Endorsements
Title page
Copyright page
Preface
Acknowledgments
Note on Further Reading and Sources
Epigraph
About the Companion Website
1: A Tricky, High-Stakes Game
1. 1.1 Where We Are Today
2. 1.2 What We Need to Do Better
3. 1.3 How Can We Do Better?
4. Questions
5. Further Reading and Sources
6. Note
7. References
2: When Nature Won
1. 2.1 The Best-Laid Plans
2. 2.2 Why Hazard Assessment Went Wrong
3. 2.3 How Mitigation Fared
4. 2.4 The Challenges Ahead
5. Questions
6. Further Reading and Sources
7. Note
8. References
3: Nature Bats Last
1. 3.1 Prediction Is Hard
2. 3.2 Forecasts, Predictions, and Warnings
3. 3.3 Earthquake Prediction
4. 3.4 Chaos
5. Questions
6. Further Reading and Sources
7. References
4: Uncertainty and Probability
1. 4.1 Basic Ideas
2. 4.2 Compound Events
3. 4.3 The Gaussian Distribution
4. 4.4 Probability vs Statistics
5. 4.5 Shallow and Deep Uncertainties
6. Questions
7. Further Reading and Sources
8. Note
9. References
5: Communicating What We Know and What We Don't
1. 5.1 Recognizing and Admitting Uncertainties
2. 5.2 Precision and Accuracy
3. 5.3 Testing Forecasts
4. 5.4 Communicating Forecasts
5. Questions
6. Further Reading and Sources
7. References
6: Human Disasters
1. 6.1 Assessing Hazards
2. 6.2 Vulnerability and Interconnections
3. 6.3 The 2008 US Financial Disaster
4. 6.4 Pseudodisasters and Groupthink
5. 6.5 Disaster Chic
6. Questions
7. Further Reading and Sources
8. Note
9. References
7: How Much Is Enough?
1. 7.1 Rational Policy Making
2. 7.2 Lessons from National Defense
3. 7.3 Making Choices
4. 7.4 Uncertainty and Risk Aversion
5. 7.5 Present and Future Value
6. 7.6 Valuing Lives
7. 7.7 Implications for Natural Hazard Mitigation
8. Questions
9. Further Reading and Sources
10. Note
11. References
8: Guessing the Odds
1. 8.1 Big Events Are Rare
2. 8.2 Time-Independent Probability Models
3. 8.3 Time-Dependent Probability Models
4. Questions
5. Further Reading and Sources
6. Note
7. References
9: When's the Next Earthquake?
1. 9.1 A Very Tough Problem
2. 9.2 Earthquake Frequency-Magnitude Relation
3. 9.3 Earthquake Cycle Model
4. 9.4 Computing Earthquake Probabilities
5. 9.5 Shaky Probabilities
6. Questions
7. Further Reading and Sources
8. References
10: Assessing Hazards
1. 10.1 Five Tough Questions
2. 10.2 Uncertainties
3. 10.3 How Is the Hazard Defined?
4. 10.4 Where Will Large Earthquakes Occur?
5. 10.5 When Will Large Earthquakes Occur?
6. 10.6 How Big Will the Large Earthquakes Be?
7. 10.7 How Much Shaking?
8. 10.8 Dealing With the Uncertainties
9. 10.9 Next Steps
10. Questions
11. Further Reading and Sources
12. References
11: Mitigating Hazards
1. 11.1 Approaches
2. 11.2 Accepting Risk
3. 11.3 Transferring Risk
4. 11.4 Avoiding Risk
5. 11.5 Mitigating Risk
6. 11.6 Combined Strategies
7. Questions
8. Further Reading and Sources
9. Note
10. References
12: Choosing Mitigation Policies
1. 12.1 Making Choices
2. 12.2 House Fire Mitigation
3. 12.3 Losses from Hazards
4. 12.4 Optimal Natural Hazard Mitigation
5. 12.5 Nonoptimal Natural Hazard Mitigation
6. 12.6 Mitigation Given Uncertainties
7. 12.7 Robust Policy Making
8. Questions
9. Further Reading and Sources
10. Note
11. References
13: Doing Better
1. 13.1 Final Thoughts
2. 13.2 Community Decision Making
3. 13.3 Improved Organization
4. Questions
5. Further Reading and Sources
6. Note
7. References
Supplemental Images
Index
End User License Agreement

