Natural Hazards and Scientific Advice: Interactions Among Scientists, Decision Makers and the Public PAOLO GASPARINI Professor Emeritus at the University of Napoli Federico II GOLDSCHMIDT CONFERENCE 2013 Plenary session 28th August Florence, Italy This presentation benefitted of contribution from Mauro Dolce, Dept. of Civil Protection Warner Marzocchi, INGV Rome Giulio Selvaggi, INGV Rome and of ideas and discussions with: Tom Jordan, SCEC Gaetano Manfredi, Università di Napoli Federico II Franco Barberi, Università di Roma 3 Gordon Woo, RMS Friedemann Wenzel, KIT Jochen Zschau, GFZ Outline of the talk WHY LOSSES DUE TO NATURAL HAZARDS ARE INCREASING THE ROLE OF SCIENTISTS AND DECISION MAKERS RECENT EXAMPLES OF FALSE OR MISSED ALARMS AN EXTREME CASE: THE L’AQUILA 2009 EARTHQUAKE AFFAIR DEALING WITH VERY LOW PROBABILITIES A POSSIBLE SCENARIO OF INTERACTIONS WHY LOSSES DUE TO NATURAL HAZARD ARE INCREASING WHY LOSSES DUE TO NATURAL HAZARD ARE INCREASING WHY LOSSES DUE TO NATURAL HAZARD ARE INCREASING The Urban Explosion In 2030 about 82% of the population of industrial countries and 57% of the population of less developed countries will be living in urban areas Urban population in developing countries Rural population in developing countries Urban population in industrialised countries Rural population in industrialised countries WHY LOSSES DUE TO NATURAL HAZARD ARE INCREASING Natural hazard risk index for megacities Worldwide, the loss potential from natural catastrophes is increasingly dominated by megacities. Source: Munich Re, 2004 WHY LOSSES DUE TO NATURAL HAZARD ARE INCREASING Increasing disaster-related fatalities and damage even if the frequency and magnitude of geophysical events remains unchanged. Virginia earthquake – August 23, 2011 Mw = 5.8 – Damages in Washington, D.C. 135 Km away Emilia earthquake – May, 2012 Mw = 5.8 Christchurch earthquake – February 22, 2011 Mw = 6.3 WHY LOSSES DUE TO NATURAL HAZARD ARE INCREASING Black Swans (Extremely rare events with huge impact on human life, structures, and socio-economy) Tohoku Tsunami triggered by the unexpected Mw=9.0 offshore earthquake of March 11, 2011 THE ROLE OF SCIENTISTS AND DECISION MAKERS Communication is a major challenge for the management of risk nowadays and in the future; We live in exponential growth time: world population is growing exponentially, cities and megacities are exponentially increasing in complexity; Cities are natural risk attractors; Communication of scientific information is a complex and still unresolved task, fundamental to prevent cities to become natural risk traps. THE ROLE OF SCIENTISTS AND DECISION MAKERS Scientific advise for natural hazard management is mainly asked for: • URBAN AND LAND USE PLANNING (Long Term Hazard and Risk Assessment) • EMERGENCY MANAGEMENT (Short Term Hazard and Risk Assessment) THE ROLE OF SCIENTISTS AND DECISION MAKERS Scientific advise for Urban and Land Use Planning is a demanding, complex but not complicated task: No urgency requirements; Fairly well defined responsibilities. Main requirements of scientific advise: Careful consideration of all possible hazard; Harmonization and probabilistic quantitative evaluation of the parameters defining hazard occurrence. THE ROLE OF SCIENTISTS AND DECISION MAKERS THE ROLE OF SCIENTISTS AND ADMINISTRATORS Risk curves of the hazards due to windstorms, floods and earthquakes for the city of Cologne considering losses at buildings and in the sectors private housing, commerce and industry (2000). (Matrix Project, Grunthal, et al, 2011) THE ROLE OF SCIENTISTS AND DECISION MAKERS SCIENTIFIC ADVISE FOR AN IMPENDING EMERGENCY SITUATION IS DRAMATICALLY DIFFERENT: Need of rapid response; No clear separation of responsibilities; No established rules for information to the public; No established protocols for real time risk mitigation actions. THE ROLE OF SCIENTISTS AND DECISION MAKERS The Challenge is for scientists to articulate uncertainty without losing credibility and to give public officials the information they need for decision-making Public officials Scientists this requires to bridge the gap between