SAR INTERFEROMETRY MONITORING OF LANDSLIDES ON THE STROMBOLI VOLCANO

Antonello G. (1), Casagli N. (2), Farina P. (2), Guerri L. (2), Leva D. (3), Nico G. (3), Tarchi D. (1)

(1) European Commission, Joint Research Centre, Institute for the Protection and Security of the Citizen, Ispra (Italy),

dario.tarchi@jrc.it (2) Earth Sciences Department, University of Firenze (Italy), paolo.farina@geo.unifi.it (corresponding author)

(3) LiSALab Ltd., a JRC spin-off company, Legnano (Italy), davide.leva@lisalab.com ABSTRACT A large landslide occurred on the steep NW flank (Sciara del Fuoco) of the Stromboli volcano on December 30th 2002, following an intensification of the explosive activity of the volcano and the descent of a lava flow within the Sciara del Fuoco. Since February 20th 2003, this landslide is monitored through an innovative radar system installed on the flank of the Sciara del Fuoco. This apparatus, which was extensively tested by our research group for landslide monitoring, consists of a ground-based SAR (Synthetic Aperture Radar) interferometer, known as LiSA (Linear SAR), capable of producing a radar image of the observed area every 12 minutes, with a pixel resolution of about 2 m x 2 m. The interferometric analysis of sequences of consecutive images allows us to derive the entire deformation field of the observed portion of the Sciara del Fuoco and of the crater with a millimeter accuracy. The collected data are used for early-warning purposes, since they provide information over a wide area (about 2 km2), with a high accuracy and observation frequency. This note describes the results related to data acquired from February up to November 2003, which allow us to follow in detail the evolution of the slope instability of the Sciara. 1 INTRODUCTION Stromboli is a volcanic island located on the NE of Sicily in the Tyrrhenian Sea. An intensification of the explosive activity of the volcano started from November 2002. An eruptive fissure was opened at the base of the NE crater feeding the descent of a lava flow within the Sciara del Fuoco, the steep NE flank of the volcano. On December 30th at 13:22 GMT+1 a large landslide occurred on the Sciara del Fuoco of Stromboli. The landslide, which has an estimated volume of more than 30 million cubic meter, partially extended below the sea level and caused a tsunami which produced severe loss along the island shores [1]. After the events of December 2002, National Civil Protection Department (DPC) and the National Institute for Geophysics and Volcanology (INGV) set up an extensive monitoring system on the Stromboli. Our research group was involved by the National Civil Protection Department for installing an innovative apparatus, developed by the Joint Resarch Centre of the European Commission, for monitoring surficial ground deformations on the flank of the Sciara del Fuoco. The system consists of a ground-based SAR (Synthetic Aperture Radar) interferometer, known as LISA (Linear SAR). The specific version of the instrument installed in Stromboli has been called InGRID-LISA, acronym of “Interferometric Ground-based Imaging Deformeter”. Such an instrumentation is able to acquire a radar image of the observed area every 12 minutes, with a pixel resolution of about 2 m x 2 m. The processing of the radar images is based on the radar interferometry technique, well known from satellite imagery but, in this case, implemented by using ground-based sensors. The interferometric analysis of sequences of consecutive images allows us to derive the entire deformation field of the observed are with an accuracy of about 1 mm. Due to its completely remote sensing character and due to the possibility of obtaining data night and day and in any weather conditions, the system was placed on a stable flank of the volcano, in order to continuously monitorize the displacement field affecting the Sciara del Fuoco slope. This note briefly describes the proposed technology and presents a synthesis of results on the deformation of the landslide on the Sciara, based on the data collected between the end of February and the first days of November 2003. 2 MATERIALS AND METHODS The InGrID-LiSA system belongs to ground-based interferometers series of prototypes, designed by the Joint Research Centre of the European Commission [2] and it has been extensively tested by our research group for landslide monitoring [3,4,5,6,7,8].

____________________________________________________________

Proc. of FRINGE 2003 Workshop, Frascati, Italy,1 – 5 December 2003 (ESA SP-550, June 2004) 107_anton

