Multisource Information Fusion, Situation Assessment, and Course of Action Selection
The first panel session on Day 2 featured three presentations on planning for and responding to an unanticipated rare event. The three speakers were Alice Hill, the David M. Rubenstein Senior Fellow for Energy and the Environment at the Council on Foreign Relations; Delores Knipp, research professor at the University of Colorado Boulder; and John Organek, director of operational architecture at the Electric Infrastructure Security Council. Vicki Bier, professor emerita at the University of Wisconsin–Madison and planning committee member, moderated a discussion period following the three presentations.
OVERCOMING FAILURES OF IMAGINATION
Alice Hill noted that although this workshop’s focus is on rare events, such as pandemics and extreme weather events—the subjects of her recent research— these are no longer rare. It is notable that infectious disease experts have been predicting that the world was due for a pandemic, but even after close calls with Ebola, AIDS, SARS, and MERS, the world was not prepared for COVID-19. Similarly, climate scientists have been raising the alarm about climate change and the dire consequences it would have, but policy makers failed to take this warning seriously until the growing number of extreme weather events provided a taste of what the future holds without action.
Hill explained there are many cognitive reasons for why humanity keeps getting caught unprepared for these events that are no longer rare, with a major reason being that humans suffer from a failure of imagination. She remarked that behavioral economists have identified cognitive biases that impede human decision-making and lead to flawed decisions. “We need to figure out how to get over our innate optimism, which causes us to believe that climate change impacts will not particularly affect us, or our tendency to judge risk based on our own experience,” said Hill. “The question is, how can we overcome these biases? Because they have meant that we are not ready even though the predictions are clear that these events will occur.”
Two means of overcoming a failure of imagination are to use scenarios and to create more usable and publicly available data. Scenario analysis, Hill explained, uses defined stories to examine possible outcomes for the future. Given that future events may not resemble what has occurred in the past, scenarios can stimulate and discipline the imagination. Originally developed for military planning, scenario analysis is now a part of the planning process for a range of industries, including the fossil fuels industry. The Shell Corporation, taking this approach early on, created a climate change scenario in 1998 centered on a huge storm pounding the U.S. Atlantic coast so badly that it sparked young people into climate activism, the government into regulating fossil fuels, and litigation over who
pays for the damages caused by climate change. “We have not seen that yet in a big scale, but we have certainly seen elements of what they imagined,” said Hill.
Climate scientists’ use of scenarios has underpinned their analysis of how different levels of emissions over time could affect projected climate impacts. Researchers have also used scenarios for pandemic planning. Hill explained that a 2006 special report in the Harvard Business Review1 outlined the risks of a pandemic and the planning needed to respond to a pandemic, information that the business world, or policy makers for that matter, did not use to prepare for the current pandemic. This report recognized that a pandemic differs fundamentally from other more traditional business continuity risks in that it is not a discrete event, but rather an unfolding global event. The authors of this special report recommended scenario planning for gauging potential effects on product demand, corporate operations, and company financials.
The COVID-19 pandemic, said Hill, has triggered a new appreciation of scenario-informed planning among companies and governments as they have tried to imagine what the new normal will be like and how soon it might arrive. Similarly, worsening climate extremes have also sparked an interest in scenario-based planning. Investigators in New South Wales, Australia, for example, recommended greater use of scenarios following the devastating summer bush fires of 2019 and 2020. The investigators explained that scenario planning would allow the government to consider the potential for more significant and more extreme events, including worst-case scenarios, to quantify the risk, to understand the preparation required to respond, and to identify the policy or other innovations that could exacerbate or reduce the risk.2
For scenarios to be useful, they must rest on accurate, usable, and relevant data. Hill emphasized that in the climate world, acquiring those data will require an overhaul of how that information is produced. Current climate models, she explained, capture broad trends and do not make localized predictions. However, such predictions are exactly what municipal planners and those running facilities like power stations and wastewater treatment plants need to anticipate potential disasters, ranging from floods in Indonesia, to heat waves in Portugal, to a severe cold snap in Texas.3
Hill recalled a conversation she had with the part-time mayor of Perdido Beach, Alabama. This small town on the Gulf of Mexico is surrounded by water on three sides, and flooding caused by Hurricane Ivan eroded big chunks of the town’s beach front. The mayor told Hill in 2013 that as a small-town mayor, she did not have a planning staff or any resources to know the size of the threats her town faces and the actions the town can take to protect its residents. “Almost a decade later, that information is still lacking about extremes at a particular location, and not surprising, adaptation to the extremes we are already seeing, much less those in our future, has lagged,” said Hill. Indeed, in June 2021 the chair of the Adaptation Committee of the United Kingdom’s Climate Change Committee described adaptation as “under-resourced, underfunded, and often ignored.”4 In short, the research response to date on adaptation has been too little, too removed, and too theoretical. “There needs to be a broader, open shift to apply science to local climate adaptation,” she remarked.
