Introduction

There is a growing worldwide appreciation of the importance of protecting against the loss of underwater cultural heritage (UCH) material, or its in situ context, at maritime archaeological sites subject to ongoing or accelerated degradation and disturbance. There have been, and still are, impediments in providing such protection through responsible in situ management practices.

In situ management of maritime archaeological sites can reduce the detrimental impacts from recognized in situ, or site-formation processes (Gregory 1996, 2009; Manders et al. 2010; Gregory and Matthiesen 2012). The range of processes, their underlying theory, and examples of applicable management approaches for submerged shipwrecks have been well described by Keith (2016). However, to understand the processes acting on and below the seabed at specific sites, and hence effectively manage and plan responsible scientific research at those sites, noninvasive baseline and periodic monitoring data are required (Gregory and Manders 2015; Oxley 2016:215).

Marine seismic reflection techniques, including sub-bottom profiler (SBP) instruments, have been used to determine the physical properties and geological information of the seafloor and materials below the seabed. SBPs send sound pulses into shallow subsea sediments, and these acoustic waves bounce off the sea floor and sub-seabed sediment, rock, or material interfaces (layers). The strength of the returning waves depends upon the differences in the density of the materials forming the interfaces and the rate at which the sound waves travel through these materials. The times for the different reflected waves to return and be recorded by the SBP indicate the depth of the seabed and the depths of sub-seabed layers. SBPs based on nonlinear (parametric) acoustics have favorable characteristics compared with linear SBPs for the detection and location of small shallow-buried materials (Wunderlich et al. 2005).

The potential capability of SBP acoustic methods to investigate a range of submerged palaeolandscapes and maritime archaeological sites has advanced significantly during the past two decades. Examples of these applications include Quinn, Bull, and Dix (1997a), Quinn, Bull, Dix, and Adams (1997b), Arnott et al. (2002), Gregory (2015), Gregory and Manders (2015:35–37), Missiaen et al. (2018), and Astrup et al. (2020). Noninvasive SBP-derived data have the potential to inform those responsible for management of UCH sites by identifying and/or confirming the presence of sub-seabed UCH material; identifying the potential level of risk of further material loss from ongoing degradation; providing key information with which to make a sound in situ management or site-research plan; and monitoring the success of in situ management actions.

Despite these apparently significant advantages, Oxley and Keith (2016:8) noted that practitioners had not widely adopted in situ management approaches, even when acknowledging the importance of the underlying site formation theory. These authors argue that the reason for this may arise from “a lack of funding, limited time and lack of access to the necessary specialists.” The current practice may also come from confusion or lack of confidence in how and when to apply in situ preservation methods, including the use of SBPs, their value and effectiveness.

This article aims to address some of these constraints and encourage practitioners to better understand the application potential of parametric SBPs to document and manage sites with shallow-buried maritime archaeological materials. It draws on the author’s unpublished (Winton 2020b) research and integrates with components that have already been published (Winton 2019). Firstly, responsible in situ management practices are examined to characterize noninvasive data requirements for the identification of shallow-buried materials and to subsequently support best practice management decisions. The performance results of a parametric SBP under controlled field conditions over purpose-buried timber and metal “sleepers,” and across a shipwreck site that had been previously excavated, surveyed, and reburied, is then presented. The reliability associated with detecting the buried sleepers and the quantified accuracy of measuring their depths of burial is summarized. The ability and associated accuracy of measuring the lateral extent, depth of burial (DoB), and material characteristics over a complex wreck site is then discussed together with an interpretation of site degradation risk. Finally, these outcomes are drawn together to assess the usefulness and efficacy of SBPs for site investigation, in situ management, and performance monitoring purposes.

Responsible in Situ Management Practice

The ratification of the 2001 UNESCO Convention on the Protection of the Underwater Cultural Heritage, together with the earlier European Valetta Convention, delivered the galvanizing impetus to protect and manage UCH material in situ. The subsequent release of UNESCO’s Manual for Activities Directed at Underwater Cultural Heritage (Maarleveld et al. 2013) provides clarification of the rules incorporated within the 2001 UNESCO convention, as well as practitioner guidance on responsible in situ management practices.

