EARTH SURFACE PROCESSES AND LANDFORMS Earth Surf. Process. Landforms 35, 1116-1121 (2010) Copyright © 2010 John Wiley & Sons, Ltd. Published online 28 April 2010 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/esp.2016
Letters to ESEX
Transverse architecture of lahar terraces, inferred from radargrams: preliminary results from Semeru Volcano, Indonesia C. Gomez1* and F. Lavigne2 Ecole Normale Superieure, Department LSH, Lyon, France 2 University Paris 1 Pantheon Sorbonne, Department Geography, Laboratory LGP, Meudon, France
1
Received 12 October 2009; Revised 9 February 2010; Accepted 15 February 2010 *Correspondence to: Christopher Gomez, Ecole Normale Superieure, Department LSH, CNRS UMR 5600, 15 Parvis Rene Descartes BP 7000, 69342 Lyon Cedex 07, France. E-mail:
[email protected]
ABSTRACT: Semeru Volcano is the highest mountain of Java (Indonesia), and a vulcanian explosion occurs every 15 minutes on average, since 1967. Thus a constantly renewed stock of material and the heavy monsoon rainfall [3700 mm yr−1 at 1500 m above sea level (a.s.l.)] provide a perfect setting for the study of lahars and their deposits. Hence, we examined the architecture of lahars’ terraces 9·5 km from the summit in the Curah Lengkong Valley. We first used ground penetrating radar (GPR) over vertical exposures of the lahars cut-bank terraces. This allowed us to better understand transversal radargrams across terraces, which are not visually accessible in the field. Preliminary results from a single radargram are very instructive, since (1) they prove that the lateral architecture does not correspond to that observed from banks only; (2) we could observe the presence of lenses and stratigraphic discontinuities; (3) the setting of the various units can also help reconstruct deposition processes and the chronology of different units. In order to finalize these preliminary results, we however need to perform multiple GPR radargrams and provide a complete set of results. Copyright © 2010 John Wiley & Sons, Ltd. KEYWORDS: ground penetrating radar (GPR); lahar terraces; radargrams; Semeru Volcano, Indonesia
Introduction Lahars are rapidly flowing mixtures of rocks, debris and water, triggered on volcanic slopes (Smith and Fritz, 1989). Sediment concentration defines two main flow types, debris-flows (sediment concentration >60% volume) and hyperconcentratedflows (sediment concentration between 20% and 60% volume) (Pierson and Costa, 1987). Most lahars are non-cohesive flows, where grain interactions control the internal dynamic (Iverson, 1997a, 1997b; Iverson and Vallance, 2001; Iverson and Denlinger, 2001), while cohesive lahars are more scarce (clays >6~7% volume) – e.g. lahars at Papandayan in 2002 (Lavigne et al., 2005). Studies show that cohesive lahars – mudflows – tend to be more voluminous than non-cohesive lahars (Crandell, 1971; Mothes et al., 1998; Vallance and Scott, 1997), with volumes ranging for instance from ~0·8 × 105 m3 for small events to ~4·6 × 106 m3 for large events in Whangaehu River on Ruapehu Volcano, New Zealand (Cronin et al., 1997). Lavigne (1998) proposed a classification of lahars according to their link to volcanic eruptions, with three classes: (1) ‘Syn-
eruptive’ lahars occur during an eruption, when the volcanic material melts and incorporates ice or snow (Major and Newhall, 1989; Trabant et al., 1994; Thouret et al., 1995; Waitt, 1989), or when pyroclastic-flows enter rivers or lakes (Waythomas and Wallace, 2002). Former ice slurry lahars evolve with a distinct behaviour, because of the presence of ice chunks (Major and Newhall 1989; Pierson and Janda, 1994; Lube et al., 2009). (2) ‘Post-eruptive’ lahars are triggered a few days to a few years after an eruption, either by rainfall (Rodolfo and Arguden, 1991; Thouret et al., 1998; Tungol and Regalado, 1996), or by lake overflow/outbreak (Lecointre et al., 1998; Thouret et al., 1998, Waythomas et al., 1996). Finally, Lavigne (1998) created a last category (3), ‘noneruptive’ lahars, which are caused either by large landslides or debris-avalanches without any prior eruption. Lahars deposits are commonly massive and poorly sorted, with boulders (where main axis can reach several metres) in a sandy gravel matrix (Scot et al., 1995). In this matrix, blocks are usually disjoined, without any preferential orientation. Lavigne and Thouret (2000) explain this latter characteristic by important shear stress in the flow. A single or a series of
TRANSVERSE ARCHITECTURE OF LAHAR TERRACES
subhorizontal units dominates their longitudinal architecture, although eroded units can modify this stratification. Units are either ‘debris-flow deposits’ or ‘hyperconcentrated-flow deposits’, and sole layers (5 to 20 cm thick of sands and/or silts) can sometimes separate these units. While the longitudinal architecture of lahars terraces has been intensively studied for more than a decade (e.g. Coussot and Meunier, 1996; Vallance and Scott, 1997), very little is
1117
known about their transversal architecture, mainly because of the lack of natural transversal outcrops and the difficulty of digging into lahars deposits. Cronin et al. (2000) provided transverse trenches and test pits of lahar deposits in Whangaehu valley on Ruapehu Volcano (New Zealand), but only in areas dominated by sandy deposits. The aim of this research has therefore been to bring insights into lahars’ terraces transversal architecture in areas where
Figure 1. The Curah Lengkong valley, at Semeru Volcano: study location of lahars and their deposits. The Semeru Volcano is located to the east of Java Island, and it is the highest mountain in Java, at 3676 m a.s.l. Despite lahars and pyroclastic hazards (Thouret et al., 2007), cities such as Sapiturang or Sumbersurip developed on the flanks of Semeru Volcano. The study area is located in the Curah Lenkgong valley, 9·5 km to the summit of Semeru Volcano. Lahars regularly flow in this valley several times a year, therefore we have been studying lahars at this location for almost a decade (Gomez, 2001; Gomez et al., 2008; Lavigne, 2004; Lavigne and Suwa, 2004; Thouret et al., 2007; etc.). We usually work on a laboratory area of 5 km (a), and for the present GPR survey we studied a deposit downstream this area (b). Copyright © 2010 John Wiley & Sons, Ltd.
