Author: By Andrei NIKONOV, Dr. Sc. (Geol. & Mineral.) Otto Schmidt Joint Institute of Physics of the Earth, RAS

Articles in this rubric reflect the opinion of the author.- Ed.

The earthquake disaster that hit northwestern Turkey near the Gulf of Izmit on August 17, 1999, stands out from among calamities of this kind at least in two respects. First, in the scale of damage done to the country's economy and in the appalling toll of human lives involved, and, second, in the seismic situation that precipitated the disaster and its consequences. While sufficient data are available on the first count, the scientific aspects of the seismology involved and prospects for the future remain relatively obscure.

Earthquakes of the intensity of 9 points (on the Richter Scale) and magnitude about 7.5 are nothing exceptional in Turkey and in the European-Asiatic sector of what we call the Alpine seismic belt. Scores of quakes have been recorded there during this century. But even then the disaster of August 17, 1999 stands out as something uncommon which can only be compared with calamities like the Messina quake of 1908, the Ashkhabad quake of 1948 and the Spitak disaster of 1988. As for Turkey itself, the scale of the tragedy has been the greatest for more than a century. The death toll exceeded 18 thousand, 30-35 thousand people were buried under the rubble, 23 thousand were injured and 500 thousand people left homeless. The material damage, estimated at 25 to 40 billion dollars, stands out from the list of similar natural disasters in the recent history of Europe and the Near East. The reasons for the losses on this scale are pretty obvious. The disaster struck a large and densely populated area with a developed infrastructure and major industries. And the standards of housing construction by and large have proved to be simply inadequate. There had been no early warnings of the impending doom which left the people and the authorities totally unprepared, to say nothing of the fact that the quake struck at night when people are usually in their beds.

FROM THE GLOOM OF AGES

As said before, for most of us earthquakes in Turkey are nothing unusual. But only experts know that most of the damage done over the centuries fell to the lot of the northwestern part of the country, as is confirmed by over two millennia of written records. And since in olden times people saw earthquakes as forebodings of social and national upheavals, they described in rather sufficient details the disasters which struck the capital city called Byzantium, Constantinople and then Istanbul. These sources tell us, for example, of a quake of the year 69 A. D, which demolished the city of Nicomedia- the capital of Bithynia of the Gulf of Izmit. According to even more detailed accounts, the same city and all of its environs were literally wiped off the face of the earth by a quake of August 24, 328 A.D. Both must have been quakes of no less than 9 points, if not more, originating in a local focus.

page 100

This evidence, however, has somehow failed to attract sufficient attention on the part of seismologists during this century. The same can be said about the more recent seismic history of the region, especially in connection with the North Anatolian fault. Even a major earthquake of 1719 (stronger than that of 1999) at the Izmit fault, described by the British seismologist Nicholas Ambraseis, only in 1995, has failed to attract due attention. And this despite the fact that there have been a minimum of 7 such major tremors there over the past 1800 years at intervals of 190 to 400 years (of 270 years on the average). The 1999 disaster struck on the same spot 280 years after the previous one. That means it was just one in a row, and not even the strongest one in the chain of such disasters over the past two thousand years. But the rate of their repetition, speaking of one and the same focal area (and not of the whole region) and of the strongest quakes, is one such event in several hundred years. This time interval seems to be quite enough for a disaster to sink into oblivion.

THE OVERSIGHT

The determining geodynamic and seismo-tectonic structure for Northern Turkey is what is known as the North-Anatolian fault. This is a major sub-latitudinal zone of ruptures, or discontinuities, stretching all across the country some 50 to 100 kilometers away from the Black Sea coast. Geological and geodetic studies have established that this is mostly a right-lateral, or dextral, fault with the southern crustal block shifting against the northern one westwards at the rate of several centimeters a year. This being so, and the fault having several bends and branches which can "catch" the moving sides, there are bound to occur powerful earth tremors, such as those which do take place in reality. And the most interesting thing is the way this happens. Let us now try and trace this process in its progression during the 20th century.

It all began with the Great Erdzinjan Earthquake of 1939 in the eastern part of the fault which had magnitude 8 and power 11. The death toll exceeded 32 thousand and more than 30 thousand homes were destroyed. A fault trench, or fissure, opened up over a stretch of 300 km, was several meters wide and the horizontal displacement measured as much as 7.5 meters.

Between 1942 and 1944 there were another three quakes of magnitude 7.1-7.3. Their epicentral zones were also stretching in the latitudinal direction, exactly along the North-Anatolian fault. At that time the intensity reached 10 points, but because of the smaller size of the disaster area and its population the summary death toll

page 101

was slightly over 10 or 11 thousand. The chain of tremors opened up a fracture over another 450 km, with every new stretch being located to the west of the previous section with a dextral shift of 1 to 3 m. Subsequent events took place in 1957 and 1967 and were located even further westwards.

