Terrestrial phenomena disasters/Earthquakes/Urban structure damage. Urban disasters/Urban structure damage.
Disaster name
Damage from the 1999 Kocaeli, Turkey Earthquake
Author of WEB conversion
Matsuoka Masashi

Case Study

No. 14

1. Analysis objective

To understand regions of structural failure

2. Analysis procedure Analysis flow chart

When changes in the earth's surface result from a disaster such as an earthquake, the altered region can be sampled by comparing SAR images before and after the earthquake because the strength and phase of backscatter will change.

Before an earthquake, the backscatter strength of microwaves irradiated by satellite are larger because of multiple reflections between buildings and roads (cardinal effect). The components of microwaves irradiated from a satellite into a damaged area after an earthquake and returned to the satellite, however, will be smaller because of multidirectional scattering from collapsed buildings. In other words it is projected that the spatial distribution of backscatter characteristics will differ before and after an earthquake, and the change can be expressed by the difference or correlation of the backscatter strength.

Moreover, from coherence analysis to evaluate the phase information interference included in SAR images from two different times we can express subtle changes in the earth's surface during the observation period. When we use phase information that is more sensitive than the microwave wavelength, however, this will impose spatial and time limitations on the images from the two different time periods. Spatial limitation refers to the fact that there is a distance between two satellites, and when this distance is large the interference for the entire image decreases, making it impossible to detect the low coherence region that causes the earth surface changes we want to sample. Time limitation refers to the fact that there is an observation interval between two periods of time, and when this is long it will become difficult to distinguish between not only the changes we wish to sample but also between these changes and the natural changes that will occur with the passage of time.

When sampling areas damaged by an earthquake, in other words, the ideal approach is to select two time periods that provide a short observation interval between the times before and after the earthquake, using observations with satellite orbital paths that are as same as possible. The same conditions are preferred when using strength information as well, but because the strength information for changes in the earth's surface is not sensitive, however, this is not considered a problem to extract an extensively damaged region (Refer to Case Study No. 16 of SAR for the Southern Hyogo Prefecture earthquake).

In this analysis we used a pair of ERS satellite images that almost completely satisfy the above conditions (pre-earthquake: August 13, 1999 and post-earthquake: September 17, 1999) to create a difference image in strength, a correlation image and a coherence image, and examined the changes in values in the structural damage zone around Golcuk.

The results showed a great deal of variation in the value for each level of damage. For the difference in strength, the decline in the value was remarkable over a wide range of damage, while the drop in the coherence value was slow even when damage was extensive. The correlation value declined as the extent of damage increased.

The above-mentioned tendency is also agreed with the results for the Southern Hyogo Prefecture Earthquake. Therefore we used these characteristics and estimated the damage distribution for the entire image area using the level slice technique.

3. Analysis results

By performing a basic study to understand the damage caused by the Kocaeli, Turkey earthquake using satellite SAR images, we quantitatively clarified the characteristics of the SAR images for the damaged areas.

As a result, for areas where collapsed structures are broadly distributed we learned that the strength of SAR image backscattering after an earthquake will decline compared to the level before the earthquake and that the speckle patterns will change.

We also learned that phase is nearly preserved before and after an earthquake in the non-damaged regions.

4. Results from using the analysis results

The importance of damage information in the initial stage following a disaster occurrence was strongly felt after the Great Hanshin-Awaji Earthquake Disaster in 1995. Given the time required to assess the damage when an earthquake occurs directly beneath an urban area and results in widespread damage, for example, active efforts are now underway to develop and introduce a real-time earthquake disaster mitigation system that will quickly estimate damage during the initial stages, based upon seismic monitoring using a seismograph network and a database prepared using the Geographical Information System (GIS). One example of an early stage damage estimation system at the national level is the Disaster Information System (DIS) for earthquakes in the Cabinet Office (Formerly, National Land Agency). The Early Estimation System (EES) for earthquake damage is part of this system and covers all of Japan using a 1km grid. Prepared with ground and buildings data, this system utilizes information from measurement seismographs the Japan Meteorological Agency is developing throughout Japan and can automatically estimate and total the scale of earthquake damage approximately 30 minutes after a disaster occurs, based on empirical equations. The goal of this system is to support national emergency response activities. In the end, however, the damage result estimates of such a system are still only estimates, and discrepancies can be quite large.

Accordingly, a broad understanding of actual damage must be carried out at the earliest stage possible. Remote sensing may be said to be one powerful technology for this purpose. The fact is, however, that there are limits to the ability to obtain images at the necessary times and in the required areas from remote sensing using earth observation satellites, because the satellites' recurrent periods are fixed, which presents problems when trying to use such satellites to assess damage during the early stages of a disaster.

Optical sensors, moreover, cannot be used when the target region is covered by clouds. Because of the inverse relationship between the degree of resolution and the number of recurrent days of a satellite returns to the same orbit, the use of high-resolution satellites is particularly limited.

In situations where a broad-range of damage conditions must be understood, however, even if the necessary information in the initial stages is not in detail but is merely general information, it can serve a useful purpose for decision-making at the national level in particular. From this point of view, there will also be situations in which satellites having current capabilities of resolution will be adequate.

In any event, by using multiple satellites, including satellite equipped with SAR that have been launched or are in the planning stages for launch by various countries in recent years, the problems of recurrent period and resolution have been greatly reduced.

5. Sources

Matsuoka M. and Yamazaki F.: Characteristics of Satellite SAR Images of Areas Damaged by the Earthquake in Kocaeli, Turkey, Proceedings of the Second Real-Time Earthquake Disaster Prevention Symposium, pp. 165-168, 2000 (in Japanese).