CHANGES IN GEOLOGICAL FAULTS ASSOCIATED WITH EARTHQUAKES DETECTED BY THE LINEAMENT ANALYSIS OF THE ASTER (TERRA) SATELLITE DATA.
Authors: A. Arellano Baeza, A. Zverev, V. Malinnikov
Department of Applied Astronautics, Moscow State University of Geodesy and Cartography, Moscow, Russia (firstname.lastname@example.org)
Received: Dec, 03, 2004
Last modifications reception: Dec, 06, 2004 Predicción de Terremotos
As it is well known, the region between South Peru and Northern Chile is one of the most seismically and volcanically active regions in South America due to a constant subdiction of the South American plate, converging with the Nazca plate in the extreme North of Chile. We used the 15 m resolution satellite images, provided by the ASTER (VNIR) instrument onboard the Terra satellite to study changes in geological faults close to earthquake epicenters in the South of Peru. Two visible bands with the spectral ranges 0.52-0.60 µm, 0.63-0.69 µm, respectively, and one near-infrared band with the spectral range 0.76-0.86 µm were analysed using "The Lineament Extraction and Stripes Statistic Analysis" (LESSA) software package to examine changes in the lineament features caused by seismic activity. We used the satellite images 5 months and 2 months before the earthquake. Fortunately, the seasonal variations in the South of Peru and North of Chile are very small, and the vegetation is very limited. This allowed us to study a zone that suffered a 5.2 magnitude earthquake in the Richter scale, and to establish changes in the location, extension, orientation and density of lineaments. This makes it possible to develop a methodology able to evaluate the seismic risk in this region for the future.
Keywords: lineament analysis, earthquake, ASTER satellite
Strong efforts have been made over the last decades to develop new techniques directed to the study and potential prediction of earthquakes with the use of satellites. It includes the study of pre-earthquake thermal anomalies using the images of the IR satellite (Ouzounov, 2004), observation of behavior of the surface latent heat flux data (Singh and Ouzounov, 2004), measurements of ionospheric precursors of earthquakes (Serebryakova et al., 1992, Liu et al., 2000).
Carver et al., (2003) used the SRTM and Landsat-7 digital data and paleoseismic techniques to identify active faults and evaluate seismic hazards on the northeast coast of Kodiak Island, Alaska.
These techniques are a good complement to the traditional ground-based observations like measurements of temporal variations of propagation times of primary and secondary seismic waves (Aggarwal et al., 1973).
Relationship between earthquakes and the geological structure of the area of earthquake due to the study of lineaments was studied by a number of authors. For example, Cotilla-Rodriguez and Cordoba-Barba (2004) studied the morphotectonic structure of the Iberian Peninsula and showed that the main seismic activity is concentrated on the first- and second rank lineaments, and some of important epicenters are located near the lineament intersections. Stich et al., (2001) obtained from the analysis of 721 earthquakes with magnitude between 1.5 and 5.0 mb that the epicenters draw well-defined lineaments and show two dominant strike directions N120-130ºE and N60-70ºE, which are coincident with known fault system in the area and with the source parameters of three of the largest events. Distances within multiplets (typically several tens of meters) are smaller than the fracture radii of these events.
It is generally accepted that the scenario of an earthquake develops similar to one of the rupture of solid body. The most common model of earthquakes is the dilatancy-diffusion model (Whitecomb et al., 1973; Sholz et al., 1973; Griggs et al., 1975). The term "dilatancy" refers to an increase in the volume of a rock deformed by pressure, caused by the expansion and extension of small cracks within the rock. This effect can be detected in strained rocks just before an earthquake, and is the basis of one type of earthquake prediction. However, it never was applied before to the satellite image analysis.
However, this model was modified recently by introducing a concept of self-organized criticality, proposed by P. Bak (Bak et al., 1988) for description of the behavior of complex system. In application to earthquakes, this approach describes an interaction between the ruptures of different rank and the collective effects of its formation before a strong earthquake. (see for example Varnes, 1989, Sammis and Sornette, 2002). Wide area around the future epicenter reaches a metastable state, and the system turns to be very sensitive to small external actions. The concept of SOC does not contradict to the concept of dilatation. However it assumes that significantly greater region is involved during the last stages of the earthquake preparation as the dilatation theory implies. As the system of faults and ruptures as a fractal structure (King, 1983) the formation of ruptures has a hierarchy, and the size of a block, in which the process of self-organization takes place, determine the magnitude of the future earthquake.