scientific output (probability) and the boolean logic (YES-NO) of decision-makers THE ROLE OF SCIENTISTS AND DECISION MAKERS Scientists responsible for hazard and risk assessment Emergency managers Decision making of emergency Education of the public Human behaviour response of the population at risk INDIVIDUAL DECISION MAKING THE ROLE OF SCIENTISTS AND DECISION MAKERS This ideal division of tasks and responsabilities is not usually respected in emergency management. a) While providing assessment of risk, scientists often suggest more or less explicitely risk mitigation actions and are asked to contribute to inform the public; b) Scientific advise is rarely given in quantitative probabilistic terms; c) Emergency time is different for each kind of hazard. Successful results for volcanic crises (Mt. Soufriere of Montserrat, Mt. Pinatubo eruption) for floods and partially for hurricanes. d) Earthquakes are an extreme case of decision making in short time and great uncertainty RECENT EXAMPLES OF FALSE OR MISSED ALARMS 370,000 New York Citizen were evacuated under the threat of two hurricanes in 2011 and 2012. Hurricane Sandy: October2012 Hurricane Irene: August 2011 It was a false alarm: several industries and private companies threatened to undertake legal actions againts Mayor Michael Bloomberg. No blame on scientists 19 RECENT EXAMPLES OF FALSE OR MISSED ALARMS A debated case. Actions taken to protect aircrafts during the Explosive eruption of Eyjafjallajökull (Iceland) in 2010. RECENT EXAMPLES OF FALSE OR MISSED ALARMS A situation of great uncertainties Scientists provided a continuous simulation of ashes density in the plume and forecast of plume direction. The British Civil Aviation Authority established threshold levels for the ash density. Several flights over Europe were forbidden and some airport were closed for several days. The CAA manager was blamed to have been too cautious. No blame on scientists. RECENT EXAMPLES OF FALSE OR MISSED ALARMS A case of missed alarm: The Madrid snow storm of January 9, 2009 45.000 people was trapped at Airports and 400 km of vehicles were trapped on Highways because authorities waited to give warning untill the false alarm probability was very low. Too Late to reach people timely! RECENT EXAMPLES OF FALSE OR MISSED ALARMS An outstanding case of missed alarm: the M9 March 2011 Tohoku Earthquake and tsunami On March 11, 2011, a M9 hits the Honshu coast Probability map for the period Jan.1-Dec.31 2011. The legend reports the annual probability map for a M≥ 8.5 earthquake in each cell of 0.1x0.1 degrees. The box is the fault of the Tohoku earthquake and the star is the epicenter. RECENT EXAMPLES OF FALSE OR MISSED ALARMS On March 9, 2011, a M7.2 occurred in the same area Weekly prob M8.5+=0.12% Prob gain > 100 Probability map for the period March 9 - March 16, 2011. The legend reports the probability map for a M≥ 8.5 earthquake in each cell of 0.1x0.1 degrees. The box is the fault of the Tohoku earthquake; the circle is around the epicenter of the M7.2 earthquake occurred on March 9; the star is the epicenter of the Tohoku earthquake. RECENT EXAMPLES OF FALSE OR MISSED ALARMS Performance EEW nel M9.0 Terremoto di Tohoku RECENT EXAMPLES OF FALSE OR MISSED ALARMS The Government of Japan and The Chief Manager of the Japan Meteorological Agency admitted the inadequacy of their actions and apologized on dedicated TV Broadcast. RECENT EXAMPLES OF FALSE OR MISSED ALARMS Christchurch Earthquake, New Zealand On September 3, 2010 a M7.1 earthquake with 11 km focal depth hit an area 40 km away from the town of Christchurch. No casualties. On January 21, 2011 a M6.3 earthquake located 5 km away from Christchurch, at a depth of about 5 km, hit the region producing 180 casualties and about 7000 injuried people. RECENT EXAMPLES OF FALSE OR MISSED ALARMS Aftershocks gradually shifted with time from the location of the M7.1 quake Eastward in direction of the city of Christchurch RECENT EXAMPLES OF FALSE OR MISSED ALARMS No alert was given during the aftershock sequence. Although some blame was on scientists and administrators, no legal actions. The only legal actions were against designers and builders of some collapsed structures. L’AQUILA 2009 EARTHQUAKE AFFAIR Seismic Hazard map of Italy (2004) in terms of acceleration expected with a return period of 475 y (10% exceedence probability in 50 years). A zoom on Abruzzo Apennine is shown in the map on the left. Expected acceleration at L’Aquila on basement rocks are in the range 0.25-0.75 g. L’AQUILA 2009 EARTHQUAKE AFFAIR 1985 1994 2003-04 Sequences with M>3.9 earthquakes in an area with 50 km radius around L’Aquila in the last 50 years. 