Distinguishing between primary and secondary volcaniclastic deposits

The distinction between primary and secondary volcaniclastic deposits, which are currently defined as the “direct” products of volcanic eruptions and the “reworked” products of the former, respectively, is the first step to interpreting volcaniclastic deposits, particularly the genetic connection with active volcanism. The distinction appears straightforward, but is not always applicable to natural deposits. During the 3.7 ka BP eruption of the Songaksan tuff ring, Jeju Island, Korea, there was an invasion of typhoon. The tuff ring was partly submerged underwater and affected by wave activity for over a day, resulting in a peculiar volcaniclastic deposit composed of both vent-derived (primary) and substrate-derived (reworked or secondary) volcaniclastic particles. We propose a new term “reprocessed” for a category of volcaniclastic deposits or particles, which originated directly from volcanic eruption but was deposited finally by nonvolcanic processes. Here we show that both reprocessed and reworked particles can coexist in the same volcaniclastic deposit, making it impossible to differentiate it into either a primary or a secondary deposit according to the current definition of volcaniclastic deposits. We thus define the secondary volcaniclastic deposits as comprising either or both of reprocessed and reworked volcaniclastic particles.

and secondary deposits according to the current volcaniclastic terminology. We suggest that some deposits or particles, which originated directly from volcanic eruption but was deposited finally by nonvolcanic processes, can be described with a new term "reprocessed", and that both reprocessed and reworked particles constitute secondary volcaniclastic deposits.

