Process-sedimentological challenges in distinguishing paleo-tsunami deposits
- First Online:
- Cite this article as:
- Shanmugam, G. Nat Hazards (2012) 63: 5. doi:10.1007/s11069-011-9766-z
- 852 Views
There has been a lively debate since the 1980s on distinguishing between paleo-tsunami deposits and paleo-cyclone deposits using sedimentological criteria. Tsunami waves not only cause erosion and deposition during inundation of coastlines in subaerial environments, but also trigger backwash flows in submarine environments. These incoming waves and outgoing flows emplace sediment in a wide range of environments, which include coastal lake, beach, marsh, lagoon, bay, open shelf, slope and basin. Holocene deposits of tsunami-related processes from these environments exhibit a multitude of physical, biological and geochemical features. These features include basal erosional surfaces, anomalously coarse sand layers, imbricated boulders, chaotic bedding, rip-up mud clasts, normal grading, inverse grading, landward-fining trend, horizontal planar laminae, cross-stratification, hummocky cross-stratification, massive sand rich in marine fossils, sand with high K, Mg and Na elemental concentrations and sand injections. These sedimentological features imply extreme variability in processes that include erosion, bed load (traction), lower flow regime currents, upper-flow regime currents, oscillatory flows, combined flows, bidirectional currents, mass emplacement, freezing en masse, settling from suspension and sand injection. The notion that a ‘tsunami’ event represents a single (unique) depositional process is a myth. Although many sedimentary features are considered to be reliable criteria for recognizing potential paleo-tsunami deposits, similar features are also common in cyclone-induced deposits. At present, paleo-tsunami deposits cannot be distinguished from paleo-cyclone deposits using sedimentological features alone, without historical information. The future success of distinguishing paleo-tsunami deposits depends on the development of criteria based on systematic synthesis of copious modern examples worldwide and on the precise application of basic principles of process sedimentology.
KeywordsPaleo-tsunami depositsPaleo-cyclone depositsProcess sedimentologySedimentological criteria
The world’s coastlines have frequently been inundated by catastrophic waves of tsunamis and tropical cyclones during the Holocene (Clifton 1988; Lockridge et al. 2002; Scheffers and Kelletat 2003; Shanmugam 2008a; NHC 2009; NGDC 2009). Our attention to recognizing tsunami-related deposits has grown from sporadic journal articles in the 1960s to thematic volumes and symposiums in the 2000s (Coleman 1968; Bourgeois et al. 1988; Yamazaki et al. 1989; Shiki et al. 1996; Bryant 2001; Tinti and Pelinovsky 2001; Tappin 2007; Tsunami Society 2008; Tsunami Field Symposium 2008). Bourgeois (2009) reports that over 500 peer-reviewed articles were published through 2006. This is remarkable considering that tsunami research has accelerated only since the early 1980s. In proportion to the growing number of publications on deposits of tsunami origin, related controversies are also mounting. Specifically, there is a serious debate on distinguishing between deposits of tsunamis and those of tropical cyclones.
For example: (1) Ballance et al. (1981) interpreted anomalous deep-water sand beds with coconuts as deposits of tsunamis in New Zealand. However, Murty (1982) argued that these sands could alternatively be explained by tropical cyclones. (2) The origin of boulders on the Lanai Island, Hawaii, originally attributed to a giant tsunami wave approximately 105,000 years BP (Moore and Moore 1988), has later been reinterpreted to be products of multiple cyclone events (Rubin et al. 2000). Similarly, the origin of boulder fields in Australia is contentious (Bryant 2001; Felton and Crook 2003; Nott 2004). (3) The controversy over the origin of rip-up mud clasts by tsunami (Morton et al. 2007) versus by cyclone (Bridge 2008) continues. (4) A flurry of discussions and replies (Le Roux and Vargas 2007; Bridge 2008; Kelletat 2008; Dawson et al. 2008; Jaffe et al. 2008) ensued following papers on sedimentological features of tsunami-related deposits worldwide (Fujino et al. 2006; Morton et al. 2007; Dawson and Stewart 2007; Kortekaas and Dawson 2007). (5) In their review of the book ‘Tsunami—The Underrated Hazard’ by Bryant (2001), Felton and Crook (2003, p. 1) state that “…unequivocal criteria for distinguishing between deposits of storms and tsunamis have yet to be developed.” In his editorial to a special issue on “Sedimentary features of tsunami deposits”, Tappin (2007, p. 151) states that “Robust criteria for the discrimination of tsunami deposits from other sources such as storms is still lacking…” These are the clearest indications yet that the science of distinguishing paleo-tsunami deposits is far from settled.
The primary reason for this precarious situation is the hasty development of distinguishing criteria and the lack of process-sedimentological principles. In drawing attention to this chronic problem, the key objective of this paper is to present an overview and critique of a wide range of sedimentological criteria that have been used by various authors to identify potential paleo-tsunami deposits and to argue that these criteria cannot be used to distinguish between paleo-tsunami and paleo-cyclone deposits. Much of the sedimentary evidence for purported paleo-tsunami deposits is equivocal (e.g., Morton et al. 2007), especially when considered in isolation. The challenge of distinguishing paleo-tsunami deposits from paleo-cyclone deposits should centre on issues related to the processes of sediment transport and deposition. Unfortunately, there are too many examples in the paleo-tsunami literature of interpretations that lack detailed consideration of the physical processes and in particular of alternative mechanisms for transporting sediment to produce the deposit under consideration. This critical review is intended to serve as a reminder of the need to consider alternative hypotheses using principles of process sedimentology in tsunami research.