13 earthquakes had magnitude greater than 4. L’AQUILA 2009 EARTHQUAKE AFFAIR Comparison between seismic swarms occurred in 1985 and in 2009 (before the Aprile 6 main shock). Number of events with M>2.5 and M> 1 (upper left) are reported. A Paganica B C (A) Epicenters of earthquakes between January 1 and April 6, 2009. (B) Location of the vertical section reported in (C) the violet curve encloses the foci of the pre-April 6 sequence. Black dots are the aftershoks the brown line is the fault plane generating the April 6 main event. L’AQUILA 2009 EARTHQUAKE AFFAIR 2009 L’Aquila Affair Sequence Seismic activity in the L’Aquila area increases in January, 2009 During the sequence Mr. Gianpaolo Giuliani, a technician of INFN, issued predictions of impending large earthquakes in the region (The predictions were not backed by his scientific institution) These unofficial predictions were based on variations of Radon concentration measured with gamma-ray detectors. The used correlation procedures were not made available, and when presented to ICEF after April 6, resulted to not have a scientific basis. At least two of the specific predictions resulted to be false alarms, but generated widespread public concern. INGV and the DPC respond with statements about “earthquake prediction” M<3.5 earthquakes were felt in L’Aquila region and continued in the following months. On March 31, The seismology section of the High Risk Commission (HRC), a consultant body to help DPC in crisis management, was convened in L’Aquila to give an opinion on Giuliani’s prediction. On Apr 1, L’Aquila mayor asks for state of emergency Damage from swarm estimated to be €15 M. Municipality plan had been activated in “attention phase”,; several schools evacuated after felt events. L’AQUILA 2009 EARTHQUAKE AFFAIR 2009 L’Aquila Affair Sequence After two quakes of M3.5 and 3.9 occurring a few hours before, a main shock of M 6.3 occurred at 3.32 a.m. of April 6, 2009. The quake killed over 300 people, injuried more than 1700 persons and destroyed or severely damaged about 20,000 buildings. On October 22, 2012 The Court of L’Aquila condemned at first level some of the participants at the March 31 meeting of the High Risk Commission. L’AQUILA 2009 EARTHQUAKE AFFAIR The verdict Seven scientists and public officials who attended the High Risk Commission meeting of March 31, 2009 were found guilty for involuntary multiple manslaughter and multiple serious injuries. The faults consisted in “negligenza, imprudenza and imperizia” (negligence, uncautiousness, unskillfulness). They were condemned to: 6 years in prison (*) Perpetual interdiction from public offices and legal interdiction during the enforcement of the sentence (*) Financial compensation to the families of the victims (€ 8ml) (**) (*) execution suspended until final level of judgment (**) immediate execution L’AQUILA 2009 EARTHQUAKE AFFAIR In spite of its considerable length (940 pages), the verdict motivation can be summarised in few lines A High Risk Commission (HRC) meeting was convened in L’Aquila on March 31, 2009, following a 6 month lasting low magnitude seismic sequence culminated with a ML 4.0 the day before; All the seven indicted people are considered member of the Commission but only four actually were; The Commission had peculiar duties of assessment of the risk situation in L'Aquila and correct information to the population; The Commission has failed in those obligations of assessment and information, and all the HRC components are responsible for the violation of such obligations; The