Earth Surf. Process. Landforms, Vol. 35, 1116-1121 (2010)
1118
C. GOMEZ AND F. LAVIGNE
deposits contains coarse materials and they are blocks bearing. In this contribution we present the preliminary results of this research. In order to reach this aim, we conducted our field survey at Semeru Volcano – Java Island, Indonesia (Figure 1). This volcano towers at 3676 m above sea level (a.s.l.) and belongs to the Tengger Caldera complex, which covers more than 900 km2. Semeru Volcano has been constantly active since 1967 (Thouret et al., 2007), with vulcanian explosions every 15 minutes on average, and pyroclastic flows every 3·5 to 7 years. Half of this erupted material is reworked within a year – mainly by lahars – (Lavigne, 2004) due to steep slopes towards the summit (>31°) and heavy monsoon rainfalls (3700 mm yr−1 at 1500 m a.s.l.) with 200 mm day−1 rainshowers every five years and 400 mm day−1 every 50 years (Lavigne and Suwa, 2004). We carried out a ground penetrating radar (GPR) survey 9·5 km from the summit of the volcano in the Curah Lengkong valley, which is located below the horseshoe-shaped summital rim, into which lahars flow several times a year (Gomez and Lavigne, 2009). Indeed the rather small catchment of the Curah Lengkong (28·5 km2) has a sediment discharge of 2·7 × 105 m3 .m2.yr−1 (data for 2000 in Lavigne, 2004), mostly because of several lahars, which can individually reach volumes of 5·7 × 105 m3 (Lavigne and Suwa, 2004). Each
event emplaces thick terraces deposits, which are the objects of our investigation.
Methdology GPR has been the main tool for this research. GPR uses characteristics of electromagnetic waves to respond to dielectric characteristics and discontinuities of the medium they travel in. Hence, it is an appropriate tool for shallow subsurface exploration, although significant challenges and limitations remain (Heggy et al., 2003; Olhoeft, 1998; Schrott and Sass, 2008). The use of this technology in earth-sciences started to increase in the early 1990s, as commercial devices became readily available. GPR has found a wide range of field applications, such as hydrostratigraphy (e.g. Kostic et al., 2005), landslides assessments (e.g. Grandjean et al., 2006), archeology (e.g. Baker et al., 1997) and tsunami deposits (Gomez et al., 2006). Investigations on volcanoes started a few years later, mainly with calibrations against outcrops (Russell and Stasiuk, 1997; Gomez-Ortiz et al., 2007) in order to determine the usability of GPR on volcanic terrain. Gomez-Ortiz et al. (2006) also coupled GPR with another geophysical method – electric resistivity imagery – in order to improve and complete retrieved datasets. GPR data remains associated with some uncertainty
Figure 2. Radargram at the terrace’s bank. This first radargram is 7 m long and slightly less than 1·5 m deep. It presents three horizontal units, although the top unit presents a discontinuity between 2 and 3 m. The different units on the radargram corresponds with stratigraphic units, which are separated by a horizon of silt and sands (white dashed line). The radargram also presents a more complex architecture in detail inside the units, with hyperbola that correspond to blocks inside the deposit. Copyright © 2010 John Wiley & Sons, Ltd.
Earth Surf. Process. Landforms, Vol. 35, 1116-1121 (2010)
TRANSVERSE ARCHITECTURE OF LAHAR TERRACES
1119
Figure 3. Transversal radagram on a terrace partially deposited by the 12 April 2006 lahar and units’ delineation in composite colours. Letter A to I corresponds to the different units and the letter J is the surface echo (created when the radio-wave enters the medium). The architecture of the terrace is dominated by a horizontal disposition of units with a discontinuity between 20 m and 30 m, which certainly marks the limit of erosion, before units A, B, C and I were emplaced. From the terrace bank at distance 0 m, and from surface observations, units A, B, C, I and H are lahar deposits, whereas G is an old pyroclastic-flow deposit from the Kali Koboan river (Figure 1). In details, unit H is presenting an internal horizontal bedding, which is not evidenced in other units. Unit C and E consist of lenticular sub-units. The question mark expresses an uncertainty upon the location of the vertical limit, because the signal was blurred by the presence of blocks indicated by hyperbolas on the radargram.
that requires calibration (e.g. detection of block-and-ash flow structures at Merapi Volcano – Gomez et al., 2008, 2009). For the present survey, we first calibrated the GPR survey using the signal we obtained from a 7 m long and 2 m deep radargram within a lahar deposit terrace that was accessible on the outcrop scale. Secondly we recorded a 42 m long and 1·5 m deep radargram across the same terrace, perpendicularly to the valley direction. The tool we used is a commercial Ramac® with a 500 MHz antenna mounted with a measuring wheel. At the laboratory, we processed the GPR signal with Reflex® (Sandmeier, 2009).
Preliminary Results The observed terrace was emplaced by the lahar of 12 April 2006. The deposited material is mainly composed of coarse clasts in a matrix of sands and gravels – typical of noncohesive lahars. The terrace deposit is organized in subhorizontal layers, separated by thin horizons of well-sorted silty to sandy material (