1.1 Process sedimentology
Process sedimentology (aptly ‘depositional process sedimentology’), a subdiscipline of physical sedimentology, is concerned with the detailed bed-by-bed description of siliciclastic sedimentary rocks for establishing the link between the deposit and the physics and hydrodynamics of the depositional process (Shanmugam 2006a, Chap. 1). It is the foundation for reconstructing ancient depositional environments and for understanding sandstone reservoir potential. Interpretation of depositional processes in the rock record is possible because physical features preserved in a deposit directly represent the physics of sediment movement that existed at the final moments of deposition (Middleton and Hampton 1973). However, one cannot interpret transportational processes or triggering events from the depositional record (Shanmugam 1996, 2006a, 2007, 2010). A combined knowledge of basic physics, soil mechanics and fluid mechanics is essential for interpreting the mechanics of various fluid–sediment–gravity processes (Brush 1965, p. 23). Sanders (1963) published the pioneering paper on process sedimementology entitled ‘Concepts of fluid mechanics provided by primary sedimentary structures’.
Process-sedimentological procedures require objective, accurate, precise, consistent and quantitative descriptions of the rocks using conventional nomenclature. This approach integrates bed-by-bed description (usually at a scale of 1:20) of the rocks with theoretical knowledge, experimental knowledge, modern analogues (physical oceanography), regional geology and logic/common sense. An important rule is that the field description of lithofacies using notations of facies models, such as the ‘Bouma Sequence’ (Bouma 1962) for classic turbidites (i.e., deposits of turbidity currents), is disallowed. This decree not only protects the basic tenet of physical sedimentology in maintaining the distinction between observation and interpretation, but also opens up a vista of alternative possibilities for process interpretations.
In process-sedimentological studies, deposits of various processes, such as turbidity currents (Sanders 1965), tidal currents (Klein 1971) and sandy-mass-transport deposits (Shanmugam 2006a, 2010), are routinely recognized in the ancient geological record using sedimentological criteria alone, without any supporting historical data. But this is not the case when it comes to distinguishing paleo-tsunami deposits, which invariably require precise dates of tsunamis. Because process-sedimentological procedures require direct examination of cm-scale features preserved in the rocks (core and outcrop), seismic profiles and wireline logs, which cannot resolve cm-scale features, should not be used for distinguishing paleo-tsunami deposits.
2 Process-sedimentological challenges
The problem of recognizing paleo-tsunami deposits is exacerbated by our failure to routinely apply the basic tenets of process sedimentology, such as (1) differences in sediment trigger versus transport (Shanmugam 2006a, 2007), (2) sediment transport versus deposition (Middleton and Hampton 1973) and (3) sediment deposition versus preservation (Nichol and Kench 2008). I address these basic issues as follows.
2.1 Triggering mechanism versus transportational processes
Tsunamis are oceanographic phenomena that represent a water wave or series of waves, with long wavelengths and long periods, caused by an impulsive vertical displacement of the body of water by earthquakes, landslides, volcanic explosions or extraterrestrial (meteorite) impacts. Earthquakes commonly generate tsunamis through the transfer of large-scale elastic deformation associated with rupture to potential energy within the water column (Geist 2005).
Physical, biological and geochemical features associated with the Holocene tsunami-related deposits
Medium to coarse sand, sharp or erosional basal contact, upward-fining and upward-coarsening sequence, and foraminiferal assemblages
Coastal beaches—2004 Indian Ocean Tsunami
Hawkes et al. (2007)
Discontinuous sand sheet (up to 80 cm thick), normally graded, unabraded marine shells and landward-fining trend
Coastal environments—2004 Indian Ocean Tsunami
Moore et al. (2006)
Normal and inverse grading, cross-laminae, rip-up clasts and diatom assemblages
Coastal beach ridges and swales—2004 Indian Ocean Tsunami
Sawai et al. (2009)
Upward-fining, upward-coarsening and upward-uniform textural trends
Coastal plain—2004 Indian Ocean Tsunami
Morton et al. (2008)
Sand sheet: up to 0.3 m thick, 20 m wide and 180 m long
Island (Atoll)—2004 Indian Ocean Tsunami
Nichol and Kench (2008)
Gravel-size corals mixed with man-made items
Coastal beaches—2004 Indian Ocean Tsunami
Whelan and Keating (2004)
Normal grading, parallel laminae, landward- and seaward-inclined cross-laminae and rip-up clasts
Coastal beaches—2004 Indian Ocean Tsunami
Choowong et al. (2008)
Erosional unconformity, horizontal planar laminae, cross-laminae, sand layers with heavy minerals, landward-fining trends and benthic foraminifera
Coastal beaches—2004 Indian Ocean Tsunami
Bahlburg and Weiss (2007)
Siliciclastic deposit (1.5 metre thick), erosional base, upper-flow regime plane beds, normal grading, inverse grading and small channels
Coastal plain—2004 Indian Ocean Tsunami
Switzer et al. (2007)
Erosive base, horizontal planar laminae, cross-laminae, sand layers with heavy minerals, landward-fining trends and reworked foraminiferal specimens
Coastal beaches—2004 Indian Ocean Tsunami
Srinivasalu et al. (2009a)
Syn-sedimentary deformation, truncated flame structures, and rip-up mud clasts
Coastal lowland—2004 Indian Ocean Tsunami