List of Illustrations

Figure 1.1 (a) Natural disasters in 2012. (Munich Re, 2013a. Reproduced with permission from Munich Reinsurance Company AG.) (b) Overall and insured losses since 1980 due to natural disasters. (Munich Re, 2013b. Reproduced with permission from Munich Reinsurance Company AG.)
Figure 1.2 More than a dozen ships were washed inland by the Tohoku tsunami in Kesennuma City, Miyagi Prefecture. The fishing trawler Kyotoku-maru came to rest on a giant debris pile on one of the main roads to City Hall. (Courtesy of Hermann M. Fritz.)
Figure 1.3 Comparison of earlier and revised estimates of possible tsunami heights from a giant Nankai Trough earthquake (Cyranoski, 2012a. Reproduced with permission from Nature.)
Figure 1.4 (a) Seismic hazard map for Haiti produced prior to the 2010 earthquake showing maximum shaking expected to have a 10% chance of being exceeded once in 50 years, or on average once about every 500 years. (b) Map of the shaking in the 2010 earthquake. (Stein et al., 2012. Reproduced with permission of Elsevier B.V.) See also color plate 1.4.
Figure 1.5 Comparison of seismic hazard maps for Haiti made before (a) and shortly after (b) the 2010 earthquake. The newer map shows a factor of four higher hazard on the fault that had recently broken in the earthquake. (Stein et al., 2012. Reproduced with permission of Elsevier B.V.) See also color plate 1.5.
Figure 1.6 Comparison of successive Italian hazard maps, which forecast some earthquake locations well and others poorly. The 1999 map was updated after the missed 2002 Molise quake and the 2006 map will presumably be updated because it missed the 2012 Emilia earthquake. (Stein et al., 2013. Reproduced with permission of Elsevier B.V.) See also color plate 1.6.
Figure 1.7 Annual deaths in the United States from earthquakes, 1812–2003.
Figure 1.8 Schematic illustrating how formulating hazard policy involves integrating assessment, mitigation, and economics.
Figure 2.1 Comparison of Japanese government hazard map to the locations of earthquakes since 1979 that caused 10 or more fatalities. Hazard is shown as probability that the maximum ground acceleration (shaking) in any area would exceed a particular value during the next 30 years. Larger expected shaking corresponds to higher predicted hazard. The Tohoku area is shown as having significantly lower hazard than other parts of Japan, notably areas to the south. Since 1979, earthquakes that caused 10 or more fatalities occurred in places assigned a relatively low hazard. (Geller, 2011. Reproduced with permission of Nature.) See also color plate 2.1.
Figure 2.2 Schematic illustration of the earthquake cycle on the locked megathrust fault at a subduction zone. Strain builds up for many years (a) until it is released in a major earthquake (b). The strain accumulation can be measured using techniques including GPS.
Figure 2.3 Comparison of the trench segments and maximum earthquake sizes assumed in the Japanese hazard map (a) to the aftershock zone of the March 11 earthquake (b), which broke five segments. (Stein et al., 2012. Reproduced with permission of Elsevier B.V.)
Figure 2.4 What went wrong at Tohoku. (a) Illustration of the relative fault dimensions, average fault slip, and average tsunami run-up for magnitude 8 and 9 earthquakes. (b) Data available in 1980, showing the largest earthquake known at various subduction zones. Magnitude 9 earthquakes had been observed only where young lithosphere subducts rapidly. Diagonal lines show predicted maximum earthquake magnitude. (c) Physical interpretation of this result in terms of strong mechanical coupling and thus large earthquakes at the trench interface. (d) Update of (b) with data including the 2004 Sumatra and 2011 Tohoku earthquakes. (e) Earthquake history for the Nankai trough area illustrating how different segments rupturing cause earthquakes of different magnitudes. Segment “D” is the presumed Tokai seismic gap. (Stein and Okal, 2011. Reproduced with permission of American Geophysical Union.) See also color plate 2.4.
Figure 2.5 Seawall destroyed by the tsunami. (Courtesy of Hermann M. Fritz.)
Figure 3.1 The number of hurricanes expected to strike various areas of the U.S. during a 100-year period based on historical data, showing areas expected to receive 20 to 40, 40 to 60, or more than 60 strikes. (USGS, 2005.)
Figure 3.2 Volcanic hazard map for the Mount Rainier, Washington area. (Hoblitt et al., 1998.)
Figure 3.3 Issues of prediction and forecasting have interested the public from the 5th century bc to the present, as shown by the ancient tale of The Boy Who Cried Wolf, and the more modern take in the 1997 film Dante's Peak in which the volcano actually erupts.
Figure 3.4 1980 eruption of Mount St. Helens. (USGS/Cascades Volcano Observatory.)
Figure 3.5 Schematic comparison of some forecast and warning methods.
Figure 3.6 The U.S. National Earthquake Hazards Reduction Program (NEHRP) was established in 1977 in response to claims like that in 1976 by the National Research Council, the operating arm of the National Academy of Sciences, that “it is the panel’s unanimous opinion that the development of an effective earthquake-prediction capability is an achievable goal.”
Figure 3.7 The Parkfield earthquake predicted to occur within five years of 1988 occurred in 2004. Black dots show the dates of earthquakes before the prediction was made, and the line shows when earthquakes 22 years apart should happen. (Stein, 2010. Reproduced with permission of Columbia University Press.)
Figure 3.8 Small perturbations grow. Comparison of two time series generated by the same equation, showing how slightly different initial conditions quickly lead to quite different values.
Figure 3.9 Tracks of North Atlantic hurricanes, tropical storms, and depressions for two very active hurricane seasons. (Ebeling and Stein, 2011. Reproduced with permission of the Seismological Society of America.)
Figure 4.1 Probability of observing specific values from a Gaussian parent distribution.
Figure 4.2 Histogram of the results of drawing N samples from a Gaussian parent distribution with mean zero and a unit standard deviation. (Stein and Wysession, 2003. Reproduced with permission of John Wiley & Sons.)
Figure 4.3 Maximum rate of motion in and across the New Madrid seismic zone shown by successively more precise GPS measurements. (Calais and Stein, 2009. Reproduced with permission of The American Association for the Advancement of Science.)
Figure 5.1 Comparison of the predicted (a) and actual (b) tracks of Hurricane Ike. ((a) NOAA, 2008. (b) NOAA, 2009.)
Figure 5.2 Comparison of the rise in global temperature by the year 2099 predicted by various climate models. For various scenarios of carbon emissions – B1, B2, etc. – the vertical band shows the different predicted warming. (IPCC, 2007. Reproduced from IPCC and Cambridge University Press.)
Figure 5.3 Measurements of the speed of light between 1875 to 1960. Vertical bars show the experimenters' assessments of the uncertainty in their measurements. (Henrion and Fischhoff, 1986. Reproduced with permission of the American Association of Physics Teachers.)
Figure 5.4 Illustration of the concepts of precision and accuracy. (Taylor, 1997. © University Science Books, Mill Valley, CA. Used with permission. All rights reserved.)
Figure 5.5 Illustration of overfitting by comparison of linear and quadratic fits to a set of data. The quadratic gives a better fit to the points but a poorer representation of the trend. (Stein et al., 2012. Reproduced with permission of Elsevier, B.V.)
