1 In 2003, a theory was proposed that a widespread coastal flood event, which affected the Bristol Channel and Severn Estuary around the coasts of south Wales and southwest England on 30^th January 1607, may have been due to the impact of a tsunami (Bryant and Haslett 2003). The theory was developed through a combination of, firstly, a rereading of contemporary historical accounts published as pamphlets or chapbooks that provide ambiguous accounts of the weather with some stating that the day was “most fayrely and brightly spred” (Bryant and Haslett 2003, p. 164) and, secondly, coastal geomorphological and stratigraphic evidence potentially linked to the event. Haslett and Bryant (2005) undertook further historical research of accounts from Devon and Cornwall, and Bryant and Haslett (2007) and Haslett and Bryant (2007a, b) reported field evidence that was utilised to evaluate the theory.

2 The impact of the 1607 flood was extensive, affecting 570 km of coastline and flooding 518 km² of coastal lowlands causing great economic loss and an estimated 2000 fatalities (Bryant and Haslett 2007; Fig. 1) making it Great Britain’s worst historical natural disaster. Although some authors considered the 1607 flood to have been caused by a storm surge (e.g., Horsburgh and Horritt 2006; see also Haslett 2007), Kerridge (2005), in a UK Government report, concluded that “the most credible source for a strong, potentially damaging tsunami reaching the UK coast is a large passive margin earthquake in the Sole Bank area (western Celtic Sea)” (p. 29) directly offshore of the Bristol Channel.

3 Disney (2005a, b) referred to a second-hand report of an earth tremor being felt on the morning of the flood in 1607; however, it has not been possible to locate the historical source to support this statement. Nevertheless, Haslett (2011) identified from historical documents two previously uncatalogued seismic events that affected the region in 1607; the first occurred two weeks after the flood on 14^th February and the second on 22^nd May, suggesting that the year was a seismically active period. It is also possible that an earthquake may have triggered a submarine slide on the steep continental slope offshore Ireland (Kenyon 1987) that may have generated a tsunami, such as happened across the North Atlantic with a 13 m-high tsunami that struck the Burin Peninsula, Newfoundland, in 1929 (Heezen and Ewing 1952; Piper et al. 1999; Moore et al. 2007).

4 Bryant and Haslett’s (2007) field evidence collected to test the hypothesis included analysis of erosional coastal landforms and the dislodgement of boulders. They also acknowledged that the dating of coastal erosion features and boulder movement is challenging but were able to suggest a chronological link with the AD 1607 event based on a suite of evidence. For example, boulder dislodgement and transport at one site appears to have occurred in the period after AD 1590 but before AD 1634–1672, which overlaps with a significant episode of coastal erosion documented throughout the Bristol Channel and Severn Estuary that has been dated to the early seventeenth century (Allen and Rae 1987; Allen and Fulford 1992). Historical sources offer corroboration of extreme wave-energy during the 1607 event, such as the inland transport of a fully laden 60-t ship and the demolition of numerous buildings. Taken together, Bryant and Haslett (2007) suggested that a single catastrophic event is likely to account for this suite of features, which temporally excludes the Great Storm of 1703 as an alternative explanation. Nevertheless, additional field studies are required to expand the number of securely dated sites to test the hypothesis further. Boulders were investigated by Bryant and Haslett (2007) through the measurement of boulder clast axes (i.e. a, b, c for the longest to shortest axes, respectively) at a number of sites from which minimum wave heights required to initiate boulder movement (dislodgement) were derived based on the hydrodynamic equation of Nott (2003), which had been developed from previously published equations (Nott 1997).

5 Since 2007, a growing number of studies have analysed boulder deposits to investigate whether storm or tsunami waves have contributed to the evolution of coastlines around the World (e.g., Hansom et al. 2008; Spiske et al. 2008; Goto et al. 2009, 2010; Barbano et al. 2010; Scheffers et al. 2009, 2010; Costa et al. 2011; Lorang 2011; Paris et al. 2011; Erdmann et al. 2015, 2017; Watanabe et al. 2016; Cox et al. 2018; Haslett and Wong 2019a, b; Abad et al. 2020). However, some boulder studies have noted shortcomings in Nott’s (1997, 2003) hydrodynamic equations (Bourgeois and MacInnes 2010; Switzer and Burston 2010; Gandhi et al. 2017; Kennedy et al. 2017; Piscitelli et al. 2017; Cox et al. 2020) and, indeed, we agree that Nott’s versions of the equations are outdated and, therefore, should no longer be used. Subsequent studies have, however, further developed the hydrodynamic equations to refine them and to correct errors in previous versions of the equations, such as the work of Nandasena et al. (2011), Kain et al. (2012) and more recently Haslett and Wong (2019a). Yet despite ongoing refinements and corrections, Cox et al. (2020) claimed that all hydrodynamic equations derived from Nott’s (1997, 2003) seminal work should be regarded as flawed.