In this work we analyze the changes in the structure of stripes and lineaments extracted from the ASTER (TERRA) images using the Lineament Extraction and Stripes Statistic Analysis (LESSA) software package (Zlatopolsky, 1992) associated with 5.2 magnitude earthquake, occurred in the South of Peru. This region is characterized by the lack of vegetation which facilitates the study of changes in stripes and lineaments associated to earthquakes. Recently this region has intensively been studied using the ground based seismic network (Comte et al., 2003)
2. Instrumentation and data analysis.
Images of Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) onboard the TERRA satellite were used. The satellite was launched to a circular solar-synchronous orbit with altitude of 705 km. The radiometer is composed by three instruments: Visible and Near Infrared Radiometer (VNIR) (bands 1-3), Short Wave Infrared Radiometer (SWIR) (bands 4-9) and Thermal Infrared Radiometer TIR (bands 11-14) which measure the reflected and emitted radiation of the Earth's surface covering the range 0.56 to 11.3 µm (Abrams, 2000).
In this work we used the ASTER Level 2 On-Demand reflectance images in all 14 bands. The images were processed using the Lineament Extraction and Stripes Statistic Analysis (LESSA) software package (Zlatopolsky, 1992, 1997), which provides a statistical description of the position and orientation of short linear structures through detection of small linear features (stripes) and calculation of descriptors that characterize the spatial distribution of stripes. The program was applied successfully to a number of fields including geodynamics, seismology, and mineral exploration (see Technical Reports of Geology Research Institute of Russia: VSEGEI 1988, 1991). Short linear structures (stripes) were extracted in each of eight directions (0º, 22.5º, 45º, 67.5º, 90º, 112.5º, 135º, and 157.5º) using a convolution between the circular masks of 7 pixels for each direction and the image.
After what we concentrated our attention to the analysis of changes in the density fields of stripes detected for each of aforementioned directions using the ASTER images obtained a few months before earthquakes. To be able to extract a weak variation in the stripe density from the strong background of stripes related to the geomorphologic and human made structures, we analysed the residuals between the stripe density fields obtained applying the same procedure to a par of images associated to earthquakes.
3. Analysis of January 27, 2004 Event.
The 5.2 M earthquake took place in the South of Peru (-17.6869 Lat, -70.6715 Long) January 27, 2004. ASTER images, with include the epicenter of this earthquake were found for September 21, 2003 (4 months before the earthquake), December 10 (1 month before the earthquake). We selected the area of 37.5 x 37.5 km common for these pair of images. Figure 1a shows the image of the area around the future earthquake five months before.
Figure 1. Zone of the earthquake in the South of Peru. The earthquake is matched by a red circle.
The knowledge of the magnitude of the 5.2 M earthquake makes it possible to estimate the length of rupture for this earthquake as (Riznichenko, 1985):
where M is the magnitude of earthquake in the Richter scale,
l is the large of rupture. For 5.2 M it gives l = 190 km.
A difference between mean value of the area of seismogenic ruptures and the area, generated by the event (Sobolev, 2003) is:
where Ko = 8.5 is the mean value of energetic class for seismic activity, K is the energetic class for the current event related to its magnitude as:
In our case k =12.5 and DS=495
These estimations allow us to suppose that selected zone of the images covers the area of earthquake formation.
Figure 2. Density fields for all directions from the image Figure 1, 5 months before the earthquake. Bands 1, 2, and 3.
Figure 3. Density fields for all directions from the image Figure 1, 1 month before the earthquake. Bands 1, 2, and 3.
Figure 4. Residuals between density fields of stripes in all directions, and of each direction in particular
obtained for the images of zone of the earthquake in the South of Peru five and one month before the earthquake.
Figure 1 shows the area of interest around the future epicenter,
marked with a circle. The image was obtained by combining the three bands measured
by VNIR instrument September 21, 2004, five months before the earthquake. As
can be seen, the area is covered by mountains, the vegetation is practically
absent. Manmade objects are very limited (one road and small villages, predominantly
in the right lower corner). Fortunately the cloud coverage is close to 0%.