Figure 5.6 (a) Number of earthquakes and their magnitudes in the L'Aquila area, during the period leading up to the large April 6, 2009 earthquake. (b) Earthquake hazard map of Italy. (Hall, 2011. Reproduced with permission of Nature Publishing.)
Figure 5.7 Comparison of earthquake hazard, described as peak ground acceleration (PGA) as a percentage of the acceleration of gravity expected with 2% risk in 50 years, predicted by various assumptions. (Stein et al., 2012. Reproduced with permission of Elsevier, B.V.)
Figure 5.8 Italian flag graphic showing an assessment of both the evidence for and against a proposition and the uncertainty involved, using the 50%, 30% and 20% percentages quoted in the text.
Figure 6.1 (a) Explosion of shuttle Challenger. (b) Final launch of shuttle Columbia. These two losses in 107 missions correspond to a loss rate of about 1 per 50 missions, much higher than the 1 in 100,000 that had been assumed by NASA management. ((a) NASA, 1986. (b) NASA, 2003.)
Figure 6.2 US house price index from 1975 to 2011. Prices are nominal, i.e. not adjusted for inflation. Shaded areas indicate US recessions. (“FRED^®” charts © Federal Reserve Bank of St. Louis, 2014. All rights reserved. All “FRED^®” charts reprinted by permission. http://research.stlouisfed.org/fred2/)
Figure 6.3 Financial Stress Index (a) and unemployment rate (b) showing the effects of the 2008 disaster. Shaded areas indicate US recessions. (“FRED^®” charts © Federal Reserve Bank of St. Louis, 2014. All rights reserved. All “FRED^®” charts reprinted by permission. http://research.stlouisfed.org/fred2/)
Figure 6.4 President Gerald Ford is vaccinated against swine flu in 1976. (Kennerly, 1976. Courtesy of Gerald R. Ford Library.)
Figure 6.5 One of many books advising how to survive the predicted Y2K disaster.
Figure 7.1 Comparison of the marginal benefit of increasing anti-submarine capability, dp(rts), to the marginal cost C′(v), as a function of the resources invested.
Figure 7.2 Marginal return on capital invested for sectors G and S. The optimum point x indicates that a* is the optimum allocation of capital between the sectors.
Figure 7.3 (a) In the absence of uncertainty and risk aversion, the maximum return less the cost of capital occurs for Z(k*) = V(k*) − C(k*). (b) This optimum occurs where the marginal return equals the marginal cost of capital, V′(k*) = C′(k*). Including uncertainty and risk aversion reduces the optimum to k**, given by V′(k**) − R′(k**) = C′(k**).
Figure 8.1 Estimation of flood frequency from a long-term record. (Eric Baer, SERC.)
Figure 8.2 Changes in flood frequency due to human activity. (Dinicola, 1996.)
Figure 8.3 A model for the probability of an event is drawing a ball from an urn filled with balls, some labeled “E” for event and others labeled “N” for none. (Stein and Stein, 2013a. Reproduced with permission of the American Geophysical Union.)
Figure 8.4 Comparison of the probability of an event as a function of time for a time-independent (solid line) and two time-dependent (dashed lines) urn models. (Stein and Stein, 2013a. Reproduced with permission of the American Geophysical Union.)
Figure 8.5 Sequence of events as a function of time for the time-independent (top line) and time-dependent (lower lines) urn model runs in Figure 8.4. (Stein and Stein, 2013a. Reproduced with permission of the American Geophysical Union.)
Figure 9.1 Frequency-magnitude plot for ∼13,000 earthquakes with surface wave magnitude M_s ≥ 5.0 during 1968–1997. The line shown, with slope b about 1, fits the data reasonably well. (Stein and Wysession, 2003. Reproduced with permission of John Wiley & Sons.)
Figure 9.2 Comparison of seismograms for the 1906 San Francisco and 2004 Sumatra earthquakes, shown on the same scale. (Richard Aster, Colorado State University. Reproduced with permission.)
Figure 9.3 Comparison of earthquakes with different magnitudes in terms of how often they happen and the energy they release. (Stein and Wysession, 2003. Reproduced with permission of John Wiley & Sons.)
Figure 9.4 Comparison of rupture areas and slip distances for earthquakes with different magnitudes. Seismic moments are given in dyne-cm.
Figure 9.5 Types of plate boundaries in oceanic lithosphere. Oceanic lithosphere is formed at ridges and subducted at trenches. At transform faults, plates slide by each other. (Stein and Wysession, 2003. Reproduced with permission of John Wiley & Sons.)
Figure 9.6 Map of major plates and earthquake locations, shown by dots. The earthquakes outline most plate boundaries. “NM” marks New Madrid. “MAR” is Mid-Atlantic Ridge. “EAR” marks the East African Rift. (Stein and Wysession, 2003. Reproduced with permission of John Wiley & Sons.)
Figure 9.7 Major plates and the relative motion at their boundaries. Arrow lengths show the speed of the motion. Diverging arrows show spreading at mid-ocean ridges. Single arrows on a subducting plate show convergence. Stippled areas are diffuse plate boundary zones. (Gordon and Stein, 1992. Reproduced with permission of American Association for the Advancement of Science.)
Figure 9.8 Some results of the earthquake on the San Andreas fault, April 18, 1906. (a) Ground breakage along the fault trace. (b) A fence offset by the earthquake. (Gilbert, 1906. Reproduced with permission of the U.S. Geological Survey.)
Figure 9.9 How elastic rebound works is shown by the history of a fence across a fault. (Stein and Wysession, 2003. Reproduced with permission of John Wiley & Sons.)
Figure 9.10 An analogy for elastic rebound: using a rubber band to pull a soap bar in a box across a mat.
Figure 9.11 (a) Very precise plate velocities can be obtained by measuring the positions of receivers over time using GPS signals. (b) Profile across the San Andreas showing GPS velocities in the direction along the fault. (Stein and Wysession, 2003. Reproduced with permission of John Wiley & Sons.)
Figure 9.12 Geological studies of the San Andreas fault. (a) The rate of motion across the fault, and thus between the Pacific and North American plates, is measured using the offset of Wallace Creek. (b) Paleoearthquake history at Pallett Creek. (Stein, 2010. Reproduced with permission of Columbia University Press.)
Figure 9.13 (a) Gaussian probability density function for recurrence times. (b) Conditional probability that the earthquake will occur in the next twenty years for two different models.
Figure 9.14 Portion of the seismic gap map used by Kagan and Jackson (1991) to test the gap hypothesis. The shaded segments of the plate boundaries had been assigned seismic potentials of high (R), intermediate (O), and low (G). Unshaded segments were regarded as having uncertain potential. During the ten years following the map's publication, ten large (M > 7) earthquakes (dots) occurred in these regions. None were in the high- or intermediate-risk segments, and five were in the low-risk segments. (Stein, 1992. Reproduced with permission of Nature Publishing.)
Figure Q9.2 Comparison of earthquakes per year and years between earthquakes. (Richard Aster, Colorado State University. Reproduced with permission.)
Figure 10.1 Comparison of the 1982 and 1996 US Geological Survey earthquake hazard maps. The predicted hazard is shown as a percentage of the acceleration of gravity. Redefining the hazard raised the predicted hazard in the Midwest from being much less than in California to being even greater than California's. (Stein, 2010. Reproduced with permission of Columbia University Press.)
Figure 10.2 Schematic illustration showing how the predicted earthquake hazard increases for longer time windows. The circles show areas within which shaking above a certain level will occur. (Stein, 2010. Reproduced with permission of Columbia University Press.)