Matsumoto et al. (2008)
Boulder fields and a-axis measuring up to 14.8 m
Offshore (5–25 m deep)—2004 Indian Ocean Tsunami
Paris et al. (2010)
Multiple sand layers, erosional base, rip-up clasts, common normal grading, rare inverse grading, ripple cross-laminae and landward thinning
Coastal beaches and stream valley—2001 Peru Tsunami
Normally graded sand with rip-up clasts
Coastal plain—1998 Papua New Guinea Tsunami
Gelfenbaum and Jaffe (2003)
Gravel fabrics, sand sheets, current ripples and dunes
Coastal beaches—1993 Hokkaido Tsunami
Nanayama and Shigeno (2006)
Shell deposits showing large vertical and lateral extent, allochthonous mixing of articulated bivalve species out of life position and high amount of fragmented valves, with angular breaks and stress fractures
Microtidal lagoon—1945 tsunami
Donato et al. (2008)
Pumice-enriched complex lithofacies
Beach—1883 Krakatau Tsunami
Carey et al. (2001)
Tephra and associated sand layer with reworked shell fragments
Shallow water—1883 Krakatau Tsunami
van den Bergh et al. (2003)
Graded pebbly sand layers over deformed bedding
Smoot et al. (2000)
A huge sand dome, composed of massive medium to coarse sand with coarser particles
Bayhead beach—1854 tsunami
Sugawara et al. (2005)
Coastal beaches—1755 AD Tsunami
Kortekaas and Dawson (2007)
Sand units with a sharp increase in K, Mg and Na, and fining-upward trends
Coastal lagoons—1755 Lisbon Tsunami
Erosional lower contact, imbricated inclusions, rip-up clasts, fining-upward trend and planar bedding
Continental shelf—Late Bronze Age (ca. 1630–1550 BCE) tsunami related to Santorini eruption
Goodman-Tchernov et al. (2009)
Sheet of gravel with cobble-sized clasts tsunami
Coastal dunes—1400 AD
Nichol et al. (2003)
Washover deposits with pumice clasts
Back beach wetland—15th century tsunami
de Lange and Moon (2007)
Erosional basal contact, landward-fining trend, abrupt thinning at margins and rip-up clasts
Coastal beaches—15th century tsunami
Goff et al. (2004)
Gravel, sand and shell debris in lagoonal muds
Coastal lagoons—6th and 3rd millennium tsunamis
Vött et al. (2009)
Basal erosional unconformity, coarse to fine sand, normally graded to massive, rip-up clasts, marine fossils and landward-fining trend
Coastal lakes—7,000 years BP
Bondevik et al. (1997)
Fine arkosic sand layers encased in non-marine organic mud, abundant marine diatoms, no grading and landward-tapering and fining trends
Coastal plains—869 Jõgan Tsunami
Minoura et al. (2001)
Anomalous sand sheets with marine microfossils encased in mud or peat, rip-up clasts and normal grading
Tidal marsh, lagoon and coastal lake
Sand and gravel layers in bay mud, erosive base, internal mud drapes, repeated turnover of paleocurrent directions, hummocky cross-stratification, normal grading, inverse grading, transported shells, rip-up clasts and wood fragments
Floating boulders in sandy matrix and sheet geometry
Coastal to shallow water
Dawson and Shi (2000)
Cita and Aloisi (2000)
2.2 Triggering mechanism versus depositional record
Triggering mechanisms of mass-transport processes in deep-water environments
Short-term events: a few minutes to several days or months
2. Meteorite impact
3. Volcanic activity
4. Tsunami (This paper)
5. Tropical cyclone
6. Monsoonal rainfall
7. Ebb tidal current
1. Tectonic oversteepeninga
Intermediate-term events: hundreds to thousands of years
2. Glacial loading
3. Salt movement
4. Depositional oversteepening
5. Depositional loading
6. Hydrostatic loading
7. Biological erosion
8. Gas hydrate decomposition
1. Lowstand of sea level
Long-term events: thousands to millions of years
In comparison with turbidity current, sandy debris flow is a sediment flow with plastic rheology and laminar state from which deposition occurs through freezing en masse (Hampton 1972; Shanmugam 1996). Because of the freezing deposition, all sandy debrites (i.e., deposits of sandy debris flows) will develop similar depositional record with massive sand showing floating quartz granules, floating mudstone clasts, inverse grading, etc. Cantalamessa and Di Celma (2005) and Le Roux et al. (2008) interpreted tsunami-triggered backwash sandy debrites. However, tsunami-related sandy debrites cannot be distinguished from earthquake-induced sandy debrites because the physics of sandy debris flows would be the same in both cases at the time of deposition.
2.3 Depositional record versus transportational process
2.4 Depositional facies versus depositional environments
Studies have documented tsunami-related deposits in various Holocene environments (Table 1), which include coastal beaches, lakes, back-barrier marshes, lagoons, shallow water (shelf), and deep-water slope and basin (Fig. 1b). As pointed out earlier, sandy debrite facies will exhibit similar features because of freezing en masse. Consequently, a depositional facies, by itself, does not reveal anything unique about the depositional environment in which it occurs, be it subaerial or submarine. Facies association would help in establishing a particular environment (Reading 2001).
2.5 Single triggering event versus multiple depositional units
A single cyclone event can trigger multiple waves and related sediment gravity flows (Puig et al. 2004), which in turn can result in multiple depositional units. But there are no reliable sedimentological criteria to accurately determine the number of deep-water depositional units associated with a single cyclone or a tsunami event. This problem is further compounded when multiple triggering mechanisms operate concurrently (e.g., earthquake, tsunami, cyclone, etc.) and generate a messy stratigraphic record.