fault consisted in approximated evaluation of the situation and in having contributed to spread reassuring messages to the population; Because of this, 29 dead people out of 309 fatalities, that were used to run away from their homes alarmed by earthquakes, remained in their homes on the night of April 6, 2009 and died as a result of the collapse of the buildings in which they had remained. L’AQUILA 2009 EARTHQUAKE AFFAIR The 31 March Meeting The meeting was held with the participation of many local authorities (including L’Aquila Mayor) and other officers all belonging to the National Civil Protection Service. Experts were neither alarmistic nor reassuring, concluding that: large earthquakes are not predictable deterministically a large event in the short-term is “unlikely” L’Aquila is one of the areas in Italy with the highest seismic hazard After the meeting, De Bernardinis and Barberi hold a press conference, reporting the above mentioned conclusions. However no audio recording of the conference is available. No press release; No minutes of the 31/3 meeting were published before April 6. L’AQUILA 2009 EARTHQUAKE AFFAIR Participants at the 31 March Meeting • Government officials • Scientists with appointment as member of the High Risks Committee (HRC) • Scientist (Selvaggi), not formally invited to the meeting, accompanying a member of the HRC (Boschi). Each of them had a different duty decided by the law Government officials of the Civil Protection Government-appointed Scientists (High Risk Committee) Scientist Bernardo De Bernardinis (Vice head of DPC) Franco Barberi (President) Giulio Selvaggi (Director of the National Earthquake Centre, INGV) Mauro Dolce (director Risk mitigation office) Enzo Boschi (member) INGV President Massimo Cialente (Mayor of L’Aquila) Gian Michele Calvi (member) Daniela Stati (Head of Regional Civil Protection) Claudio Eva (member) Preventive Action (including informing the public) Evaluation of Scientific data Provision of Scientific data L’AQUILA 2009 EARTHQUAKE AFFAIR The HRC main faults according to the sentence: HRC and some of its members in several interviews stated that the seismic sequence in course at L’Aquila had to be considered normal and the probability of an impending higher M event was minor; HRC did not behave as a unitarian body because in the meeting only statements were produced without explaination of scientific basis. Hence not all the members were aware of the reliability of the statements; The virdict’s motivations indicate what a legitimate behaviour should have been: « it would be sufficient to not say sentences as:… It is not possible make predictions, ….(seismic) predictions have no scientific base…, the simple obsrevation of several small earthquake is not in itself a precursor, ..some strong earthquakes have been preceded by small shocks in many cases small earthquakes have not been followed by strong events …and so on» In other words, obvious scientific data should be dismissed and hidden. L’AQUILA 2009 EARTHQUAKE AFFAIR Recurrence of Large Earthquakes Preceded by Swarms 698 seismic sequences or single shocks with 3.5<M>5.0 occurred in Italy in the period 1.1.1983-31.3.2009 None of them was followed by a M>5.5 earthquake within 20 days and anda radius of 200 km According to Marzocchi and Zhuang, (2011) about 8/1000 M4 earthquakes are followed by strong events. L’AQUILA 2009 EARTHQUAKE AFFAIR Forecasting the mainshock (made retrospectively) L’AQUILA 2009 EARTHQUAKE AFFAIR The impact on society From Discharge of Energy to Discharge of Responsibility Schools evacuated after a 2.1 magnitude earthquake in two towns 2,000 earthquakes with magnitude larger than 2.1 each year in Italy L’AQUILA 2009 EARTHQUAKE AFFAIR Seismic sequence in Garfagnana (Tuscany) M4.8 on Jan 25 2013, then M3.2 on Jan 30 1. In the weekly seismic bulletin to DPC, INGV reports the largest aftershock of the sequence (M3.2), and its slightly offset location, stating that other aftershocks were likely 2. After several hours, the report is transmitted from DPC as it is to Regione Toscana and to the municipalities Final result: the twit from the Mayor “Following the communication received from the Prime Minister office, we suggest to leave your house and stay outside” 3. At least 25,000 stayed out of their houses overnight. a L’AQUILA 2009 EARTHQUAKE AFFAIR Consequences of L’Aquila Verdict A defensive attitude from emergency managers meaning to give out decisional responsibility; A general loss of credibility; Indiscriminate application of the “precautionary principle” will be encouraged; Precautionary principle can give catastrophic consequences if applied without a clear statement of the risk that a given community is keen to accept; The precautionary principle in an absolute sense cannot operate; Demand for a technology to be put into operation without risk is a metaphysics desire; We have to take risks, but risks must be balanced with a calculation; An absolutist use of the precautionary principle will stop forms of scientific investigations and application of innovative techniques even for the protection of humanity. DEALING WITH VERY LOW PROBABILITY International Commission on Earthquake Forecasting (ICEF) Charged on 11 May 2009 by Dipartimento della Protezione Civile (DPC) to: 1. 2. Report on the current state of knowledge of short-term prediction and forecasting of tectonic earthquakes Indicate guidelines for utilization of possible forerunners of large earthquakes to drive civil protection actions ICEF report: “Operational Earthquake Forecasting: State of Knowledge and Guidelines for Utilization” Findings & recommendations released by DPC (Oct 2009) and endorsed by IASPEI (July 2011) Final report published in Annals of Geophysics (Aug 2011) Members (9 countries): T. H. Jordan, Chair, USA Y.-T. Chen, China P. Gasparini, Secretary, Italy R. Madariaga, France I. Main, United Kingdom W. Marzocchi, Italy G. Papadopoulos, Greece G. Sobolev, Russia K. Yamaoka, Japan J. Zschau, Germany http://www.annalsofgeophysics.eu/index.php/annals/article/view/5350 DEALING WITH VERY LOW PROBABILITY Issues with Operational Earthquake Forecasting While the probability gains of short-term, seismicity-based forecasts can be high (> 100 relative to long-term forecasts), the daily/weekly probabilities of large earthquakes typically remain low (< 1% per day) Standardization of OEF procedures is in a nascent stage of development Preparedness actions appropriate in such high-gain, low-probability situations have not been systematically investigated Incremental benefits of OEF for civil protection (e.g., relative to long-term seismic hazard analysis) have not been convincingly demonstrated Under these circumstances, governmental agencies with statutory responsibilities for earthquake forecasting have been cautious in developing operational capabilities DEALING WITH VERY LOW PROBABILITY Almost all short-term earthquake forecasting models are based on seismic clustering. ETAS (Epidemic-Type Aftershock Sequence) is probably the most diffuse Poisson background Power-law spatial decay Omori (modified) law Utsu law Gutenberg-Richter law A POSSIBLE SCENARIO OF INTERACTIONS No singles models are used! ENSEMBLE forecasting model. Each model is weighted according to its forecasting performances Selected models (constraints) They must be submitted to at least one CSEP experiment DEALING WITH VERY LOW PROBABILITY Forecast Validation ICEF Criteria for “Operational Fitness” Quality of the forecast Reliability and skill Retrospective and prospective testing Consistency among forecasts Across temporal and spatial scales Need for integrated model development Value of the forecast to users Economic cost/benefit analysis; psychological value Pre-set alert thresholds Transparent messaging system; public education DEALING WITH VERY LOW PROBABILITY Collaboratory for the Study of Earthquake Predictability CSEP goal is rigorous testing of predictability hypotheses and forecasting models: Automates blind, prospective testing in a standardized, controlled environment Current testing is only “quasi-prospective” owing to catalog latencies Has established experiments in a variety of tectonic environments and on a global scale