Figure 2 shows the stripe density fields for all directions, obtained from the 1-3 band of previous image. As can be seen, the near-infrared band shows density field distribution different from the visible light bands. Figure 3 shows the same field for the image, which covers the same area, but obtained one month before the earthquake. As can be seen, both images have distribution of stripe densities, which do not correlate with any visible feature in the Figure 1. Comparing the Figures 2 and 3 it is possible to see that the density fields differ from band to band. They also change in time. However, it is difficult to extract the information related to the epicenter of the forming earthquake.
Figure 4 shows the residuals between density fields for all directions and for each direction separately, obtained using the 3 band. Each color axis was normalized by the maximum value of the density field for corresponding direction. It varies between -1 and 1. As shown in Figure 4 (all directions), the total stripe density does not vary significantly. But the directional densities suffer significant variations. The most important feature is the presence of enlarged zone in which the stripe density increases in direction 90º (red) and decreases in all another directions (blue) that may indicate about the reorientation of stripes during the final stages of earthquake formation.
Figure 5. Zone of the absence of earthquake in the South of Peru.
Figure 6. Density fields for all directions from the image Figure 5. Bands 1 (up) and 3 (down).
Figure 7. Density fields for all directions from the image of the same zone four months later. Bands 1 (up) and 3 (down).
4. Analysis of images in the absence of earthquakes.
It is very important to check whether the features analysed in the previous section are associated with earthquakes. To do this we repeated the same procedure for another par of images obtained in the same region. There were no earthquakes in the zone analysed. Figure 5 shows the image obtained again by combining the three bands measured by VNIR instrument May 25, 2004. As can be seen, the area is also covered by mountains, the vegetation is practically absent, there are no manmade objects, and the cloud coverage is close to 0%.
Figures 6 and 7 show the stripe density fields for all directions, obtained from the 1-3 bands using the previous image, as in case of Figure 2, and from the image, obtained for the same zone four months later. As can be seen, in this case the variations in stripe density fields from band to band are not so strong. Density fields do not suffer also significant alterations with time. Therefore it is not surprising that the resulting residuals for all directions and for each of eight directions are close to zero, and there are no specific features observed in the previous set of images associated with the earthquake.
Figure 8. Residuals between density fields of stripes in all directions, and of each direction
in particular obtained for the images of zone without earthquakes in the South of Peru.
5. Discussion and conclusions.
Analysis of variations in the stripe density fields extracted from the ASTER (TERRA) images five and one month before the January 27, 2004 earthquake in the South of Peru showed that these residuals show the reorientation of stripes, forming a long trace in which the stripes take predominantly the direction of 90º. This behavior agrees with the dilatancy models of earthquakes, according to which during the last stage of the earthquake formation, the fractures are aligned predominantly in one direction. Of course the results obtained are preliminary, and it is necessary to check carefully the possibility to detect the aforementioned processes in a significant number of earthquakes. It is necessary also to understand the mechanism of registration of fractures by the satellite. Probably, the fractures are more visible before the earthquake, because the proximity of the earthquake alters the underground water, which goes to the surface. Another possibility is that according to Serebryakova et al. (1992), during the last stages of earthquake formation, fractures emit electromagnetic waves with increasing frequency able to penetrate into the ionosphere and magnetosphere, which could be registered in infrared images.
We plan to continue research in this direction during the next years to develop a new technique for early earthquake warning, which could be a very promising complement to the other techniques of earthquake forecast.
We acknowledge the Hiroji Tsu (Geological Survey of Japan - GSJ) - ASTER Team Leader, Anne Kahle (Jet Propulsion Laboratory - JPL) - US ASTER Team Leader and the Land Processes Distributed Active Archive Center for providing the ASTER level 2 images. We acknowledge Zlatopolsky for providing the Lineament Extraction and Stripes Statistic Analysis (LESSA) software package.
- Abrams, M., The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER): Data products for the high spatial resolution imager on NASA's Terra platform, International Journal of Remote Sensing, 21(5): 847-859, 2000.
- Aggarwal Y. P., Sykes L. R., Simpson D. W., Richards P. G., Space and temporal variations of ts/tp and P waves residuals at Blue Mountain Lake, J. Geophys. Res., Vol. 80, P. 718-732, 1973.
- Bak, P., C. Tang, and K. Wiesenfeld, Self-organized critucality, Physical Review A, 38(1), 364-374, 1988.
Carver, G., J. Sauber, W. R. Lettis, R. C. Witter, Use of SRTM and Landsat-7 to evaluate seismic hazards, Kodiak Island, Alaska, Abstract #4513 of EGS-AGU-EUG Joint Assembly, Nice, France, 6-11 April, 2003.
