3. Extrusive volcanism of TVG
Yen et al. (1984) was the first to notice that products of extrusive volcanism in the TVG strongly prevailed over those of explosive volcanism. We have investigated the morphology of well-preserved lava domes and flows in order to determine eruptive parameters of the most recent extrusive eruptions. Details of the petrology of TVG lavas are given in Chen (1975), Lo (1982) and Wang et al. (2004).

3.1. Lava domes

Examination of available outcrops has shown that the vast majority of well-preserved volcanic cones of the E–W ridge are composed of massive, poorly-to-moderately vesiculated crystal-rich andesitic lava with no internal layering. This fact, together with the dome-like steep-sloped morphology of the edifices, indicates that thedominant style of recent volcanism of the TVG was extrusion of viscous lava.

 Some domes (Shamao, Miantian, and Dajian) have very steep slopes, reaching 50° in their upper parts (Fig. 2a). These domes were formed by extrusion of almost completely solidified magma (plug domes or Pelean-type domes). Less steep slopes of some other domes (Datun, Dajianhou, and Huangzuei) indicate that they were formed by extrusion of less viscous lava, which was still able to deform plastically. Most edifices are close to half-spherical with one summit, indicating formation during single episodes of continuous magma extrusion. Thus, most domes may be considered monogenetic.

Available outcrops show that the domes are surrounded by short, steep-sided aprons of very coarse, clast-supported gravel deposits. The constituent rock clasts are of the same lithology as the domes; many have radial cooling joints and are oxidized to various degrees, indicating high temperatures during their deposition.

These are smallvolume deposits of hot avalanches accompanying dome extrusion. Few deposits of block-and-ash flows were found, and explosive episodes as well as Merapi-type dome collapses therefore had small volumes and were probably rare.

Based on observations of historical dome-forming eruptions (e.g., Walker, 1973; Zharinov and Demyanchuk, 2008; Vallance et al., 2008) as well as on physical modeling of dome-building eruptions (Fink and Griffiths, 1998), many studies have shown that volcanic domes form when the average discharge rate of the magma is low; commonly b1 m3/s.

This implies that each of the dome-forming eruptions of the TVG lasted several years to decades, in order to form edifices with volumes of 0.05–0.3 km³. However, structures of some domes of the TVG (Cising, Huangzuei, Siangtian) are more complicated, with deviating thick lava flows, indicating that the discharge rate of lava fluctuated during the course of some dome-forming eruptions, periodically exceeding 1 m3/s.

Several domes (e.g. Huangzuei and Siangtian) have well- or partially-preserved crater-like depressions on their summits. No pyroclastic deposits were found on the crater rims, suggesting that these craters are not likely to be of explosive origin.

Instead, they may have formed by subsidence as a result of magma receding back into the conduit (due to flank outbreak of lava, intrusion of magma at depth, or collapse of bubbles in the conduit through late-stage degassing).

Notably, the edifice of Cising volcano (Fig. 2b), which is the largest in the TVG, is different from the other domes: it was formed by several extrusive episodes, separated by periods of repose and/or periods of explosive activity.

The structure of Mt. Cising is composed of lava domes and thick lava flows. Thus, Cising volcano can be classified as a polygenetic effusive lava dome, being structurally transitional to a stratovolcano. The youngest lava flows of Mt. Cising will be described in the next section.

3.2. Lava flows

In the TVG, there are nine well-preserved lava flows (Figs. 1 and 2b). The flows have clearly defined outer boundaries allowing easy tracing from their frontal parts to their source. Most of the flows start at the slopes and summit areas of lava domes. Thicknesses of the flows are great (80–150 m), and the frontal parts are very steep, indicating high magmatic viscosities and yield strengths. Most flows consist of only one long branch (so-called single flows).

Nanhu and Siacao flows have several branches. DEM and topographic maps show that many flows have various large-scale surface structures common for viscous lava flows: longitudinal lateral levees and transverse pressure ridges (Yang et al., 2004). Structures typical for blocky andesitic lava flows are observed. Upper surfaces of the flows are composed of up to 15 m thick autobreccia — large vesicular lava blocks (meters across) surrounded by sandy gravel of the same composition. Contacts of brecciated and non-brecciated lava are sharp, but very irregular. The upper parts of lava flows are often vesicular (average 30%), while internal parts are dense (5–10%) and massive or display platy joints.