CSEP components: Natural laboratories comprising active fault systems with adequate, authorized data sources for conducting forecasting experiments Testing centers with validated procedures for registering and evaluating prediction experiments Model classes with common target events, forecasting regions, and forecast updating intervals A POSSIBLE SCENARIO OF INTERACTIONS The Valuation Problem Economic valuation is one basis for prioritizing how to allocate the limited resources available for short-term preparedness In a low-probability environment, only low-cost actions are justified Cost-Benefit Analysis for Binary Decision-Making (e.g., van Stiphout et al., 2010) Suppose cost of protection against loss L is C < L. If the short-term earthquake probability is P, the policy that minimizes the expected expense E: - Protect if P > C/L - Do not protect if P < C/L Then, E = min {C, PL}. However, many factors complicate this rational approach Monetary valuation of life, historical structures, etc. is difficult Valuation must account for information available in the absence of forecast Official actions can incur intangible costs (e.g., loss of credibility) and benefits (e.g., gains in psychological preparedness and resilience) ⇒ A POSSIBLE SCENARIO OF INTERACTIONS Short-Term Earthquake Risk Assessment (2) Time-dependent Probabilistic Seismic Hazard Assessment Loss Estimation Time-dependent Probabilistic Seismic Risk Assessment Given the probability P that an earthquake has a certain impact, ... ... should mitigation actions be taken or not? C is financial effort to mitigate an earthquake risk: #Evacueed People x socio-economic cost per day x day PLC L is financial loss due to an earthquake: #Fatalities x Willingness to pay for human life saved Optimal policy for decision making: P> C mitigation action is favorable L P< C mitigation action is NOT favorable L CBA A POSSIBLE SCENARIO OF INTERACTIONS Evolution of Probabilistic Risk in L'Aquila Probabilistic risk of having 100 fatalities as a function of time in the last month prior the L'Aquila MW6.3 earthquake Seismicity within 50 km distance to L'Aquila in the last month prior the MW6.3 earthquake A POSSIBLE SCENARIO OF INTERACTIONS In September 2009, during construction work for a new section of roadway, a crack was discovered in an eyebar on the eastern span of the San Francisco – Oakland Bay Bridge. According to Caltrans, this problem was significant enough to have closed the bridge on its own. During a seismic sequence, additional infrastructure inspections might be scheduled to detect seismic defects. A 50-foot section of the Bay Bridge was damaged in the 1989 Loma Prieta earthquake A POSSIBLE SCENARIO OF INTERACTIONS During a seismic sequence, existing seismic retrofit programs might be accelerated at a rate commensurate with the increase in the threat level. None of the recently retrofitted structures at the Anheuser-Busch brewery in Van Nuys, California were damaged in the 1994 Northridge earthquake A POSSIBLE SCENARIO OF INTERACTIONS Earthquake hazard is a low-probability high-consequence international peril like pandemic and terrorism. There are public pandemic and terrorism alerts, which are advisory not compulsory, and do not cause panic. Increase in seismic risk advisories may also be provided to the public. CONCLUSION Something to remember: In all sciences, there is always uncertainty, and under conditions of uncertainty, we have to take our decisions; “We take our decisions in the twilight of probability, and not into the clear light of certainties”(John Locke); “The calculus of probability is essential to decide upon our conduct” (Blaise Pascal); Dealing with natual hazard we deal with very low probabilities and high uncertainty; The concept of probability is not included in the legislation of many countires, including Italy. In Italy there is no legislation on insurance for NH; The final aim should be for people being able to make decisions individually within a probabilistic framework. Data and protocols must be transparent and readily available to the public. AN UNCOMMENSURABLE BUT PRECIOUS TASK!