The famous Report of the State Earthquake Investigation Commission on the 1906 San Francisco earthquake is the basis for much of modern seismology (see p. 4, "Earthquakes"). Professor Andrew C. Lawson chaired the seminal study of the tectonic source of seismic waves. Key findings were that elastic rebound along the San Andreas fault caused the heavy shaking; and secondly, the intensity of the strong ground motion was related not only to distance from the ruptured fault, but also to the superficial geology and soil conditions. Recently, the author visited the sites of two devastating earthquakes that struck along plate margins in highly seismic areas of the world.

The Izmit, Turkey, Earthquake

The first was the Turkey earthquake of 17 August 1999 (M_w = 7.4); the second was the Chi-Chi earthquake (M_w = 7.6) in central Taiwan on 21 September 1999. With the help of local colleagues, I went to observe the faulting and damage, and to study crucial records of strong-ground motion accelerometers near the heavily shaken zones. In 1994, I had lectured for a semester at the Bosphorus University, Istanbul, and, by means of a NSF grant, installed a large-scale seismic array (then the largest in the world) at Lotung in northeastern Taiwan (see p. 63) in the 1980s.

The Izmit earthquake killed over 16,000 people. It was generated by a right-lateral strike-slip of the Western Anatolian fault, whose tectonic displacement results from westward extrusion of Turkey in the collision between the Arabian and Eurasian plates (see p. 138). The August 1999 bilateral fault rupture started at a hypocenter near the city of Gölcük and produced surface offsets on land for a distance of 110 km. The aftershock distribution indicates that the rupture may have extended another 50 km to the west under the Marmara Sea. The average fault slip was about 3 meters horizontally (see Figure 1), with a maximum of about 5 meters. (The maximum 1906 San Andreas offset was 6 meters.) A short fault section showed dip-slip of a meter or so, down to the north.

On the drive from Istanbul to Izmit, the first noticeable consequences were the dense clusters of tents where refugees from the calamity were temporarily housed. The devastation was concentrated and severe along the fault zone. Many, but not all, buildings were damaged by the heavy shaking and special nature of the near-source ground-motion, which includes an energetic ground velocity "pulse." In 1971, I described this pulse as a "fling" of the elastic rebound, enhanced by the energy concentration as the rupture propagates (see p. 272). Also, the surface fault offsets destroyed hundreds of straddled buildings, damaged port facilities, pipe lines, roads, and industrial facilities. Along the southern coast of Izmit Bay were large areas of subsidence and, particularly in the city of Adapazari, lakebed sediments, which contained water-saturated layers of sands and silts, liquefied. The result was the settling and tipping of hundreds of buildings. Strong-motion seismographs in the shaken areas measured horizontal accelerations of over 30 percent of gravity.

The earthquake was a prime example of lack of proper construction practices and design, even though the seismic hazard was well known (see p. 33). In this century alone, the North Anatolian fault caused numerous destructive earthquakes–a remarkable series of seven with magnitude over 7 occurred between 1939 and 1967, the latter being adjacent to the area of the Izmit earthquake. After my visit, a second devastating earthquake (M_w = 7.2) on 12 November was generated by a further rupture of the Northern Anatolian fault eastward toward Bolu from the terminus of the 17 August slip. Again, there were many hundreds of deaths and similar collapses.

The Chi-Chi Earthquake, Taiwan

U.C. Berkeley and the Academia Sinica in Taipei organized a joint symposium in early November 1999, on improved seismic hazard reduction. After a presentation there, I made a field study of the extraordinary faulting that generated the September earthquake and examined some of the strong-motion recordings.

In my opinion, the Chi Chi earthquake (2,200 deaths) rates second only to the 1906 San Francisco earthquake (see p. 4) as the most important in understanding the cause of earthquakes, the detailed fault rupture process, and the generation of seismic waves. The reasons are that the surface rupture of the Chelungpu fault is easily accessible on land, as is that of the San Andreas fault, and there are capable local geologists, seismologists, and earthquake engineers on hand. But above all, there are many modern instrumental measurements. For example, there is a network of systematic pre- and post-earthquake GPS geodetic readings of the crustal displacements; there is a reliable record of seismicity with well-located earthquake foci; and the earthquake was well recorded by broadband digital seismographs around the world (see p. 56). Uniquely, the seismic waves were recorded near their source by over 500 digital strong-motion accelerometers. (The peak ground accelerations were almost 100 percent of gravity.) It is likely that seismographic studies of this earthquake will continue for many years.

The Chelungpu thrust fault movement (see p. 84) was dramatically evident for over 80 km from Shihgang in the north to Tungtou in the south. The hanging wall dipped about 30º toward the east, with uplift ranging from over 2 meters in the south to 7 meters in the north (see Figure 2). Left-lateral slip of several meters also occurred at many locations.

This enormous surface elevation is the consequence of the strain developed along the convergent plate boundary of the Eurasian and Philippine Sea Plates underlying the central portion of eastern Taiwan. The convergence raises the Taiwan central mountain range, the second highest, after the Himalayas, in Asia. The Chelungpu fault strikes along the margin of the western foothills and associated north—northeast trending folds.

I was awestruck by the results of the large-scale uplift. Where the fault cuts through cities and lifelines, hundreds of buildings, a major dam, numerous roads and levees, and at least five major bridges were ruptured. Nevertheless, as in the 1906 San Andreas and 1999 Anatolian fault dislocations, well-built structures, even though adjacent to such major offsets, were left standing, often with little damage. Across one previously level street, I encountered the fault scarp with undamaged houses on one side, elevated by 6 meters so that they now looked down on their neighbors; only the house on the lot through which the faulting occurred was destroyed. From the riverbank on the side of this street, one could observe a newly created waterfall in the riverbed exposing at least a 100-meter zone of gouge. The Chelungpu fault had previously been mapped as active, but no building zoning was enforced. A lesson is the value of the Alquist-Priolo legislation that regulates construction along mapped active faults in California (see p. 233).

For the student of earthquakes, whenever in the region, it is now obligatory to visit the Chelungpu fault in Taiwan.

Figure 1. View along Anatolian fault offset near Izmit and running through school playground.

Figure 2. Hanging wall of Chelungpu fault across running track at Wufeng high school. East side up.