2.6 Short-term versus long-term triggering events
There are at least 17 triggering mechanisms of sediment failures that can initiate mass-transport deposits (MTD) in deep water (Table 2). These mechanisms are divided into three types depending on their duration: (1) short-term events that last for only a few minutes to several days, (2) intermediate-term events that last for hundreds to thousands of years and (3) long-term events that last for thousands to millions of years (Table 2). There are no established criteria to discriminate whether MTDs in deep water were triggered by short-term events (e.g., earthquakes, tsunamis, tropical cyclones, etc.) or by long-term sea-level changes. More importantly, there are no tools to measure the precise duration of gravity-emplaced units in a matter of hours or days in the geological record (Shanmugam 2009). Such factors are seldom addressed in tsunami research.
2.7 Tsunami- versus cyclone-related marine deposits
Tsunami-related deposition in open-marine environments may be explained in four progressive steps (Shanmugam 2006b, his Fig. 1): (1) triggering stage (i.e., wave generation offshore), (2) tsunami stage (i.e., wave propagation landward), (3) transformation stage (i.e., wave inundation) and (4) depositional stage (i.e., backwash flows and related gravity-driven processes seaward) on the shelf, slope and basin. Like tsunamis, tropical cyclones also emplace sediment on the shelf, slope and basin by gravity-driven processes (Shanmugam 2008a). The challenge of distinguishing tsunami deposits from cyclone deposits remains.
2.8 Sediment source of marine deposits
The source of sediment (provenance) for deposition in marine environments is commonly the land in cases of both tsunamis (Shanmugam 2006b, his Fig. 1) and cyclones (Shanmugam 2008a, his Fig. 6). In both cases, backwash flows and related gravity-driven processes are the primary means of downslope sediment transport. In this scenario, one might consider the use of provenance and related lithofacies characteristics as a criterion for distinguishing paleo-tsunami deposits. But such considerations are complicated by other issues, such as triggering event and flow transformation (Fig. 2).
2.9 Lack of tsunami-related empirical data
Unlike robust empirical data on velocities of bottom flows of tropical cyclones in shelf, slope and submarine canyon settings (Shanmugam 2008a, his Table 1), measured velocities of tsunami-related bottom flows are lacking. For example, maximum measured velocities of cyclone-triggered bottom flows are commonly in the range of 100–300 cm s−1 on the shelf and 200–7,000 cm s−1 in submarine canyons and troughs (see Shanmugam 2008a, his Table 1). At these high velocities, even boulder-grade grains would be eroded and transported by cyclone-triggered bottom currents.
In rare cases, bottom flow velocities of tsunamis have been estimated. For the 2004 Indian Ocean Tsunami, near bottom threshold velocity calculated from the biggest dune morphology shows a range of flow velocity of 1.74–1.03 m s−1 (Choowong et al. 2008). Nanayama and Shigeno (2006) reported that the outflow velocity of the 1993 Hokkaido tsunami in northern Japan was 234 cm s−1. Paris et al. (2010) estimated flow velocity of greater than 750 cm s−1 for transporting largest boulders during the 2004 Indian Ocean Tsunami. The problem here is that the lack of empirical data on velocities of bottom flows necessitates one to speculate on the transport of large boulders using sophisticated theoretical models. Nevertheless, elegant mathematical models are not a substitute for empirical data. In summary, the lack of empirical data and other process-sedimentological challenges are the primary impediments in distinguishing paleo-tsunami deposits.
3 Reliability of sedimentological criteria
The ultimate goal of studying Holocene sediments is to develop a viable set of sedimentological criteria for recognizing paleo-tsunami deposits (Table 1). In this context, Mamo et al. (2009, their table 1) claim that the existing catalogue of features, which include basal unconformity, intraclasts, fining-upward sequence, imbricated boulders, etc., can be used as reliable criteria for recognizing paleo-tsunami deposits. But there is nothing unique about this assortment of features, either individually or collectively, that points to deposition from tsunami-related processes.
3.1 Basal erosional surfaces (Fig. 3a)
Bryant and Young (1996) attributed the bedrock sculpting to tsunamis. But Bourgeois (2009, p. 18) argues that “There is no fundamental basis for the argument that tsunamis are more powerful sculptors than storm waves.” Like tsunami deposits, cyclone deposits also have basal erosional surfaces (Kortekaas and Dawson 2007). Erosional surfaces associated with the 2004 Indian Ocean Tsunami in Tamil Nadu (India) have been termed “erosional unconformity” (Bahlburg and Weiss 2007). According to the American Geological Institute’s Glossary of Geology, an unconformity is “A substantial break or gap in the geologic record…It results from a change that caused deposition to cease for a considerable span of time, and it normally implies uplift and erosion with loss of the previously formed record.” (Bates and Jackson 1980, p. 675). Because an erosional unconformity commonly encompasses millions of years (Shanmugam 1988), the synonymous use of the term ‘unconformity’ for an ‘erosional surface’ formed only very recently in 2004, without a substantial time gap, is inappropriate and misleading. By nature, all erosional unconformities represent erosional surfaces, but not all erosional surfaces are qualified to be erosional unconformities. The misuse of conventional nomenclature further confuses the issue of criteria in the tsunami literature. In open-marine environments, events (e.g., bottom currents) other than tsunamis can cause true erosional unconformities of great areal extent (Berggren and Hollister 1977; Howe et al. 2001). Thus, the tsunami versus cyclone origin of a deposit can only be delineated from the preserved depositional record (fill), not from a basal erosional (empty) surface.
3.2 Anomalously coarse sand layers (Fig. 3a)
Tsunami-related deposits occur as anomalously coarse sand layers (i.e., event bed) in comparison with the overlying and underlying fine-grained (mostly muddy) lithofacies (Ballance et al. 1981; Cantalamessa and Di Celma 2005; Dawson and Stewart 2007). These event beds are composed of many exotic fragments (e.g., plants, wood, beachrock, corals, etc.), which are absent in the overlying and underlying muddy units (Takashimizu and Masuda 2000; Fujino et al. 2006). Analogous to tsunami deposits, cyclone deposits are also composed of anomalous sand layers (Ager 1974) with exotic wood fragments (Siringan and Anderson 1994).