Rule 1 of the convention states that the “protection of underwater cultural heritage through in situ preservation shall be considered as the first option before allowing or engaging in any activities directed at this heritage.” This has been described as the most debated and least understood rule of the convention (Maarleveld et al. 2013:21), which led to a dichotomy of practitioner interpretations of in situ preservation actions ranging from “no excavation allowed” to “do nothing” (Ortmann 2009:79; Ortmann et al. 2010:36–38). UNESCO clarified in its manual that in situ preservation be considered as the first (not the only) option, ahead of but not to the exclusion of other activities directed at the UCH. In situ preservation does not limit archaeological excavation and scientific study where the intention is to make a significant contribution to the protection, knowledge, and enhancement of the UCH located at that site. For sites where UCH material is deemed at risk, a conservation plan addressing in situ protection needs to be prepared and implemented (Rule 24). UNESCO also recognized that research and in situ management are dependent upon data, and that nondestructive techniques to collect those data come first and are preferred over intrusive methods (Rule 4)—intrusive methods remain important, however, and their efficacy will be significantly improved if informed by preliminary data gathered using noninvasive techniques.

As a consequence of the implementation of the UNESCO and Valetta conventions, extensive scientific research was necessary and undertaken to effectively implement the intentions of both conventions, for example, Godfrey et al. (2005), Manders (2004), Bergstrand and Nyström Godfrey (2007), Manders et al. (2010), Nyström Godfrey et al. (2012), Richards et al. (2013), Veth et al. (2013), and Gregory and Manders (2015). Specifically, data were needed to answer questions relating to the mechanisms and speed of shipwreck deterioration, the time period that shipwrecks could be protected in situ, and the validity of approaches for long-term management. To determine the degradation risk for buried organic and metallic materials, the relationship between sediment depth, sediment chemistry, and biological and chemical activity was needed. Based on this accruing scientific data and in accordance with the principles of the 2001 UNESCO and Valetta conventions, Gregory (2009) and Gregory and Matthiesen (2012) argued for a process-driven approach to in situ preservation—one that identifies the site threats and identifies the extent of the site to be preserved, uses in situ data to assess and quantify these threats, and uses baseline and ongoing data to identify whether the site is safe and if mitigation measures or modification to the in situ management plan are required.

Consistent with the above process-driven approach on sites potentially containing shallow-buried material, noninvasive data are required to provide key site information on the lateral extent, DoB, and degradation risk to that material, initially to inform and subsequently to assess in situ management strategies. The following sections describe results from research at a purposely prepared sub-seabed site, and on the adjoining James Matthews (1841) wreck site in Western Australia, and demonstrate how noninvasive SBP-derived data support responsible management of UCH sites.

SBP Site Surveys

Multiple modern European oak (Quercus robur), pine (Pinus radiata), and jarrah (Eucalyptus marginata) archaeological specimens (known as “sleepers”), each 12.5 cm wide by 12.5 cm deep and 50 cm long, were buried at shallow depths (10, 20, 30, and 50 cm) in medium-fine grained calcareous sediments adjacent to the James Matthews wreck site. Ferrous sleepers were also buried with 20, 30, and 50 cm sediment coverage. This site is located on the northern side of Woodman Point, approximately 7 km south of Fremantle, Western Australia (Fig. 1). The fabrication and burial of these sleepers is described by Winton (2019:4–6, 2020b:66–74) and their purpose was to enable validation, under controlled in situ conditions, of the detection reliability and DoB measurement accuracy using a parametric SBP.

Fig. 1
figure 1

Location map of James Matthews wreck site (Winton 2019:figure 1).

Full size image

The SBP measurements over the buried sleepers were undertaken multiple times during two half-day surveys in 2017 and 2018. An Innomar SES-2000 compact SBP transducer was initially pole-mounted on a survey vessel, and subsequently on an autonomous seabed sled, with survey speeds of 2.0 and 0.15 m/s, respectively. A Trimble POS MV Surfmaster GNSS G2 real time satellite positioning antenna and heave correction sensor recorded the position and motion of the transducer head while vessel-mounted. Otherwise, a Leica CS20 was used to log the position of a Leica GS16 High Precision GNSS RTK Rover mounted to the transducer pole above the seabed sled (Winton 2019:6–8, 2020b:97–102). Extraction and interpretation methods for the SBP data are detailed by Winton (2019:8–13, 2020b:102–196).

During the 2017 vessel-based survey, 89 SBP transects were also recorded within a two-hour survey period over the vicinity of the James Matthews wreck site. This site was excavated and reburied annually from 1974 to 1977 by maritime archaeologists from the Western Australian Maritime Museum (WAM), and during the extensive 1975/1976 excavation it was archaeologically surveyed in three dimensions (Henderson 1975, 1976, 1977b; Baker and Henderson 1979). This site is now surrounded by a plastic road crash-barrier cofferdam installed in 2013 (Fig. 2) and partially backfilled with clean sand to protect it against seabed erosion (Richards et al. 2014). The WAM annotated plan drawing of the excavated shipwreck material and surrounding cofferdam is shown in Figure 2a, with the stern to the northwest and the key slate cargo mound feature depicted in black. Of the 47 selected SBP survey transects that directly crossed the site, 10 runs tracked “loosely parallel” and longitudinally along the shipwreck remains (northwest–southeast orientation) at an average spacing of 1.4 m, and 37 tracked transversely (southwest–northeast) across the enclosed site (Fig. 2b) with an average spacing of 0.8 m.

Fig. 2
figure 2

James Matthews wreck site: (a) WAM plan drawing of 1976 excavated remains of James Matthews, including the 2013 surrounding cofferdam (courtesy of WAM); (b) 2016 aerial image of James Matthews site showing cofferdam and the selected 2017 SBP transects (Winton 2020b:figure 4.44).

Full size image

SBP Survey Results

SBP Detection Reliability and DoB Accuracy

The detailed examination of the results from the controlled SBP survey over the purpose-buried sleepers has been published by Winton (2019). Here, those results have been summarized in order to incorporate measurement reliability and accuracy into the overall discussion. The estimates of detection reliability and accuracy of SBP measurements were based on 104 independent and controlled SBP observations along the line of the timber and iron buried sleepers. Winton (2019:13–15) reported the 95% confidence level range associated with correctly identifying sleeper locations for all burial depths was 88%– 97%, around a mean value of 94%. These outcomes were based on the Watson and Petrie (2010:1169–1171) method agreement analysis for categorical variables. Excluding the sleepers with DoB <11 cm, these confidence levels rose to a range 94%–100% around a mean of 99%.

The Passing-Bablok linear regression relationship derived from the Bland and Altman (1999) method comparison analyses quantified the accuracy of the SBP-derived sleeper burial depth estimates against their known DoB (Winton 2019:15–21). This included sleepers with flat upper surfaces and upper surfaces inclined 22.5° to the horizontal. Within a 95% confidence range of 0.6–2.5 cm, the true burial depth (in centimeters) derived from velocity-corrected parametric SBP depth measurements (dDoBcorr) of materials buried from 10 to 50 cm below the seabed can be derived using the following equation:

$$\mathrm{True burial depth }(\mathrm{cm})=1.63+1.01\times ({\mathrm{d}}_{\mathrm{DoBcorr}})$$

In addition to the detection of the upper face of the sleepers, two or three separate interfaces associated with the bottom face of single sleepers and the gaps between multiple stacked sleepers were also identified. Vertical resolution was estimated to be at least 10 cm, estimated sleeper thicknesses ranged from 13.9 to 17.9 cm vs. actual 12.5 cm, and the gaps between stacked sleepers were estimated to range from 8.7 to 10.3 cm vs. 7.5 cm actual.

Determination of Lateral Site (Plan) Extent on a Complex Wreck Site

Surface and sub-seabed reflectors interpreted from the 47 SBP runs across the James Matthews wreck site identified the complexity of the aerial extent of the buried shipwreck remains, as well as isolated buried features. One transverse SBP transect (run 20170608 031531) and one longitudinal SBP transect (run 20170608 042734) surveyed across the James Matthews wreck site (locations shown in Figure 2b) are presented as Figures 3 and 4. These echo plots display low frequency (12 kHz) data in grayscale above and in 10-color scale, with interpretations, below—the raw data have been smoothed (each value averaged with its neighboring values) and water column noise reduction added to enhance visual interpretation. Figure 3 displays the wreck site cross section looking toward the bow, with the 0.9 m high plastic cofferdam shown on both sides. The depth of the seabed outside the cofferdam varies between 2.3 and 2.6 m below the sea surface. The remains of the slate mound, which protrudes up to 0.6 m above seabed, and adjacent seabed scour are evident. The maximum depth of sub-seabed reflectors is approximately 0.8 m on the bulwarks side of the slate mound—those located within the WAM survey plan correspond to timber ceiling planking with underlying frames. Isolated reflectors are also noted outside of the known plan area. The longitudinal cross section shown in Figure 4 displays a profile of contiguous subsurface shipwreck reflectors from the bow to the stern, with a maximum depth of approximately 0.9 m located immediately forward of the slate mound. Also identifiable are reflectors isolated from the main hull remains near the crash barrier located at the stern. These isolated reflectors may represent the damaged stern post or other displaced shipwreck material.

