Papers by Goran Trendafiloski
The worldwide ratio of injured to fatalities in earthquakes, R = Inj/Fat, has increased over time... more The worldwide ratio of injured to fatalities in earthquakes, R = Inj/Fat, has increased over time. This shows that it is more likely by approximately a factor of 2 that a person survives an earthquake today than 50 years ago. However, any meaningful analysis of R requires (as a minimum) separation by type of country and by location of epicentres (land or offshore). R in earthquakes beneath land is typically half of that for events offshore. R in the industrialised world is about two to three times larger than in the developing world. The countries that have made the greatest progress in protecting their population are Japan and China. Countries where R has not increased with time include Iran, Turkey, and Greece. The basic trends are clear, but the data sets for some individual countries are too small for the averages to be considered firm. We propose to use R to adjust the casualty matrices for estimation of human losses due to earthquakes worldwide.
Currently, we are constructing our second-generation loss estimation tool QLARM (earthQuake Loss ... more Currently, we are constructing our second-generation loss estimation tool QLARM (earthQuake Loss Assessment for Response and Mitigation) and upgrading the input database to be used in real-time and scenario mode. Our tool and database are open to all scientific users. The estimates include: (1) total number of fatalities and injured, (2) casualties by settlement, (3) percent of buildings in five damage grades, and (4) a map showing mean damage by settlement. The QLARM worldwide database of the elements-at-risk consists of city models constructed with the following parameters: (1) soil amplification factors, (2) distribution of building stock and population into vulnerability classes of the European Macroseismic Scale (EMS-98). We calculate damage and losses using vulnerability curves, regionally-based collapse models, and casualty matrices pertinent to EMS-98 vulnerability classes as a function of the seismic intensity. We calibrate our tool for different countries and regions world...
A Tsunami Model for Chile for (Re) Insurance Purposes
Egu General Assembly Conference Abstracts, May 1, 2014
Catastrophe models help (re)insurers to understand the financial implications of catastrophic eve... more Catastrophe models help (re)insurers to understand the financial implications of catastrophic events such as earthquakes and tsunamis. In earthquake-prone regions such as Chile,(re)insurers need more sophisticated tools to quantify the risks facing their businesses, including models with the ability to estimate secondary losses. The 2010 (M8.8) Maule (Chile) earthquake highlighted the need for quantifying losses from secondary perils such as tsunamis, which can contribute to the overall event losses but are not often modelled. This paper presents some key modelling aspects of a new earthquake catastrophe model for Chile developed by Impact Forecasting in collaboration with Aon Benfield Research partners, focusing on the tsunami component. The model has the capability to model tsunami as a secondary peril - losses due to earthquake (ground-shaking) and induced tsunamis along the Chilean coast are quantified in a probabilistic manner, and also for historical scenarios. The model is implemented in the IF catastrophe modelling platform, ELEMENTS. The probabilistic modelling of earthquake-induced tsunamis uses a stochastic event set that is consistent with the seismic (ground shaking) hazard developed for Chile, representing simulations of earthquake occurrence patterns for the region. Criteria for selecting tsunamigenic events (from the stochastic event set) are proposed which take into consideration earthquake location, depth and the resulting seabed vertical displacement and tsunami inundation depths at the coast. The source modelling software RuptGen by Babeyko (2007) was used to calculate static seabed vertical displacement resulting from earthquake slip. More than 3,600 events were selected for tsunami simulations. Deep and shallow water wave propagation is modelled using the Delft3D modelling suite, which is a state-of-the-art software developed by Deltares. The Delft3D-FLOW module is used in 2-dimensional hydrodynamic simulation settings with non-steady flow. Earthquake-induced static seabed vertical displacement is used as an input boundary condition to the model. The model is hierarchically set up with three nested domain levels; with 250 domains in total covering the entire Chilean coast. Spatial grid-cell resolution is equal to the native SRTM resolution of approximately 90m. In addition to the stochastic events, the 1960 (M9.5) Valdivia and 2010 (M8.8) Maule earthquakes are modelled. The modelled tsunami inundation map for the 2010 Maule event is validated through comparison with real observations. The vulnerability component consists of an extensive damage curves database, including curves for buildings, contents and business interruption for 21 occupancies, 24 structural types and two secondary modifies such as building height and period of construction. The building damage curves are developed by use of load-based method in which the building's capacity to resist tsunami loads is treated as equivalent to the design earthquake load capacity. The contents damage and business interruption curves are developed by use of deductive approach i.e. HAZUS flood vulnerability and business function restoration models are adapted for detailed occupancies and then assigned to the dominant structural types in Chile. The vulnerability component is validated through model overall back testing by use of observed aggregated earthquake and tsunami losses for client portfolios for 2010 Maule earthquake.