3.3 Exotic boulders (Fig. 3a)
Cyclones are also known for their ability to transport large boulders. Etienne and Paris (2010), for example, described boulder accumulations along the volcanic rock coast of Reykjanes (south-west Iceland). These accumulations consist of large boulders, up to 70 tonnes in weight, and were transported up to 65 m inland by powerful cyclones with wave heights of more than 15 m. Goff et al. (2004, their Fig. 6) have documented cyclone-transported boulders in New Zealand. Characteristics and hydrodynamics of cyclone-wave-related processes that transported boulders have been discussed by Goto et al. (2009). The 2004 Hurricane Ivan has been interpreted to be the force behind overturning a 3-m-long boulder in the Caribbean (Morton et al. 2006). Goto et al. (2011) claim that the size of boulders associated with cyclones is comparable to the size of boulders emplaced by tsunamis. Richmond et al. (2011) suggest that tsunami deposits “…are characterized by boulder-strewn gravel fields with large clasts and thin sediment accumulations with sheet-like sand and gravel deposits in topographic lows.” But gravel fields in topographical lows can also be associated with deposits of cyclones.
3.4 Imbricated boulders (Fig. 3b)
Imbricated boulders (Bryant et al. 1992, 1996; Bryant and Nott 2001; and Kennedy et al. 2007) and gravel clusters with imbrications (Yamazaki et al. 1989) have been interpreted as products of tsunami. Nonetheless, imbricated boulders have also been attributed to tropical cyclones (Williams and Hall 2004; Saintilan and Rogers 2005). Nott (2003) claims that because tsunami waves travel at much higher velocities than cyclone waves at the shore, tsunamis can exert a much greater force on a boulder per unit wave height. Nott (2003) calculated that a tsunami only needs to be approximately one-quarter the size of a cyclone wave to transport the same size boulder at the shore. But Morton et al. (2006) dispute these calculations. Because tropical cyclones and ‘meteorological tsunami’ (Rabonovich and Monserrat 1996) tend to behave like tsunami waves, imbricated boulders are not unique products of tsunamis. The flawed notion that tsunami is more powerful than cyclone is not based on empirical data (see discussion under Sect. 2.9). The belief that only tsunami can develop imbricated boulders is biased. In addition to cyclones, imbricated boulders have been attributed to sediment-outbursts from advancing glaciers (Davies et al. 2003).
The mode of emplacement of boulders, either from suspended load or from bed load (traction), is also the subject of ongoing debate. Some authors have proposed suspended load for the emplacement of boulders along the SE coast of Australia (Bryant et al. 1992, 1997; Bryant and Nott 2001; Bryant 2008). But others have argued that a suspended load origin is physically unrealistic based on theoretical grounds (Goff et al. 2010). Conventionally, imbricated clasts and boulders have been related to bed load (traction) deposition (Harms et al. 1982).
An alternative explanation for imbrications of clasts and boulders is by mass-transport processes, such as debris flows. Experimental studies of debris flows have shown the development of pebble imbrications (Major 1998). Imbrications have also been attributed to compressional shear straining of debris during emplacement (Nemec 1990). In the light of these alternative depositional mechanisms, the use of imbricated boulders as evidence of tsunami deposition is not compelling.
3.5 Chaotic bedding (Fig. 3c)
Chaotic bedding and slumps are associated with tsunami-related deposits (Coleman 1968). Michalik (1997, his Figs. 4.2, 4.3, 4.8) suggested that sigmoidal deformation features (i.e., “rope-ladder texture”) are the evidence for tsunami deposition. These deformation features are analogous to imbricate slices deposited by sandy debris flows in flume experiments (see Shanmugam 2000, his Fig. 18B) and to duplex-like structures formed by slumps in deep-water channels (Shanmugam et al. 1988). In addition, a large slump mass of about 100,000 m3 (328,000 ft3) in size with internal chaotic bedding was triggered by the May 1975 cyclone in the Scripps Submarine Canyon, off San Diego, California (Marshall 1978). Further, slumping features are commonly associated with mass movements in both subaerial and submarine environments due to rapid deposition. These examples illustrate that slumps and related chaotic bedding are not unique to tsunami-related deposition.
3.6 Rip-up mud clasts (Fig. 3d)
Rip-up mud clasts or intraclasts are reported from the tsunami-related deposits in coastal lowlands (Jaffe et al. 2003; Matsumoto et al. 2008), bay settings (Fujiwara et al. 2000), shallow-marine environments (Bourgeois et al. 1988; Cantalamessa and Di Celma 2005) and deep-marine environments (Takayama et al. 2000; Le Roux et al. 2004). Morton et al. (2007) and Jaffe et al. (2008) have argued that cyclone deposits are devoid of rip-up mud clasts because of the turbulence of the cyclonic water and prolonged vigorous agitation that disaggregates and disperses the mud clasts. But Tuttle et al. (2004, their Figs. 3, 4) have documented mud clasts in cyclone deposits on the foreshore beach side and on the back-barrier lagoonal side with washover deposits. These mud clasts were produced by the 30–31 October 1991 cyclone, known as “the Perfect Storm” (a popular movie was made after the same name). In the modern Monterey Canyon, offshore California, a cyclone on 20 December 2001 triggered a sediment gravity flow, which emplaced a pebbly sand unit (Paull et al. 2003). Rip-up clasts are ubiquitous in deep-water sandy debrites in both modern and ancient submarine canyons (Shepard and Dill 1966; Shanmugam 2006a). These rip-up clasts are unrelated to tsunami-related deposition. In other words, rip-up clasts are not restricted to tsunami-related deposits.