Fig. 3
figure 3

Transverse SBP transect 20170608 031531 with starboard bulwarks to the right and keel to the left: (a) grayscale, uninterpreted; (b) 10-color scale with interpretations (Winton 2020b:figure 4.47).

Full size image
Fig. 4
figure 4

Longitudinal SBP transect 20170608 042734 with bow to the right and stern to the left: (a) grayscale, uninterpreted; (b) 10-color scale with interpretations (Winton 2020b:figure 4.51).

Full size image

All 47 SBP echo plots crossing the James Matthews wreck site were analyzed, and the location and extent of sub-seabed reflectors in each were determined and mapped in GIS. Lines of near-contiguous sub-seabed reflectors along each echo plot are depicted in red in Figure 5, and isolated sub-seabed reflector lines depicted in yellow. Also shown is the surficial crash barrier cofferdam, marking the outer limits of the site protection works, and the plan outline (dashed white) of the intact buried shipwreck material from Figure 2a.

Fig. 5
figure 5

James Matthews site plan showing extent of contiguous and isolated reflectors across the wreck site (Winton 2020b:figure 5.19).

Full size image

As can be seen in Figure 5, the lines of contiguous buried reflectors identified from the multiple SBP runs are situated mostly within the limits of the cofferdam, with the exception on the northeastern side where they extend approximately 5 m farther out. On the southwestern side, reflectors extend beyond the plan extent of the shipwreck remains, but generally lie within the line of the crash barrier. Isolated reflectors were identified up to 20–30 m away from the wreck site to the southwest and to the east, but most are located within 5 and 10 m of the southwestern and northeastern sides of the known shipwreck extent. Without knowledge of the history of this site, interpretation of reflectors from the SBP runs would overestimate the dimensions of the wreck site by 5–10 m.

Following the extensive 1975/1976 excavation, survey, and backfill of this site, WAM archaeologists excavated all sides of the known hull remains in 1977 in an attempt to find and raise more material, and fully define the site (Henderson 1977a:97–113, 1977b:78; Baker and Henderson 1979:228–229). These excavations resulted in three trenches 5 m wide and 1–1.5 m deep along the stern, bow, and starboard sides of the remaining intact hull, and their locations are shown as orange boxes on Figure 6. Rope from the rigging was found along and underneath the starboard side of the wreck, but the bow and stern trenches proved to be almost sterile. During the 1975/1976 excavation, broken timbers in quantity were noted to lie along the port side of the vessel remains, and ballast stones were relocated from the hull to this area shown in shaded blue on Figure 6. Consequently, only a 1 m wide by 1 m deep clearance trench, located 5 m out from the keel side of the hull, was excavated in 1977 to determine the lateral extent of the broken timbers. Once completed, all clean trenches were then backfilled.

Fig. 6
figure 6

James Matthews site plan showing extent of contiguous and isolated reflectors and historical excavation limits across the wreck site (Winton 2020b:figure 5.20).

Full size image

With few exceptions, the red and yellow buried reflector lines shown in Figure 6 correspond to the excavated dimensions of the intact remains of the James Matthews shipwreck and surrounding clearance trenches. The seven isolated reflectors (A–G) that are located outside of the areas excavated in 1975/1976 and in 1977 may or may not be associated with material lost from James Matthews during the wrecking event, or subsequently due to wreck-site formation processes. Within the excavation areas, the SBP survey identified and mapped the buried hull remains and the broken spars and ballast stones on the port side. The SBP survey has also identified either density differences between the unexcavated sand and the looser backfilled sediment at the base of the clearance trenches or weak density layers within the backfilled trenches. Quinn, Bull, Dix, and Adams (1997b) found similar sediment density interfaces, resulting from backfilled scour holes, on the Mary Rose site using a Chirp SBP. Considering acoustic reflections from the contiguous hull, isolated timbers and ballast, and excavation trenches, this parametric SBP survey mapped the extent of the complex James Matthews site to 38 × 16 m within ±1 m, equating to an error margin of approximately 3%–6%.