This what we call migration of foci is generally nothing new to specialists. In 1993 some of them placed into what we call "hazard category one" the whole stretch of the North-Anatolian fault over more than 1,000 km as prone to quakes of 9 points. But in a 1997 publication the probability of a major tremor in the area of Izmit was assessed at ...only 12 percent.

But bearing in mind the "westward trend" of the propagation of this seismic process along the fault, it would be only natural to anticipate the next tremor to hit an area to the west of the 1967 quake of magnitude 7.1-which happens to be the area of the Gulf of Izmit. Unfortunately no such conclusion had been drawn which must be due to a mistake in the location of the epicenter of the 1894 earthquake. On the basis of some unconfirmed data, it was assumed to be near Izmit, where over a distance of 225 km the fault was traced along the rupture, although that was not corroborated by direct evidence. On the strength of these assumptions experts thought that any seismic stresses in the western part of the North-Anatolian fault near the Gulf of Izmit were off. In actual fact, however, the focal zone of the 1894 quake passed, as was shown back in 1904, along the shore-line of the north-western strike, that is between the mouth of the Gulf and the western edge of the Bosporus. And that means that this particular tremor had no direct link with the area of the North-Anatolian fault. Its Izmit stretch remained covered up, so to say, or unruptured. That is why in the conditions of a clear-cut migration of foci along this zone it had to be regarded as potentially hazardous for the 20th century.

As we can all see now (and could have been foreseen in due time with due consideration for historical data), it was just there that a gap was formed between the 1963 focus in the west and the 1967 one in the east. Knowing the mean rate of directional migration of foci and the length of the unruptured stretch, one could have figured out the location, magnitude and even the mechanism of the forthcoming major seismic event, calculate the size of the tremor-prone area and make appropriate preparations for both research and civil defense activities..

MISSED CHANCE OF PROGNOSTICATION

In actual fact, however, we had some experience of successful long-term prognostication of strong events along major faults using the migration model which the author of this article suggested back in 1975-1984, although for a different region. In 1970 and 1993 some foreign authors (above all N. Ambraseis) made an analysis of data on devastating quakes of the past along the North-Anatolian fault, and calculated the intervals between them, but stopped short of any confident prognostications. After the 1999 disaster the author collected and systematized the available data from various sources on devastating earthquakes that hit the region in the past. The task was, first, to try and trace the probable rows of foci migration along the North- Anatolian fault up to the present century and then to decide whether or not this can be regarded as a regularity inherent in the given structure. Second, it was necessary to determine with all possible accuracy the intervals of recurrence of large-magnitude events in certain segments of the zone itself and in the adjacent areas. This analysis included only the strongest quakes of the magnitude equal or exceeding 7 (using modern assessments for events of remote past). For the earliest of these, however, we have no data on the size of the focal areas, their extension and the opening up of the corresponding fault stretches. Nevertheless, since the damage data apply to ancient cities which existed within the belt of this fault, it appears to be appropriate to relate the above quakes to the zone under consideration. And from the Middle Ages we get more reliable data for the identification and qualification of earthquakes as originating exactly in the North-Anatolian fault.

page 102

Over some 1.5 thousand years there appeared (with different degrees of completeness) four or five rows (cycles) of westward migration of foci along this zone. Over a stretch of 600 to 1,000 km from east to west the fault was opened up over 50-70 years at a rate of some 11 ?2 km/year. The intervals of relative calm between successive rises of activity reached from 270 to 400 years (or divisible values), but not 75-150 or 435 years as has been suggested by some foreign researchers.

WHAT NEXT?

If the present seismological views on the cyclic nature of the process of accumulation and discharge of stresses in the earth crust are correct, we should hardly expect any catastrophic earthquakes along the zone of the North-Anatolian fault in the 21st century. That does not mean, however, that upon the fault itself and its branches there can be no seismic events of strength 8 because some patches of the fault could have remained unruptured.

At this stage it seems to be more worthwhile for specialists to focus their attention on other areas. Danger-prone in the next century may be regions of European Turkey and the Mediterranean coast. Always threatened is the southeastern part of that country, especially around Lake Van. Also deserving of attention is the Black Sea coast where there were repeated strong quakes in the past, although information about them is scarce.

The earth crust in the Alpine mobile belt, including the greater part of Turkey, is so active that we simply cannot lull ourselves into thinking that the foci gathering strength deep under would fade away all by themselves.

The nagging question now is how high is the probability of a powerful quake in the area of Istanbul over the next few years? This probability can be assessed by various methods, including complex mathematical calculations. We, however, would like to turn to some of the simplest methods, based again on the knowledge of seismic history.