- Comte, D., H. Tavera, C. David, D. Legrand, L. Dorbath, A. Gallego, J. Perez, B. Glass, H. Haessler, E. Correa, A. Cruz, Seismotectonic characteristics around the Arica bend, Central Andes (16ºS-20ºS): preliminary results, Abstract #S41A-05 of American Geophyisical Union Fall Meeting, San Francisco, USA, 2003.
- Cotilla Rodriquez, M. O., D. Cordoba Barba, Morphotectonics of the Iberian Peninsula, Pure and Applied Geophysics, 161(4), 755-815, 2004.
- Dey, S., S. Sarkar, R. P. Singh, Anomalous changes in column water vapor after Gujarat earthquake, Adv. Space. Res., 33(3), 274-278, 2004, doi: 10.1016/S0273-1177(03)00475-7.
- Griggs, D. T., D. D. Jackson, L. Knopoff, and R. L. Shreve, Earthquake prediction: Modeling the anomalous Vp/Vs source region, Science, 187, 537-540, 1975.
- Liu, J. Y., Y. I. Chen, S. A. Pulinets, Y. B. Tsai, Y. J. Chuo, Seismo-ionospheric signatures prior to M=6.0 Taiwan earthquakes., Geophys. Res. Lett., 27(19), 3113-3116, 2000.
- Okada, Y., S. Mukai, R. P. Singh, Changes in atmospheric aerosol parameters after Gujarat earthquake of January 26, 2001, Adv. Sp. Res., 33(3), 254-258, doi: 10.1016/S0273-1177(03)00474-5, 2004.
- Ouzounov, D., F. Freund, Mid-infrared emission prior to strong earthquakes analyzed by remote sensing data, Adv. Space Res., 33(3), 268-273, doi: 10.1016/S0273-1177(03)00486-1, 2004.
- Riznichenko, Y. V., Problems of seismology, Moscow, Nauka, 408 p. 1985.
Sammis, C. G., and D. Sornette, Positive feedback, memory, and predictability of earthquakes, Proceedings of the National Academy of Sciences, 99, 2501-2508, 2002.
- Serebryakova, O. N., S. V. Bilichenko, V. M. Chmyrev, M. Parrot, J. L. Rauch, F. Lefeuvre, O. A. Pokhotelov, Electromagnetic elf radiation from earthquake regions as observed by low-altitude satellites, Geophys. Res. Lett., 19(2), 91-94, 1992.
- Scholz, C. H., Sykes, L. R., and Aggarwal, Y. P., Earthquake prediction: A physical basis, Science, (181), 803-809, 1973.
Singh, R. P., and D. Ouzounov, Earth processes in wake of Gujarat earthquake reviewed from space, EOS Transactions, AGU, 84(26), 244-244, doi: 10.1029/2003EO260007, 2003.
- Sobolev, G. A., A. V. Ponomarev, Physics of earthquakes and its precursors, M. Nauka, 270 p., ISBN 5-02-002832-0, 2003.
- Stich, D., G. Alguacil, J. Morales, The relative locations of multiplets in the vicinity of the Western Almeria (southern Spain) earthquake series of 1993-1994, Geophysical Journal International, 146(3), 801-812, 2001.
- Varnes, D. J., Predicting earthquakes by analyzing accelerating precursory seismic activity, PAGEOPH. 130(4), 661-686, 1989.
VSEGEI, Automated analysis of the natural lineament sets. Technical Report of the Geology Research Institute of Russia (VSEGEI), Leningrad, 131 p., 1988.
- VSEGEI, Digital processing of the natural lineament sets: Technical Report of the Geology Research Institute of Russia (VSEGEI), Leningrad, 137 p., 1991.
- Whitcomb, J. H., J. D. Garmany, and D. L. Anderson, Earthquake prediction: variation of seismic velocities before the San Fernando earthquake, Science 180, 632-641, 1973.
- Zlatopolsky, A. A., Program LESSA (Lineament Extraction and Stripe Statistical Analysis): automated linear image features analysis - experimental results, Computers & Geosciences, 18(9), 1121-1126, 1992.
- Zlatopolsky, A. A., Description of texture orientation in remote sensing data using computer program LESSA, Computers & Geosciences, 23(1), 45-62, 1997.