Figure 10.3 Seismicity along the North Africa plate boundary for 1963–2004. Simulations using a frequency-magnitude relation derived from these data predict that if seismicity is uniform in the zone, about an 8000-year record is needed to avoid apparent concentrations and gaps. (Swafford and Stein, 2007. Reproduced with permission of the Geological Society of America.) See also color plate 10.3.
Figure 10.4 Global Seismic Hazard Map (1999) for North Africa, showing peak ground acceleration in m/s² expected at 10% probability in 50 years. Note “bull's-eye” at site of the 1980 M_s 7.3 El Asnam (EA) earthquake. The largest subsequent earthquakes to date, the May 2003 M_s 6.8 Algeria and February 2004 M_s 6.4 Morocco events (stars) did not occur in the predicted high hazard regions. (Swafford and Stein, 2007. Reproduced with permission of the Geological Society of America.)
Figure 10.5 (a) Seismic hazard map for China produced prior to the 2008 Wenchuan earthquake, which occurred on the Longmenshan Fault (black rectangle). (b) Seismicity in the region. The hazard map showed low hazard on the Longmenshan fault, on which little instrumentally recorded seismicity had occurred before the Wenchuan earthquake, and higher hazard on faults nearby that showed more seismicity. (Stein et al., 2012. Reproduced with permission of Elsevier, B.V.) See also color plate 10.5.
Figure 10.6 Episodic, clustered, and migrating large earthquakes. In many continental fault systems, it appears that rather than one main fault staying active for a long time (a), many faults turn on and off (b). (McKenna et al., 2007. Reproduced with permission of the Geological Society of America.)
Figure 10.7 Earthquake history of North China. Solid circles are locations of events during the period shown in each panel; open circles are locations of events from 780 bc to the end of the previous period (1303 ad for panel A). Bars show the rupture lengths for selected large events. (Liu et al., 2011. Reproduced with permission of the Geological Society of America.) See also color plate 10.7.
Figure 10.8 Comparison of the 1985 and 2005 Geological Survey of Canada earthquake hazard maps of Canada. The older map shows concentrated high hazard bull's-eyes along the east coast at the sites of the 1929 Grand Banks and 1933 Baffin Bay earthquakes, whereas the new map assumes that similar earthquakes can occur anywhere along the margin. (Stein et al., 2012. Reproduced with permission of Elsevier, B.V.) See also color plate 10.8.
Figure 10.9 (a) Schematic comparison of time-independent and time-dependent models for different seismic zones. Charleston and New Madrid are “early” in their cycles, so time-dependent models predict lower hazards. The two model types predict essentially the same hazard for a recurrence of the 1906 San Francisco earthquake, and time-dependent models predict higher hazard for the nominally “overdue” recurrence of the 1857 Fort Tejon earthquake. The time-dependent curve is schematic because its shape depends on the probability distribution and its parameters. (b) Comparison of the conditional probability of a large earthquake in the New Madrid zone in the next 50 years, assuming that the mean recurrence time is 500 years. In the time-independent model the probability is 10%. Time-dependent models predict lower probabilities of a large earthquake for the next hundred years. (Hebden and Stein, 2009. Reproduced with permission of the Seismological Society of America.) See also color plate 10.9.
Figure 10.10 Comparison of hazard maps for the New Madrid zone. Shading shows peak ground acceleration as percentages of 1 g. Compared to the hazard predicted by the time-independent model, the time-dependent model predicts noticeably lower hazard for the periods 2000–2050 and 2100–2150, but higher hazard if a large earthquake has not occurred by 2200. (Stein et al., 2012. Reproduced with permission of Elsevier, B.V.) See also color plate 10.10.
Figure 10.11 Results of numerical simulations of earthquake sequences. Rows show results for sequences of different lengths. Left panels show the log-linear frequency-magnitude relation sampled, with dots showing the resulting mean recurrence times. Center panels show the parent distribution of recurrence times for M ≥ 7 earthquakes (smooth curve) and the observed mean recurrence times (bars). Right panels show the fraction of sequences in which a given number of M ≥ 7 earthquakes occurred. (Stein and Newman, 2004. Reproduced with permission of the Seismological Society of America.) See also color plate 10.11.
Figure 10.12 Comparison of ground motion (peak ground acceleration and 1 Hz) as a function of distance for three different earthquake magnitudes predicted by three models for the central US. For M_w 8, the Frankel et al. (1996) model predicts significantly higher values than the others. (Newman et al., 2001. Reproduced with permission of the Seismological Society of America.) See also color plate 10.12.
Figure 10.13 Comparison of the predicted hazard (2% probability in 50 years) showing the effect of different ground motion models and maximum magnitudes of the New Madrid fault source. (Newman et al., 2001. Reproduced with permission of the Seismological Society of America.) See also color plate 10.13.
Figure 10.14 Comparison of the hazard at St Louis and Memphis predicted by the different hazard maps of the New Madrid zone shown in Figures 10.10 and 10.13. For example, Frankel/M8 indicates the Frankel et al. (1996) ground motion model with a maximum magnitude of 8 in Figure 10.13, and TI indicates the time-independent model in Figure 10.10. (Stein et al., 2012. Reproduced with permission of Elsevier, B.V.)
Figure 10.15 Logic tree combining the four models in Figure 10.13 to predict shaking at St. Louis. (Stein, 2010. Reproduced with permission of Columbia University Press.)
Figure 11.1 Intensity map for the first of the three major New Madrid shocks, on December 16, 1811. Boxes label different damage zones corresponding to intensity contours. Some of the sites from which reports exist are shown. (Stein, 2010. Reproduced with permission of Columbia University Press.)
Figure 11.2 How vulnerable buildings are depends on the material used in their construction. (Stein and Wysession, 2003. Reproduced with permission of John Wiley & Sons.)
Figure 11.3 The Memphis Veterans' Administration hospital during reconstruction and seismic retrofitting. (Courtesy of Joseph Tomasello.)
Figure 12.1 Comparing mitigation options. (a) The optimal mitigation level, n*, minimizes the total cost, the sum of expected loss and mitigation cost. (b) n* occurs when the reduced loss –Q′(n) equals the incremental mitigation cost C′(n). Including the effect of uncertainty and risk aversion, the optimal mitigation level n** increases until the incremental cost equals the sum of the reduced loss and incremental decline in the risk term R′(n). (Stein and Stein, 2012b. Reproduced with permission of the Geological Society of America.)
Figure 12.2 Illustration of the effects of overmitigation and undermitigation.
Figure 12.3 Selecting mitigation levels. (a) Comparison of total cost curves for two estimated hazard levels. For each, the optimal mitigation level, n*, minimizes the total cost, the sum of expected loss and mitigation cost. (b) In terms of derivatives, n* occurs when the reduced loss –Q′(n) equals the incremental mitigation cost C′(n). If the hazard is assumed to be described by one curve but is actually described by the other, the assumed optimal mitigation level causes nonoptimal mitigation, and thus excess expected loss or excess mitigation cost. (Stein and Stein, 2013b. Reproduced with permission of the Geological Society of America.) See also color plate 12.3.

