Deformation Characteristics of Unzen Lava Dome based on Long Range Displacement Monitoring

Transcription

1 Deformation Characteristics of Unzen Lava Dome based on Long Range Displacement Monitoring Yasuyuki SATOU, 1 Tadanori ISHIZUKA, 2 Senro KURAOKA, 3,* Yuichi NAKASHIMA, 3 and Takanori KAMIJO 3 1 Unzen Restoration Work Office Ministry of Land, Infrastructure, Transport and Tourism (7-4 Minamishimokawajiri machi, Shimabara-shi Ngasaki Prefecture , Japan) 2 Erosion and Sediment Control Research Group, Public Works Research Institute (1-6, Minamihara, Tsukuba-shi, Ibaraki Prefecture, , Japan) 3 R&D Center, Nippon Koei Co, Ltd. (234 Inarihara, Tsukuba-shi, Ibaraki Prefecture, , Japan) *Corresponding author. a4982@n-koe.co.jp Unzen volcano, in Shimabara City of Nagasaki Prefecture Japan, erupted in 199, causing recurrent formation and collapse of the lava domes. Although these hazardous failures of the domes have almost ceased in 1996, the large remaining lava domes exist along the steep slope. Unzen Restoration Work Office, Ministry of Land, Infrastructure, Transport and Tourism has been monitoring the movements of the lava dome by two electronic distance measurements (EDMs) and one ground based synthetic aperture radar (GBSAR) in order to ensure the public safety. The analysis of the EDMs measurements indicates that the 8 reflecting mirrors installed on the lava dome are moving towards east-south direction. On the other hand, the surficial displacement distribution, obtained by GBSAR, indicates that lower region of the lave dome may be moving more than the upper region. The mechanisms of these trends are assessed, taking into account the structure of the lobes and the underlying layers, in order to aid in developing the early warning system. Key words: Unzen volcano, electric distance measurement, ground based synthetic aperture radar, 1. INTRODUCTION Unzen volcano is located in Shimabara City of Nagasaki Prefecture which is in the western region of Kyushu Japan. The volcano erupted in 199, causing the recurrent formation and collapse of lava domes that almost ceased by Although these hazardous failures of the domes have almost subsided, the large remaining lava dome (approximately 1 million m 3 ) that exist along the steep slope is still posing threat to the public (Photo 1). The Unzen Restoration Work Office, Ministry of Land, Infrastructure, Transport and Tourism (MILT) has been monitoring the movements and localized failure of the lava dome in order to ensure the safety of the public. The monitoring instruments include surveillance cameras, vibration sensors, electronic distance measurements (EDMs), and the ground based synthetic Aperture Radar which is a relatively new method to be used in Japan. Among these methods, EDMs and the Ground Based Synthetic Aperture Radar (GBSAR) are the systems that have been continuously monitoring the movements of the lava dome. Two EDMs at two different locations have been monitoring the movements of the 8 reflecting mirrors, placed on the lava dome since 27 [Tamura et al., 212]. The results show that the change in the distance between the station and the mirror, which will be referred to as displacement, is between 38 mm and 69 mm per year. Furthermore, directions of the movements of the mirrors can be analyzed, using the displacements obtained by two EDM stations. The GBSAR, on the other hand, was installed in 211 following the fundamental verification tests performed in 21 [Mizoguchi et al., 211; Tamura et al., 212]. While EDMs measures the local movements of the mirrors, GBSAR measures the surficial displacement of the areas covered by the radar wave, thus allowing detection of the area that may move faster than the other areas [Bozzano et al. 28]. 92