3.7 Normal grading (Fig. 3d)
Tsunami-related deposits showing normal grading have been reported from coastal (Gelfenbaum and Jaffe 2003) to deep-water environments (Cita and Aloisi 2000). In contrast, cyclone-triggered combined flows have been suggested for explaining the origin of normal grading in shallow-marine shelf environments (Gagan et al. 1990; Allison et al. 2005).
In elucidating the origin of normal grading by tsunamis, Gelfenbaum and Jaffe (2003) ascribed it to settling from suspension, using a turbidity current analogy (Sanders 1965). But the normally graded sand of Gelfenbaum and Jaffe (2003) does not represent a true ‘normal grading’ because the sand bed has floating rip-up mud clasts in the middle (see their Fig. 9). Floating mud clasts can be better explained by freezing of laminar sandy debris flows with plastic rheology (Fisher 1971; Hampton 1975; Enos 1977; Shanmugam 1996; Shanmugam et al. 2009).
The prevailing notion is that floating mudstone clasts in normally graded sands can be explained by suspension settling from turbidity currents. This is based on the false belief that all mudstone clasts are pure in composition and that they have a lower density than that of sand matrix composed of quartz. But the reality is much more complex. For example, the average grain density of quartz is 2.65 g/cm3 at 0°C (Folk 1968), but modern deep-sea clays have a wide range of density values (2.41–2.72 g/cm3), depending on composition (Opreanu 2003–2004). Further, mudstone clasts are invariably composed of: (1) detrital quartz grains, (2) igneous, sedimentary and metamorphic rock fragments, (3) fossil fragments, (4) wood or coaly fragments and (5) quartz veins. In other words, the settling velocity of mudstone clasts is affected not only by the clast size but also by clast composition, shape, bulk density and diagenesis (Shanmugam 2002a). Also, if the mudstone clast is made up of quartz mineral, then the density of the clast would be that of quartz (2.65 g/cm3) (see Parsons et al. 2001, p. 435). For these reasons, the argument that floating mudstone clasts can be explained by their lower density (in comparison with quartz sandy matrix) is flawed. The freezing en masse of debris flow with strength is the most viable mechanism for explaining floating mudstone clasts in ‘normally graded’ sands. The alternative origin of floating clasts by high-density turbidity currents (Postma et al. 1988) is problematic for reasons discussed elsewhere (see Shanmugam 1996, 2002b).
3.8 Inverse grading (Fig. 3e)
Inverse grading has been associated with deposits of both tsunamis (Jaffe et al. 2003) and tropical cyclones (Spiske and Jaffe 2009). The origin of inverse grading is controversial. Inverse grading of sandy matrix has been explained by dispersive pressure in grain flows (Bagnold 1954), traction carpets in high-density turbidity currents (Lowe 1982) and kinetic sieving in gravity flows (Middleton 1970). These controversies have presented an additional challenge to the issue of distinguishing paleo-tsunami deposits.
3.9 Multiple upward-fining units
Cantalamessa and Di Celma (2005) proposed that multiple upward-fining units in tsunami-related successions reflect decreasing transport energy with time during deposition from multiple successive waves. But multiple upward-fining units have also been interpreted to be products of cyclones (Hobday and Morton 1984, their Fig. 3, p. 208; Keller et al. 2007).
Morton et al. (2008) reported that the 2004 Indian Ocean Tsunami had generated multiple patterns in vertical textural trends that include (1) upward fining, (2) upward coarsening and (3) uniform, which they have interpreted to represent flow acceleration, initial unsteady pulsating flow, relatively stable and uniform flow, flow deceleration, slack water, return flow or flow redirection, etc. Because of these complex upward-textural trends, caution should be exercised in using a particular trend for distinguishing paleo-tsunami deposits.
3.10 Landward-fining trend
Grain size of tsunami-related sediment has been observed to show a landward-fining trend from the coast (Foster et al. 1991; Dawson 1994; Minoura et al. 1996). Nevertheless, Bahlburg and Weiss (2007) reported that some deposits of the 2004 Indian Ocean Tsunami do not show a landward-fining trend. Also, Goto et al. (2009) reported landward-fining trend of boulders by tropical cyclones.
3.11 Horizontal planar laminae (Fig. 3f)
3.12 Cross-stratification (Fig. 3g)
Low-angle cross-stratification has been reported from the deposits of the 2004 Indian Ocean Tsunami in Tamil Nadu, India (Srinivasalu et al. 2009a). But cross-laminae have also been reported from the deposits of the 2003 Hurricane Isabel (Morton et al. 2007, their Table 3). In fact, Morton et al. (2007, p. 204) consider cross-stratification and climbing ripples associated with bed load (traction) transport are indicative of cyclone deposits.
Morton et al. (2007) suggested that tsunami deposits are characterized by deposition from suspended load. To the contrary, Bahlburg and Weiss (2007, their Fig. 11d) documented evidence for bed load (traction) deposition by the presence of cross-bedding in deposits of the 2004 Indian Ocean Tsunami in Tamil Nadu, India.
Cyclic turnover of flow directions (i.e., basal landward flow and upper seaward flow) is considered to be an important aspect of tsunami-related deposits (Massari et al. 2009; Takashimizu and Masuda 2000; Fujiwara 2007). However, adjacent beds with opposing cross-stratification are common in tide-dominated estuarine facies (Shanmugam et al. 2000).
Tsunamis and tidal currents are two genetically unrelated entities.