Determination of DoB on a Complex Wreck Site

A quantitative assessment of the above-seabed exposure levels, and depth of burial estimates of material from the James Matthews wreck site, was achieved by direct comparison of interpreted SBP echo plots to the archaeological survey conducted during the 1975/1976 excavation. Based on the archival survey records held by WAM, the author constructed a three-dimensional (3-D) AutoCAD digital model of the buried hull and associated remaining cargo from around amidships through to the stern (Winton 2020a) (Fig. 7). It is important to note that for preservation and display purposes a number of artifacts, and an extensive quantity of the slate, were removed from the site following the underwater survey and prior to the site’s reburial in 1976 (Henderson 1977b:77,79). The AutoCAD model replicates what is currently reburied on site and also reflects the inadvertent physical damage to the stern section during commercial sand dredging operations (Pat Baker, 2016, pers. comm.). Oblique virtual cross sections were cut through the AutoCAD model aligning with two transverse 2017 SBP transects (runs 0607 051953 and 0608 031531). These runs were selected for direct comparison as they cross over or cross forward of the slate mound, and over iron cargo bars, the keel, a veneer of slate, and multiple layers of hull timbers (Winton 2020b:191–196). The cross sections for these runs are shown in Figures 8b and 9b.

Fig. 7
figure 7

Plan view of stern section of James Matthews AutoCAD model showing the 2017 SBP run locations (Winton 2020b:figure 4.55).

Full size image
Fig. 8
figure 8

DoB assessment: (a) comparison between SBP reflectors identified along run 20170608 031531 and outline of corresponding 3-D model cross section; (b) details within outline of 3-D model cross section (Winton 2020b:figure 5.21).

Full size image
Fig. 9
figure 9

DoB assessment: (a) comparison between SBP reflectors identified along run 20170607 051953 and outline of corresponding 3-D model cross section; (b) details within outline of 3-D model cross section (Winton 2020b:figure 5.22).

Full size image

The horizontal and depth positions of reflectors representing the seabed surface, exposed material, and buried interfaces along SBP runs 0607 051953 and 0608 031531 were extracted from every 5th or 10th trace in the SBP echo plots. The complexity of the site, with varied buried materials at different depths, the effects of excavation and backfilling, subsequent erosional protective works placed on the site, and seasonal emergent algal growth resulted in significant variability in the strength and depth of reflectors along both echo plots. Reflectors were interpreted as representing buried interfaces only when their amplitude magnitudes were relatively large and consistent across a number of adjacent traces (Winton 2020b:242–246). All extracted depths were corrected using respective water column and sediment velocities. The position of seabed surface and sub-seabed reflectors from SBP runs 0608 031531 and 0607 051953 were co-plotted at the same horizontal and vertical scale with the corresponding virtual cross section cut through the 3-D digital model of the buried remains of James Matthews (Figs. 8a, 9a). The gray dots represent the seabed surface and the top of the exposed slate mound and iron rods, and the blue dots represent reflector interfaces at varying depths below the seabed. The vertical datum of the 3-D digital model, and hence the model cross section, was based on the arbitrary height of the underwater survey frame. The height datum of the model cross section was consequently adjusted until a match occurred with the upper profile of the stable slate mound. The relative accuracy of this datum alignment was estimated to be ±5–10 cm. For visual interpretation purposes, the virtual cross-sectional outlines based on the AutoCAD model are shown in both components of Figures 8 and 9.

SBP reflectors in Figure 8 identified the outline of the exposed slate mound, the buried keel, and frames with ceiling and outer planking within a horizontal accuracy of 0–15 cm, and within the ±5–10 cm relative vertical accuracy. Note that the keelson, sitting on top of the keel but under the iron rods, was not depicted in the model cross section. However, SBP reflectors were identified above the keel, and these may well represent the acoustic wave reflecting from the top of the keelson. Importantly the DoB, from 0 to 30 cm over the frames/inner planking, was closely aligned to the upper side of the model cross section. On the starboard side of the exposed slate mound, the SBP data indicated a higher surface level that may represent less localized scour on this side of the mound or a sandbag more recently placed as part of WAM’s site protection activities. Beyond the starboard bulwarks, a number of reflectors were identified between 30 and 55 cm below the seabed level. These may reveal the legacy of the 1977 excavation and backfilling activities along and under the starboard bulwarks.