“This is truly an amazing book! The product of a unique collaboration between a renowned economist and renowned seismologist (who happen to be father and son), Playing against Nature lays out a clear story, in easy-to-read prose, of what natural disasters are, what the limitations of risk prediction can be, and how society's response to them has to account for the reality that we have limited economic resources. The authors present fascinating case studies to illustrate examples of where predictions have failed, and why. They also take a bold step by showing how natural disasters and economic disasters provide similar challenges, and provide a clear description of how risk should be assessed, and how it can be mitigated reasonably. This is a book that researchers, policy makers, and the general public should read. It can even serve as valuable text for the new generation of interdisciplinary college courses addressing the interface between science and social science.” – Stephen Marshak, Professor and Director of the School of Earth Society and Environment, University of Illinois at Urbana-Champaign

“I very highly recommend this book for anyone dealing with or interested in natural hazards assessment and mitigation. It is a tour de force with examples, descriptions, illustrations, reference lists, and explanations for understanding natural disasters and negotiating the often perilous and misguided approaches for hazards mitigation. This book is a huge achievement in that it has collected an enormous amount of relevant information, case studies, economics and engineering factors, loss statistics, references, and even study guides and questions for students. It is both highly technical with all the probability and statistics formulations needed to express necessary relationships but on the other hand, so well written that professionals in government, business, and education will find it exceedingly readable. In my everyday work experience, I attempt to communicate principles of hazard occurrences and risks. This book gives me far more useable material than I have ever had to achieve my goals for advising public officials, teaching university students, and educating citizens. This is the best resource in existence for understanding natural hazards and hazard mitigation.” – James C. Cobb, State Geologist and Director, Kentucky Geological Survey, University of Kentucky