2 Approximately 7 m Photo 1 View of the south side of the Unzen lava dome (Photos taken on October, 211) This manuscript firstly presents the main results and characteristics of the long range displacement monitoring by EDMs and GBSAR. Secondly, the potential mechanisms of the movements of the lava dome are discussed based on the results of these measurements and previous investigations that describe the formation of the lobes and debris deposits. displacements shown in Fig. 3 is attributed to the noise associated with the variation of atmospheric conditions; humidity, atmospheric pressure, and temperature. The amplitudes of the short term fluctuation are in the order of 1 mm to 2 mm. The annual fluctuation component of the other mirrors may not be as clear as in the case of P3 due to differences in the intensity of scatter. For example, the daily displacements obtained with P5 of Tenguyama EDM station show clear correlation with the monthly temperature from 27/9/1 to 21/9/1 (Fig. 4), whereas the trend is subdued from 21/9/1 to 214/3/1, during which the annual trend may have been masked by the short term fluctuation. Nevertheless, it may be assumed that the seasonal climatic effect is present in most of the results obtained with the EDMs. 2. AUTOMATIC ELECTRONIC DISTANCE MEASUREMENTS One of the ongoing main monitoring methods for detecting the movements of lava dome is the automatic long distance EDM, using the eight reflecting mirrors installed near the foot end of the lava dome. The change in the distances between the mirrors and the station are obtained with the two EDM stations set at two different locations since September 1 st of 27(27/9/1). The distances from the EDM station to the mirrors are ranging between 2.5 km to 4. km, depending on the mirrors. For example, distances from the two stations to the mirror, designated as P3, are shown in Fig. 1. The change in the distance will be referred to as the displacement in this manuscript. 2.1 Long term trends of displacement velocity The displacement rates of the reflecting mirrors between 27 and 214 do not show significant changes except for the periodic fluctuations. The fluctuations contain short and annual periodic components as can be seen in the example of P3 that is observed from Tenguyama EDM station (Fig. 2 and Fig. 3). The displacements between September 1 st of 21(21 /9/1) and 212/9/15 is expanded and plotted together with the monthly temperature as shown in Fig. 3, which shows that the annual fluctuation of the displacements appears to be correlated with the variation of the monthly temperature. The short term fluctuation of the Fig. 1 Locations of the EDM stations and reflecting mirrors Changein distance (mm) TenguyamaP3 Expanded in Fig.3 Change in distance 65 Mean monthly temperature 55 Dates Fig. 2 Displacements of reflecting mirror (P3) measured from Tenguyama EDM stations (September 1 st, 27(27/9/1)to 214/3/1) Temperature (Degrees) 93

3 Change in distance (mm) Sep-1 1-Nov-1 1-Jan-11 3-Mar-11 3-May-11 3-Jul-11 2-Sep-11 Tenguyama P3 55 Mean monthly temperature 45 Fig. 3 Displacements of reflecting mirror (P3) measured from Tenguyama EDM stations (September 1st, 21(21/9/1) to 212/9/15). Change in distance (mm) TenguyamaP5 Date Fig. 4 Displacements of reflecting mirror (P5) measured from Tenguyama EDM stations (September 1st, 27(27/9/1) to 214/3/1). The cumulative displacements taken over 6.5 years (from September 1 st, 27(27/9/1) until March 1 st, 214) are ranging between 15 mm to 37 mm, depending on the mirrors and the EDM stations as shown in Fig. 5. The two separate displacements measured by the two EDM stations can be combined in order to obtain the displacement vector of the mirror by vector summation, given the orientation of the line of sight from the EDM station to each mirror. The resulting displacements will be denoted as the combined displacement vector as shown in Fig. 6. The average displacement velocities calculated from the combined displacements are range between 38 mm/year and 69 mm/year. 2.2 Variation of the annual displacement velocity Assuming that the annual seasonal fluctuation starts every September 1 st (9/1), as shown in Fig. 7, the mean annual displacement velocity is computed for each mirror and EDM station in order to examine the degree of the variation in the annual velocity and the effects of rainfall on the annual velocity. 2-Nov-11 Dates 2-Jan-12 3-Mar-12 3-May-12 3-Jul-12 2-Sep Temperature (degrees) 65 Change in distance Meanmonthly temperature Temeprature (Degrees) Oonokoba Sabo Observatory 4 Tenguyama Station P1 P2 P3 P4 P5 P6 P7 P8 Mirrors Fig. 5 Cumulative displacements of each reflecting mirror measured from the two EDM stations (Period: September 1st, 27 (27/9/1) to 214/3/1). Cumulative displacment (mm) P1 P2 P3 P4 P5 P6 P7 P8 Mirrors Fig. 6 Combined cumulative displacements of each reflecting mirror (Period: September 1, 27(27/9/11) to 214/3/1). Combined displacement (mm) Fig. 7 Beginning and end of a one year period for the computation of the annual displacement velocity Potential correlation between the maximum monthly rainfall depth (Fig. 8) of a year and the displacement velocity was examined. It was found that the displacement velocities of the three mirrors (P3, P4, and P7), in the south-east end, tend to be higher for higher maximum monthly rainfall depths measured between 28 and 212 as shown in Fig. 9. However, the velocities from 212/9/1 to 213/9/1 are high even though the maximum rainfall depth is relatively low (Marked as 213/8 in Fig. 9). On the other hand, there are mirrors that show no clear correlation as shown in Fig. 1. Although no conclusive remark can be made regarding the correlation between the annual displacement velocities and the maximum monthly rainfall up to 213, there is a possibility that movements of the locations such as P3, P4, and P7 94

4 16 14 Monthly rainfall P6 P Month Fig. 8 Monthly rainfall from 27 to P1 P2 P7 P3 P5 P4 Fig. 11 Displacement vectors corresponding to the period from September 1, 27(27/9/1) to 213/3/1. Fig. 9 Annual mean displacement velocities of P3, P4, and P7 (Oonokoba Sabo Observatory) versus maximum monthly rainfall from 27 to 213. The dashed line indicates possible correlation between the displacement velocity and the maximum rainfall of the year with exception of data on 213/8. 65 Oonokoba Sabo Observatory 213/ /6 211/6 29/ /6 5 28/ P2 P Maximum monthly rainfall of the year(mm) Fig. 1 Annual mean displacement velocities of P2 and P6 (Oonokoba Sabo Observatory) versus maximum monthly rainfall. Displacement velocity (mm/year) may be affected by the rainfall. Further monitoring and the factors affecting the displacement rate need to be assessed taking into account the geological structure of a lava dome and the underlying layers. 2.3 Orientation of the combined displacement The directions of the combined displacement vectors of the eight mirrors, measured from September 1 st 27(27/9/1) to 214/3/1, showed that the movements of mirrors are, in general, directed towards the east-south direction (Fig. 11). The directions of the vectors of P1, P2, P6, and P8 deviate from the dipping direction of the surface contour: vectors of P6 and P8 are rotated towards the south and those of P1 and P2 are rotated towads the north with respect to the dipping direction of the contour. Two factors affecting the directions of the displacement vectors are considered. Firstly, the center of gravity of the 11 th lobe (lobe 11B in Fig. 12) is in the south side of the ridge of the slope before the eruption such that the mass may be pulled by the gravity in the southeast direction (Fig. 12). Secondly, the distinct old valley region of the slope, before the eruption, extends from the south-east end of 11B towards the east as indicated by the blue region in Fig. 13. The thickness of the debris deposit in the old valley regionis is in the order of 1 m to 2 m (Fig. 22). Consequently, lobe 11B may be deforming towards the old valley, considering that the deformability of the debris deposit in the valley is likely higher than that of the original slope before the eruption. 3. GROUND BASED SYSNTHETIC APERTURE RADAR (GBSAR) 3.1 Outline of the system The Ground Based Synthetic Aperture Radar (GBSAR) has been scanning the lava dome of Unzen Volcano since October 211 (Photo 2). This radar can measure displacements of arbitrary locations of the scanned scenario, and possibly detect the fast moving area [Bozzano et al. 28]. However, there are limitations as follows: (a) Displacements of vegetated areas cannot be measured. 95

5 Lobes 6 to 9 Surface before eruption (1991) Surface before formation of lobe 11B (1993) Surface of 28 Lobe 11B 5 m View from east to west Lobes 11A Lobes 11B Cross section shown in Fig (d) Direction of the displacement cannot be measured with one system. 3.2 Outline of the basic principles The displacement is analyzed, using the phase variation of the two scenarios measured at different times (Fig. 14). The synthetic aperture and the two dimensional resolution are realized by moving the antenna (radar sensor) along the 2 m long rail, which is referred to as the linear scanner, while emitting/receiving wave signals (Fig. 15 and Photo 2). If the radar sensor is fixed in one location, the resolution will be limited to the range direction. The center frequency of the radar wave is GHz, and the approximate wave length is 17 mm. The atmospheric effects are reduced by the method similar to the Persistent Scatterer (PS) technique which is used in the satellite based SAR [Farina, 26]. In the case of GBSAR, PS points can be considered as the stable ground control points, present on the scenario, with good SN ratio and coherence. A spatial mathematical function of the atmospheric effects is modeled, using the phase variation of the PS points, assuming that the phase variation of the PS points is due to the atmospheric effects. Then, the atmospheric noise of the rest of the pixels is removed with this spatial function. Fig. 12 Lateral cross section of the central part of the 11 th lobe (11B) shown together with the surface (green line) before formation of lobe 11B. (a) Radar location Antenna Fig. 13 Surface contour of the slope before eruption (1991). The location of the lava dome is indicated by the solid line, showing that the old valley extends from the southeast end of the lava dome. Control unit Liner scanner (b) Although it can detect the precursor of failure, it is difficult to measure the rapid movements such as rock fall and sudden collapse. (c) It may become difficult to remove the noise due to the atmospheric effects in case of long range monitoring. (b) Radar unit Photo 2 (a) Location of the GBSAR (b)view of the GBSAR (IBIS-L) inside the shelter. 96