The cross section in Figure 9 lies approximately 1 m forward of the slate mound, and was chosen to comparatively assess any interpretation influence from the slate mound. With the exception of the midsection of the ribs/planking, the SBP reflectors identified the upper and lower sides of the model cross section within the estimated relative vertical accuracy. Extensive seasonal algal growth rooted in the sediments above the midsection of this cross section resulted in numerous high amplitude water column reflections, which quickly decreased below the seabed. This effect may have resulted in the loss of acoustic wave energy at deeper depths, as here reflections were too weak to identify the lower face of the ribs and outer planking. The DoB ranged from 14 to 60 cm above the slate veneer and the ceiling planking/frames. A shallower section, 0.75 m in length, was indicated on the upper side of the planking in the same area affected by the red algae. This difference from the surveyed cross section may result from the interference associated with the red algae, or from items moved and left in situ after the survey was complete. In a similar fashion to the previous cross section, a number of reflectors approximately 80 cm deep were identified outside the starboard bulwarks.

Material Characteristics

Density differences between fully saturated/degraded oak timbers forming parts of the James Matthews buried hull remains and adjacent unsaturated recently buried oak timbers were characterized from SBP data using derived reflection coefficient values. The potential to remotely identify material characteristics of the buried remains of James Matthews and the adjacent recently buried timbers was evaluated using a slightly modified reflection coefficient method based on Plets et al. (2008). The reflection coefficient (KR) is the numerical expression for the amount of acoustic wave energy reflected vs. transmitted from a boundary (seabed surface, the interface between two sedimentary layers, or a buried object). As previously stated, KR values are a function of the material properties (density and compressional P-wave velocity) in each of the two adjoining layers. Reflection coefficient values have mostly been calculated using measured intrinsic properties of both boundary media. Plets et al. (2008:appendix A) demonstrated, however, that KR values for a deeper buried reflector can also be calculated based on the combination of SBP acoustic properties of that reflector, and those of the seabed surface and its multiple reflection (which form a calibration constant, x), for that same trace. While this latter method allows KR values to be calculated for each individual acoustic trace, significant variability in the acoustic reflection amplitudes between adjacent traces from the seabed surface strongly influenced intratrace variability of the derived reflection coefficient (Winton 2020b:224).

In order to reduce the variability associated with individual x values for each reflector, Plets et al. (2008) calculated x values for a large number of trace locations along the SBP run, selected the 50th percentile value from the distribution of those x values, and applied that value as the constant for the calculation of KR on each trace. This author used a slightly modified approach by adopting the modal value from the distribution of x values due to the distribution’s highly negatively skewed nature. This provided greater consistency when compared to KR values independently derived from the respective in situ impedance properties of the buried timbers (Winton 2020b:220–238). Reflection coefficient magnitudes calculated in this manner for the R1–R4, K1, K2, and S1 reflectors associated with the James Matthews site shown on Figures 8 and 9, together with the isolated reflectors A–J shown on Figure 6, are listed in Table 1.

Table 1 Reflection coefficient magnitude values for James Matthews site and adjacent buried timbers
Full size table

Examination of the KR values alone in Table 1 does not provide a great insight into the possible characteristics associated with the materials forming the various wreck-site reflectors. For the same type of buried material, however, differences in derived reflection coefficient values can be associated with differences in timber density, as shown in Figure 10. During a 2000 WAM conservation predisturbance survey of the James Matthews wreck site, Richards (2001:13) identified that the keel and keelson were constructed using beech and white (European) oak timbers, and the ceiling planks, ribs, and outer planking were constructed of white oak timber and were all waterlogged, with some significantly deteriorated. The mean values of the specific gravity (a unitless measure for density) of these beech and oak timbers determined by Richards varied from 0.58 (beech wood in keel) to 0.47–0.45 (oak timber in ribs and inner planking). Winton (2020b:134–149) determined the density of the partially saturated European oak timber sleepers that had been recovered 20.5 months following burial adjacent to the James Matthews wreck site. The reflection coefficient values for these buried oak sleepers determined using the same approach as above, together with their density expressed as specific gravity, are plotted in Figure 10. Likewise, Richards’s specific gravity values for the saturated oak timbers and corresponding KR values for reflectors R1, R2 and R3 are also plotted in Figure 10. Following 160 years of submergence the waterlogged and (partially) degraded oak shipwreck timbers have a significantly lower specific gravity, reflecting the loss of cellulose in the timber (Grattan 1988:65–66), and corresponding lower reflection coefficient values compared to the partially saturated oak sleepers.