- 2.The origin of HCS is itself controversial:
HCS has traditionally been considered as evidence for storm-wave deposition by oscillatory flows in shallow-marine (<200 m water depth) environments (Harms et al. 1975).
Duke et al. (1991) proposed the origin of HCS by combined flows (i.e., combination of unidirectional flows and oscillatory flows).
Mulder et al. (2009a, b) explained the origin of HCS-like structures as deep-sea antidune stratification formed by turbidity-current standing waves, relating to Kelvin–Helmholtz instability occurring between high-density basal layer and low-density upper layer in stratified flows (Prave and Duke 1990). Higgs (2010) disputes this deep-sea origin.
Morton et al. (2007) claim that climbing ripples are diagnostic of cyclone-induced deposits.
“Chevron” bed forms with subcritically climbing translatent stratification, which were originally considered to be products of tsunami (e.g., Bryant 2001), have been reinterpreted to be deposits of wind (Kindler and Strasser 2000; Bourgeois and Weiss 2009). In short, the genesis of kinds of cross-stratification further complicates the debate over distinguishing between tsunami and cyclone deposits.
3.13 Rich in marine fossils (Fig. 3h)
Tsunami-related massive sands are rich in marine fossils (Peters et al. 2007; Massari et al. 2009). On the other hand, abundant shell fragments are also considered characteristic of cyclone deposits (Morton et al. 2007; see also Saito 1989). Tuttle et al. (2004) reported that both tsunami- and cyclone-related deposits contain mixtures of diatoms. Clearly, the abundance of marine fossils cannot be a distinguishing criterion of tsunami deposits.
3.14 Changes in chemical elements (Fig. 3i)
Costa (2004) observed that tsunami-related massive sands show an increase in K, Mg and Na elemental concentrations. In contrast, Jayakumar and Siraz (1997) observed higher concentrations of Na, Mg, K and Cl elements in the coastal area of Tamil Nadu in India due to cyclone-induced penetration of seawater. Srinivasalu et al. (2009b) reported that in the deposits of the 2004 Indian Ocean Tsunami, organic carbon is lacking because of the transport of fine-grained sediment by backwash flows into deep water in Tamil Nadu, India. Because post-depositional digenetic alterations tend to change the original (depositional) sediment composition, the use of chemical elements to discriminate tsunami-related deposits is meaningless. Font et al. (2010) suggested that a very low magnetic susceptibility of sand as a criterion for distinguishing tsunami deposits based on one case study of 1755 Lisbon tsunami in Portugal. Because particle size and carbonate concentration affect the magnetic susceptibility (Feng and Johnson 1995), this method for distinguishing paleo-tsunami deposits is also of no practical value.
3.15 Sand injection and soft-sediment deformation (Fig. 3c)
Le Roux et al. (2004) interpreted sand injections in deep-marine environments as products of tsunamis in Chile. Some of these tsunami-induced sand injections exceed 15 m in lengths (Le Roux and Vargas 2005), suggesting very high dynamic pressures. Similar high-pressure conditions are viable during cyclone-related deposition. Furthermore, earthquake-induced clastic dikes, which are post-depositional features, have been reported (Levi et al. 2006). Distinguishing between syn-depositional and post-depositional sand injections in the ancient record has been a challenge.
Different types of soft-sediment deformation features, associated with tsunamis, have been reported from Japan (Takashimizu and Masuda 2000), France (Schnyder et al. 2005) and Egypt (Salem 2009). By contrast, Alfaro et al. (2002) reported soft-sediment deformation structures (load casts, ball-and-pillows and pipes) in cyclone-induced deposits, related to liquefaction and/or fluidization caused by cyclic effect of storm waves on unconsolidated sediments.
In summary, our understanding of processes associated with tsunamis is highly muddled. The assortment of features claim to be associated with tsunami deposition suggests extreme variability in processes that include erosion, bed load (traction), lower flow regime currents, upper-flow regime currents, oscillatory flows, combined flows, bidirectional currents, mass emplacement, freezing en masse, settling from suspension and sand injection. At the present level of knowledge, sedimentological features alone are unreliable for distinguishing paleo-tsunami deposits from paleo-cyclone deposits, without historical information. In the absence of historical data, the use of a non-specific term “catastrophic inundation event” is more appropriate than the term “tsunami event” when interpreting ancient deposits.
A multitude of tsunami-related deposits has been termed as “tsunamites,” irrespective of their true depositional origin (Shanmugam 2006b). At present, the term tsunamite represents turbidite, debrite, tempestite, fluxoturbidite, undaturbidite, seismite, seismoturbidite, gravitite, gravite, densite, tractionite, hyperpycnite, tidalite, unifite, homogenite and injectite. Although the term “tsunamite” is still being used (e.g., Pratt and Bordonaro 2007; Dawson and Stewart 2007; Shiki et al. 2008; Massari et al. 2009), it is sedimentologically meaningless (Shanmugam 2006b; Bourgeois 2009; Mulder et al. 2010). Because the term “tsunamite” represents neither a single depositional process nor a single depositional unit, it should be abandoned.
4.2 Genetic facies models
The ‘Bouma Sequence’ lacks the theoretical foundation (Hsü 1964, 1989; Sanders 1965; Van der Lingen 1969; Shanmugam 1997). No one could explain how a suspension turbidity current (Ta) transforms into an upper-flow regime traction current (Tb) in forming the vertical (Ta–Tb) sequence (Leclair and Arnott 2005, p. 4).