6 Hence, it is noted that the displacements obtained after this process are the displacements relative to the PS points that are not necessarily the absolute immovable points. The relative displacements can be measured for each pixel which is defined by the range ( R) and cross-range ( CR) resolutions (Fig. 15). The range resolution is determined by the bandwidth B of the transmitted stepped frequency signal and the speed of light c: c B R = 2 (1) The cross range (azimuth) resolution is directly related to the synthetic antenna length L by: λ CR = r (2) 2 L where r is the distance between the target and the sensor and L is the length of the linear scanner. In the case of monitoring the lava dome of Unzen, R is approximately 1m, and CR is approximately 15 m. Linear scanner Pixel Sensor Object ΔR ΔCR Range direction Azimuth direction 3.3 Location of the system and general performance GBSAR is installed on the concrete foundation of the shelter, which is located approximately 3.5 km away from the lava dome in the southeast direction (Photo 2). The radar is oriented so that the line of sight is close to the direction of the movements of the lava dome. The data (pixel number, phase, backscattered power, etc.) is acquired every seven minutes, corresponding to the time of the radar sensor to make one cycle of movement on the linear scanner. The relative displacements are calculated through the phase unwrapping and correction for the atmospheric effects. When the system was operated for the first time in October 211, the displacements were calculated λ d = 4π Fig. 14 Principle of the interferometric technique ( ϕ ϕ ) 2 1 Fig. 15 Illustrations of GBSAR resolutions and mapping of the pixels on DEM. every seven minutes. After monitoring over a month, a significant random fluctuation in the time-series plot was found. The problem was due to the fact that the atmospheric phase components are much larger than the actual displacements which are in the order of.1 mm per day. Moreover, the projected scenario may have been lacking sufficient number of PS points for the correction of atmospheric effects. Therefore, a new method was applied to reduce the atmospheric effects, using the two dimensional unwrapping method which is based on the rule that the integration of the phase along a closed loop will be null. Furthermore, the time interval for the unwrapping was set to two days so that the incremental displacements are larger than the margin of error. After this modification, clear increase in the displacement velocity in response to rainfall was detected for areas, particularly, I and K in Fig. 16, where it can be seen that the velocity of the areas I and K clearly increased in response to the rainfall by over 2 mm/day on June 16 and June 24. These areas are steep and the materials consist of soft clinker. Therefore, the movements of these areas may have become active in response to the rainfall. 3.4 Results of the GBSAR from 212 to 213 The relative displacement maps (accumulated 97

7 I Period A: 212/4/18 ~ 212/6/17 (Approx. 2 months) Fig. 16 An example of time variation of the relative displacements obtained by the GBSAR. The velocity of areas I and K in the upper figure clearly increased in response to the rainfall by over 2 mm/day on June 16 and June 24. every two months) from April 212 to October 212 are shown in Fig. 17. The cumulative rainfall depths corresponding to these periods are shown in Fig. 18. These figures show that the areas with relatively large displacements are found to occur in the lower east-south region of lobe 11B. Moreover, the displacements in this area tend to become larger during the period of high rainfall than in relatively dry period: period A, B, and C shown in Fig. 18. A similar trend can be found with the relative displacement maps obtained from June 213 to December 213, i.e. the areas with relatively large displacements over 5 mm, corresponding to the orange and red pixels, are found to occur in the period with high rainfall period B as shown in Fig. 19 and Fig. 2. These results indicate that the areas of relatively large displacements during period B are located above the thick debris deposit, whereas the areas with relatively small displacements are located above other lobes which were formed before lobe 11B (Fig. 21 and Fig. 22). Therefore, it is implied that the thick deposit of the debris may be one of contributing factors that are inducing the relative large displacements. Period B: 212/6/17 ~ 212/8/18 (Approx. 2 months) Period C: 212/8/18 ~ 212/1/18 (Approx. 2 months) Fig. 17 Plan view of the displacement maps by GBSAR from April 212 to October 212. Raifall (mm) Period A Period B Period C Fig. 18 Cumulative rainfall depths corresponding to three periods shown in Fig