“Playing against Nature is a virtuoso performance by a father-son duo. A distinguished economist and seismologist have produced a pioneering work that promises to enhance our ability to integrate assessment science, cost-benefit analysis and mitigation design and engineering. The result will be more informed, bottom-up, hazard mitigation policies. This outstandingly researched book is highly readable and destined to become a classic.” – Steve H. Hanke, Professor of Applied Economics, The Johns Hopkins University

“Elegantly written in Seth Stein's usual memorable prose, Playing against Nature treats jointly seismic and economic catastrophes in a thought-provoking and readable way. How blindingly obvious something can be after the event! Ringing oh so very true, it provides insight into why science and scientists don’t get things right all the time. Enriched with gems of quotes, and an unusual mix of hard science and philosophy, Playing against Nature will make a great supporting text for any course on hazards – geologic, engineering, political or economic – and judging from current trends, we could all use as much understanding of this topic as possible.” – Gillian R. Foulger, Professor of Geophysics, University of Durham

“Authored by a remarkable father and son team, Playing against Nature is a comprehensive, lucid assessment of the interplay between natural hazards and economics of many kinds. As world population continues to increase to more and more unsustainable numbers, and demand for economic growth plagues the world, human activities continue to place us in more and greater vulnerability as Earth processes go on, as they have over deep time. We need to better recognize and thus more responsibly prepare for inevitable natural events. Blunt, forceful, and true statements (e.g., ‘Humans have to live with natural hazards’ and ‘Hazards are geological facts that are not under human control’) characterize Playing against Nature and make reading this contribution, by anyone, a sobering and enlightening experience. I highly recommend Playing against Nature to those who care about the future of the human race.” – John Geissman, Professor of Geosciences, University of Texas at Dallas

“In the wake of recent natural disasters and economic crises, the authors question the inability of specialists – of earth and planetary sciences on one side and economists on the other – to predict such events. Beyond these two spheres, this work also reveals a bridge between seemingly distinct fields of science, which meet as soon as one starts to focus on concepts that are fundamental for both, such as hazard, risk or vulnerability. This book discusses the laws of probability and the most appropriate models for predicting rare events; it also offers strategies to optimize mitigation plans. Playing against Nature thus is an innovative work that should encourage researchers in different disciplines to collaborate. It may also become a useful tool for graduate students. This book furthermore constitutes an ideal reference work for policy makers.” – Serge Rey, Professor of Economics, University of Pau

“Insightful and provocative, Playing against Nature by Stein and Stein explains in a brilliant yet playful way why experts missed many of the recent natural and manmade disasters, from the 2011 Tohoku earthquake to the 2008 financial crisis. It makes an enjoyable read for anyone who has ever wondered how society prepares and responds to natural disasters. The authors, an economist father and a geophysicist son, provide a unique perspective of how scientific study of natural disasters interplays with policy making for hazard mitigation. As a student of earthquake science, I found many arguments and facts in the book compelling and intriguing. Facing many unknowns and with limited resources, we are gambling with nature in hazard preparation and mitigation, as the authors put it. We may not expect to win every hand, but we need to understand our odds. Playing against Nature offers a fresh way to look at nature's games. It should be helpful to professionals, and delightful to everyone who opens the book.” – Mian Liu, Curators' Distinguished Professor in Geological Sciences, University of Missouri