It was first shown by Walker (1973) that lengths of a lava flows depend mostly on magma discharge rate. Later, several methods to determine eruptive parameters based on the geometry of lava flows were devised (e.g. Moore et al., 1978; Kilburn and Lopez, 1991; Lyman et al., 1994; Stevenson et al., 1994; MacKay et al., 1998; Lyman et al., 2004).

We have determined geometrical parameters of the TVG lava flows from a 1:50,000 topographic map of the area, and have calculated the parameters of their eruptions (Table 2). The data show that effusive eruptions at TVG occurred with average magma discharge rates up to 10 m3/s and lasted up to several years, producing lava flows with lengths up to 5.6 km and volumes up to 0.6 km3.The edifice of Hunglu volcano is exceptional among TVG edifices, because a significantpart of it is composed of multiple, relatively thin (of the order of a few meters) “aa” type lava flows of basaltic composition.

4. Explosive volcanism of the TVG

Few primary pyroclastic deposits are preserved in TVG. Even if it is assumed that a significant part of pyroclastic units was removed by erosion, volumes of magma erupted in the form of pyroclasts constitute less than a few percent of the total volume of magma erupted in the TVG. Thus, eruptions of the TVG were less explosive than eruptions of subduction-related volcanoes in general.

Pyroclastic products of TVG are dominated by tephra fall units with rare block-and-ash pyroclastic flow and surge deposits. No ignimbrites (welded or unwelded) were found. Among the fallout deposits of the TVG we have distinguished: (1) pumice fallout deposits of plinian eruptions; (2) lithic ashfall deposits of vulcanian eruptions; and (3) explosive breccias of wet phreatomagmatic and phreatic eruptions.

4.1. Deposits of pumice fallout of Plinian eruptions

With rare exceptions, all deposits of this type were found on the slopes of Cising volcano. A significant part of the deposits was remobilized by lahars soon after deposition (see Section 6). The characteristics of both pumice and lithics in the Plinian deposits differ significantly from one outcrop to another, thus several (at least 4) different eruptions, which likely originated from different vents, occurred at Mt. Cising.

A vent of only one of these eruptions has been discovered. We refer it as Siaoyoukeng tephra ring because it is located near Siaoyoukeng recreational area (Figs. 1 and 2c). This is the only volcanic edifice of the TVG that is completely composed of pyroclastic material. The vents of the other Plinian eruptions may have been located somewhere in the summit area of Mt. Cising and are now completely buried by the youngest lava flows of this volcano.

Deposit of the Siaoyoukeng tephra ring. Only about 1/4 of the area of the tephra ring is exposed: its eastern part is buried/destroyed by deposition of the Cising 2 lava flow, and its southern part is buried by the Cising debris avalanche. The preserved part allows an estimate of the dimensions of the ring: crater rim diameter — 300 m, base diameter — 500 m, and maximum relative height — 40 m. Outer slopes of the ring edifice have inclinations of around 20°.

The ring is composed of unconsolidated well-sorted, clastsupported coarse lapilli (Fig. 3). The material is represented by strongly vesicular pumice (vesicularity 49–66%, average 59%) evenly intermixed with a large percentage (up to 40%) of lithics. The pumice is grey andesite, with thin alternating bands of darker, moderately vesicular, and lighter (almost white) highly vesicular material.

Lithics are represented by uniform, nonvesicular, grey andesite, as well as by a very small percentage of other rock types. Close to the base the
deposit is especially coarse: clasts of pumice reach 10 cm in diameter, and lithics up to 15 cm. The deposit contains ballistic blocks of angular dense andesite up to 50 cm in diameter, which likely represent underlying rocks ripped off from the conduit walls.

Outside the tephra ring, fallout deposit of the same eruption with a thickness of about 40 cm was found at a distance 1.3 km from the source, where it is represented by coarse lapilli with clasts of pumice up to 8 cm in diameter.
SEM images of the pumice show that it has round bubbles tens to hundreds of microns across, separated by thin 1–10μm walls of glass (Fig. 4a,b). Bubbles frequently coalesced, forming larger irregular vesicles (up to a few mm across). The character of broken bubble walls indicates that the magma was ductile during fragmentation.

Vesicularity indices of the pumice as well as grain-size characteristics of the tephra are similar to those of deposits of dry Plinian eruptions (Walker, 1981; Houghton and Wilson, 1989). The erupted magma volume can roughly be estimated at 0.1–0.3 km3. Based on the grain size and volume of the deposit, we suggest that the eruption was of VEI 4 scale, with an eruption column approximately 15–20 km high (Newhall and Self, 1982; Pyle, 1989).

Deposits of other Plinian eruptions of the TVG have grain-size distributions similar to the deposit of the Siaoyoukeng tephra ring. Vesicularity indexes of pumices from these deposits are lower (vesicularity 36–56%, average 48%), and bubbles are smaller than those of the tephra ring pumice. The eruptions that ejected these relatively dense pumices may have been of phreatomagmatic type, in which case the vesiculation of magma was arrested by chilling when it came into contact with water.

Fission track dating of zircons from the Siaoyoukeng tephra ring deposit yielded an age of 0.5±0.08 Ma (Wang and Chen, 1990). However, good preservation of the tephra ring edifice, composed of friable and easily-erodible pyroclastic material, as well as the presence of fresh glass in the pumice, suggests a much younger age. Based on stratigraphic relations with other deposits in the area, we suggest that all Plinian eruptions of Mt. Cising, including that which formed the Siaoyoukeng tephra ring, were closely spaced in time and occurred at the very end of Pleistocene (see Section 7).

4.2. Lithic ashes of vulcanian eruptions

Lithic ashes were found in two principle locations: (1) the area immediately to the south of the cones of Dajianhou and Huangzuei volcanoes; and (2) the area near the southern foot of Shamao lava dome.

In the Mt. Dajianhou–Huangzuei area a layer of light-brown ash of 30–40 cm thickness mantles the ground surface. It rests on brown clay (strongly weathered lava) and is covered only by soil layer of 20–30 cm thickness. The ash is medium-coarse grained, well-sorted (Fig. 3a) and has poorly developed parallel lamination. In many locations it is moderately indurated (turned into tuff). The ash particles are poorly vesiculated and have an angular, blocky morphology.

The ash layer becomes notably thicker and coarser grained towards Mt. Huangzuei, indicating that the ash may have been erupted by that volcano. Moreover, near the foot of Huangzuei numerous ballistically deposited blocks several tens of cm across are scattered on the surface of the ash. The blocks are composed of dense, dark-grey andesite with poorly developed “bread-crust” surfaces, and are dissected by radial cooling joints. The material of the blocks is similar to that of the ash.

Multiple xenoliths of Miocene sandstones up to several cms across are enclosed in the andesite of the ballistic blocks.
Morphologies of both the ash particles and the ballistically deposited blocks indicate that they were formed as a result of fragmentation of very viscous (possibly already solid) poorly vesicular magma.

The explosions were probably phreatomagmatic, and occurred during the initial and/or intermediate stages of extrusion of the Huangzuei lava dome. Relatively good sorting of the fallout ashes suggests that these were relatively dry phreatomagmatic eruptions (i.e. that the magma/water ratio was high). We refer to that explosive activity as “vulcanian” in a broad sense (Heiken and Wohletz, 1985; Morrissey and Mastin, 2000).

Based on the degree of preservation of the ash layer, as well as the thickness of soil above it, this ash may be of the very end of
Pleistocene–Early Holocene age.

In the Mt. Shamao area, four ash layers (designated by numbers 1 to 4 from bottom to top) are intercalated with very fine-grained and finely laminated deposits of a paleolake (Fig. 5). This small lake was formed when a landslide from the southern flank of Shamao lava dome dammed the narrow valley of Huangsinei Creek.

The lake sediments (partially chemogenic) were deposited in oxygen-poor low-energy conditions, suggesting that the lake was rather deep in addition to being partially fed by thermal springs. The lake therefore provided favorable conditions for burying and preservation of ash layers. The exposed ash layers are accordingly 5, 6, 20 and 3 cm thick,and have sharp contacts with the lake sediment. The uppermost layer 4 is traceable over considerable distance.

Morphology and grain-size characteristics of the ash particles (although they are notably finer grained) are similar to the ash found in the Mt. Dajianhou–Huangzuei area (Figs. 3a and 4c,d) and probably have originated from similar vulcanian-type explosions. Radiocarbon dating has shown that the ash layers were deposited in the very end of Pleistocene (see Section 7).

4.3. Explosive breccias of wet phreatomagmatic eruptions

A deposit of this type makes up a small well-preserved tuff cone at the summit of Hunglu volcano. The major (and older) part of the Mt. Hunglu edifice represents a shield-like volcano composed of thin lava flows of basaltic composition. The lava edifice had a crater of about 350 m width, which is now almost completely buried by the tuff cone. Part of the rim of the old crater is recognizable in the W-NW parts of the summit area.

