1Mountain rivers with steep slopes and Quaternary stratovolcanoes are characteristic landforms in subduction zones such as Japan. Stratovolcanoes generally experience large-scale collapses, that are larger than those which occur in nonvolcanic mountains (Voight, 1978; Machida, 1984). Such collapse is called a volcanic sector collapse (Ui, 1983; Ui et al., 1986; Siebert, 1984, 1992; Siebert et al., 1987). In the case of volcanoes located inland in a subduction setting, material produced by a sector collapse may move down steep river valleys and reach lowland plains (Machida, 1984). As a result, sector collapse-related deposits can sometimes contribute to landform development and aggradation in the lower reaches of drainage basins. This possibility is potentially serious and should be taken into account in natural hazard assessment in Japan, the Philippines, Indonesia, and other countries in similar environments where populations and infrastructure are concentrated along river valleys and on plains. However, detailed investigation of these large-scale events, which generally do not occur with high frequency, has been restricted to relatively recent historical events such as Mount St. Helens (United States) eruption in 1980 (Voight et al., 1983; Glicken, 1996) and Bandai volcano, Japan, in 1888 (Nakamura, 1978; Moriya, 1988).
2Studies are needed on how and where primary material produced by prehistoric collapses traveled. In Japan, the Kisogawa volcanic mudflow is one of the few known examples of material traveling a long distance after a sector collapse of a volcanic edifice (The Quaternary Research Group of the Kiso Valley and Kigoshi, 1964; Takarada et al., 1999). This sediment transport occurred at ca. 50 ka, and the event-related deposits are traceable as far as 46 km from the source area as debris-avalanche deposits, which are then progressively transformed to debris-flow deposits traceable as far as 144 km from the source (Takarada et al., 1999). The source volcano is the 3063-m-high Ontake volcano located 210 km west of Tokyo in central Japan. As the event-related sediments are overlain by alluvium in the plain, a complete picture of the sediment transport and depositional processes has not been obtained.
3On the other hand, sediments derived from the sector collapse of Asama volcano at ca. 24 ka (fig. 1; Aramaki, 1963; Unozawa and Sakamoto, 1972; Nakamura et al., 1997; Takemoto and Kubo, 2003) are exposed continuously along the present river course, as far as 90-100 km from the source.
Fig. 1 – Shaded relief map of the region of the study area showing the distribution of deposits derived from the sector collapse of Asama volcano (modified after Yoshida and Sugai, 2007).
Fig. 1 – Carte en relief de la région étudiée montrant la distribution des débris provenant de l’effondrement de flanc du volcan Asama (modifiée d’après Yoshida et Sugai, 2007).
1: locations in fig. 2; 2: cross sections in fig. 3; 3: presumed distribution of deposits.
1 : localisations sur la fig. 2 ; 2 : coupes sur la fig. 3 ; 3 : répartition supposée des débris.
4The Asama event enables us to investigate the nature of catastrophic material transport by examining the lithofacies of the sediments and the depositional landforms. Asama volcano, a 2568-m-high edifice in the northwestern Kanto district, central Japan, is one of the most active Quaternary stratovolcanoes in Japan (fig. 1). Its assumed pre-collapse height was ca. 2900 m a.s.l. (Aramaki, 1963). The volcanic edifice is composed of an alternation of pyroclastic beds and lava flows composed of pyroxene andesites. Deposits originated from the sector collapse have been discovered in the surrounding region, and the deposits have been correlated with high accuracy using tephrochronology (fig. 1; Aramaki, 1963; Unozawa and Sakamoto, 1972; Nakamura et al., 1997; Takemoto and Kubo, 2003). We focus here on how and where the collapsed materials from Asama volcano traveled, paying particular attention to the topography along the flow path.
5This research deals with the material traveling away from the northern flank of Asama volcano. The lithofacies of the sector collapse deposits depends on travel distance from the source. In the Ohkuwa region, up to about 20 km from the source (fig. 1), large blocks with many jigsaw cracks are exposed in the sediment and the matrix is composed of clasts of various sizes with many wood fragments. This sediment is similar to reports of typical debris-avalanche deposits (a in fig. 2; Ui, 1983; Siebert, 1984). The blocks are considered to be derived from the source volcanic edifice.
Fig. 2 – Representative lithofacies of the sector collapse deposits of Asama volcano.
Fig. 2 – Lithofaciès représentatifs des débris de l’effondrement de secteur du volcan Asama.
Locations of exposures are shown in fig. 1.
La localisation des affleurements est indiquée sur la fig. 1.
6The absence of evidence of heating in the lithofacies indicates that when transported deposits were cold, although a phreatic explosion might have led to the collapse (Aramaki, 1963). Yoshida and Sugai (2006) recognized more than 100 hummocks in the Ohkuwa region (fig. 1). In Nakanojo, ca. 45 km from the source (fig. 1), there is a deposit consisting of brittle andesite blocks with jigsaw cracks, surrounded and supported by a matrix containing rounded fluvial gravels (b in fig. 2). Several hummocks are found in Nakanojo (Yoshida and Sugai, 2006). At an exposure located ca. 86 km from the source in the northwestern corner of the Kanto Plain (fig. 1), the proportion of matrix relative to blocks is at least two or three times as large as farther upstream (c in fig. 2). The event-related deposits in this area are un-bedded and un-sorted and contain large, fragile blocks composed of non-welded pyroclastic rocks (fig. 2; Yoshida and Sugai, 2006) as well as wood fragments and rounded gravels. No hummocks are observed.
7The block size ranges from several tens of centimeters to 15 m. Observations indicate that a single gravity current transported the material as a debris avalanche (Yoshida and Sugai, 2006). The material derived from a single sector collapse at Asama volcano would have arrived in the northwestern corner of the Kanto Plain after an extremely short time, because there is no evidence of a lake being created behind a dam of debris deposits.
8Schematic topographic and geological cross sections from where the material entered the Agatsuma River valley (Naganohara) to the northwestern corner of the Kanto Plain (figs. 1 and 3; Yoshida, 2004a and b; Yoshida and Sugai, 2005) were produced from field survey data and existing borehole column data, using topographic maps (scale 1/25,000) and aerial photographs (scale 1/20,000) published by the Geographical Survey Institute, Japan. A remarkable difference may be recognized in the elevation of the transverse depositional surface in the Agatsuma River valley, especially at Naganohara and Nakanojo (sections 1-1’ and 2-2’ in fig. 3).
Fig. 3 –Topographic and geological cross sections (modified after Yoshida, 2004a and b).
Fig. 3 – Coupes topographiques et géologiques (modifiées d’après Yoshida, 2004 a et b).
1: width of valley; 2: relative height difference; 3: borehole; 4: bedrock; 5: gravel layer (before event); 6: gravel layer (after event); 7: deposits derived from the event.
1 : largeur de vallée dans la zone de dépôt ; 2 : différences de hauteur relative entre la surfaceet la base du dépôt ; 3 : forage ; 4 : substrat ; 5 : couche de graviers (avant l’effondrement) ; 6 : couche de graviers (après l’effondrement) ; 7 : débris provenant de l’effondrement de flanc du volcan Asama.
9In contrast, the cross-sectional profile of the depositional surface is flat in the northwestern corner of the Kanto Plain (sections 4-4’, 5-5’, and 6-6’ in fig. 3). The relative height difference between the top depositional and the basal surfaces of the sector collapse deposits (i.e. the difference between the lowest altitude and the inferred highest altitude in each cross section) is 70 m at Naganohara, 40-50 m at Nakanojo and Komochi, and 15-25 m in the northwestern corner of the Kanto Plain (fig. 3).
10We reconstructed the projected longitudinal profiles of the approximated basal and depositional surfaces of the deposits derived from the sector collapse event of Asama volcano along the present river valleys (fig. 4A).
Fig. 4 – Longitudinal changes of geomorphological characteristics along the Agatsuma-Tone River valleys.
Fig. 4 – Évolution longitudinale des caractéristiques géomorphologiques le long des vallées des rivières Agatsuma et Tone.
A: longitudinal profiles of the present river bed and the depositional and basal surfaces of the event-related deposits from Asama volcano (modified after Yoshida et al., 2006); B: average channel slope of the basal surface at ca. 24 ka; C: width of the valley within the depositional area; D: relative height difference between the depositional and basal surfaces.
A : profils longitudinaux du lit actuel de la rivière et surfaces de dépôt et de base des débris se rapportant à l’événement du volcan Asama (modifié d’après Yoshida et al., 2006) ; B : pente moyenne du chenal de la surface de base de 24 ka ; C : largeur de vallée dans la zone de dépôt ; D : différences de hauteur relative entre la surface des dépôts et leur base.
11Before the sector collapse, the Agatsuma and Tone River valleys were steeper than they are at present. The average slope from the upper to the lower reaches was ca. 13% from Naganohara to Nakanojo, ca. 9% from Nakanojo to Komochi, ca. 6% from Komochi to northern Maebashi in the northwestern corner of the Kanto Plain, ca. 3% from northern Maebashi to southern Maebashi, and ca. 2% from southern Maebashi to Okabe, where the sector collapse deposits are no longer traceable (fig. 4B). Thus, the channel slope suddenly decreased at the northwestern corner of the Kanto Plain.
12The width of the river valley before the Asama volcano collapse event was also investigated by using topographic and geological cross sections (fig. 3). We define the pre-collapse valley width as the transverse distribution of the sector collapse deposits in each section. Thus, based on the estimated horizontal distribution of the event-related deposits, the width of the valley from Naganohara to Komochi was measured at ca. 1-km intervals along the present river course using 1/25,000 topographic maps, except for the section covered by younger sediments from Haruna volcano (Soda, 1989, 1996). The Agatsuma River valley above Komochi is narrow and less than 2-3 km wide at its widest point. In contrast, the valley width increases rapidly to several times, or ten times the width, in the northwestern corner of the Kanto Plain. Longitudinal changes in the relative height difference between the depositional and the basal surfaces of the deposits correspond well to the variations in the width of the valley (fig. 4C, D). The relative height difference decreases in the northwestern corner of the Kanto Plain, where the valley opens out. This greatly expands the horizontal space available to the flow. These clear differences in the characteristics of the deposits and in the topography of the valleys show that the Agatsuma River acted as a corridor allowing passage of the material to the northwestern Kanto Plain. The finding is consistent with our previous report, that ca. 80% of the event-related material in the northern flank of the volcano was deposited in the northwestern corner of the Kanto Plain (Yoshida and Sugai, 2007).
13The undulating and rugged cross-sectional profile of the depositional surfaces in the Agatsuma River valley suggests that the sediment flow crashed violently into the valley walls as it traveled for a distance of ca. 50 km. At Nakanojo and Komochi, the deposits cover part of the preexisting fluvial terraces, which were formed during an interstadial of the last glacial period (fig. 5).
Fig. 5 – Presumed flow behavior in the Nakanojo and Komochi regions and the impact on the existing fluvial terrace at Komochi, indicating erosion of top soil and loam from the ground surface along the main route of the flow (modified after Yoshida and Sugai, 2005; Yoshida et al. 2005). The assumed main part of the flow is schematically shown by the large arrows.
Fig. 5 – Comportement présumé de l’écoulement dans les régions de Nakanojo et Komochi et impact sur la terrasse fluviatile de Komochi, indiquant l’érosion de la terre végétale et du limon le long de la trajectoire principale de l’écoulement (modifié d’après Yoshida et Sugai, 2005 ; Yoshida et al., 2005). La partie principale supposée de l’écoulement est matérialisée schématiquement par les plus grandes flèches.
1: depositional surface on the existing terraces; 2: depositional area; 3: assumed flow direction.
1 : surface de dépôt sur la terrasse fluviatile ; 2 : zone de dépôt ; 3 : direction supposée de l’écoulement.
14The orientation of the major axes of the hummocks at Nakanojo shows the approximate flow directions (fig. 5A; Yoshida and Sugai, 2005; Glicken, 1996) and indicates that the material ran violently on to the existing terraces when they struck the valley side, so creating depositional ramps (section 2-2’ in fig. 3). At Komochi loam and topsoil, with a thickness of several meters on the existing terrace surface, were in some places completely stripped (fig. 5B-2; Yoshida et al., 2005); in particular, the flow eroded several meters from the ground surface at the bottom of the southern part of the terrace, which was situated directly in the assumed flow path. This interpretation is supported by the captured fluvial gravels in the event-related sediments from Nakanojo to the northwestern corner of the Kanto Plain (Yoshida and Sugai, 2007). In another place, an undisturbed loam layer overlain by the event-related deposits is preserved on the terrace surfaces farther from the river channel (fig. 5B-3; Yoshida et al., 2005), indicating that the erosional force of the flow was restrained there by the local topography.
15The flow spread laterally as the valley widened in the northwestern corner of the Kanto Plain, where a fan gravel (i.e.“Maebashi Gravel layer”) is extensively covered by event-related deposits (Yoshida, 2004b). Sand and silt layers a few meters thick found between the sector collapse deposits and the Maebashi Gravel layer (fig. 6) are evidence that the ground surface in the northwestern Kanto Plain was not so strongly eroded by the flow, in contrast to the flow behavior in the Agatsuma River valley.
Fig. 6 – Distribution of sand and silt between the Maebashi Gravel layer and the event-related deposits of Asama volcano in the northwestern corner of the Kanto Plain, interpreted from borehole column data.
Fig. 6 – Répartition des sables et des silts entre la couche de graviers Maebashi et les débris liés à l’événement du volcan Asama, à l’extrémité nord-ouest de la plaine de Kanto, interprétée à partir de données de forages.
16From Komochi to the northwestern corner of the Kanto Plain, the relative height difference between the depositional and basal surfaces of the material decreased to at most one-third of the upstream value, despite the rapid increase in the valley width to even ten times the upstream width. Moreover, the depositional thickness becomes significantly greater in an area farther from the source, farther out on the Kanto Plain (figs. 3 and 4). These findings suggest that the flow was a debris avalanche, and was not completely transformed into a lahar.
17In volcanic arcs, the distribution of material from sector collapse events depends on the topography in the area of the source volcano. If the volcano stands on a broad plain, a large portion of the collapsed materials will be deposited close to the source area, as in the case of Bandai volcano. If the source volcano is near the coast, a sector collapse can cause devastating tsunamis, as was the case at Ritter volcano, Papua New guinea and Mount St. Augustine, southcentral Alaska (Sigurdsson, 2000). A tsunami caused by the sector collapse of a volcano at Unzen-Mayuyama, Japan, in 1792 (Hoshizumi et al., 1999) was responsible for as many as 15,000 people being killed or missing, making this the worst volcanic disaster in Japanese history. Moreover, large amounts of material episodically produced by the collapse of inland volcanoes are likely to flow along valleys and, thus, quickly reach remote areas. The present study demonstrates that when a V-shaped valley with an average slope of 10% or so channels the flow, the sector collapse sediments can travel as far as 90-100 km, as happened in the case of Asama volcano sector collapse.
18The Kisogawa volcanic mudflow traveled 144 km from Ontake volcano down a similarly steep valley. However, there have previously been few similar studies, except for reports of cold lahars, which were saturated with water, such as that which occurred at Nevado del Ruiz, Colombia, in 1985 (Sigurdsson and Carey, 1986; Pierson et al., 1990). One reason for the lack of such reports is that it is difficult to find evidence of debris avalanche-related deposits from prehistoric events inside steep river valleys. Another reason is that such deposits are often overlapped by alluvium in the lower depositional area, as in the case of the Ontake sediments. The physical processes resulting from volcanic sector collapses are potentially as hazardous as catastrophic tsunamis. We emphasize the necessity of re-examining the regional settings of inland volcanoes, especially in island arcs, from the point of view of both the regional topography and the effect of water on sediment transport.
19Volcanoes and mountain rivers with steep slopes are characteristic landforms in volcanic arcs. We examined the topography along the path of the 24 ka-old sector collapse sediments from Asama volcano, together with the depositional landforms, which were traceable as far as 90–100 km from their point of origin. Geological and geomorphological interpretation of the deposits and previous tephrochronological researches suggest strongly that they were transported as a single gravity flow, which was intense in the steep and narrow Agatsuma River valley, over a distance of about 50 km. On the other hand, the event-related material were more quietly deposited in the main depositional area, where both the channel slope and the relative height difference between the depositional and the basal surfaces of the deposits decreased as the valley width expanded. The greater thickness of the deposits in the more distant downstream area, farther out on the Kanto Plain, suggests that the flow still maintained the rheological characteristics of a debris avalanche, although the effect of water on the sediment flow dynamics is still obscure. A V-shaped valley, with an average slope of ca. 10%, can thus transport large quantities of material produced in volcanic terrains to a downstream plain within an extremely short period, as occurred with both the Asama volcano sector collapse and the Kisogawa volcanic mudflow.