AIR Currents

August 11, 2009

August 11, 2009

Editor's Note: This month, the residents of Turkey commemorate the 10th anniversary of the Izmit earthquake (also known as the Kocaeli earthquake), described by Turkey's then prime minister Bülent Ecevit as "the heaviest in Turkish history." In this article, senior research engineer Dr. Guillermo Franco and research associate and engineering seismologist Dr. John Alarcon review the historical event, discuss the current scientific debate surrounding whether and to what extent the Izmit earthquake increased seismic risk in Istanbul, and examine the potential impact of a large earthquake in the Istanbul area.

The Izmit and Düzce Earthquakes

Before dawn on August 17th, 1999, a moment magnitude 7.51 earthquake ripped through western Turkey, setting the country's biggest oil refinery aflame and toppling crowded apartment buildings as residents slept. The epicenter was located near the port city of Izmit, the capital of Turkey's Kocaeli province. At a relatively shallow depth of 17 km (10.5 mi), the earthquake ruptured the surface in several locations along its path. Severe ground shaking destroyed thousands of buildings, damaged a naval base, severed underground pipelines, and collapsed bridges along the major highway between Istanbul and Ankara. The friction generated between the two sides of the fault led to a roaring explosion of methane gas, which had been trapped in the Gulf of Izmit. While lasting for just 45 seconds, the event ruptured more than 150 kilometers (93 mi) from Düzce all the way to the Sea of Marmara, cutting a swath of destruction across seven provinces, from Istanbul to Bolu.

Three months later, on November 12th—while residents were rebuilding and many were still living in tent cities—another major earthquake struck western Turkey. This time, the M7.2 quake occurred in the Bolu province near the town of Düzce, about 112 km (70 mi) east of the M7.5 August event. Although this less-densely populated region was not as severely affected as Izmit, it was still devastating by any measure: the event leveled the towns of Kaynaşlı and Düzce and killed nearly 1,000 people.

Cat Bond Figure 1 Event 1
Figure 1. Epicenters of Izmit and Duzce Earthquakes (Source: AIR)

In total, the earthquakes were responsible for more than 19,000 fatalities, 48,000 injuries, and the displacement of approximately half a million people. Property loss was extensive; some 214,000 houses and 30,500 businesses suffered damages ranging from slight to complete collapse. In the end, the earthquakes were estimated to have caused more than €1.4 billion (US$2 billion) in insured losses and some €14 billion (US$20 billion) in total damage in 1999 currency (Swiss Re, 2000).

Dr. Guillermo Franco and Dr. John Alarcon

By: Dr. Guillermo Franco and Dr. John Alarcon

Tectonic Setting and Seismic History

Turkey lies on a slab of continental crust—known as the Anatolian Block—sandwiched between the Arabian, African, and Eurasian plates. Most earthquakes here are caused by the northwards motion of the Arabian plate against the Eurasian plate, pushing the Anatolian Block westwards. This motion occurs along two major strike-slip faults, the North and East Anatolian Faults. The North Anatolian Fault system spans about 1,300 km (800 mi) from eastern Turkey to the Sea of Marmara.

Like California's San Andreas Fault, the North Anatolian Fault is a right-lateral strike-slip fault—which means that if you straddle the fault, the right side moves towards you. Unlike the San Andreas Fault, however, the North Anatolian Fault has shown an apparent directional pattern of earthquake epicenters. Since the devastating 1939 M8.2 Erzincan earthquake, a sequence of large-magnitude (6.7 and larger) ruptures—1939, 1942, 1943, 1944, 1951, 1957, 1967—have progressed in a domino-like pattern moving westward along the North Anatolian Fault system, culminating with the 1999 Izmit quake.

Cat Bond Figure 1 Event 1
Figure 2. Map of historical earthquakes and crustal movements along the North Anatolian Fault in the 20th century (Source: AIR)

Did the Izmit and Düzce Earthquakes Put Istanbul at Risk?

This apparent relentless westward march of earthquake epicenters has fueled an active and sometimes lively scientific debate—still ongoing today—about where the next earthquake along the North Anatolian Fault will occur and how big it will be.

Indeed, two years before the occurrence of the Izmit earthquake, Stein et al. (1997) proposed that, as a result of the progression of stress transfer since the 1939 event, the city of Izmit was possibly next in line (Figure 3). The prediction was soon realized.

Cat Bond Figure 1 Event 1
Figure 3. Cumulative Stress Evolution Resulting from the Sequence of North Anatolian Earthquakes from 1939 to 1992 (Source: Stein, 1997)

Soon after the devastating earthquakes of 1999, some seismologists suggested that the Izmit event indeed triggered the subsequent M7.2 Düzce earthquake and loaded stress on westerly faults near Istanbul. Parsons et al.(2002) suggested that two fault segments—central Marmara and Princes' Island (Figure 4)—were particularly at risk and, as a result, Istanbul now had a 62 percent chance of experiencing a large seismic event (M7 or higher) in the next 30 years.2

Cat Bond Figure 1 Event 1
Figure 4. Locations of the Central Marmara and Princes' Island Fault Segments (Source: AIR)

To estimate the likely magnitude of a possible event in the Istanbul area, Meade et al. (2002) used a block model to estimate fault slip rates and depths in the Marmara Sea region. The model was consistent with evidence that strain has accumulated at a significantly shallower depth in the Marmara Sea region than previously thought, which would imply a smaller potential rupture area. Thus, the expected energy release (magnitude) for an earthquake would be about 2.3 times less than earlier estimates, indicating that Istanbul could experience an earthquake as large as M7.2 and that an event of magnitude 7.4 would be unlikely in the Marmara faults.