InGrID-LiSA is constituted by a continuous-wave step-frequency (CW-SF) radar and by a linear rail 3.0 m long and two antennas that move on it every 5 mm. The microwave transmitter produces, step-by-step, continuous waves at 1601 discrete frequency values, sweeping the bandwidth from 17.0 and 17.1 GHz. The receiver acquires the in-phase and the quadrature components of the microwave signal backscattered by the target. Range and cross-range synthesis of complex images is obtained by coherently summing signal contributions relative to different antenna positions and different microwave frequencies. As radar images are obtained through sampling techniques, frequency and spatial steps have to be selected in order to avoid ambiguity in range and cross-range. With these operational parameters, InGrID-LiSA produces a synthesized radar image of the observed area every 12 minutes, night and day and in any condition of visibility, with a pixel resolution of about 2 m in range, and 2 m on average in cross range. The observed area is shown in Fig.1 and it contains five sectors (numbered from 1 to 5), separated by morphological depressions, visible by the SAR system. In the same figure, a radar power image of the target scene, projected on a horizontal plane, is shown; the color scale expresses the power of the backscattered signal. The above mentioned five sectors, separated by shadow zones, are evident on the image and correspond to the five zones sketched on the photo. The upper part of the Sciara is represented in sectors No. 2 and 3. The processing of the radar images is based on the radar interferometry technique, well known from satellite applications in the field of Earth observation [9]. Interferograms are calculated using pairs of sequential images taken at different times exactly from the same position (zero baseline condition). A coherence threshold between an image and the other has been fixed, and pixels with lower coherence, have been rejected. The interferometric analysis of sequences of consecutive images allows us to derive the entire displacement field of the observed portion of the Sciara del Fuoco and of the crater with an accuracy of about 1 mm. The displacement is calculated from the phase difference between the waves received in different times, through the cross-correlation between two images taken at different times. The result represents the displacement occurred along the LOS (line of sight) in the time interval and it does not contain topographic information, because position of the antennas is the same between the two acquisitions. The phase values are not unwrapped and therefore displacement values are affected by the intrinsic ambiguity of phase: if ground displacement towards the observer exceeds the end of the scale, i.e. -0.25λ = -4.4 mm, the successive values will restart from the opposite scale end, i.e. +0.25λ = 4.4 mm. After a ground displacement of 0.5λ = 8.8 mm, the image pixels are in phase (value = 0) again. The typical effect of phase wrapping is shown in the interferogram of Figure 2, where two interferometric fringes are evident on the third interferograms. By counting the number of phase cycles, it is possible to assess the amount of displacement (2 × 8.8 mm = 44 mm in this case). Deformation maps can be derived from interferograms by cumulating the displacements measured at each couple of consecutive images. Because of the short time interval (12 min) between each couple of images, these interferometric displacements are always smaller than half-wavelength, so unwrapping procedures are not necessary.

Figure 1. Photo of the target area observed from the InGrID-LiSA showing the five sectors where it is possible to obtain radar images and synthesized SAR power image showing the five sectors.

3 DISCUSSION OF RESULTS The systematic analysis of the entire sequence of SAR images produced by InGrID-LiSA since February 2003 (about 120 images per day and a total of more than 35,000 images), allows us to follow in detail the evolution of the displacements on the Sciara del Fuoco slope.

Through the radar data it is possible to observe two zones with different movements: the Sciara del Fuoco zone and the flank of crater zone (sectors 2-3 & 4). In particular, the portion of the Sciara del Fuoco exhibits a complex displacement pattern, deriving by the superimposition and interference of the following geomorphic processes: 1. lava flows which move at a high speed rate usually channelled into morphological depressions and sometimes diverting over the slope; 2. gravitational sliding of the volcanoclastic materials on the Sciara along a deep-seated slip surface, related to the land-slide of December 2002, identified by Tommasi and Marsella (2003) as landslide α; 3. gravitational slow viscous flow of cooling lava masses accumulated on the Sciara. The portion of the flank of crater exhibits a displacement deriving by a planar slide and by several small scale slope movements and erosional processes.

Figure 2. From left to right:12 m- interferogram showing the lava flows over the slope;1h-interferogram showing the gravitational sliding of the volcanoclastic materials; 24h-interfrograms showing the gravitational slow viscous flow of