8 1 11A 4 Line of cross section shown in Fig. 22 Area of relatively large displacement during period B in Fig. 17 and B 9 2 Period A: 213/6/21 ~ 213/8/2 (Approx. 2 months) m Fig. 21 Schematics of the remaining lobes generated during the formation of the lava dome. The lines are drawn based on the original drawing made by Oota, m Period B: 213/8/2 ~ 213/1/19 (Approx. 2 months) Lobe 11B Area where the movement is relatively large Lobe 2 and 4 Natural ground before eruption Debris deposit 18 m Fig. 22 Cross section of the lava dome and the underlying debris deposit derived from the recurrent formation and collapse of the lobes. Period C: 213/1/19 ~ 213/12/17 (Approx. 2 months) Fig. 19 Plan view of the displacement maps by GBSAR from June 213 to December 213. Rainfall (mm) Period A Period B Period C Fig. 2 Cumulative rainfall depths corresponding to three periods shown in Fig SUMMARY The random and seasonal fluctuation components present in the displacements measured by the long range monitoring are the major obstacles of long range monitoring. These fluctuations are mainly induced by the variation of atomospheric conditions that cumulatively impose effects on the beams and waves while travelling through a long distance. In the case of monitoring at Unzen Volcano, these fluctution components in the displacements obtained by the EDM are relatively large compared with the annual displacement velocity such that the incremental displacements of several months cannot be differentiated from the noise. In spite of these limitations, distinct characteristics of the direction of the displacement and the velocities have become 99

9 discernible, owing to the continous monitoring over 6.5 years. The main results, providing useful insight for understanding the mechanisms of the movements, are outlined below. The displacement vectors of the eight mirrors, measured from 27/9/1 to 214/3/1, are directed towards east-south. Two potential mechanisms pertaining to the direction of the movements are proposed: (1) lobe 11B is pulled by gravity towards east-south, due to the non-symmetric geometry of the lobe 11B; (2) lobe 11B may be deforming towards the thick deposit of debris extending from the southeast end of lobe 11B towards the east. The Ground Based Synthetic Aperture Radar (GBSAR) also suffers from the variation of the atmospheric conditions which was reduced, owing to the modifications of the unwrapping algorithm of the interferograms. After two years of continuous monitoring, areas with relatively large displacements are found to occur in the lower east-south region of lobe 11B. Moreover, the displacements in this area appear to become larger during the period of high rainfall than in relatively dry period. The results of both EDM and GBSAR suggest that one of the main factors affecting the direction and the non-uniform movements of lobe 11B may be the differences in the deformability of the materials underlying 11B such as the debris deposit and other lobes. Further monitoring with EDM and GBSAR will be performed (1) to assess the effects of rainfall on the displacement velocities; (2) to improve the reliability of the measurements; and (3) to develop methodologies to evaluate the stability based on the results of measurements. Works Research Institute). Special thanks are due to engineers and administrators of IDS Ingegneria Dei Sistemi for the efforts in improving the performance of GBSAR. REFERENCES Bozzano, F., Mazzanti, P. and Prestininzi, A. (28): A radar platform for continuous monitoring of a landslide interacting with a under-construction infrastructure, Italian Journal of Engineering Geology and Environment, 2, pp Farina, P., Colombo, D., Fumagalli, A., Marks, F. and Moretti, S. (26): Permanent scatterers for landslide investigations: outcomes from the ESA-SLAM project, Engineering Geology 88, pp Mizoguchi, Y., Matsui, M., K, Tamura, Maeda, A., Ishizuka, T., Yamakoshi, T., Kuraoka, S. and Mayer, L. (211): Preliminary application of ground based synthetic aperture radar for surveillance, Proceeding of the 6 th Japan Society of Erosion Control Engineering (in Japanese). Oota, K. (1996): Review of eruption of the Unzen Volcano, Jinetsu, Vol. 33, No.4, pp (in Japanese). Tamura, K. and Maeda, A. (212): Monitoring of the lava dome of Unzen Volcano by EDM and ground based synthetic aperture radar, Trans. Journal of the Japan Society of Erosion Control Engineering, Vol.65 No.1, Ser. No.3, pp (in Japanese) ACKNOWLEDGMENT: Authors wish to express appreciation for the supports and guidance, concerning the licensing of the GBSAR station, provided by the Kanto Bureau of Telecommunications, Ministry of internal affairs and communications. For the long range monitoring, in general, authors are grateful for the kind advices provided by Dr. Takeshi Matsushima of Institute of Seismology and Volcanology, Faculty of Sciences, Kyushu University. During the initial phase of the monitoring with GBSAR, valuable suggestions were given by Dr. Takao Yamakoshi of Ministry of Land, Infrastructure, Transport and Tourism, Water and Disaster Management Bureau (Formally Erosion and Sediment Control Research Group, Public 1