In a later study, Erdik et al. (2004) carried out a probabilistic hazard assessment using both time dependent and time independent models. The authors assign probabilities of occurrence to "characteristic" earthquake scenarios for different fault segments in the Marmara Sea; from these, the largest probability (40% probability in the next 50 years) is given to a M7.2 event rupturing a single segment south of Istanbul. They also propose, without assigning a probability, a "credible worst case" scenario of a M7.5 event that would rupture three fault segments in unison.

Yet another recent study by Utkucu et al. (2008) examined the historical record of unruptured segments in the Marmara region and presented three possible sequences of rupture and a range of magnitude for each scenario: 7.1-7.2, 7.3-7.4 and 7.4-7.5. According to the study, a large earthquake has the potential to strike the eastern part of the Sea of Marmara within the next two decades.

While there is little question about the region's seismic risk, there is still significant uncertainty about when an event will manifest near Istanbul and how big it will be. Nevertheless, the scientific community continues to make advances in its understanding of the tectonic setting and seismic hazard of the Marmara region to gain insight into which scenario is most probable.

The Potential Impact

If a segment or segments of the North Anatolian Fault beneath the Marmara Sea were to rupture, what would be the impact? We have some insight from the events of 1999. In the aftermath of the Izmit earthquake, an AIR post-disaster reconnaissance team assessed the damage in several affected locations. The team observed the collapse of modern engineered buildings resulting from poor workmanship, a lack of rigorous structural design, and the use of inferior materials such as concrete mixtures with inappropriate ratios of cement to sand. In addition, many buildings still standing had suffered partial failures of beam to column connections (Figure 5), rendering the structures uninhabitable.

Cat Bond Figure 1 Event 1
Figure 5. Inadequate Beam and Column Connection Revealed During AIR's Post-Disaster Survey in Izmit (Source: AIR)

So-called "soft story" collapses were common where large open spaces on the ground floor used to accommodate commercial or parking space lacked adequate lateral reinforcement (Figure 6).

Cat Bond Figure 1 Event 1
Figure 6. Soft Story Failure in Golcuk, Turkey (Source: AIR)

Most of the deaths and injuries resulted because of the collapse of relatively modern residential structures typical of the region—five or six story apartment buildings composed of reinforced concrete frame with unreinforced masonry infill walls. Yet, at the time, the Turkish building code included strict regulations for seismic resistant design. So why did so many recently-constructed buildings collapse so easily?

The Great Erzincan earthquake of 1939 led to the development of the first seismic regulations in Turkey, beginning in the early 1940s. Since then, numerous revisions have been made, most significantly in 1975 and 1998, and more recently in 2007. Although seismic provisions in Turkey have always been on par with the state of the knowledge at the time, their implementation has often been jeopardized by the lack of available government resources to adequately oversee seismic design and enforce the quality of construction.

This situation was exacerbated in the period preceding the Izmit and Düzce earthquakes. During the previous quarter century, Turkey had experienced a construction boom, driven by the rapid growth of the Turkish economy. Nearly two-thirds of all buildings in Izmit had been constructed during this period and nearly 49% of buildings in the entire country were less than 25 years old (Nunez, 2000).

As a result of the lessons learned from Izmit, a new legislative scheme passed in 2000 outlining the role of government-licensed but private "supervision firms." These supervision firms held the responsibility for verifying the accuracy of building designs and ensuring conformity of the final construction to the original specifications for all private buildings constructed in Istanbul and 27 other at-risk provinces, including those impacted by the 1999 earthquakes.3 Likely to prove the most ambitious to date, the seismic code was revised again in 2007 to include a focus—for the first time—on seismic assessment and retrofitting of the existing building inventory.

Nevertheless, despite significant improvements in building practices and building code enforcement, there is still a considerable amount of work to do both to ensure compliance of future buildings and to complete the retrofit of existing ones, not only in Istanbul but in areas of concentrated exposure along the length of the North Anatolian Fault. Consequently, seismic risk in these areas may remain quite high.

Large Loss Scenario in Istanbul

To investigate the potential financial impact of an earthquake near Istanbul, AIR simulated a M7.1 event on the Princes' Island segment of the North Anatolian Fault system—not a worst case scenario, but certainly a plausible one. This segment last ruptured in 1776 and is commonly associated with a recurrence period of around 270 years.