The National Oceanic and Atmospheric Administration (NOAA) reported5 that the United States experienced a record-breaking 20 climate disasters in 2021 that each caused at least $1 billion in damages. Globally, Hill explained, the tally for climate change–exacerbated disasters reached $210 billion in 2020. Investing in risk reduction before disaster strikes, however, can save enormous sums of money. In higher-income countries, every dollar spent on risk reduction saves approximately $6 in damages when done effectively. In lower-income countries, every dollar invested in more resilient infrastructure yields $4 in benefits. An alarming trend in the United States, however, is that migration is strongest into some of the nation’s regions with the highest risk for climate-associated disasters, such as flooding and fires.
According to the Federal Emergency Management Agency, 65 percent of U.S. counties do not have disaster-resistant building codes, even though every dollar spent on constructing a building that meets stronger codes
1Harvard Business Review, 2006, “Preparing for a Pandemic,” May, https://hbr.org/2006/05/preparing-for-a-pandemic.
2 United Nations Office for Disaster Risk Reduction, 2020, Final Report of the NSW Bushfire Inquiry, July 31, https://www.unisdr.org/preventionweb/files/73305_finalreportofthenswbushfireinquiry.pdf.
3 A. Hill, 2021, “COVID’s Lesson for Climate Research: Go Local,” Nature 595:9, https://doi.org/10.1038/d41586-021-01747-9.
5 NOAA, 2022, “U.S. Billion-Dollar Weather and Climate Disasters,” National Centers for Environmental Information, https://www.ncei.noaa.gov/access/billions, doi: 10.25921/stkw-7w73.
saves $11 in damages following a bad event, observed Hill. Equally worrisome, she asserted, is that the United States does not have a model climate-resilient building code available for communities to adopt. She mentioned that without information about where and how damaging events are likely to unfold and how to build resiliently, choosing the right adaptations to invest in can resemble guesswork. However, the current 100 to 150 km2 resolution of climate models is too large for planning purposes because an area that large can span several towns and different risks from extreme events.
It is not just discrete communities that need access to localized climate predictions, she remarked, given the growing vulnerability of supply chains to disruption that extreme weather events and the COVID-19 pandemic have demonstrated. This is not a newly recognized problem, either. In 2011, severe flooding in Thailand closed factories that produce 40 percent of the world’s computer hard drives, which caused prices to double and squeezed computer manufacturers.6 During the pandemic, the fact that China produces 90 percent of a key component needed to make penicillin caused worldwide shortages.
In Hill’s view, investments in localized climate risk information should be a public good. She highlighted the fact that wealthier communities and businesses are already hiring their own expensive consultants to provide tailored climate risk information. For example, a client wanting information about its exposure to hazards such as floods, fires, and extreme heat could pay upward of $1 million for 1 year of services, while a large corporation could pay a much steeper price, she explained. Such for-profit systems leave poor communities without access to the information they need to prepare for climate risks. “Governments must work with academia, nongovernmental agencies, and the private sector to develop publicly available models and tools that give decision makers the basic information they need to keep themselves safe,” said Hill.
