1. Operational Earthquake Loss Forecasting in Italy: Preliminary Results I. Iervolino, 1 E. Chioccarelli, 1 M. Giorgio, 2 W. Marzocchi, 3 A. Lombardi, 3 G. Zuccaro, 1 F. Cacace. 1 1 Università degli studi di Napoli Federico II, Naples, Italy. 2 Seconda Università degli Studi di Napoli, Aversa (CE), Italy. 3 Istituto Nazionale di Geofisica e Vulcanologia, Rome, Italy. Second European Conference on Earthquake Engineering and Seismology – Istanbul (Turkey), August 25 th 2014.

2. I. Iervolino – 2ECEES (TR), August 25 th 2014. IntroductionHazard, vulnerability, and exposure models for Italy Passing from M 4+ rates to loss MANTIS K. 1/10 The CASSANDRA system of INGV provides weekly rates of events with magnitude (M) 4+ for a 0.1° grid including the whole country. (Updated daily.) The Italian Civil Protection asked to investigate whether it is possible (and useful) to use the INGV data to produce consequence estimates to be used as a decision support system for seismic risk management. The framework is that of performance-based earthquake engineering, that is including probabilistic measures of hazard, vulnerability and exposure at a National scale. Introduction

3. I. Iervolino – 2ECEES (TR), August 25 th 2014. IntroductionHazard, vulnerability, and exposure models for Italy Passing from M 4+ rates to loss MANTIS K. 1. Probabilistic seismic hazard analysis based on OEF 2/10 Area of interest Site of interest (w,z) Source cell (x,y) R(x,y,w,z) The goal here is to pass from OEF rates at a cell to rates of events causing some seismic intensity (MS) at a site of interest (for example one where exposure to seismic risk exists) Rates of events at (w,z) with MS=ms because of earthquakes occurring at (x,y) OEF rates at (x,y) MS prediction equation (attenuation law) Distribution of M for earthquakes at (x,y) Indicating dependence on recorded history Indicating that varies with time (OEF updates) Passing from M 4+ rates to loss

4. I. Iervolino – 2ECEES (TR), August 25 th 2014. IntroductionHazard, vulnerability, and exposure models for Italy Passing from M 4+ rates to loss MANTIS K. The goal here is to pass from OEF rates at a cell to rates of events causing some damage state (DS) in a building of certain structural typology (k) at site of interest 2. Weekly rates of events causing building damage 3/10 Area of interest Site of interest (w,z) Source cell (x,y) R(x,y,w,z) Rates of events at (w,z) causing DS=ds because of earthquakes occurring in the whole area OEF rates at (x,y) Same as per the previous slide Probability of some damage state to a structure of typology K given ms intensity Summing up over all source cells gives the total damage rate at (w,z) site Passing from M 4+ rates to loss

5. I. Iervolino – 2ECEES (TR), August 25 th 2014. IntroductionHazard, vulnerability, and exposure models for Italy Passing from M 4+ rates to loss MANTIS K. The goal here is to pass from OEF rates at a cell to rates of events causing some individual consequence to an occupant of a building of a given structural typology (k) at site of interest 3. Weekly rates of events causing individual loss 4/10 Area of interest Site of interest (w,z) Source cell (x,y) R(x,y,w,z) Rates of events at (w,z) causing casualty in buildings of typology k OEF rates at (x,y) Same as per the previous slide Probability of casualty in a building of typology K given damage state DS=ds Summation is to account that casualty can be caused by any DS Passing from M 4+ rates to loss

6. I. Iervolino – 2ECEES (TR), August 25 th 2014. IntroductionHazard, vulnerability, and exposure models for Italy Passing from M 4+ rates to loss MANTIS K. Expected losses in the week after the OEF rates release 5/10 In the short-term, it may be assumed that the rates just shown are constant, that is the occurrence of events follows a point-wise Poisson stochastic process, the parameter of which is updated at each OEF release. Expected number of buildings of typology k at (w,z) in damage state DS=ds in the week after OEF rates release Buildings of typology k at (w,z) Rate of events causing damage to typology k at (w,z) One week Expected number of casualties due to damage to buildings of typology k at (w,z) in the week after OEF rates release Occupants in buildings of typology k at (w,z) Rate of events casualties in buildings of typology k at (w,z) One week Expected number of displaced and shelter- seeking people Expected number of fatalities and injuries Passing from M 4+ rates to loss