Volcaniclastic Deposit at Songaksan
Jeju Island is an intraplate alkali basaltic volcano built on the southeastern Yellow Sea continental shelf 9,10 ( Fig. 1A). Songaksan is a young phreatomagmatic volcano, which erupted ~3.7 ka BP at the southwestern coast of the island 11,12 (Fig. 1B), providing a complete cross-section of a tuff ring (Fig. 1C). Recent studies reveal that the tuff ring resulted from a single continuous eruption 13 possibly in a month when the sea level was almost identical to that at present 14 . The tuff ring formed mostly above high tide level by pyroclastic surges and fall 15 , but contains three interbeds of horizontally laminated, low-to high-angle cross-stratified, and hummocky to swaly cross-stratified volcaniclastic deposits in the middle of the tuff sequence ( Fig. 2) up to an altitude of ~5.5 m, i.e., ~4.5 m above the high tide level. These interbeds, named units R1, R2, and R4, are interpreted to have formed by wave activity in a swash to surf zone when the sea level rose several meters above normal high-tide level during a storm event 16 . The triple intercalation of the wave-worked deposits is interpreted to reflect three tidal cycles during the storm event that is inferred to have lasted ~1.5 day. In this paper, we focus on unit R2 because it occurs along an interface between two tuff sequences that have contrasting accidental componentry and juvenile tephra composition, thereby making it possible to assess the relative proportions of the vent-derived (i.e., primary) and substrate-derived (i.e., reworked or secondary) particles in the deposit.
Detailed sedimentological observations of unit R2 reveal that the unit comprises both primary and wave-worked deposits. The former consists of poorly sorted and crudely stratified (lapilli) tuff and occurs in the proximal part (to the east of loc. 4; Fig. 1C), at altitudes higher than ~5.5 m above sea level, and at a distal locality (loc. 14), where the deposit accumulated upon a lava bulge, about 1 m higher than the surrounding areas. Stratification in these deposits is generally planar but shows subtle undulations and thickening/thinning of layers over the bedform reliefs and other topographic irregularities of the underlying unit T2 (Fig. 3A). The overall deposit features and its lateral continuation into thicker and coarser-grained deposit toward the crater rim suggest primary deposition from a pyroclastic surge 15 .
At altitudes between ~5.0 and 5.5 m (between loc. 4 and 7), unit R2 shows peculiar vertical facies changes from (1) well sorted very fine sand (ash) intercalated with mud drapes at the base, (2) seaward-migrating ripple cross-laminated and high-angle cross-stratified deposit composed of well-sorted sandy to granular materials, (3) horizontally laminated deposit composed of well-sorted sandy materials, to (4) poorly sorted and crudely stratified deposit at the top (Fig. 3B). The facies units 1 and 2 are interpreted to have formed in the surf zone affected by storm waves and return flows driven by coastal setup together with periodic suspension settling of fines 16 . The facies unit 3 suggests deposition on a beach face by swash and backwash of breaking waves; the facies unit 4 is interpreted to be primary pyroclastic surge deposit. The facies transition at these localities suggests gradual emergence of the depositional site from a surf zone to a subaerial surface associated with falling sea level during the ebb tide 16 .
At other distal localities in lower altitudes, unit R2 comprises hummocky to swaly cross-stratified deposits passing upward into horizontally to low-angle stratified deposits (Fig. 3C), suggesting deposition by wave-induced combined flows in the surf zone followed by deposition on a beach face by breaking waves 16 . The coexistence of evidently primary volcaniclastic deposits and wave-worked deposits and their lateral transition within the same depositional unit (Fig. 3D) provide strong evidence for the contemporaneity of volcanic eruption and storm wave-working of volcaniclasts during deposition of unit R2. In addition, the laterally extensive erosion of the underlying unit T2 suggests partial incorporation of unit T2 tephra into unit R2. We thus attempt to assess the relative proportions of the tephra that was reworked from unit T2 (i.e., secondary volcaniclasts) and the tephra that was derived from contemporaneous eruption (i.e., primary volcaniclasts) based on the analyses of the chemistry of juvenile tephra particles and the accidental componentry.
Previous studies reveal that the Songaksan tuff ring can be subdivided into four tuff sequences (A to D), which resulted from four magma pulses with marked chemical variations, particularly across the boundary between the tuff sequences B and C 13,17 . Above all, MgO content is useful for distinguishing juvenile particles from different tuff sequences, as is the case in other mafic volcanoes 18,19 . The MgO contents of juvenile tephra particles from tuff sequence B range between 2.83 and 3.33 wt% with a mean at 3.13 wt% (Fig. 4A), whereas those from tuff sequence C range between 4.00 and 4.90 wt% with a mean at 4.42 wt% 13 (Fig. 4B).
In order to obtain more representative values of MgO contents of unit T2 tephra, we analyzed the chemical composition of 342 juvenile particles from the unit ( Table 1). The analysis shows that the tephra particles of unit T2 have a wide range of MgO contents with some tephra particles having intermediate MgO contents (Fig. 4C). We interpret that the tuff unit comprises (1) low-Mg tephra (MgO content ≤ 3.33 wt%) from the earlier magma batch that was left in the diatreme, (2) high-Mg tephra (MgO content ≥ 4.00 wt%) from the later magma batch, and (3) intermediate-Mg tephra (3.33 wt% <MgO content <4.00 wt%) which resulted from mixing of the two magmas in the feeder dike. Geochemical and petrographic evidence for the magma mixing in Songaksan volcano is provided in a former publication 20 , and similar magmatic processes are also reported in other monogenetic volcanoes 21,22 . The analysis shows that unit T2 comprises ~30% low-Mg tephra, ~27% high-Mg tephra, and ~43% intermediate-Mg tephra (Table 1). A simple mathematical calculation yields a solution that ~55% of low-Mg magma and ~45% of high-Mg magma contributed to form unit T2.
As for unit R2, similar analyses were performed for 1,517 particles from 9 localities ( Table 1). The distribution of MgO contents of individual tephra particles is strongly dependent on the deposit facies. The primary deposits mostly comprise high-Mg tephra, of which the MgO contents are higher than the average MgO content of tuff sequence C but are similar to those of unit T3 (Fig. 4D-H). The tephra composition thus attests to derivation of the tephra mostly from the later high-Mg magma and subordinately from the diatreme-filling low-Mg tephra. On the other hand, wave-worked deposits have strongly bimodal distribution of tephra composition (Fig. 4I-R), attesting to their derivation from both the underlying unit T2 and the newly erupted high-Mg magma. A simple mathematical calculation suggests that ~60 to 90% of tephra in these deposits originated from the reworking of unit T2, and the rest from contemporaneous eruption of high-Mg magma ( Table 1).
The contents of accidental particles, composed mostly of detrital quartz grains from subsurface sedimentary strata, provide further evidence for the dual sources of unit R2 tephra. Unit T2 contains abundant accidental particles (34.3% quartz plus minor amounts of lithics and other crystals) (Table 1), whereas the tuff sequence C contains much smaller amount (an average of 8.0%) of accidental particles 13 . The componentry analysis shows that the primary facies of unit R2 contains less than 10% quartz, whereas the wave-worked facies contains ~19 to www.nature.com/scientificreports www.nature.com/scientificreports/ 32% quartz ( Table 1), suggesting that the wave-worked facies of unit R2 comprises significant proportions of both vent-derived and substrate-derived tephras whereas the primary facies of the unit comprises only vent-derived tephra.   13 , in which the magnesium contents were given in magnesium number.

Discussion
Volcaniclastic materials that were emitted from a vent are transported through air, water, granular debris or some combination thereof 8 , as witnessed in recent eruptions of Mount St. Helens, Washington, 1980 23 , Nevado del Ruiz, Columbia, 1985 24 , and Soufrière Hills volcano, Montserrat, 2003 25 among others. In addition to the changes in the transport medium, they can be subject to different processes and deposited finally by a process that is commonly completely unrelated with a volcanic eruption, such as streamflows, ocean currents 26 , waves/tsunamis 27 , tidal currents 28 , and winds 29,30 . The term "resedimented syn-eruptive volcaniclastic" was formerly proposed to describe a class of similar volcaniclastic deposits that are syn-eruptive but are not, or do not appear to be, primary 3 . However, those deposits resulting from such syn-eruptive, hybrid or combined volcanic-sedimentary processes have been poorly explored so far, and have been largely regarded as deposits of uncertain origin or ambiguous deposits 8 . Currently, deposits of such hybrid processes are defined as primary volcaniclastic deposits as long as the volcanic materials did not involve interim storage and reworking 8 .
The wave-worked facies of unit R2 is evidently syn-eruptive because it passes laterally into a primary facies toward the crater rim, and is interpreted to have formed by the entrance of a pyroclastic current into a stormy sea. The term "resedimented syn-eruptive" is not, however, appropriate to describe the wave-worked facies of unit R2 because only part of the deposit was resedimented from an earlier deposited primary deposit. So we propose a new term "reprocessed" for a category of volcaniclastic deposits or particles, which originated directly from a volcanic eruption but were deposited finally by nonvolcanic processes. The word "reprocess" has the meaning of "process again or differently" and seems to be an appropriate term to describe such deposits or particles. The terms such as "wave-modified but syn-eruptive" or "current-modified but syn-eruptive" can be used to describe the depositional processes of individual deposits, but the term "reprocessed" is proposed here as a comprehensive term for describing a category of syn-eruptive volcaniclastic deposits deposited finally by nonvolcanic processes, just as the term "reworked" is used, irrespective of the exact reworking processes involved.
We also propose to classify reprocessed volcaniclastic deposits as a subcategory of secondary volcaniclastic deposits because of a practical reason that such deposits would hardly be described as primary deposits by geologists because their depositional structures, which are the key criterion to distinguish between primary and secondary volcaniclastic deposits, would indicate a nonvolcanic process. The case study at Songaksan also shows clearly that reprocessed and reworked particles can coexist in the same deposit, making it practically and conceptually impossible to differentiate it into either a primary or a secondary deposit according to the current volcaniclastic terminology. We thus propose to define the secondary volcaniclastic deposits as comprising either or both of reprocessed and reworked volcaniclastic particles.
The new distinction proposed here will remove the ambiguity in the current distinction between primary and secondary volcaniclastic deposits because the final depositional processes can be more readily interpreted from a deposit than the interim processes of temporary deposition and resedimentation. According to our definition, we only need to interpret the "final" depositional processes of a deposit to distinguish between primary and secondary volcaniclastic deposits because all volcaniclastic deposits that were finally deposited by nonvolcanic processes are secondary volcaniclastic deposits. Where additional interpretation of the contemporaneity of volcanic eruption and the occurrence of interim storage and reworking of volcanic debris is possible, the secondary deposit can further be classified into a reworked or a reprocessed deposit.

Methods
Sedimentological observations. Sedimentological observations were made at fourteen sites along the western shore of the Songaksan tuff ring. Grain size, sorting, and clast shape, bed thickness, depositional and erosional structures, post-depositional deformation structures, bed geometry, and lateral bed continuity were described in the field. Individual beds or units were correlated by tracing them along the continuous coastal exposures. The altitudes of key stratigraphic surfaces were obtained by the South S82T RTK GPS surveying unit. componentry analysis. A total of eighteen specimens were obtained from these sites. About 50 grams of each specimen were immersed in water for a day and treated in an ultrasonic vibrator for 10 min. Those fractions coarser than 4 Φ (1/16 mm) were then selected by wet sieving, dried, and dry-sieved at 1 Φ interval. The ash grains between 0 Φ (1 mm) and 1 Φ (0.5 mm) were then impregnated with epoxy and prepared for polished thin sections for componentry analysis. Backscattered electron (BSE) images were obtained from the polished sections using a JEOL JXA-8100 electron microprobe at the Center for Research Facilities of Gyeongsang National University. 125 to 182 particles were counted from the BSE images of each polished section to obtain the percentages of accidental quartz grains in each specimen. chemical analysis of tephra. The MgO contents of more than 100 juvenile tephra particles from each specimen were also obtained from the same polished thin sections. A field-emission electron probe micro-analyzer (Model JXA-8530F PLUS, Jeol) at the same institute was used to obtain the MgO contents. Energy dispersive spectroscopy (EDS) analysis (Model X-max, Oxford) was conducted at 15 kV voltage, 10 nA current with the focal distance of 11 mm. The percentages of unit T2-derived and vent-derived tephras of unit R2 were calculated with the equation MgO specimen = (1 − x)MgO B + xMgO C , where MgO specimen is the average MgO content of a specimen, and MgO B and MgO C are the average MgO contents of tuff sequences B and C, which are 3.13 wt% and 4.42 wt%, respectively. Tephra particles of the primary facies of unit R2 have MgO contents larger than MgO C , and all of them are interpreted to be vent-derived.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.