The 1000-km-long North Anatolian fault separates the Eurasian plate from the Turkey plate, whose westward motion at 2 cm/yr is facilitated incrementally in earthquakes. Eleven damaging Mw>7 earthquakes on the fault started in 1939 and have sequentially ruptured westwards towards Istanbul (Figure 1). Previous sequences in the 10th and 16th centuries concluded with an earthquake near Istanbul with catastrophic damage to the city. None doubt that a damaging earthquake will soon occur near Istanbul, with its population of 8 million. Numerous scientific projects have been initiated to characterize the probable size and timing of this earthquake.

As on the San Andreas fault in California, slip on the North Anatolian fault occurs principally in major earthquakes but partly by processes of aseismic creep, a process whereby the fault slips slowly. Slow slip at depth precedes some major earthquakes, which suggests that it may be possible to monitor processes preceding the occurrence of a future Istanbul earthquake. These slow processes require sensors capable of monitoring very small, long duration signals in the Earth: creepmeters and borehole strainmeters.

The  strainmeters are part of a collaborative project between Bogacizi University, UNAVCO, Boulder, and GFZ, Potsdam, and are  being installed in 100-200 m deep boreholes surrounding Istanbul. The first of these to start operating in 2015 recorded the September Chilean M8.3 earthquake; the others will come on line in 2016.  

The creepmeters consist of 15-m-long graphite rods anchored to one side and buried obliquely across the surface fault. Displacement of the free end of the rod indicates fault slip to a precision of 0.01 mm with a range of 1.7 m. Data are recorded every 30 minutes and transmitted to Boulder via the Iridium satellite once per hour.

The creepmeter at Izmit shows no slip has occurred, but the creepmeter at Ismetpasa, which has a history of slow creep, recorded a double creep event in April 2015 with a total of offset of 3.4 mm (Figure 2). Although the duration of each event was more than a week, 50% of the slip occurred in the first 5 hours and an additional 25% occurred in the first day). Creep here followed surface ruptures of two Mw=7.4 damaging earthquakes on the Anatolian fault in 1943 and 1944. We surmise that creep here has been decaying slowly in rate since then and that the surface fault is now creeping at an average rate of approximately 3.62 mm/year.

We note that a 4 month period of slow slip at 3.6 mm/yr followed by a brief 6 week period of quiescence, preceded the 2015 creep event. Thus the surface fault is capable of three types of behavior: no slip, slow slip, and event-like creep.  Following a creep event,  stress near the fault reduces briefly to a relaxed state that is incapable of inducing fault slip (dashed line A). Hence from May to Sept 2015 we observed no slip on the fault. After a sufficient time has elapsed, however,  the stress in the shallow fault will rise to a level that will again induce slip on the fault.  We suspect the fault will attain this stress level (dashed line B) by January or February 2016. At this time the 2015 sequence may repeat (slow slip followed by a creep event in about August 2016). Alternatively, if no slip occurs on the fault until April 2015, stress may occur as a creep event similar to the 2015 double event.

The importance of our observations is that they emulate the known stick-slip behavior of faults, and they permit forecasts of fault behvior.  The slow slip process (B) occurs in the top tens of meters of the fault and creep events (C) in the top hundred meters.  Below these depths damaging earthquakes occur at longer intervals and release tens of cm or meters of slip.  By studying these harmless processes at Ismetpasa we hope to gain insight into pro...

Please report errors in award information by writing to: awardsearch@nsf.gov.