AIR estimates that a M7.1 event near Princes' Island would result in catastrophic losses for Istanbul and the surrounding region, causing damage to buildings and contents (insurable losses) of about €42 billion (US$60 billion). The residential exposure, which includes single-family and multi-family apartments, accounts for the majority of the losses (see Figure 7). It should be pointed out, however, that the percentage of residential buildings in Turkey actually insured against the earthquake peril is relatively low—at roughly 20%.

Cat Bond Figure 1 Event 1
Figure 7. Insurable Losses from a Simulated M7.1 Event on the Princes' Island Fault (Source: AIR)

The majority of the building stock in the Istanbul area is comprised of reinforced concrete of varying quality. Reinforced concrete buildings with unreinforced masonry infill walls, such as the typical Beşkat (in Turkish, "five-stories"), are expected to sustain most of the damage, as they did in the 1999 Izmit event. Reaching mean damage ratios of almost 20% (depending on site conditions and proximity to the epicentral area), these buildings are expected to fare poorly (see Figure 8).4 With a weak lateral resisting mechanism, deformations are expected to be high, causing severe damages to connections and infill walls.

Cat Bond Figure 1 Event 1
Figure 8. Distribution of Damage to Reinforced Concrete with Unreinforced Masonry Infill Walls (Source: AIR)

Reinforced concrete buildings with shear walls that have been constructed more frequently in the last 10 years—and that comply with stringent code requirements—provide strong lateral resistance and are expected to perform significantly better. Mean damage ratios for this construction type in this scenario reach values of about 7% (see Figure 9) on the southern Marmara coast, and east and west of the Bosphorus (also known as the Istanbul Straight).

Cat Bond Figure 1 Event 1
Figure 9. Distribution of Damage to Reinforced Concrete with Shear Walls (Source: AIR)

Closing Thoughts

Since the 1999 Izmit earthquake, several key partnerships between the Turkish government and the World Bank have been forged. One of their more high-profile initiatives consisted in establishing the Turkish Catastrophe Insurance Pool (TCIP), which has provided coverage to more than 2 million households, becoming among the largest catastrophe insurance pool in the world. With the World Bank's support, the Turkish government has moved forward with plans to upgrade high-risk buildings, bridges and other elements of the city's critical infrastructure. Many of these long-term mitigation endeavors throughout the region are still ongoing.

From a risk management perspective, sophisticated modeling tools are now available to help insurers and reinsurers fully understand the scale of risk they face in this seismically active region—and to develop effective strategies for assessing and managing the potential losses from a high-impact event.


Editor's Note: AIR released an update to its Mediterranean Earthquake Model in August 2009 that incorporates the latest research findings on building vulnerability in the region.

Erdik,M, Demircioglu,M, Sesetyan, K Durukal,E, and. Siyahi,B (2004); Earthquake hazard in Marmara Region, Turkey, Soil Dynamics and Earthquake Engineering, 24, 605–631

Meade,B, Hager,B, McClusky,S, and Reilinger,R (2002); Estimates of Seismic Potential in the Marmara Sea Region from Block Models of Secular Deformation Constrained by Global Positioning System Measurements, Bulletin of the Seismological Society of America, 92, 1, 208–215

Nunez, Ian (2000); Compound Growth or Compound Seismic Risk of Destruction? Some vulnerability lessons from the Izmit, Turkey, Earthquake of 17 August 1999, Presented at Second EuroConference on Global Change and Catastrophe Risk Management: Earthquake Risks in Europe

Parsons, T, Toda, S, Stein, R.S., Barka, A and Dieterich, J. H. (2000); Heightened odds of large earthquakes near Istanbul: An interaction-based probability calculation, Science, 661-665.

Parsons, T. (2004). Recalculated probability of M ≥ 7 earthquakes beneath the Sea of Marmara, Turkey. Journal of Geophysical Research 109 (B05304), pp 21.

Stein, R.S., A. A. Barka and J. H. Dieterich, Progressive failure on the North Anatolian fault since 1939 by earthquake stress triggering, Geophys. J. Int., 128, pp. 594-604, 1997.

Swiss Re, Sigma, No.2, 2000.

Utkucu, M, Kanbur, Z, Alptekin,O, and Sünbül, F (2009); Seismic behaviour of the North Anatolian Fault beneath the Sea of Marmara (NW Turkey): implications for earthquake recurrence times and future seismic hazard, Natural Hazards, 50:45–71


1 The magnitude of this event as estimated by various reporting agencies ranges from 7.4 to 7.6.

2 Interestingly, four years later when new data became available, Parsons (2004) revised this estimate downward to a 41% probability.

3 This provision was later rescinded on the grounds that public duties could not be transferred to private entities.

4 A mean damage ratio of 20% should be interpreted to mean that, on average, most buildings will sustain less than 20% of damage but, on the other hand, some buildings may collapse. Since collapses are less likely, the mean value remains at 20%. Secondary uncertainty implemented in the AIR models accounts for this natural dispersion observed in actual earthquake events.

 

Loading...

Close

You’re almost done.
We need to confirm your email address.
To complete the registration process, please click the link in the email we just sent you.

Unable to subscribe at this moment. Please try again after some time. Contact us if the issue persists.

The email address  is already subscribed.