Fig. 10
figure 10

Specific gravity and reflection coefficients for waterlogged European oak shipwreck and partially saturated sleeper timbers. (Figure modified from Winton [2020b:figure 5.24a].)

Full size image

Interpretation of Site Degradation Risk

The combination of SBP-derived estimates of DoB (in reality a surrogate measure for redox potential and degradation conditions in shallow sediments) and reflection coefficients can be used to identify the degradational risk of shallow-buried materials. These combinations may also provide insight into the interpretation of reflectors from other materials or sub-seabed boundaries. Reflection coefficient values for the timber reflectors in Table 1 are plotted against their respective DoB and are shown as “Group 1: timbers” in Figure 11. Here, KR values decrease with decreasing DoB, or increasing sediment aerobic conditions. In timbers buried for an extensive time, white and brown rot fungi would more aggressively degrade the timber (reducing its density) in the shallower aerobic and suboxic conditions. This contrasts with the much slower rates of degradation associated with the activity of anaerobic tunneling and erosion bacteria in deeper suboxic-anaerobic conditions (Gregory 2007:4; Gregory et al. 2008:206).

Fig. 11
figure 11

James Matthews site interpretation (Winton 2020b:figure 5.25).

Full size image

To assist interpretation of buried material on the James Matthews site, KR values and DoB estimates for the isolated reflectors derived from the SBP survey were also plotted on Figure 11, and sediment dissolved oxygen (DO) and redox potential (Eh) profiles established. The latter were derived from DO and Eh microelectrode analyses of 60 cm long sediment cores collected adjacent to the shipwreck material (Winton 2020b:124–128) and showed oxidizing but variable conditions in the surficial sediments. Two isolated reflectors (A, C) had properties similar to those of Group 1—one of these isolated reflectors was located just beyond the 1977 excavation off the bow of the shipwreck remains at a burial depth of 32 cm, the other located almost 30 m away with a burial depth of 44 cm. This interpretation suggests that these two isolated reflectors may be pieces of timber separated from the hull of James Matthews during or after the wrecking event. It is possible that isolated reflectors E and D also belong to this group. The second grouping of isolated reflectors had low KR values (0.04–0.11) with burial depths generally in excess of 55 cm (Group 2). At these depths environmental conditions within the sediments would be conducive to timber conservation, so their low KR values are unlikely to be associated with degraded timber. One of these isolated reflectors (S1) was identified on a cross section from the AutoCAD model (Fig. 9a) located outside the hull bulwarks, possibly at the bottom of the 1977 excavation trench. This suggests that the second group of low-reflection coefficient with higher DoB reflectors may be associated with weak sediment interfaces resulting from excavation and backfill of trenches. The third grouping (Group 3) of isolated reflectors are characterized by higher KR values (0.2–0.5) with burial depths in the range 18–35 cm. Reflection coefficients in the upper range of Group 3 exceed all values derived from recently buried nondegraded oak, jarrah, and pine timber sleepers on the adjacent sleeper site (Winton 2020b:227) and hence isolated reflectors G and J are unlikely to be associated with timber. Reflector G spans the location where relocated ballast stones were placed during WAM’s 1975/1976 excavation/survey. With the current SBP data it is difficult to interpret the material characteristics of Group 3 with any confidence, other than noting the higher KR values are more likely to represent acoustic returns from highly reflective buried materials, such as stone or metal.

Discussion

Consistent with current best practice aligned with binding international conventions, noninvasive data should be collected on sites potentially containing shallow-buried material to initially confirm the likely presence, lateral extent, DoB, and degradation risk of that material. This baseline information informs in situ management and research strategies for that site. As appropriate, subsequent monitoring data are used to assess site stability and success of any implemented actions. A limited number of research studies have explored the potential applicability of SBPs to provide noninvasive sub-seabed data, but quantitative assessment of SBP performance and degradation risk assessment of materials at maritime archaeological sites is currently lacking.