What is important, above all (as was first pointed out by the German researcher Johann Duke in 1904) is that in this case "fits" of seismic activity, including some very strong events, are followed by periods of relative calm. In other words, powerful earthquakes in this region occur in "clusters"-447-448, 542-558, 732-740, 865-870, 975-989, (1063), (1296)-1344, (1509), 1633-1659, 1719-1754, 1878-1894. The length of each was 5-58 years (29 years on the average). And the quiet intervals (between the end years of activity) lasted from 60 to 175 years (108 on the average). Since the last strong quake in the Istanbul

page 103

area occurred in 1894, or 106 years back, it seems likely that the Izmit catastrophe of 1999 opened up the next period of seismic activity in this region.

And one can also take a different approach to seismic prognostication in this region-to try and trace how seismic events developed there after similar strongest earthquakes near the Gulf of Izmit.

Over the past millennia devastating quakes in Istanbul (Byzantium-Constantinople) followed similar events in the area of the Gulf of Izmit at intervals of 5-35 years. One can include here the events of the year 363 (after 358), 562 (after 554), 1754 and 1762 (after 1719). In all probability this happens not by chance, but reveals a tangible link of the northwestern coastal fault with the geodynamics in the main area of the North- Anatolian fault; "waking up" of the western end of the zone provokes seismic activity in regions closest to Istanbul.

Between the mouth of the Gulf of Izmit and Bosporus (a distance of some 40 km) there were many strong quakes, such as in 489, 554, 989, 1509 and 1894. In 1894 the epicenter area must have been stretched along the coast; and the same seems to have been the case in 1509 and probably even earlier. This makes it possible to assume the existence of a major active rupture along the coast. It should be seen as a north- western branch of the North-Anatolian fault with a right-hand shift. As for the author of this article, he is not very clear on the seismo-tectonic situation at Bosporus itself. But judging by the foci of the very strong quakes along the northern coast of the Sea of Marmara, which occurred in 447 (740?), 1344 and 1766, it is possible to assume that here, too, a near-latitudinal active faults zone stretches along the coast.

If we know, at least approximately, the location of seismo-generating faults in the southern part of the Sea of Marmara and can trace them along the northern coast, we can try and "prognosticate", which sections will be ruptured up an and generate strong quakes in some near future. In doing that we should proceed from four of the most probable assumptions.

Foci of magnitude 6.5 and more in the region relate to the zones of active faults ofdextral, or right-lateral nature.

Along the sub-latitude fault zones (some of which are probably located under the Sea of Marmara and little investigated) there takes place what we call a directionally oriented (perhaps with some exceptions) westward migration of foci.

By analogy with historical events (the quake clusters), the period of relative seismic calm is to be followed by an active period in the eastern part of the Sea of Marmara ushered in by the 1999 disaster.

Strong events of this kind are most likely to occur at sections of fault zones which remained unruptured for the past one hundred years and more.

On the strength of the above one can try and identity areas with a low probability of strong quakes for the first half of the new century, and also danger-prone ones.

The northern branch of the North-Anatolian fault westward of Izmit ruptured up near the city of Karamursel in 1909 (magnitude 5.8) and near the town ofYalowa in 1963 (6.3). That being so, the next chain of strong quakes in this zone can be expected only westward of 29 E within the water area of the Sea of Marmara. The southern branch of the North-Anatolian fault at the Izmit stretch seems not to have generated strong tremors for the past several centuries which makes it potentially hazardous between 30.5 and 28 E. But its southernmost stretch-Ulubatan-still has unruptured stretches to the west and east of the city of Bursa after the two strong quakes of 1855. In the area between the Gulf of Izmit and Bosporus there were earthquakes in 1878 and 1894 (of magnitudes 6.7 and 6.5 respectively) and the probability of strong tremors there in the coming decades is low. This, however, can hardly apply to the northern coast of the Sea of Marmara to the west of Istanbul. For example, the last very strong (magnitude 7.7) catastrophe on the western edge near the city of Tekirdag occurred 244 years ago (back from 1999) and on the stretch nearest Istanbul there was one several centuries before. It is this section that appears to be most danger-prone for the near future.

Apart from the above general considerations, the author of this article also regards as signs of an approaching threat the following evidence: during the Izmit quake of August 17, 1999, and apart from the main narrow epicentral belt along the gulf and further to the east, powerful-up to points 9-tremors were also registered along the northern coast of the Sea of Marmara some 10 to 40 kilometers west of Istanbul. This effect of what one can call local amplification of tremors away from the main focus points not only to the effect of ground conditions, but also to a high-stress condition of the whole stretch and its capacity for the next major seismic event.

The above prognostication estimates need more detailed analysis and assessment, of course. Further more, adequate monitoring should be organized in this region with emphasis on the above areas. The way I see it, the region around the Sea of Marmara can be made, with international support, into a case study for the existing methods of medium and short-term seismic prognostication methods leading to the development of more reliable and practicable forecasts.

EDITORIAL NOTE:

On November 12, 1999, the north-western region of Turkey was hit by yet another devastating earthquake. There are reasons to believe that in the future seismic activity will shift towards the coast of the Sea of Marmara.

Permanent link to this publication: