Understanding the Principles of Earthquake Modeling from the 1999 Athens Earthquake Event
Callum HigginsSeptember 03, 2015
The 1999 Athens Earthquake occurred on September 7, 1999, registering a moment-magnitude of 6.0 (USGS). The tremor’s epicenter was located approximately 17km to the northwest of the city center. Its proximity to the Athens Metropolitan Area resulted in widespread structural damage.
More than 100 buildings including three major factories across the area collapsed. Overall, 143 people lost their lives and more than 2,000 were treated for injuries in what eventually became Greece’s deadliest natural disaster in almost half a century. In total the event caused total economic losses of $3.5 billion, while insured loss was $130 million (AXCO).
Losses from such events can often be difficult to predict; historical experience alone is inadequate to predict future losses. Earthquake models can assist in effectively managing this risk, but must take into account the unique features that the earthquake hazard presents, as the 1999 Athens Earthquake event highlights.
Background seismicity must be considered to capture all potential earthquake events
The 1999 event took Greek seismologists by surprise as it came from a previously unknown fault. Such events present a challenge to (re)insurers as they may not be aware of the risk to properties in the area, and have no historical basis for comparison. Effective earthquake models must not only incorporate events on known fault structures, but also capture the background seismicity. This allows potential events on unknown or complicated fault structures to be recorded, ensuring that the full spectrum of possible earthquake events is captured.
Hazard can vary greatly over a small geographical distance due to local site conditions
Soil type had significant implications in this event. Athens has grown tremendously with the expansion of the population into areas of poorer soil in the suburbs, with many industrial areas concentrated along the alluvial basins of the Kifissos and Ilisos rivers. This has increased the seismic hazard greatly with such soils amplifying the ground motions of an earthquake.
The non-uniform soil conditions across the Athens region resulted in an uneven distribution of severe damage in certain regions. The town of Adames in particular, located on the eastern side of the Kifissos river canyon, experienced unexpectedly heavy damage wheras other towns of equal distance to the epicenter, such as Kamatero, experienced slight damage. (Assimaki et al. 2005)
Earthquake models must take such site-specific effects into account in order to provide a local view of the hazard. In order to achieve this, high-resolution geotechnical data, including information on the soil type, is utilized to determine how ground motions are converted to ground shaking at a specific site, allowing for effective differentiation between risks on a location level basis.
Building properties have a large impact upon damageability
The 1999 Athens event resulted in the severe structural damage to, in some cases the partial or total collapse of, number of reinforced concrete frame structures. Most of these severely damaged structures were designed according to older seismic codes, only able to withstand significantly lower forces than those experienced during the earthquake. (Elenas, 2003)
Earthquake models must account for such differences in building construction and age. Variations in local seismic codes and construction practices the vulnerability of structures can change greatly between different countries and regions, with it important to factor these geographical contrasts in. It is important for earthquake models to capture these geographical differences of building codes and this can be done through the regionalization of vulnerability.
Additionally, the Athens earthquake predominantly affected both low and middle rise buildings of two to four stories. The measured spectral acceleration (a unit describing the maximum acceleration of a building during an earthquake) decreased rapidly for buildings with five stories or more, indicating that this particular event did not affect high rise buildings severely. (Anastasiadis et al. 1999)
Spectral response based methodology most accurately estimates damage, modeling a building’s actual response to ground motions. This response is highly dependent upon building height. Due to the smaller natural period at which low and middle rise buildings oscillate or sway, they respond greater to higher frequency seismic waves such as those generated by the 1999 Athens event; while the reaction of high rise buildings is the opposite, responding the most to long period seismic waves.
The key features of the RMS Europe Earthquake Models ensure the accurate modeling of events such as the 1999 Athens Earthquake, providing a tool to effectively underwrite and manage earthquake risk across the breadth of Europe.
You May Also Like
October 06, 2022
Hurricane Ian: The Challenge of Getting Insights into Your Business Faster
From our numerous client conversations, climate change as a business issue has risen high on the agenda, and this has certainly escalated over the last twelve months. There is a growing recognition of the need to quantify the impact that climate change will have on your business. But – where do you start with this? One of the major challenges is knowing what question to ask. With the inclusion of climate change scenarios within the General Insurance Stress Test (GIST 2019), which the larger U.K. insurers and Lloyd’s syndicates are required to respond to, the Bank of England Prudential Regulation Authority (PRA) is outlining one approach.
RMS is particularly well placed to support insurers in responding to the “Assumptions to Assess the Impact on an Insurer’s Liabilities” portion of the climate change section within GIST, which examines how changes in U.S. hurricane and U.K. weather risk under different climate change scenarios may affect losses.
Stochastic models are the perfect tools to evaluate such physical climate change risk to liabilities, with the ability to reflect changes to the hazard under different climate change views and providing a clear link between cause and effect. Our contribution to the landmark “Risky Business” report in 2014 looking at sea-level rise in the U.S. to 2100 is a key example of this. As such, RMS has developed internally adjusted views of its U.S. hurricane, U.S. flood, U.K. windstorm and U.K. flood models to reflect most of the assumptions and scenarios from the PRA, detailed in the table below:
The PRA is asking for the potential impact of these assumptions and scenarios on the Annual Average Loss (AAL) and 1-in-100 Aggregate Exceedance Probability (AEP) loss for all relevant U.S. and U.K. insurance contracts. Getting to a reasonable assessment of these numbers however is not a trivial exercise, requiring the appropriate adjustment of model data in up to 18 possible assumption scenario combinations, and then the analyses of the relevant exposure against these.
To help insurers start thinking about how to respond to the PRA request, RMS can provide broad industry-wide factors derived by running industry exposure over the adjusted models. This “Industry Factors Package” will be made available to RMS clients, while others who wish to access these will be able to license them separately. The industry-wide factors will allow for the approximation of losses under the assumptions and scenarios laid out in the table above, however there could be significant limitations to this approach for individual companies and portfolios. Your exposure, or risk profile, will not reflect that of the industry and therefore the application of industry-wide factors may not reasonably reflect your own risk. The uncertainty around this approach means you may decide this is not a satisfactory solution for your submission.
For a more detailed bespoke view, we are offering to run insurers’ own exposures through the adjusted models, via RMS Analytical Services, to better satisfy the PRA’s requirements. This “PRA-ready Package” provides unique results for submission to the PRA which reflect your book of business and allow for comparisons with those of the industry.
Even if you fall out of the larger U.K. insurers and Lloyd’s syndicates for whom this applies, you should take note. This might be the start of a new wave of analytical rigor around climate change, and more regulators are likely to follow. Beyond regulation, it is also becoming fundamental to understand the impact of climate change for business decisions. For example, to answer what is insurable in 2050 and whether you need to adjust your underwriting and portfolio management strategy accordingly.
RMS can assist in getting answers to such questions through a customized climate change consulting engagement as part of an “Enhanced Climate Change Package”, utilizing advanced climate change analytics to provide more detailed results based on the PRA or other similar scenarios. This package can include the PRA-ready results and basic PRA climate change submission information or be a separate engagement depending upon your needs.
RMS clients who are interested in these solutions should reach out to their Client Success Manager for details on how they can be accessed, while other insurers can email firstname.lastname@example.org. With the submission date of October 31 looming, and many with tighter internal deadlines, it is important not to delay! Indeed, we are already engaging with several clients on how we can help them.…
Callum is the product manager for RMS Event Response, including RMS HWind, and is based in London. Most recently he has been focused on improving client workflows through the integration of event response with the RMS Intelligent Risk Platform.
Previously, Callum has worked on climate change initiatives at RMS as well as updates to the India Agricultural Model and Australia Cyclone Model.
Callum is a Certified Catastrophe Risk Analyst and holds an integrated master’s degree (MEarthSci) in Earth Sciences from Oxford University.