Landslides Triggered by the Chi-Chi Earthquake
Hsuan-Wu Liao1, Chyi-Tyi Lee2
1 Mr., Institute of Geophysics
2 Professor, Institute of Applied Geology
National Central University
32045 Chung-Li, Taiwan
Tel: +886-3-4253334 Fax: +886-3-4223357
E-mail: swliao@gis.geo.ncu.edu.tw
, ct@gis.geo.ncu.edu.tw
TAIWAN
Keywords
Landslides, Earthquake, SPOT, GIS
Abstract
A large number of landslides were triggered by the Chi-Chi Earthquake (ML7.3) on September 21, 1999. These landslides have been mapped from SPOT images in this study. By using GIS (Geographic Information System) as a tool, we construct a GIS table of landslides triggered by the earthquake and analyze their characteristics, including types, distribution, areas and numbers, …etc. The distribution and areas of landslides have been compared with the distance to earthquake foci, distance to faults, distance to roads, distance to rivers, rock types, hill slopes and slope directions.
There were 9272 larger landslides (area greater than 625 square meters or 4 pixels on a SPOT image) occurred during earthquake shaking. The total area of landslides is 127.8 square kilometers. There were 8843 landslides located within the area of PGA value 250 gal and above. These landslides were distributed in an ellipse-shaped region with the major axis striking NNE, coinciding with the trend of regional faults.
Statistics shows the following results: (1) Landslides most located within the area that PGA greater than 250 gal, especially within the area that PGA greater than 300 gal. (2) A lot of landslides occurred within 20 km from the fault rupture plane. The most distant landslide locates at about 60 to 70 km away from the fault rupture plane. (3) The Toukoshan Formation, The Chinshui Shale and the Tachien Sandstone were more easily influenced by the earthquake shaking, especially at Huoyenshan Facies of the Toukoshan Formation. (4) Slopes with inclination larger than 100% involve more landslides. (5) At the hanging wall, most landslides locate at S and SE facing slopes. At the footwall, most collapsed slopes face to S, SE and SW, whereas slopes facing to other directions were not so seriously damaged. It shows that the movement of the faulting during the Chi-Chi earthquake was from SE to NW.
1. Introduction
At 1:47 a.m. local time on September 21, 1999, an ML7.3 earthquake struck the central Taiwan. The hypocenter located at a depth of 8 km beneath the Chi-Chi area. The surface rupture was along the existing Chelungpu fault. High peak ground accelerations were recorded over a broad region. The strong ground shaking induced not only considerable structural damages but also thousands of landslides.
We documented about 9000 landslides triggered by the Chi-Chi earthquake. This article provides a preliminary result, including a brief description of how we mapped the landslides, an overview of landslides distribution, and basic statistics of some factors that associated with landslides. Detailed analyses will be completed in the future study.
Method
This study utilized the use of SPOT satellite images and aerial photographs to identify the landslides. MapInfo GIS software was used to digitize the location and extent of landslides. Numbers and areas of landslides triggered by the Chi-Chi earthquake were then calculated. Spatial functions in GIS were used to analyze the relationship among the landslide distribution and factors that triggered the landslides.
Mapping the Landslides
The Chi-Chi earthquake provides a good opportunity for studying of earthquake triggered landslides. Since the influenced area was large, SPOT satellite images are used to identify the landslides. Areas that were covered by cloud were re-examined by using aerial photographs. Landslide maps before the Chi-Chi earthquake and after the earthquake were identified separately. The images used are only 6 days after the earthquake on September 26 and 27, and about 5 months before the earthquake on April 1.
We compared the location and extent of the landslides between two maps, and then separated the landslides that were really triggered by the Chi-Chi earthquake. We obeyed the following principles:
- Landslides that can be found on both maps before and after the Chi-Chi earthquake are assumed triggered by the earthquake.
- Landslides that can be found only on the map after the earthquake were assumed triggered by the earthquake.
- Landslides that can be found on the map before the earthquake but cannot be found at the map after the earthquake are assumed not triggered by Chi-Chi earthquake.
Combining the data we got from satellite images, aerial photographs and field check, a landslides map triggered by Chi-Chi earthquake were prepared (Figure 1). The Chi-Chi earthquake triggered over 9000 landslides, which are larger than 625 square meters in area or are more than 4 pixels in a SPOT image. These landslides distribute in an ellipse-shaped region with the major axis striking NNE, coinciding with the trend of regional faults.
Characteristics of Landslides and Discussion
This study analyzed some relations between the landslide distribution and the factors that influence the landslide. The results are discussed as follows.
Distance to earthquake epicenter
Main shock and aftershocks locations were acquired from the Central Weather Bureau. We calculate the distance between the epicenter of main shock and the edge of a landslide. We found that most landslides occurred within 30 km from the epicenter. There are few landslides occurred over 60 km from the epicenter. The greatest distance between the landslide and the epicenter is 117 kilometers, which is smaller than the result of Keefer (1984). (Figure 2)
Distance to the fault rupture plane
We used fault rupture plane (Cheng et al., 2000) to calculate the distances between landslides and the fault plane. We found that most landslides occurred in the distance of 20 km from the fault plane. Landslides occurred in the distance large than 30 km are rare. The greatest distance between the landslide and the fault rupture plane ranged from 60 to 70 km, which is also smaller than the result of Keefer (1984). (Figure 3)