The intracrater tuff cone has a base diameter of approximately 400 m, its maximum relative height is 40 m and the diameter of its crater is 150 m. The tuff cone is composed of very poorly sorted weakly indurated breccia (massive or with crude layering parallel to the cone slope), which is composed of angular or slightly subrounded coarse lapilli in a deposited blocks up to 1.5 m across are scattered along the crater rim of the tuff cone.

The blocks are composed of dense, dark-grey basalt with poorly developed “bread-crust” surfaces, and dissected by radial cooling joints.Very poor sorting, coarse grain size, blocky morphology of particles and low vesicularity of juvenile clasts in these deposits are similar to explosive breccias of wet phreatomagmatic eruptions (Wohletz and Sheridan, 1983; Belousov and Belousova, 2001).

The occurrence of a wet phreatomagmatic eruption from the summit of a rather high cone composed of basaltic lava flows is a rather unusual phenomenon. We speculate that before the formation of the tuff cone, the old Hunglu crater may have contained a crater lake or swamp (similar to that of the modern Hunglu crater).

When basaltic magma found its way to the surface on the bottom of the old crater, water would have interacted with the magma, resulting in the wet phreatomagmatic character of this eruption.
Lava flows underlying the tuff cone yield a K–Ar age of 0.11 Ma (Tsao, 1994). However, based on the degree of preservation of the tuff cone, as well as on the thickness of soil above it, this deposit may be significantly younger (end of Pleistocene to early Holocene).

4.4. Explosive breccias of phreatic eruptions

These deposits surround several small explosive craters with diameters of up to 170 m, located along two prominent fissures dissecting summit area of Cising volcano. The fissures are sub-parallel in N–S orientation. The western fissure is 2000 m long and up to 120 m wide, while the eastern fissure is 1000 m long and up to 100 m wide (Liu et al., 2007). The fissures are probably of tectonic origin, but their opening caused phreatic explosions which formed the craters. Along the western fissure a large part of a lava flow of Mt. Cising detached and slid downslope in the form of a low mobility debris avalanche (see Section 5).

Aprons of the breccias around the craters are up to several meters thick and a few hundred meters wide. The deposits are massive, or display poorly developed parallel layering. They are composed of very poorly sorted angular blocks in a matrix of fine ash. In many locations the deposits are weakly indurated. Ash particles have a blocky morphology, indicating fragmentation of solid lava (Fig. 4e,f).

The material represents fragments of internal parts of the lava flows of Mt. Cising (poorly-to-moderately vesicular grey andesite).
Significant amounts of hydrothermally altered rocks were not found in the breccias. Thus, the explosions that ejected these breccias were probably not of hydrothermal origin. The explosions may have occurred when fissure opening caused rapid lithostatic unloading of variously vesicular rocks of internal parts of the youngest lava flows of Mt. Cising.

This may only have happened soon after deposition of the lava flows, while they were still hot and contained pressurized gas bubbles. Similar phreatic eruptions occurred at Avachinsky volcano in 2001, when a fissure crossed a lava flow deposited in 1991 (Venzke et al., 2009, and unpublished data of Belousov and Belousova).
The explosions occurred simultaneously with a gravitational collapse of Mt. Cising, which took place about 6000 BP (see Sections 5 and 7).

5. Gravitational mass movements

The slopes of the youngest volcanic edifices of the TVG are smooth and weakly dissected by erosion, but many of them have broadly opened (up to 140°) horseshoe-shaped depressions, 0.5–1 km across (Figs. 1, 6 and 7). The morphologies of these depressions are identical to scars formed by large-scale gravitational collapses.

The scars are shallow-incised (100–200 m), and failure planes did not intersect the volcanic conduits. In most cases, the scars are not filled by younger volcanic edifices. Thus, there was no volcanic activity following these collapses, which probably occurred some period of time after the volcanoes had ceased to erupt. Possible exceptions are Mt. Bailaka, in the scar of which the cone of Hunglu volcano is located, and Mt. Dajian, in the scar of which Mt. Nioubei is located.