7. I. Iervolino – 2ECEES (TR), August 25 th 2014. IntroductionHazard, vulnerability, and exposure models for Italy Passing from M 4+ rates to loss MANTIS K. Operational earthquake loss forecasting (OELF) procedure summary Seismic swarm area Site (municipality) of interest (w,z) Source cell (x,y) Rate of events in the source cell of interest from OEF Probability of any macroseismic intensity level given the event Inventory data for buildings at the site of interest Exposure data for residents at the site of interest Vulnerability model Intensity attenuation law Probability of casualties or injuries for any damage level in any vulnerability class given the event Sum-up over all source cells Probability of any damage level for any vulnerability class given the event 6/10 Expected number of damaged buildings Expected number of injures, fatalities, displaced people Hazard, vulnerability, and exposure models for Italy

8. I. Iervolino – 2ECEES (TR), August 25 th 2014. IntroductionHazard, vulnerability, and exposure models for Italy Passing from M 4+ rates to loss MANTIS K. 7/10 Vulnerability based on damage probability matrices Class MSDS0DS1DS2DS3DS4DS5 A50.34870.40890.19190.04500.00530.0002 B50.52770.35980.09810.01340.00090.0000 C50.65910.28660.04980.00430.00020.0000 D50.85870.13280.00820.00030.0000 A60.28870.40720.22970.06480.00910.0005 B60.44370.39150.13820.02440.00220.0001 C60.59050.32810.07290.00810.00050.0000 D60.77380.20360.02140.00110.0000 Hazard, vulnerability, and exposure models for Italy LossStructural TypologyVulnerability ClassDS0DS1DS2DS3DS4DS5 FatalitiesMasonryA or B or C00000.040.15 FatalitiesR.C.C or D * 00000.080.3 InjuriesMasonryA or B or C00000.140.7 InjuriesR.C.C or D * 00000.120.5 Probabilities of casualty given structural damage Exposure by municipality CodeNameProv.ABCDab_Aab_Bab_Cab_D 1001Agliè001222163286186697535350990 1002Airasca00175601521384973513572350 1003Ala di Stura001186209220472181006495 1004Albiano d'Ivrea0011921478084646419199432 1005Alice Superiore001136121857617711651270 1006Almese00126131851179210067415473364 1007Alpette001144125122421165132101 1008Alpignano00122228862083212141400157312461 1009Andezeno00115311083116512315164714 1010Andrate001141131123442501046162 Buildings per vulnerability class Residents per vulnerability class

9. I. Iervolino – 2ECEES (TR), August 25 th 2014. IntroductionHazard, vulnerability, and exposure models for Italy Passing from M 4+ rates to loss MANTIS K. 8/10 MANTIS K. MANTIS K. The system for continous OELF in Italy

10. I. Iervolino – 2ECEES (TR), August 25 th 2014. IntroductionHazard, vulnerability, and exposure models for Italy Passing from M 4+ rates to loss MANTIS K. Local OELF at 00:00 of August 24 2014 (around the cell with maximum OEF rate) Distance from the maximum rate Total number of buildings Total number of residents Collapsed buildings DisplacedInjuriesFatalities < 10km 4,60727,4440.050.670.0360.010 < 30km 69,142425,3560.376.210.2400.063 < 50km 298,6281,891,6371.1423.260.8030.212 < 70km 456,730264,908,31.7431.621.0970.289 Expected weekly losses (Aug. 24-31) Area of interest 9/10 MANTIS K.

11. I. Iervolino – 2ECEES (TR), August 25 th 2014. IntroductionHazard, vulnerability, and exposure models for Italy Passing from M 4+ rates to loss MANTIS K. Summary and conclusions The study discussed the feasibility of probabilistic short-term seismic loss (risk) assessment in Italy, based on OEF. Risk metrics investigated are the expected number of fatalities, injuries, and displaced residents in one week. Probabilistically-consistent short-term seismic risk assessment in Italy appears to be feasible, yet it is conditional to the OEF and vulnerability/exposure models available. Risk measures seem to be sensitive to the short-term seismicity variations inferred by OEF, which provide the largest seismicity right after the mainshock. A prototypal system for continuous OELF analysis, MANTIS K., is currently under experimentation. It automatically receives OEF data from the CASSANDRA system of INGV and performs OELF for the whole country as well as for the region around the location where the largest seismicity is observed. 10/10

12. Operational Earthquake Loss Forecasting in Italy: Preliminary Results I. Iervolino, 1 E. Chioccarelli, 1 M. Giorgio, 2 W. Marzocchi, 3 A. Lombardi, 3 G. Zuccaro, 1 F. Cacace. 1 1 Università degli studi di Napoli Federico II, Naples, Italy. 2 Seconda Università degli Studi di Napoli, Aversa (CE), Italy. 3 Istituto Nazionale di Geofisica e Vulcanologia, Rome, Italy. Second European Conference on Earthquake Engineering and Seismology – Istanbul (Turkey), August 25 th 2014.