Early-warning of imminently approaching strong shaking that could have fatal consequences is a re... more Early-warning of imminently approaching strong shaking that could have fatal consequences is a research field that has made great progress. However, its potential to save lives has a serious Achilles heel: The time for getting to safety is 5 to 10 s only, in those cities that need warnings most. Occupants of the upper floors cannot get out of their buildings and narrow streets are not a safe place in strong earthquakes. The official advice, to crawl under a table, protects from falling objects only, but not from collapsing buildings. Thus, only about 10 % of a city's population can benefit from early-warnings, unless they have access to an earthquake protection unit that is strong enough to improve their chances of survival and not being injured by factors of 1,000 to 30,000. The purpose of this paper is to generate discussion on how to use early-warning to improve the safety for occupants of buildings exposed to strong earthquake shaking.
Since 2003, 335 loss reports after major earthquakes worldwide have been posted with an average d... more Since 2003, 335 loss reports after major earthquakes worldwide have been posted with an average delay of 40 minutes. Loss estimates we compute contain a map showing the average degree of damage in settlements near the epicenter, the total number of fatalities, the total number of injured, and the detailed list of the casualties and damage rates in these settlements. We are in the process of upgrading the computer code and data base to generate an open source tool: QUAKELOSS2. The input parameters that need to be calibrated on a regional to local basis include: (1) Attenuation relationship, (2) soil conditions, (3) building stock properties, and (4) the human losses matrix. Progress on points and is reported in a parallel abstract, where calculated results are compared to observed ones for M6.6 Bam earthquake. Here, we present the preliminary results on points (1) and .
The ratio of injured to fatalities in earthquakes, estimated from intensity and building properties
ABSTRACT
The Consequences Of The 2009 M7.6 Padang Intermediate Depth Earthquake: Rethinking Loss Estimation In Subduction Zones
ABSTRACT
Loss estimates we compute contain a map showing the average degree of damage in settlements near ... more Loss estimates we compute contain a map showing the average degree of damage in settlements near the epicenter, the total number of fatalities, the total number of injured, and a detailed list of the casualties and damage rates in these settlements.
Earthquake Spectra, 2009
We propose a method to calculate damage and human losses for cities in the developing world by av... more We propose a method to calculate damage and human losses for cities in the developing world by averaging over an entire city, or its administrative districts. Bucharest, Romania, serves as an example. First, we modeled this city as located at a single coordinate point. We transformed the census information on building types, ages and height into EMS-98 vulnerability classes and distribute the population into them. We assumed a seismic load of MSK Intensity= 8 (M7.4 1977 Vrancea earthquake). Validating our model by comparison with casualties reported in 1977, we find differences of 20% to 30%. We reduced these errors to about 4%, by adjusting the distribution of building types into vulnerability classes, based on their performance in the 1977 earthquake. Calibrations of this type will be necessary for most developing countries. In a second step, we modeled Bucharest with six districts, in which the distribution of people into building types and the average soil conditions are known. This is our preferred model. We also calculated the soil properties that would be estimated from topography, if microzonation would not be available. The agreement was satisfactory. We propose this method to model important cities in earthquake prone areas of the developing world.
Teleseismic Loss Estimates in Near-Real-Time After the M8 Wenchuan Earthquake of May 12, 2008
After large earthquakes, the true extent of losses is not known for days and weeks. In the Wenchu... more After large earthquakes, the true extent of losses is not known for days and weeks. In the Wenchuan case, the sum of fatalities and missing persons reported by news agencies began to reflect the true disaster only from the 6th day on. Therefore, it is important to estimate losses quantitatively in near-real-time, based on world data sets on population, building
Uploads
Papers by Goran Trendafiloski