cooling lava masses. 3.1 The Sciara del Fuoco deformations An automatic procedure has been implemented for the extraction of the velocity history in some selected points located in different sectors of the Sciara. Fig.3 shows the location of 4 points selected in the upper sector of the Sciara del Fuoco (on the left) and their velocity history plotted versus time (on the right). The employed algorithm analyzes four different 1 hour-interferograms each day and considers a spatial average of the displacement rate over an area of 5 pixels around the selected points. For all the four points, it is possible to observe wide fluctuations of the velocity with a short-period (ca. daily) which have been interpreted as effects of rapid shallow movements, such as lava flows, small debris slides and rock falls. All the selected points show two different trends related to different periods: one from February to July and one from July to November. In the former the velocity undergoes wide fluctuations, reaching the highest value. In this period the displacement rate greatly changes for the effects of shallow processes (lava flows, lava cooling, rock falls, debris slides), for the intensification of the effusive activity on the upper part of the Sciara and for a progressive destabilization of the slope related to the increase of load on the landslide head caused by the accumulation of lava and by the lateral pressure of magma within the eruptive fracture. In the second period the displacement rate shows a considerable decrease probably related to termination of effusive activity. The diagram of Fig.3 has the advantage to be based on an objective procedure and it is automatic updated every 6 hours for early-warning purposes. The main drawback of this type of representation is that it does not permit to clearly separate the effects of shallow processes, from the deep-seated ones (global landslide movement). It is however possible to interpret the sequence of 1 hour - interferograms produced each day, with the objective of pointing out the deformation pattern related to deep-seated deformations. In periods of scarce lava activity, these can be easily recognized since they are characterized by regular and homogeneous patterns of interferometric fringes which extend over the entire Sciara. During phases of intense lava activity this recognition is however difficult and sometimes impossible, since the pattern of fringes is disturbed by shallow processes and it requires a strong interpretation effort. In the diagram of Fig.4, the average velocity interpreted in this way is plotted versus time. This representation is based on a preliminary and subjective interpretation of the entire interferogram sequence (since February 2003) aimed at the detection of the displacement rate of the landslide on the Sciara. Deformations due to shallow processes, such as lava flows and rock falls, are therefore neglected on purpose. During the entire investigated period, the upper part of the landslide in the Sciara has shown displacement rates fluctuating between 0.6 and 10 mm/h (1.5 to 240 mm/day), with accelerating phases coincident with episodes of intense effusive activity. As previously mentioned, this could be interpreted as a consequence of the destabilising effects of

magma pressure and lava load on the landslide head. On the basis of this interpreted deformation rate (ranging from 10-6 to 10-7 m/s) the movement can be defined as “slow” according to the landslide velocity scale proposed by IUGS/WGL (1995) [10]. These interpreted data can be used to forecast a possible collapse of the landslide by using empirical models such those proposed by Fukuzono [11] and Voight [12]. For example, the same data are plotted in Fig.4 as inverse of velocity versus time. This diagram permits a clear representation of the deceleration phases of the landslide and can be employed for a “graphical” prediction of the time of failure.

Figure 3. Interferogram spanning a time interval of 1 h showing the location of four selected points whose velocity

history is plotted in the diagram which represents the velocity in the selected points plotted versus time at steps of 6 hrs and 4-points moving averages.

Figura 4. Interpreted average velocity of the landslide plotted versus time at steps of 24 hrs and Fukuzono analysis.

3.2 The flank of crater slide Through the interferometric data it has been possible to recognise a displacement situated on the flank of crater. It represents a translational slide moving between 0.4-0.7 mm/g, which can be classified as “very slow” according to the landslide velocity scale proposed by IUGS/WGL [10] (Fig.5). The explosion of 5 April 2003 caused a fast increase in velocity, bolting from 0.7 mm/g to 72 mm/g and it returned to normal values in two months.The collapse of the slide (about 106 m3) could generate an increase of the load on the head of the landslide affecting the Sciara, with disastrous consequences. For this reason it has been necessary to monitoring this area also (Fig.6). In addition, below the crater several small-sized and fast-moving slope movements were detected by the radar system. These deformations, which caused several small rock falls, have shown a retrogressive activity.

P3

P2

P1

P4P3

P2

P1P3

P2

P1

P4

0

5

10

15

20

25

18/2/03 9/4/03 29/5/03 18/7/03 6/9/03 26/10/03

Velo

cità

(m

m/h

)

P1P2P3P4Media Mobile su 4 per. (P1)Media Mobile su 4 per. (P3)Media Mobile su 4 per. (P2)Media Mobile su 4 per. (P4)

0,0

2,0

4,0

6,0

8,0

10,0

12,0

18/02/03 09/04/03 29/05/03 18/07/03 06/09/03 26/10/03

velo

cità

(mm

/h)

0,0

1,0

2,0

3,0

4,0

5,0

6,0

18/02/2003 09/04/2003 29/05/2003 18/07/2003 06/09/2003 26/10/2003

Rec

ipro

co d

ella

vel

ocità

1/

v (h

/mm

)

Figure 5. 24h-interferograms represent the displacements in the flank of the crater.

0

10

20

30

40

50

60

70

80

18/02/2003 09/04/2003 29/05/2003 18/07/2003 06/09/2003 26/10/2003

velo

cità

(mm

/gio

rno)

Figure 6. Velocity history versus time.

