Introduction

Our field inspection started on Saturday, 1 Oct. with orientation meetings with local geoscientists at the Institute of Earth Sciences in Taipei and observation of structural failure in Taipei, about 150 km from the epicenter. Traveling southward, we have so far spent two days exploring the northern and central portion of the Chelungpu Fault and the damage resulting from the ChiChi earthquake. We have surveyed locations along the fault between 24° and 24°17' latitude.

The first day's observations focused on the northernmost portion of the rupture in the vicinity of the Dachia river which drains the central mountains of Taiwan to the west. The second day we briefly toured the TaiChung harbor vicinity then examined the Chelungpu fault to the northeast of TaiChung, the third largest city in Taiwan, with a population of about 3 million. This field report will present a summary of damage and evidence of the surface rupture, then conclude with some preliminary thoughts about lessons learned.

Damage Description

Overall, we have been surprised with the performance of structures and the construction quality. Considering the levels of ground motions the structures experienced, based on preliminary data provided by the Taiwanese colleagues, seismic damage of structures have been surprisingly light. Even some of the structures located directly on the fault were relatively undamaged. In most cases, damage was concentrated on the hanging wall of obvious thrust faults.

Rupture Description

The trace of the surface rupture closely follows the frontal slope of the local mountain range where the range trends north-south. To the northern end of our current inspection, the range is cut by the Dachia River and the earthquake rupture appears to split into at least two strands. One strand (probably a minor one) tends to wrap around the incised range and trend toward the east- northeast. Although it is still unclear, this strand of the fault may connect to an exposure of the surface rupture that lies 4 km north of Donshi, where approximately 500 people died. A second, larger rupture scarp cuts across a local bridge where it has formed a 10m waterfall. From the waterfall, the fault scarp can be traced toward the failed Dachia dam to the northeast, where it is associated with a significant keystone (or dropped block) structure on the crest of the reverse fault. It is the keystone structure that is responsible for the destruction of the northern end of the dam.

The vertical offset across the rupture ranges from 2 to 4m, but lateral offset is very difficult to estimate reliably, since the main fault dips at a relatively low angle (see Lessons Learned). In two locations, however, both the vertical and horizontal offsets could be measured. At each of these sites, vertical offset was ~3.2m and horizontal shortening was ~2.65m. It was also possible at each of these sites to estimate a slip vector, since linear features crossed the fault scarp without lateral offset (showing that the hanging wall block must have moved parallel to the linear feature. In each case, the slip vector was due west. These geometrical constraints suggest that the rupture plane dips toward the east at ~32° and that the slip was due west.

Lessons Learned

While it is too early to present conclusive scientific results, a few lessons are clear. First, a society which plans in advance to live with and survive earthquakes, can limit losses when the inevitable occurs. Well designed, and well built structures which reflect an understanding of faulting and subsurface site conditions will reduce losses and enhance society¹s ability to recover from a major earthquake.

Based on our observations of structural damage, it is clear why structures should not be built immediately upon the surface trace of an active fault. However, in certain tectonic environments, the exact location of future surface traces may be difficult to pinpoint within a few tens or hundreds of meters.

It is proving difficult to evaluate with any certainty the direction of slip of the fault. This is because the fault dips at a relatively shallow angle and features that show lateral offset cross the trace of the fault at various orientations. Thus the sense of lateral offset is dictated by the relative orientation of the offset feature, and lateral offsets should properly be termed apparent lateral offsets. So far, we have unequivocally determined both the lateral and vertical offset at only two locations (see Rupture Description).

We have observed that structures on the hanging wall experienced greater damage than those on the footwall structures.