High-income nations, including the United States, can catalyze the science of practical climate predictions, just as they did with vaccine development for COVID-19, explained Hill. The need for such tools is great, particularly among low-income countries, and it is time to apply science to develop local solutions to the global climate crisis, she concluded.
SEVERE SPACE WEATHER AS A SOURCE OF RARE EVENTS OF MAJOR SIGNIFICANCE
To remind the workshop of the material that Jeffrey J. Love discussed on the workshop’s first day, Delores Knipp explained that Earth is subject to the whims of the Sun’s atmosphere and the solar wind. Bursts of solar flares on the Sun’s surface and solar wind (high energy, sub-nuclear particles that are ejected from the solar surface) can have a major effect on Earth’s magnetic field and atmosphere. These originate in active regions, which humans have long observed as sunspots. The troublesome events that occur on Earth typically happen in a matter of minutes to 1 to 3 days after a solar flare. Active regions, Knipp described, are nests of complex magnetisms that fuel solar eruptions of energetic particles that can travel at relativistic speeds and penetrate spacecraft and Earth’s atmosphere minutes later, followed by slower waves of magnetized ejected coronal mass. She noted that some of the worst disruptions occur when a series of multiple interacting bursts and ejections reach Earth.
Over the past 11 years, the NASA-European Space Agency jointly operated Solar Heliospheric Observatory has provided images of the active regions. Scientists are trying to use the information that this and other observatories produce to power AI and machine learning approaches to understand which of these active region events will develop into super events. One such event occurred in October 2003 when the flares from the active region ionized Earth’s upper atmosphere for hours.
Another major event, in May 1967, produced powerful X-ray bursts and extreme radio emissions that triggered a blackout of the radar and radio communication systems that served as an early warning system against nuclear attack by the Soviet Union. While the Intelligence Community and combat command officers were trying to determine whether this blackout was an act of war, forecasters in the solar space weather forecasting center in the Cheyenne Mountain Complex—which had opened weeks before—were determining that a solar flare and associated radio burst of unprecedented strength was occurring and was the likely cause of the blackout. From speaking with Russian colleagues, Knipp learned that this incident led the government there to fund solar research for decades afterward.
6 A. Hill, 2021, “COVID’s Lesson for Climate Research: Go Local,” Nature 595:9, https://doi.org/10.1038/d41586-021-01747-9.
A similar situation developed in August 1972 when a strong solar flare produced the fastest ejecta on record. The energetic solar particles blinded the satellites monitoring the Anti-ballistic Missile Treaty, produced strong disturbances in the U.S. electrical grid, and detonated some 4,000 mines that the United States had deployed around Vietnam during the war.
What worries Knipp is the realization over the past 5 years that these extreme energetic solar particle events can be much larger and more expansive, and the particles they produce can in all likelihood penetrate spacecraft. Of particular concern is that these events can occur from active regions as they move to the backside of the Sun. Her worry is that with the increasing number of satellites in low Earth orbit, one of these eruptive events will interfere with the monitoring systems that, among other tasks, track orbiting satellites to prevent collisions with other spacecraft or with orbiting debris. In her mind, such solar eruptions would be weapons of mass destruction for the low Earth orbit environment without ever directly harming a human. Following an event in 1989, for instance, it took weeks for satellite operators to reacquire some 2,000 orbiting satellites and pieces of space debris.
In closing, Knipp warned there are multiple threats from individual solar active regions and that humanity’s increasing level of dependence on technology creates vulnerabilities that might extend to lower levels of space weather storms. At the same time, there is little shared knowledge about these vulnerabilities because much of this information is either classified or kept confidential by companies out of concern for revealing their vulnerability to competitors. In addition, the multi-year lulls between solar cycle maxima mean that operators and decision makers are under-experienced in terms of dealing with these events.
EXPERIENCES FROM BLACK SKY PLANNING
A black sky event, explained John Organek, is an electric grid outage that extends for more than 30 days and that affects a wide area spanning at least one of the three major grid interconnections. There can be both natural and intentional hazards that cause a black sky event, including a small disturbance if the grid reaches what is called the state of self-organized criticality. The blackout of August 2003 was an example of such an event.
The electric industry refers to a black start event as the need to restart a major sector of the grid after a complete shutdown. In Texas, following the February 2021 cold snap, the grid was 4 minutes, 37 seconds from being able to witness such an event, said Organek. Even though the industry has planned and frequently exercises for a black start event, there has never been an occasion to see what happens if it were to be executed. In the Texas event, as a worrying example, only 7 of the 13 emergency generators needed for a black start were available during the cold snap. He added that the communications sector shares much in common with the electric in terms of ubiquity and consequence, and it may be more conducive to study.
The two primary hazards that can cause a black sky event are the geomagnetic disturbances that Knipp and Love addressed and a high-altitude EMP from a nuclear detonation. Organek indicated that other hazards, such as a cyber-attack, a coordinated physical attack, or intentional electromagnetic interference, could further aggravate a black sky event, which fortunately did not happened during the Texas blackout. Such intentional, malicious attacks might also be more conducive to study in terms of being able to anticipate rare events, he said.
A black sky event, Organek warned, would have a major effect on DoD’s mission to project force, both directly and through the cascading effects on other infrastructure, such as communications, transportation, and supply chains, as well as on the military enterprise itself. Fort Bragg, for example, is not only a critical installation for force projection, but it also sustains the equivalent of a city of 250,000 people and is a part of a larger community of people in the surrounding area. Beyond taking care of its own, DoD also plays a major role in helping civil authorities respond and recover, which is an undertaking that will certainly not only affect force projection capabilities, but also affect the internal view of what happens on installations. A black sky event would quickly blur the distinction between these two missions, warned Organek.
Regarding risk, his organization breaks it down into the threat, vulnerability, and consequences of an event to be able to assess risk more accurately. He noted that likelihood and probability play into each of these components. Emergence, Organek explained, is a special category of events that does not fit into the categories of natural and intentional threats. Emergence in the context of the electric grid is a sudden, no-notice collapse of a major portion of the electric grid. The August 2003 black sky event, which affected a large part of the Eastern United States and Canada,
was an example of an emergent event. In that case, the system reached a state of self-organized criticality, and the collapse was triggered by some seemingly harmless event that caused and reinforced failures throughout the system.
Organized criticality, Organek explained, is relevant to what the Electric Infrastructure Security Council refers to as the meta-grid—the tightly connected infrastructures, supply chains, and communities that exist. “Our lives and our society have become nearly completely dependent on this meta-grid, which is not capable of being fully understood,” said Organek. “Though progress is being made in understanding adaptive complex systems and self-organizing criticality, there is still a long way to go in terms of building the capacity to detect the signs of a complex system approaching criticality.” As an aside, he mentioned that the work climate scientists are doing regarding the increasing occurrence of extreme weather events may be applicable to understanding organized criticality and the electric grid.
Valuing the consequences of an event is important because resources are limited, so decision makers need to prioritize actions. Organek noted that NOAA employs an approach to determine and optimize the value of individual programs and the overall program portfolio, which is based on the contribution a program makes to improve a warning about weather events and the overall portfolio. He suggested that this approach is something that DTRA should study.
Turning to information fusion and decision-making, Organek explained that the electric sector has long collected data from critical infrastructures, including the use of time-synchronized sensors that provide wide-area situational awareness, to manage its operations. As modeling of complex adaptive systems and computational capabilities has improved, researchers and grid operators have explored how to use those data to expand the scope and improve the quality of research. Toward that end, grid modelers are developing and refining innovative and more effective models that enhance the operational capabilities that can anticipate, assess, and react to disturbances. Organek expects more of this will be automated going forward. As the grid undergoes a major topological and physical transformation, these models need to transform accordingly. As they do, the insights they provide should prove valuable to other critical infrastructures, including the meta-grid.
Modeling, Organek emphasized, should be supplemented by ground truth and developed independently of the supervisory control and data acquisition system. It should also be informed by the engineers who design and build the physical systems that models are trying to reproduce. “You have to understand the physical processes that are taking place that have been built into the systems,” he explained. Incorporating ground truth throughout the information cycle is vital not only to the security of the grid, but also to moving closer to a goal of being able to anticipate rare events taking place in a complex network of networks, he added.
Too often, Organek expressed, the focus is on monitoring a computer-generated image instead of the real thing. Attackers take advantage of this by exploiting the networks that gather fused data, including the sensors, to look for gaps in detection using adversarial machine learning techniques. They then use this knowledge to disrupt, damage, or destroy equipment that may require a long lead time to replace. Today, none of the cybersecurity protocols for major infrastructure components address the sensors, which they assume are functioning correctly.
Organek concluded his remarks by describing the following risks—gray rhinos, pink flamingos, black swans, and bias toward action. He cited that those risks lead to self-inflicted decision-making wounds.
The term gray rhinos, originally coined by Michele Wucker, refers to high-impact, highly probable, nonrandom, threats that are ignored despite leaving a trail of warnings and evidence. The 2021 Texas power grid outage was such an event.
Pink flamingos refer to known knowns that are often discussed and ignored by leaders who are trapped in organizational cultures and rigid bureaucratic decision-making structures.7 The term was used to describe the failure of leadership to view military installations as combat platforms that support force projection instead of being simply industrial complexes where people work.
Black swans refer to unknown unknowns or an event or situation that is unpredictable as a major effect8 and about which everyone agrees afterward was something for which there were actions that could have and should been taken to prevent it.
7 Mad Scientist Initiative, 2018, “Black Swans and Pink Flamingos,” May 10, https://madsciblog.tradoc.army.mil/51-black-swans-and-pinkflamingos/?replytocom=4029.
Finally, there is bias toward action, which has the opposite effect of a black swan. Bias toward action occurs when operators become too familiar with and overly confident about dealing with an incident and fail to detect a rare event in the making when they take action prematurely. “We call it the ready, fire, aim syndrome,” he stated.
Organek noted that he has been working with DTRA on EMP-related vulnerability, both for power stations and substations. As part of this work, his organization was able to show that if an 1859-sized event occurred, it would not, contrary to common belief, melt everything electronic. “We were able to determine how well the circuit breakers could withstand an EMP or how an EMP would diffuse itself within the structure of a power generator,” he explained.
The discussion began with a question for Knipp about how to prepare for millennial events—that is, those expected to occur once in a thousand years. For a solar energetic particle event, she suggested there should be discussions with satellite designers, mission designers, and operators who have had experiences with spacecraft operating in the Van Allen radiation belts, a known harsh environment. In her opinion, those experiences can provide lessons in terms of what space operators think their systems can handle. While those discussions may be happening in the classified world, that knowledge needs to be shared with the broader audience that would be affected by an outage of the global navigation satellite system.
Hill, responding to the same question, reiterated her earlier comment that many of the events that concern those attending the workshop are no longer as rare as they use to be, which she believes is something that planners and strategists have not yet internalized. The Texas grid failure, for example, was not a surprise event, given that there had been previous freezes. Rather, she asserted, a choice had been made not to be prepared. “I think the psychological and political barriers to us addressing particular events worsened by climate change is holding us back,” said Hill. “I think it is a critical area of study because those events will happen with rapidity.” Similarly, she remarked that building codes, land use choices, and infrastructure investments are still being made based on the assumption that the climate of the past is the climate of the future. “We now have an unstable climate that will continue to become more unstable,” she emphasized. “We are making choices today that are just plain poor.”
As an example, she cited the $1 billion DoD invested in the U.S. Army Garrison on Kwajalein Atoll in the Marshall Islands. When building it in 2013, Knipp stated, that both DoD and its contractor said there was no need to worry about sea level rise. Shortly afterward, they realized there would be sea level rise that would inundate freshwater supplies with salt water.