“How can policy defend society better against natural disasters whose probabilities are uncertain and in flux? In Playing against Nature, Seth Stein, a geologist, and his late father Jerome, an economist, joined forces. Their book is a clear Guide for the Perplexed, combining scholarship and exposition to show how to prepare more wisely for hurricanes, earthquakes, and tsunamis.” – Shlomo Maital, Professor Emeritus, Institute for Advanced Studies in Science and Technology, Technion-Israel Institute of Technology

“What do natural disasters and economic disasters have in common, and how is it possible to efficiently mitigate their effects? You will find the answer in this scholarly book. But there is more to it than meets the eye: this important monograph is based on what I call ‘the Steins' synergy’ (after the late Jerome Stein, an economist, and his son Seth Stein, a geoscientist). The interaction between these two scientists has been such that the combined result of their joint research, reported in this book, is much greater than the sum of the individual results: the quintessential example of what interdisciplinarity can achieve.” – Giancarlo Gandolfo, Professor, Accademia Nazionale dei Lincei, Rome, Research Fellow, CESifo, Munich, Professor of International Economics, Sapienza University of Rome (retired)

“ ‘Nature's smarter than us’ might be a good subtitle for this well-written and illustrated tome by a father-son team. Reviewing numerous natural disasters from Katrina to Haiti to Sandy to the Japan earthquake, the authors find most disaster responses to be seriously wanting. Their accounts of nature at its most violent range from humorous to appalling. The solution: a better understanding of the uncertainties of disaster response, free of politics, tradition and too narrow science.” – Orrin H. Pilkey, Professor Emeritus of Earth and Ocean Sciences, Duke University

Jerome and Seth Stein, spring 2012. Photo by Hadassah Stein.

This work is a co-publication between the American Geophysical Union and Wiley

Registered office: John Wiley & Sons, Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

Editorial offices: 9600 Garsington Road, Oxford, OX4 2DQ, UK

The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

111 River Street, Hoboken, NJ 07030-5774, USA

For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com/wiley-blackwell.

The right of the author to be identified as the author of this work has been asserted in accordance with the UK Copyright, Designs and Patents Act 1988.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher.

Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book.

Limit of Liability/Disclaimer of Warranty: While the publisher and author(s) have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. It is sold on the understanding that the publisher is not engaged in rendering professional services and neither the publisher nor the author shall be liable for damages arising herefrom. If professional advice or other expert assistance is required, the services of a competent professional should be sought.

Library of Congress Cataloging-in-Publication Data

Stein, Seth.

Playing against nature : integrating science and economics to mitigate natural hazards in an uncertain world / Seth Stein and Jerome Stein.

pages cm

Includes bibliographical references and index.

ISBN 978-1-118-62082-3 (cloth)

1. Hazard mitigation. 2. Hazard mitigation–Economic aspects. 3. Natural disasters–Economic aspects. 4. Natural disasters–Risk assessment. 5. Emergency management. I. Stein, Jerome L. II. Title.

GB5014.S83 2014

363.34'6–dc23

2013046062

A catalogue record for this book is available from the British Library.

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books.

Cover image: Photo 1 from Moment of the tsunami slideshow. A wave approaches Miyako City from the Heigawa estuary in Iwate Prefecture after the magnitude 8.9 earthquake struck the area, March 11, 2011. Reuters/Mainichi Shimbun.

Cover design by Steve Thompson

Preface

This book considers how to make policy to defend society against natural hazards more effective. Recent events including Hurricane Katrina in 2005, the 2011 Tohoku earthquake and tsunami, and Hurricane Sandy in 2012 show that in its high-stakes game of chance against nature, society often does poorly. Sometimes nature surprises us, when an earthquake, hurricane, or flood is bigger or has greater effects than expected from detailed natural hazard assessments. In other cases, nature outsmarts us, doing great damage despite expensive mitigation measures being in place, or causing us to divert limited resources to mitigate hazards that are overestimated.

This situation may seem surprising because of the steady advances being made in the science of natural hazards. In our view, much of the problem comes from the fact that formulating effective natural hazard policy involves using a complicated combination of science and economics to analyze a problem and explore the costs and benefits of different options, in situations where the future is very uncertain. In general, mitigation policies are chosen without this kind of analysis. Typically, communities have not looked at different options, and somehow end up choosing one or having one chosen for them without knowing how much they're paying or what they're getting for their money. This is like buying insurance without considering how much a policy will cost and what the benefits would be. Not surprisingly, the results are often disappointing. Thus it is worth thinking about how to do better.

This book explores these issues, taking a joint view from geoscience and economics. My view is that of a seismologist interested in the science of large earthquakes and earthquake hazard mitigation. My coauthor and late father, Jerome Stein, brought the view of an economist interested in public policy.

As my father told the Brown Daily Herald in November, 2012, he viewed this book as derived from the day in 1960 that he took his 7-year old son to hear a lecture about the new discoveries of continental drift that would soon transform modern geology. Apparently I was intrigued by the idea, and asked the speaker whether drifting continents were like bars of soap floating in the bathtub.