6 Notwithstanding these ongoing developments, the present study aims to recalculate minimum wave heights derived from boulder clasts investigated by Bryant and Haslett (2007) using the revised formulae of Haslett and Wong (2019a) and to consider the significance for understanding the historical high-magnitude flood event of January 1607 in the Bristol Channel and Severn Estuary. The presentation of updated results, in light ofthe revised hydrodynamic equations, serves to better inform other researchers and the public. For example, public sources, such as newspaper articles (e.g., Barnes 2020), continue to refer to previous estimates of wave heights for the 1607 event; those earlier results are now refined and superseded by the present study.

7 The study area is shown with principal sites in Figure 2, and details of the sites are presented in Table 1, categorised using the three zones defined by Bryant and Haslett (2007): the outer Bristol Channel, the inner Bristol Channel, and the Severn Estuary. Bryant and Haslett (2007) reported that 136 boulders were measured as part of their study, 129 of which appear to have been wave transported and seven clasts comprising an untransported boulder lag deposit derived from cliff retreat at Dunraven Bay (boulders DbL1-7), the significance of which for inclusion in their study was for dating boulder dislodgement (Bryant and Haslett 2007 pp. 264–266). From this dataset, Bryant and Haslett (2007) presented data for nine boulders only, being those from sites where the boulder was estimated to have the greatest resistance to flow within an area of each zone, both from storm and tsunami waves, as indicated by the formulae they used. Given that formulae are likely to continue to be developed and refined, the full dataset is presented here enable future use of these data (Appendix A, Table A1).

8 For clarity, the equations used in this study are given here as equations 1 and 2 for boulders in submerged and joint-bound pre-transport settings (equations 6b and 8d of Haslett and Wong, 2019a) respectively:

Display large image of Equation 1

and

Display large image of Equation 2

where H is wave height (m); b and c are the measurements (m) of the intermediate and shortest axes of a boulder; C_l is the lift coefficient of 0.178; C_d is the drag coefficient of 1.2; ρ_s is the density of a boulder (kg m^-3); ρ_w is the density of water (1020 kg m^-3); θ is the shore slope angle, estimated here to be 10˚; μ_s is the static friction coefficient of 0.7; and δ is a wave-type parameter, which is δ = 4 for tsunami and δ =1 for storm waves (for a full explanation see Haslett and Wong 2019a). The minimum speed/velocity (u, m s^-1) at which a wave impacts a boulder is also calculated for a submerged pretransport setting derived from Equation 6a of Haslett and Wong (2019a).

9 In their analysis, Bryant and Haslett (2007) referred to observed wave height data to provide the modern context for comparison with calculated minimum wave heights derived from boulder measurements. In the eastern North Atlantic area, the maximum deep-water wave height is 35 m with a 50-yr return period (NERC 1991), but the continental shelf shallows eastwards so that wave heights become ≥10 m in the outer Bristol Channel. However, in the shallower and more protected inner Bristol Channel and Severn Estuary, 50-yr return period waves of 4.7 m and 3.5 m, respectively, occur (McLaren et al. 1993). These observed wave heights are utilised in the present study for comparative purposes. However, the caution of Stephenson and Naylor (2011) should be noted in that some “boulders used to reconstruct past energy regimes [along the South Wales coast] may underestimate wave energies, if the blocks have been broken since detachment” (p. 23).

10 Dimensions of the 136 boulders measured in the field are presented in Appendix A (Table A1) along with the measured imbrication direction for each boulder, as reported by Bryant and Haslett (2007). Imbrication is considered an indication of the direction from which the flow responsible for transport, and subsequent deposition, originated (see Fig. 3 for examples). Mean imbrication was analysed by Bryant and Haslett (2007) who concluded that at their nine sites “wave refraction is not efficient, a feature of tsunamis more than storm waves” (p. 263); the matter is not revisited further in the present study. From these data the range of recalculated minimum wave heights under storm (H_storm) and tsunami (H_t) scenarios are derived for each site and presented in Table 2 along with wave velocity (u). These wave height data are also plotted in Figure 4 against approximate distance (km) seaward from the tide limit at Gloucester, along with observed wave heights for the outer and inner Bristol Channel and Severn Estuary from McLaren et al. (1993). In the outer Bristol Channel five sites were investigated. Boulders measured at Hele Bay, Coombe Martin, and Croyde indicate a similarly low range of minimum wave heights required to initiate movement, whereas boulders at Lee Bay and Tears Point indicate a much higher range. However, in each case the threshold for boulder dislodgement in a submerged pretransport setting is less than observed storm wave heights for the zone of ≥ 10 m (McLaren et al. 1993). Therefore, although all the boulders at the outer Bristol Channel sites may have been dislodged by H_t ≥ 2.3 m, dislodgement at each site may also be explained by H_storm ≥ 9.3 m.

11 Bryant and Haslett (2007) also concluded that it is theoretically possible that boulders at the outer Bristol Channel sites may have been dislodged by storm waves but recognised that in the “inner Bristol Channel and in the Severn Estuary, the required H_storm [for boulder dislodgement] is significantly greater than observed storm wave heights” (p. 263). The present study corroborates this observation; in these two zones, results from all sites yield H_storm greater than observed storm wave heights reported by McLaren et al. (1993) and the dislodgement of all boulders may be explained by H_t ≥ 4.2 m or H_storm ≥ 16.9 m (Fig. 4). This result is derived from Brean Down, located on the boundary between the inner Bristol Channel and the Severn Estuary, and as such may reasonably be taken to represent both zones.