Many of the scars are situated where volcanic edifices are intersected by tectonic faults. In some cases (Mts. Datun and Cising) these are NNESSW-oriented faultsclearly visible aslineaments on DEM-basedshadow images of the TVG (Figs. 6 and 7). Some collapses (e.g., Mt. Siaoguanyin) may have been facilitated by EW-oriented faults, which are less prominent in the images. Some of the collapse scars (Mts. Cigu, Dajianhou, Datun West Peak) are asymmetric in plain view; one of the branches of their“horseshoe” is notably shorter than the other (Fig. 1).

Their shorter branches may have formed along —E–W trending faults, while their longer branches occurred along NNE-SSW trending faults. Debris avalanche deposits connected to the formation of the horseshoeshaped collapse scars of Datun, Siaoguanyin, Cising (Fig. 6), Cigu, and Dajianhou volcanoes were investigated; their characteristics are summarized in Table 3.

The avalanches of Datun, Siaoguanyin and Cising volcanoes were confined by the deep valley of Siahu Creek, where they cover one another (Fig. 7).
The largest of the avalanches, with a volume ∼0.1 km3, was formed as a result of collapse of the E part of Mt. Datun which was detached by the NNE-SSW-oriented Chihshan fault. The collapsed material formed a typical debris avalanche deposit composed mainly of block facies (terminology after Glicken, 1998): meters-sized domains of strongly shattered, deformed, but not completely intermixed material of the former volcanic edifice.

The dominant rock type of the avalanche consists of variously vesicular porphyritic hornblende andesite (lightgrey, grey, or reddish due to oxidation) similar to the andesites of Mt. Datun. The deposit is coarse-grained and poorly sorted (Fig. 3b); particles have blocky morphology (Fig. 4g,k) with microcracks common for debris avalanche deposits (Komorowski et al., 1991). The avalanche was moderately mobile with H/L ∼0.2 (Table 3). The rear slide blocks of the avalanche did not travel far from the source; they stopped high inside the collapse scar, forming multiple narrow toreva blocks descending downslope.

The Siaoguanyin debris avalanche deposit is in most locations represented by uniformly massive, very coarse grained, fines-poor, gravelly material (Fig. 3b) with boulders up to several meters across. At the northern slope of Mt. Huangsi the deposit is represented by avalanche blocks (oval domains a few meters in size, composed of monolithologic, weakly fragmented dark-grey dense andesite).

The blocks are surrounded by heterolithologic thoroughly intermixed material containing abundant clasts of Miocene sandstone picked up from the avalanche substrate. This is a “bulldozer facies” (terminology after Belousov et al., 1999), which was formed when the avalanche collided with the topographic obstacle.

The dominant rock type of the avalanche consists of dark-grey, dense (vesicularity 7–20%), subaphyric andesite, similar to andesites of the southern flank dome of Mt. Siaoguanyin.

Other rock types are rare in the deposit. The Siaoguanyin debris avalanche must have been hot during emplacement, because its deposit contains carbonized wood. Andesite boulders within the avalanche deposit frequently have radial cooling joints, and in rare cases “bread-crust” surfaces, giving the deposit the appearance of a lithic-rich pyroclastic flow.

The paucity of fine fractions in the deposit can be connected with elutriation of fines into the convective cloud when the hot avalanche travelled down slope. Thus the deposit bears features of both debris avalanches and lithic-rich Merapi-type pyroclastic flows. The characteristics of the avalanche deposit indicate that crystallized, degassed, but still hot material of a newly extruded lava dome was involved in the collapse.

The Siaoguanyin debris avalanche was rather mobile (H/L ∼0.16), despite its small volume (Table 3); its speed reached 40 m/s at adistance of 5 km from the source. This calculation is based on 80 m high runup of the avalanche on Mt. Huangsi.The youngest collapse with a volume of ∼0.05 km3 occurred at Mt. Cising.

The collapse took the form of numerous retrogressive landslide blocks, which only partially transformed into a debris avalanche of low mobility with H/L ∼0.25 (Fig. 8; Table 3). The leading snout of the landslide collided with the lower eastern slopes of Mt. Datun (where the avalanche formed a bulge — Mt. Jhuzihhu with a relative height of 70 m) then made a 90° left turn to the valley of Siahu Creek, along which it traveled for about 1 km.

Rear sliding blocks of the collapse traveled only several hundred meters and stopped near the landslide source (Fig. 7). The former lava flow that was involved in the collapse underwent only weak disintegration: material of the collapse is represented by big angular boulders with little fine-grained matrix. The dominant rock type of the avalanche is medium-vesicular (vesicularity 10–37%, average 24%), porphyritic, light-grey hornblende andesite (sometimes slightly reddish due to oxidation).

Around the distal snout of the avalanche, as well as along its southern margin,a “bulldozer facies” is well developed(Figs. 7 and 8).The deposit is poorly sorted, coarse-grained (Fig. 3b), and lacks internal layering. It forms a band of up to 200 m width and N10 m thickness, consisting of various types of material scraped by the avalanche from the substrate, displaced over a distance of several hundred meters, and thoroughly, although not completely, intermixed.

Clasts of dense dark-grey andesite picked up from the Siaoguanyin debris avalanche deposit prevail in the facies (hence the grain-size characteristics of several samples of the facies are similar to those of the Siaoguanyin debris avalanche); reddish material of former autobreccia of lava flows dominates locally. Some pumice as well as material of lithic ashes (in the form of weakly cemented layered blocks) admixed in minor amounts. Such a diverse composition results in the broad scattering of grain-size characteristics of this deposit (Fig. 3b).

Collapses of Mt. Cigu and Dajianhou volcanoes had small volumes of ∼0.01 km³ (Table 3), and the character of their deposits is transitional to large rockfalls (composed of large boulders with little matrix).
The studied collapses occurred after the volcanoes had ceased to erupt, and thus were not directly associated with volcanic activity.

 Hydrothermally altered rocks do not make up significant parts of the studied debris avalanches, although hydrothermal fields are common in the scars of the collapses. Probably weakening of mechanical properties of the volcanic edifices due to hydrothermal alteration did not play a key role in the studied collapses, but elevated fluid pressure and hydrothermal alteration in the foundations of the volcanoes may have had some role. The collapsed parts of the volcanic edifices were detached by tectonic motions, with collapses possibly triggered by seismic activity.

6. Lahars

A significant proportion of the volcaniclasts of the TVG is composed of lahar deposits. Deposits of each lahar are commonly massive, poorly sorted, and have thicknesses of 2 to 10 m.

Based on grain size and texture of the deposits, two types of lahars can be distinguished: poorer sorted debris flows (having broad polymodal grain-size distributions), and better sorted hyperconcentrated flood flows (the grain-size distributions of which have one prominent coarse mode) (Fig. 3c).

Deposits of debris flows are matrixsupported: large boulders are evenly distributed in a fine-grained silty to sandy matrix; these deposits are commonly weakly cemented. Deposits of hyperconcentrated flood flows are clast-supported: large boulders are distributed in gravelly matrix; these deposits are commonly friable. Individual debris flows have produced large volume deposits, but were relatively rare. Individual hyperconcentrated flood flows have produced smaller volume deposits, but occurred more frequently.

Lithic-rich and pumice-rich varieties of lahars can be distinguished based on density/vesicularity of constituent rock fragments. In some of the pumice-rich lahars the pumice content reaches 70% by volume. Density/vesicularity distributions of pumice from the lahars (vesicularity 49–64%, average 56%) are similar to density/vesicularity pyroclastic material which cannot be dated directly (see Section 7).

7. Stratigraphy and timing of the most recent volcanic events in the TVG

The stratigraphy of the youngest volcaniclastic deposits within one of the major drainage systems of TVG was established. The drainage system starts in the broad depression between Datun, Siaoguanin, and Cising volcanoes, from which the narrow deep valley of Siahu Creek emerges (Fig. 1). Going south, the valley takes in multiple small tributaries and becomes broader (as well as changing its name to Dakeng Creek, and then to Nanhuang Creek) and finally enters northern suburbs of Taipei City.

Two stratigraphic columns, formally called Nanhuang Creek stratigraphic column (key location A; Fig. 1) and Huangsinei Creek stratigraphic column (key location B; Fig. 1), were constructed for this area (Fig. 5c). Although direct tracing of deposits between the two columns is impossible due to the lack of outcrops, stratigraphic correlations have been made based on the lithologies of the deposits and five radiocarbon dates (all our samples were dated in Beta Analytic Laboratory, USA).

The Nanhuang Creek stratigraphic column covers a longer period of time. The oldest exposed volcaniclastic deposit in the area is the Datun debris avalanche. No organic material suitable for radiocarbon dating was found in the deposit.

The Datun debris avalanche is directly covered by the deposit of the Siaoguanyin debris avalanche that contains multiple charred fragments of wood dated at 22,660–23,780 BP (here and below 2 Sigma calibrated 14C ages are given; Calibration Database INTCAL04).