Over the years, my father and I often talked about science. We discussed natural hazards, starting in 1998, when I became skeptical of widely-touted claims that parts of the central US faced earthquake hazards as high as California's, and that buildings should be built to the same safety standards. To my surprise, it turned out that the federal government was pressing for these measures without undertaking any analysis of the huge uncertainties in the hazard estimates or of whether the large costs involved would yield commensurate benefits to public safety. To my further surprise, my father said that this was typical, in that economists had found that many health and safety regulations were developed without such analysis or consideration of alternative policies. In such cases, no one knew whether these policies made sense or not. I became interested in this question, and started working with colleagues and students to investigate how large the uncertainties in earthquake hazard estimates were.

Our discussions on this topic ramped up in 2011, following the Tohoku earthquake and tsunami. Japanese hazard planners had assumed that an earthquake and tsunami that big could not occur there, whereas my colleague Emile Okal and I had found before the earthquake that they could. At the same time, my father was studying how the 2008 US financial disaster had occurred, despite the fact that both Wall Street and the US government had been sure – based on economic models – that it could not. We realized that although one disaster was natural and the other was economic, they had much in common. Both resulted from overconfidence in how well hazards could be assessed, both had vulnerabilities that were not recognized, and the result in both cases was poorly formulated policies.

We decided to explore these issues in a series of journal articles that became the basis of this book. Because there are many fine books on natural hazard science and on economics, we focused on the interface between the two fields. Our discussions of the challenging questions involved and how to present them had special intensity because we started the book after my father's illness was diagnosed and knew we had only a short time to finish it.

For simplicity, we decided to primarily use earthquake and tsunami hazards as examples, although the approach applies to other natural hazards. Our goal is to introduce some key concepts and challenges, and illustrate them with examples and questions that we pose at the end of each chapter. We decided to introduce some relevant mathematics, which can be skipped by readers without losing the key themes. We illustrate the key themes with examples and questions at the end of each chapter. As is typical for natural hazards, many of the questions are difficult and few have unique or correct answers.

In this sprit of looking toward the future, we hope the book will help researchers, especially younger ones, to develop an interdisciplinary outlook as they work at the interface between the two fields. Hopefully their work, both about hazards and how to make better policies, will help society fare better in its game against nature.

Seth Stein

Glencoe, Illinois

April 2013

Royalties from this book go to the Division of Applied Mathematics at Brown University to support the Jerome L Stein award, which recognizes undergraduate students who show outstanding potential in an interdisciplinary area that involves applied mathematics.

Acknowledgments

Although science is always a human endeavor, this book is especially so because of its father–son collaboration. It would not have been completed, given my father's illness, without the support of Hadassah and Carol Stein. Their encouragement when the task seemed too big and progress slowed is even more impressive given that both went through it all for our previous books.

This book grew from ideas developed over many years via research carried out with coworkers, fruitful discussions with them and other researchers studying these or related problems, and knowledge from the broad communities of geoscientists, economists, and others interested in natural hazards. In that spirit, I would like to thank many people. All should feel free to share credit for parts of the book with which they agree, and disclaim parts with which they disagree.

I thank my coauthors on the research papers discussed here, including Eric Calais, Carl Ebeling, Robert Geller, Richard Gordon, James Hebden, Qingsong Li, Mian Liu, Jason McKenna, Andres Mendez, Andrew Newman, Emile Okal, Carol Stein, John Schneider, Laura Swafford, and Joe Tomasello.

I have also have benefited from discussions with John Adams, Amotz Agnon, Rick Aster, Roger Bilham, Eric Calais, Thiery Camelbeeck, Dan Clark, Nick Clark, Sierd Cloetingh, Jim Cobb, Mike Craymer, Bill Dickinson, Tim Dixon, Roy Dokka, Joe Engeln, Andy Freed, Tom James, John Geissman, Bob Hermann, David Hindle, Tom Holzer, Sue Hough, Ken Hudnut, Alex Forte, Anke Friedrich, Alan Kafka, Steve Kirby, Jonas Kley, Cinna Lomnitz, Mike Lynch, Steve Marshak, Ailin Mao, Glenn Mattiolli, Miguel Merino, Brian Mitchell, Antonella Peresan, Orrin Pilkey, Hans-Peter Plag, Paul Rydelek, Jim Savage, Gary Searer, Giovanni Sella, Norm Sleep, Bob Smalley, Bob Smith, Bruce Spenser, Ross Stein, Roy van Arsdale, David Wald, Zhenming Wang, John Weber, Steve Wesnousky, Rinus Wortel, Michael Wysession, Dave Yuen, and many others.

I thank Gillian Foulger, Peter Clark, Serge Rey, Hadassah Stein and an anonymous reader for reviewing the manuscript. Finally, I thank the Alexander von Humboldt Foundation for supporting my stay at the University of Gottingen, where the book's editing was completed.

Note on Further Reading and Sources

Technical information on the scientific topics discussed here is often more easily accessible from textbooks than from research papers written tersely by scientists for scientists familiar with the topics under discussion. We list several textbooks for specific chapters. Research papers mentioned, including those from which a figure is used, are listed in the references by their authors.