12 Given observed storm wave heights (50 yr return period) of 4.7 m and 3.5 m in the inner Bristol Channel and Severn Estuary, respectively (McLaren et al. 1993), which are significantly lower than Hstorm ≥ 16.9 m, the recalculations presented here agree with Bryant and Haslett’s (2007) conclusion that boulder dislodgement in these zones was “achieved not by storm waves but by tsunamis” (p. 263). Clearly, if this is the case for the inner Bristol Channel and the Severn Estuary, then boulders in the outer Bristol Channel could also have been dislodged by H_t ≥ 4.2 m but reworking by subsequent storm waves and/or meteorological tsunami remains a possibility for boulders at these outer Bristol Channel sites (see Haslett and Bryant 2009).

13 Accepting that boulder dislodgement in the Bristol Channel and Severn Estuary is more likely to have been due to a tsunami than storm waves, Bryant and Haslett (2007) were able to recognise an up-channel increase in H_t, due to wave height amplification conjectured to have been caused by the overall funnel-shape of the embayment, an observation that is fundamentally replicated in this study. However, details vary between the two studies in that Bryant and Haslett (2007) recognised an increase from H_t ≥ 4 m in the outer Bristol Channel, to H_t ≥ 5 m in the inner Bristol Channel, to H_t ≥ 6 m in the Severn Estuary. In the present study, H_t increases from ≥ 2 m in the outer Bristol Channel to ≥ 4 m in the inner Bristol Channel and the Severn Estuary.

14 Data from Sker Point, however, suggest that H_t ≥ 3 m was experienced close to this site near the boundary between the outer and inner Bristol Channel. From these data, interpolation of minimum wave heights between the key sites of Lee Bay, Sker Point, and Brean Down is plotted on Figure 4. Extrapolation beyond the end sites permits reconstruction of the possible up-channel increase in wave height, relocating the 5 m and 6 m theorised tsunami wave height of Bryant and Haslett (2007) further up-channel to approximately 55–60 km and 25–30 km seaward from Gloucester, respectively. These results, embracing those of Bryant and Haslett (2007) that were derived from an earlier outdated version of the equation (Nott 2003), represent the minimum threshold for boulder dislodgement and, therefore, do not discount the possibility that a higher tsunami wave affected these sites.

15 Shape analysis of these wave-transported boulders (excluding the seven boulders from the Dunraven Bay lag deposit) is presented here using the Zingg (1935) method, c/b and b/a results derived from the measurements of 129 wave-transported boulders in Table 2 are plotted on a Zingg diagram in Figure 5. It is clear from these data that disc-shaped boulders dominate the sample with 71 clasts (55.04% of sample), followed by 49 blades (37.98%), with minor components of seven rods (5.43%), and two spheres (1.55%). The dominance of discs and blades within the sample is not surprising given the mainly sedimentary geology of the region and the prevalence of well-bedded stratigraphic units that are predisposed to produce tabular clasts upon erosion/quarrying from bedrock. Furthermore, disc-shaped clasts are more likely to be evidently imbricated, meeting the field criteria for wave-transported boulders and, therefore, selected for measurement.

16 The results derived from this study overall are consistent with those of Bryant and Haslett (2007) and the historical contemporary accounts of the 1607 flood (e.g., Anon 1607). Although some locations in the outer Bristol Channel were badly affected, such as north Devon where H_t ≥ 2 m (Haslett and Bryant 2005), the most catastrophically affected areas were the coastal lowlands of the Somerset Levels in the inner Bristol Channel (H_t = 3–4 m), as far inland as Glastonbury (Fig. 2), and the Severn Estuary (H_t ≥ 4 m, possibly rising to 5–6 m up-channel), from around Cardiff and Weston-Super-Mare up-channel towards Gloucester (Fig. 2; see also Skellern et al. 2008). It is in these areas that most of the ca. 2000 fatalities occurred on the morning of 30th January, 1607, due to the direct impacts of the catastrophic flood and its aftermath.

17 The entire dataset of 136 boulder clasts employed by Bryant and Haslett (2007) as part of their investigation into the impact of the 1607 flood in the Bristol Channel and Severn Estuary are used here to recalculate minimum wave heights for both storm waves and tsunamis using the revised hydrodynamic equations of Haslett and Wong (2019a). The results provide overall corroboration of the Bryant and Haslett (2007) study in that the data suggest that boulder dislodgement in the region was achieved not by storm waves but by tsunamis. The present study also corroborates an upchannel increase in tsunami wave height most likely due to wave amplification in the funnel-shaped embayment. The new results suggest that a tsunami ≥ 4.2 m high accounts for the dislodgement of all measured boulders (excluding a lag deposit) in the sample. Extrapolation suggests that a tsunami wave height up to 5–6 m is possible within the Severn Estuary where the impact of the 1607 flood on the coastal lowlands was most catastrophic.