The Siaoguanyin debris avalanche is covered by a thick sequence of lahars, which is composed of three units with different lithologies. The lowermost unit is lithic-rich; it resulted from redeposition of the Siaoguanyin debris avalanche material.

The middle unit is pumicerich; it resulted from redeposition of pumice fallout deposits of several plinian eruptions of Mt. Cising. The upper unit is lithic-rich; it resulted from redeposition of Merapi-type block-and-ash flows originated from the youngest lava flows of Mt. Cising.

The uppermost unit of the Nanhuang Creek stratigraphic column is the Cising debris avalanche deposit. Fragments of plants from the bulldozer facies of the avalanche deposit were dated at 6010–6080 BP.

The Huangsinei Creek stratigraphic column is complimentary to the Nanhuang Creek stratigraphic column. It spans part of the same period of time, but shows a more detailed record of eruptions because Huangsinei Creek contained a lake in which volcanic ashes were deposited and preserved.
Two radiocarbon dates were obtained from the lake deposit (Fig. 5).

Finely dispersed organic material in the lake sediment taken from the middle of the interval between ash layers 1 and 2 yielded an age of 15,850–16,580 BP; a piece of wood from the lake sediment just below ash layer 4 yielded an age of 13,290–13,640 BP.

The lake deposit is covered by a 12-m thick lithic-rich lahar deposit. The lithology of the lahar deposit is similar to lithology of the upper lithicrich lahars of the Nanhuang Creek stratigraphic column. A piece of wood from the lahar deposit resting immediately above ash layer 4 yielded an age of 13,300–13,660 BP.

Deposits of pumice fallouts of Plinian eruptions have not been found intercalated with the paleolake deposits, and thus, no Plinian eruptions occurred in the lake area between approximately 16,000 and 13,000 BP. Thus, the lake deposit is probably younger than the pumice-rich lahars of the Nanhuang Creek stratigraphic column.

Stratigraphy and lithologies of the deposits (Fig. 5c) allow us to suggest the following succession of the most recent geological events of the area:

(1) The earliest and largest event in the area was a seismicallytriggered gravitational collapse of Mt. Datun with a volume of 0.1 km3. The resulting debris avalanche completely filled the valley of Siahu–Nanhuang Creek down to Mt. Huangsi. The collapse was followed by some period of time, during which there were no significant depositional events in the main valley; a deep canyon was eroded into the avalanche deposit.

(2) Approximately 23,000 years ago, a flank dome was extruded on the southern slope of Mt. Siaoguanyin. Shortly after its formation (tentatively b 100 years), part of the still hot dome (0.02 km3) collapsed with deposition of a high speed debris avalanche. Soon after deposition, part of the avalanche was extensively reworked, leading to the deposition of multiple lithic-rich lahars.

(3) In the period between 23,000 and 16,000 years ago several Plinian eruptions occurred at Mt. Cising. One of the largest eruptions (VEI 4) formed the Siaoyoukeng tephra ring. Multiple, relatively small lahars, originating at slopes of Mt. Cising, redeposited the erupted pumices.

(4) In the period between 16,000 and 13,000 years ago several small-to-moderate scale vulcanian eruptions occurred in the TVG area (preceding extrusive eruptions of Mt. Cising and possibly Mt. Huangzuei). Thin layers of lithic ashes of the eruptions were deposited in the area and preserved in the lake, which existed at that time near Shamao lava dome.

(5) One of the last magmatic eruptions of Mt. Cising occurred approximately 13,000 years ago. It started with small-to-moderate scale vulcanian explosive activity, which was followed by extrusionof alavaflow;deposits oftheaccompanying small-scale Merapi-type block-and-ash pyroclastic flows were redeposited by lahars that completely filled the lake near Shamao dome.

(6) Approximately 6000 years ago, possibly as a result of a strong earthquake, a large fissure dissected the summit area of Mt. Cising. This caused small-scale phreatic explosions from several locations along the fissure, as well as provoking gravitational collapse of the western slope of Mt. Cising. The collapsed material produced a debris avalanche of low mobility, with a volume of 0.05 km3. The explosive activity indicates that summit lava flows of Mt. Cising could have been extruded not too long before the opening of the fissure. If so, this may suggest that youngest magmaticeruptionof Mt. Cising occurred around 6000 years ago.