the capacity of a society to manage internal stress is to be subjected to detailed scrutiny, the motivating purposes must be accepted as constructive, access to the data must be equitable, and the benefits derived must be mutual. Given the historical legacy of security concerns, those conditions will not be easy to achieve, but they will certainly be essential.
As a practical matter these conditions would have to be established though a process of evolution as the details of monitoring arrangements are worked out. A mature system would almost certainly have to be achieved in a series of incremental steps. Nonetheless, transparency would be a central principle from the outset. To the maximum extent possible, both the methods used and the data resulting from a monitoring system must be open to global scrutiny as the best and, ultimately, the only way to establish legitimacy and to assure accuracy. That does not mean that access would be completely unrestricted. It means rather that the rules of access would be based on criteria that are broadly accepted at the outset, universally accepted in a mature system, and subject to collective reconsideration over time.
Of course, U.S. government agencies will continue to need to gather some kinds of information that will not be openly shared, and there will be questions about which data- and information-gathering methods can and should be openly shared. There will also be suspicion of the involvement of U.S. intelligence agencies in international information-gathering efforts related to security. Such issues will need to be addressed in ways that we have not had the opportunity to consider in this study. Nevertheless, the benefits of open, international data development and sharing should be taken seriously as work on monitoring systems proceeds.
Recommendation 6.3: The intelligence community should establish a system of periodic “stress testing” for countries, regions, and critical global systems regarding their ability to manage potentially disruptive climate events of concern. Stress tests would focus on potentially disruptive conjunctions of climate events and socioeconomic and political conditions.
The intelligence community presumably already uses an analogous process to consider the ability of foreign governments and societies to withstand various kinds of social and political stresses. This recommendation calls on the community to incorporate climate risks and the associated exposures and vulnerabilities into such exercises. The concept of a climate stress test provides a framework for integrating climate and social variables more systematically and consistently within national security analysis.
A stress test is an exercise to assess the likely effects on particular countries, populations, or systems of potentially disruptive climate events to which they have some likelihood of exposure in the coming decade. The recommended stress tests would involve analyzing the likely effects of an event at some projected time of occurrence in terms of key variables affecting susceptibility, coping, response, and recovery or the failure thereof, and the likely responses within regions or countries of interest in the event that these actions are perceived to be inadequate. The tests would draw on knowledge about the potential events and each of the other types of phenomena and would provide a major way of making knowledge about climate events, exposures, and vulnerabilities operational in security analysis.
Stress tests should consider two kinds of climate events of potential security concern: those that climate scientists can say with some confidence are increasingly likely to occur or become more severe, and those that seem increasingly likely to occur based on a fundamental understanding of climate dynamics but about which available evidence is not yet sufficient for climate scientists to attach confidence to such projections. Stress tests might also be triggered by assessments indicating that event likelihood, exposure, or susceptibility is increasing or that the capacity to respond adequately to certain kinds of climate events is declining in a region or country of concern.
The results of stress tests would inform decision makers about places that are at risk of becoming security concerns as a result of climate events and could be used by the U.S. government or international aid agencies to target high-risk places for efforts to reduce susceptibilities or to improve coping, response, and recovery capacities. The stress testing process would also help examine and refine hypotheses, such as those presented in Chapter
5, about the characteristics of climate events and of the affected societies that determine whether or not potentially disruptive climate events turn into security threats. Over time an accumulation of data on potentially disruptive events and their social, political, and security consequences will improve understanding and feed back into improved monitoring processes and improved skill in stress testing.
Countries, regions, and systems of particular security interest should be prime targets for periodic stress testing. Given the joint criteria of significant potential for climate change impacts and importance to U.S. national security, it is likely that no more than 12 to 15 countries will need to be monitored and subjected to periodic stress tests over the next decade, many of which are likely to be in critical, and often shared, watershed areas in South Asia, the Middle East, and Africa. If the criteria for importance to the United States are expanded to include foreign policy and humanitarian concerns, the number of countries to be monitored and stress tested regularly over the next decade may rise to between 50 and 60. Stress testing should also be applied periodically to global systems that meet critical needs, including food supply systems, global public health systems, supply chains for critical materials, and disaster relief systems, as well as to international emergency response systems.
Decision science techniques should be used and further developed to ensure that the stress tests make the best use of the available information. Stress testing might draw on various methods, including qualitative interpretation of available knowledge, formal modeling, and interactive gaming approaches. Research analysts, area experts, and others might contribute in various ways, such as conducting analyses, developing models, and playing roles in gaming exercises. Decision science techniques should be employed to design the processes and interpret the input from different kinds of expertise and modes of analysis in order to make the best possible use of information. The stress-testing exercises should themselves be monitored and critically evaluated so that stress-testing methods can be improved over time.