Home
About ElarmS
Application? - tell us
California
South Korea
Turkey
Methodology
Research Highlights
Technical Publications
Press Highlights
Contact us
Acknowledgements


Istanbul, Turkey

Istanbul is located adjacent to the western end of the North Anatolian Fault which runs east-west across Turkey. Just like the San Andreas Fault in California, the North Anatolian Fault is a right-lateral strike-slip fault and about 600 miles long. During the last century a sequence of large earthquakes occurred along the North Anatolian Fault starting at the eastern end and migrating west as shown in the map below. The most recent was the Mw 7.4, August 17, 1999 Izmit earthquake which killed an estimated 20,000 and ruptured the segment of the fault just east of the Marmara Sea.

Above: Map of past earthquakes along the North Anatolian Fault. Note the progression of earthquakes toward Istanbul. Image provided by the USGS.

Following the Izmit earthquake the Kandilli Observatory of Bogazici University installed a strong motion seismic network throughout Istanbul on the northern shores of the Marmara Sea. One goal of the network is to provide earthquake early warning to the population of Istanbul. Here we show the results of applying the ElarmS methodology to data from three earthquakes recorded by the strong motion network in Istanbul.


M5.2 October 24, 2006 in the southern Marmara Sea


This is the largest event for which data is available. One second after the event was detected the magnitude was accurately estimated to be 5.2 and the error in the predicted ground shaking intensity was 0 +/- 0.2 (MMI scale). This was 5-10 sec before peak shaking in Instanbul. The second-by-second predictive AlertMaps output by Elarms can be viewed.


M4.2 September 29, 2004 in the central Marmara Sea


This event is much closer to Istanbul, located in the central Marmara Sea just south of the city. The initial magnitude estimate is 4.5 and increases to 4.8 with time. The error in the initial intensity estimate is -0.1 +/- 0.3 and remains fairly constant for the duration of the event. There are about 4 sec from the initial ground motion prediction until peak shaking in Istanbul. The second-by-second predictive AlertMaps output by Elarms can be viewed.


M4.3 May 16, 2004 in the eastern Marmara Sea


This event occured just west of the Izmit earthquake. ElarmS struggles with the magnitude estimate for this event initially estimating a magnitude of 5.0 but then increasing the estimate to 5.7 several seconds later. This results in an initially high ground shaking estimate with an average error of 0.6 +/- 0.3 (MMI scale) which increases to 1.3 +/- 0.3 when the magnitude estimate increases to 5.4. However, 1 sec later the ground shking prediction is corrected as the first peak observations are made reducing the MMI error to a more normal 0.4 +/- 0.2. The second-by-second predictive AlertMaps output by Elarms can be viewed.


Some concluding observations

  • This test was done without any regional tuning. ElarmS would likely perform better if region-specific scaling relations were developed and incorporated.
  • Even though the error in the magnitude estimate was unusually large for the May 16, 2004 earthquake, the error in the ground shaking, which is what matters to people, was small.
  • Istanbul would greatly benefit from a few sea-floor seismometers close to the fault.

Acknowledgements
The data from the Istanbul Strong Motion Network was made available by Mustafa Erdik and Hakan Alcik of the Kandilli Observatory, Bogazici University. The data was processed by Richard Allen at UC Berkeley.