4 CONCLUSIONS InGrID-LiSA is a completely remote-sensing technique, implemented via ground-based sensors installed in a stable zone in front of the area to be monitored. This fact and the possibility of obtaining multi-temporal images showing deformation fields, represent elements of absolute innovation in the realm of deformation monitoring. The technique has been successfully tested in a series of applications on active landslides in Europe but it is applied for the first time in a volcanic environment. InGrID-LiSA is continuously active since February 20th 2003 and produces about 120 images per day of the area under investigation. During this period it has been possible to follow in detail the deceleration phase of the landslide of December 30th. The collected data can be employed in the framework of a global monitoring network for early-warning purposes, since they provide information over a wide area (about 2 km2), with a high accuracy (ca. 1 mm) and observation frequency (12 minutes). These results from the Stromboli volcano show how the new technique is suitable for an operational use aimed at monitoring ground deformations produced by mass movements on active volcanoes. Furthermore, the ground-based radar can remotely operate in ground conditions where satellite images are ineffective, such as steep slopes, and allows to keep under constant control even very rapid displacements concentrated in small areas. These characteristics encourage the application of the technique both as scientific tool able to significantly improve the knowledge of the volcano dynamics, and as a fundamental support to the conventional landslide monitoring and early-warning systems.

ACKNOWLEDGMENTS This work has been sponsored by the National Civil Protection Department (DPC). B. De Bernardinis and its group at the DPC are acknowledged for the support to the project and for the permission given to the publication. The authors wish to acknowledge M. Rosi (University of Pisa), P. Tommasi (University of Rome), G. Puglisi (INGV, Catania) for the useful discussions on the slope instability conditions, and M. Marsella (University of Rome) for providing the ortho-photo of the Sciara del Fuoco. REFERENCES 1. Tommasi, P. and Marsella, M., Relazione sulle condizioni di stabilità del fianco NE della Sciara del Fuoco al 26 maggio 2003. Unpublished Report. Civil Protection Department, June 2003. 2. Rudolf, H., Leva, D., Tarchi, D., Sieber, A.J., A mobile and versatile SAR system. Proc. IGARSS 99, 595-594, 1999. 3. Canuti, P., Casagli, N., Leva, D., Moretti, S., Sieber, A.J., Tarchi, D., Some applications of ground-based radar interferometry to monitor slope movements, Int. Symp. Landslide Risk Mitigation and Prot. of Cult. and Nat. Heritage, Kyoto 21-25 Jan. 2002, 357-374, 2002. 4. Canuti, P., Casagli, N., Farina, P., Leva, D., Tarchi, D., Nico, G., Some examples of slope movements monitored by ground-based SAR interferometry, Proc. Int. Conf. on Fast Slope Movements, Sorrento, Italy, May 11-13, 2003. 5. Casagli, N., Farina, P., Leva, D., Nico, G., Tarchi, D., Monitoring the Tessina landslide by a ground-based interferometer and assessment of the system accuracy. , Proc. IGARSS 2002, Toronto, 21-26 April, 2002, 2915-2917, 2002. 6. Casagli, N., Farina, P., Leva, D., Nico, G., Tarchi, D., Landslide monitoring on a short and long time scale by using ground-based SAR interferometry, 9th Int. Symp. on Remote Sensing for Env. Monitoring, GIS Applications & Geology. Aghia Pelagia, Greece, 23 - 27 Sept. 2002, SPIE, 2002. 7. Casagli, N., Farina, P., Leva, D., Tarchi, D., Application of ground-based radar interferometry to monitor an active rockslide and implications for emergency management, in: R. Hermanss et al. (eds.), Massive Rock Slope Failure. NATO Sc. Series Book. Kluwer, in press. 8. Tarchi, D., Casagli, N., Fanti, R., Leva, D., Luzi, G., Pasuto, A., Pieraccini, M., Silvano, S., Landslide monitoring by using ground-based SAR interferometry: an example of application to the Tessina landslide in Italy, Eng. Geology. 68(1-2), 15-30, 2003. 9. Zebker H.A., Goldstein R.M., Topographic mapping from interferometric Synthetic Aperture Radar observations, J. Geophys. Res., 91, 4993-4999, 1986.. 10. IUGS Working Group on Landslides, A suggested method for de-scribing the rate of movement of a landslide, IAEG Bull., 52, 75-78, 1995. 11. Fukuzono, T., A method to predict the time of slope failure caused by rainfall using the inverse number of velocity of surface displace-ment, Journal Japanese Landslide Society, 22, 8-13, 1985. 12. Voight, B., Material science law applies to time forecast of slope failure, Landslide News, 3, 8-11, 1988.