No one could generate the complete ‘Bouma Sequenec’ with five divisions in laboratory experiments. Breien et al. (2010) claim that they have reproduced four divisions (Ta, Tb, Td and Te) in experiments by laminar ‘fluidized flows.’ Regardless, the conventional wisdom has been that the ‘Bouma Sequence’ represents deposit of a turbulent turbidity current (Bouma 1962; Sanders 1965; Middleton and Hampton 1973).
In field application, describing a sand unit as Ta (Turbidite a division) is like describing a cross-bedded sand unit as a ‘braided fluvial deposit.’ Such a description with a built-in turbidite interpretation defeats the basic tenet of process sedimentology in maintaining a distinction between description and interpretation (Shanmugam 1997).
In conflict with the model, field studies of the Annot Sandstone in the type locality in SE France show that the sequence represents multiple processes, such as sandy debris flows, sandy slumps and sandy tidal currents (Shanmugam 2002b, 2003).
In reflecting problems associated with genetic facies models, even Reading (2001, p. 101) acknowledges that “Each environment and rock sequence is unique”. Therefore, genetic facies models, such as the “Bouma Sequence”, are inconsequential for interpreting paleo-tsunami deposits.
Despite these shortcomings, Fujiwara and Kamataki (2007) proposed a vertical facies model for tsunami deposits with four units in ascending order (Tna, Tnb, Tnc and Tnd), emulating the ‘Bouma Sequence’ (Ta, Tb, Tc, Td and Te). These authors claim that the four units represent a single depositional event of a tsunami. In this model, the Tnb unit is composed of both traction carpet and HCS (Fujiwara and Kamataki 2007). Lowe (1982, his Fig. 7) explained the origin of traction carpets by intergranular dispersive pressure and freezing of basal sand layer in high-density turbidity currents. On the other hand, Harms et al. (1975) explained the origin of HCS by storm-wave-generated oscillatory flows in shallow-marine environments. Dispersive pressure and oscillatory flow are two genetically unrelated processes. Thus, the origin of traction carpet and HCS in Tnb by a single depositional event by tsunami is hydrodynamically untenable.
Floquet and Hennuy (2003) report Ta, Tb and Tc divisions of the ‘Bouma Sequence’ in megaturbidites triggered by earthquakes and associated tsunamis. But the term ‘megaturbidite’ represents debrites, not turbidites (Labaume et al. 1987). As mentioned earlier, depositional record does not reveal anything about the triggering events (e.g., earthquakes, tsunamis, etc.).
Takashimizu and Masuda (2000) proposed a three-part facies model for earthquake-induced tsunami deposits. In this model, the lower convoluted part represents earthquake event. The middle part with traction structures and upper part with suspension deposits represent tsunami event. The problem here is that convoluted bedding can also be generated by syn-depositional deformation unrelated to earthquakes (see discussion under subheading ‘3.5 Chaotic bedding’). Importantly, traction structures are ubiquitous in deposits of deep-water bottom currents (e.g., contourites and tidalites) that are unrelated to tsunamis (Shanmugam 2008b).
The concept of genetic facies model is obsolete for tsunami. This is because a ‘tsunami’ event is not a single process that can generate a unique vertical sequence.
4.3 Depositional setting
Finally, the subject of ‘context’ (i.e., depositional setting) is of extreme relevance for interpreting paleo-tsunami deposits. Although most of the tsunami-related sedimentary signatures can be produced by one or more other mechanism(s), it can also be argued that by carefully considering the depositional setting for a given anomalous deposit, the evidence for a paleo-tsunami event can become compelling. An example would be coarse-grained sheet deposit (gravel) in sand dunes at an elevation above storm-surge limits. In such a case, the depositional context (sand dunes) combined with the elevation is compelling evidence for tsunami, whereas a gravel deposit forming a strandline within the reach of storm surge is (obviously) less compelling. Nevertheless, interpretation of these complex depositional settings in the ancient geologic record is still challenging.
Incoming tsunami waves (landward) and outgoing backwash flows (seaward) can trigger a variety of processes, which emplace sediments in lacustrine, coastal, shelf, slope and basinal environments. But there are no reliable sedimentological criteria for distinguishing paleo-tsunami deposits in various environments.
Tsunamis (oceanographic phenomena) and tropical cyclones (meteorolocal phenomena) are two genetically unrelated catastrophic wave events. Despite their differences in origin, both tsunamis and tropical cyclones can generate identical depositional processes and related sedimentary features. Therefore, the type of triggering event (tsunami versus cyclone) cannot be distinguished by examining the depositional record alone, without historical information.
The future success of distinguishing paleo-tsunami deposits depends on (1) the development of criteria based on systematic synthesis of many modern examples worldwide, (2) the precise application of basic principles of process sedimentology and (3) the careful consideration of depositional setting.
My interest on tsunamis-related deposition was triggered by the 2004 Indian Ocean tsunami, which hit the coast of Tamil Nadu in south-eastern India on 26 December. My hometown (Sirkali), which is located about 12 km inland from the tsunami-devastated coast, provided immediate shelter for tens of thousands of tsunami victims. I would like to thank N. Swedaranyam, T. Saraswathi (my sister), S. Thambidurai and S. Murugan for their assistance during my 2005 field study of coastal deposits of the 2004 Indian Ocean Tsunami in Tamil Nadu. I thank Guest Editor Arun Kumar for inviting me to contribute this article. I wish to thank Journal Editor T. Murty for his suggestions on content during early stages of manuscript preparation in 2009. My sincere thanks to two anonymous reviewers for their detailed, critical and helpful comments on the manuscript. Jean Shanmugam (my wife) is thanked for her general comments. I am grateful to Cliff Frohlich, The University of Texas at Austin, for providing photographs of tsunami-emplaced boulders in Tongatapu Island, south-west Pacific.