Organek warned that one of the humbling factors of an extreme event is that past performance is not an indicator of the future. This has been true of the COVID-19 pandemic, where experts continued to think they could make predictions when this virus was something novel. When he served in the military, he learned to use the word forecast instead of prediction because it is impossible to accurately predict anything that has a component of probability. On the other hand, he said, it is possible to prepare to be in a position to better understand an event and take more effective and appropriate action. In his view, one such action would be to make better use of the growing Internet of Things and the sensing information it can generate. It should be possible, he said, to create a monitoring system that is more interactive and more capable of observing infrastructure in real time. The data that networked sensors generate can also power simulations of alternative futures that could inform actions to respond to future extreme events.
Wucker asked Hill to talk about whether responses are included in a feedback loop when considering the potential costs and consequences of an event. As an example, if Texas had been prepared for the freeze, how much would that have reduced the costs and consequences of the resulting power outage? One lesson from the pandemic, noted Hill, is that the economy’s emphasis on efficiency left supply chains vulnerable, in part because that approach failed to build redundancy and resilience into the system so it could respond better when a situation unfolds differently than anticipated. Modelers, she noted, have not been able to identify the cascading compounding effects that follow from weather extremes or rare events. It was clear that the electric grid in California was not ready for the wildfires that have increasingly affected the state, and the grids shut down when there is a major fire, producing a cascading effect on the state’s economy. The Northern California fires caused a major displacement
of people, and 20,000 people have moved into Chico, which had a population of 100,000, resulting in crowded classrooms and roads. Current models, she explained, cannot quickly forecast these cascading events, which results in an underinvestment in preparation and in educating the public to understand that maybe they should not live in a fire-prone area where insurance is no longer available.
A workshop participant asked the panelists if they think that analysts, technologists, and decision makers should receive training in strategic foresight and how to act on past real-life experiences. Hill replied that under the U.S. Constitution, decisions about where and how people build rests with state, local, tribal, and territorial governments. That has created a moral hazard in terms of the choices about where to build and how to build because the local communities want that tax base. Today, local decision makers may assume the federal government will take care of any consequences that develop, and Hill suggested that the federal government could send strong signals that if communities want the maximum economic help after a disaster, they need to invest in risk reduction. In fact, the federal government decided in the 1960s that states are responsible for developments on coastal barrier islands if they want to allow that to happen. Vicki Bier added that individuals are compounding the moral hazard problem through their intolerance of being told where they cannot build.
Christopher Barrett asked the panelists to comment on how the nation can move from simulation and forecasting to action. Knipp replied that in the satellite protection world, the new generation of companies that are launching large constellations of small satellites are less concerned about losses. They assume that they will simply launch a replacement if they lose one satellite, even though destruction of one of their satellites could generate enough debris to trigger a cascade of collisions that would render specific orbital ranges unusable going forward.
Organek explained that the small water systems, one of which experienced a cyberattack in early 2021, do not have the capacity to conduct the analyses and take the necessary preparatory actions to deal with such an event. In his opinion, there is a need for pooling modeling resources, something the Environmental Protection Agency and the U.S. Department of Agriculture do, to help these smaller systems with preparedness. One approach for doing so would be to create a regional grant system that would enable groups of smaller systems to pool resources. He also mentioned that DoD has funding to help communities address some of these preparedness issues.
Bier noted that the social sciences have done research on how better to communicate risk to decision makers. Recently, she recalled, Francis Collins, the director of the National Institutes of Health (NIH), said he wished that NIH had invested more in studying human behavior,9 and a noted sociologist pointed out that sociologists have been saying that for a long time.
9PBS News Hour, 2021, “Dr. Collins Reflects on Career at NIH, COVID Response Effort, Work on Genome Sequencing,” December 20, https://www.pbs.org/newshour/show/dr-collins-reflects-on-career-at-nih-covid-response-effort-work-on-genome-sequencing.