This research demonstrates that parametric SBPs can be used to efficiently and effectively provide baseline and monitoring data at sites with shallow-buried maritime archaeological material. In this study survey speeds of 2.0 and 0.5 m/s were achieved from vessel-based and autonomous seabed sled-based SBPs with a vertical pole-mounted combined transducer/receiver. In combination with precise navigation and heave sensors, the position and motion of the SBP transducer, as well as the concurrent reflected acoustic wave returns, were continuously recorded along 87 transects over a shallow open-water coastal wreck site in an aggregate of two hours across two consecutive days. Using a remotely controlled sled, 33 repeated transects were recorded on the adjacent site in an aggregate of 1.5 hours on two days almost one year apart.

Controlled surveys over the purpose-buried sleepers demonstrated high levels of detection reliability—the 95% confidence level range associated with correctly identifying sleeper locations for all burial depths (5.5–46 cm) was 88%–97%, around a mean value of 94%. Excluding the four sleepers with DoB <11 cm, these confidence levels rose to a range of 94%–100% around a mean of 99%. The accuracy associated with measurement of burial depths (derived from velocity-corrected parametric SBP depth measurements) below the seabed surface was quantified to within a 95% confidence range of 0.6–2.5 cm. Multiple reflectors identified sleeper thickness to range from 13.9 to 17.9 cm (12.5 cm actual) and gaps between stacked sleepers to range from 8.7 to 10.3 cm (7.5 cm actual). These results were determined from 104 independent and controlled in situ measurements over the purposely buried single and multiple-stacked European oak, jarrah, pine, and iron sleepers. Three iron sleepers were buried in a horizontal position. The 23 green timber sleepers were fabricated and buried such that their upper faces were either horizontal or inclined at 22.5o to the seabed surface, and with the timber’s longitudinal grain oriented horizontally or vertically. These validated outcomes demonstrated the potential suitability of parametric SBP to detect and measure DoB for different types of shallow-buried material, without need for a priori information on timber grain orientation.

These results confirm the parametric SBP’s efficacy to satisfy two of the above data requirements on a shallow clear site with medium-fine grained calcareous sediments—namely the ability to (1) identify and/or confirm the presence of acoustically reflective sub-seabed UCH material and its associated DoB and (2) provide monitoring data. To be of value to practitioners this performance also needs to be demonstrated on complex wreck sites and/or submerged paleo-landscapes, where degraded timbers or occupational sites may be difficult to identify, and the risk of further degradation of UCH materials challenging to discern.

Using the 87 SBP site transects, verified measurement accuracies of ±1 m were achieved for the subsurface plan extent of the complex James Matthews (1841) wreck site. While the archaeologically surveyed dimensions of the intact and reburied hull remnants are 6 × 27 m, the full sub-seabed extent of the site, as determined from SBP recorded subsurface reflectors, was 16 m in width and 38 m in length. This extended site comprises the remnant intact starboard hull and keel, unincorporated broken timbers, displaced ballast stones, and surrounding backfilled clearance trenches—the last being legacies from archaeological excavation and backfill surveys conducted in the mid- to late 1970s. The interpretation of the SBP survey extent matched the known dimensions of the full complex site within an error margin of 3%–6% and also identified isolated reflectors further distant from the hull that may be associated with postwrecking event material. However, without the a priori knowledge of the historical disturbances at this site, and the ability to differentiate between SBP reflections from degrading timbers, ballast stone, and sediment interfaces, simple interpretation of the reflectors from all SBP runs would overestimate the dimensions of the intact wreck remains by up to 10 m.

The depths of key features on the James Matthews site were also verified to within ±5–10 cm by comparison of the velocity-corrected SBP reflector depths with known cross-sectional features—the latter based on a 3-D AutoCAD model derived from WAM’s archaeological survey measurements. These features included the keel, keelson(?), starboard ceiling planks, frames and external planking, and the cargo slate mound and iron bars. The DoB of the upper face of the starboard hull remains ranged from 0 to 55 cm on one cross section, and from 14 to 60 cm on the other. These very shallow burial depths indicate a high potential of long-term timber degradation, and this interpretation has been independently confirmed by WAM’s previous predisturbance survey results. Patches of emergent algae in the water column locally reduced the SBP’s acoustic wave seabed penetration efficiency. While this caused a loss in definition of some buried features in affected areas, these were isolated, and broad characteristics could still be inferred from adjacent measurements.

Comparison of reflection coefficient values, derived for selected reflectors, against their respective DoB estimates provides the basis for a preliminary risk interpretation for the James Matthews site. This construct shows promise with regard to assessing preservation potential, site risk, and mitigation strategies, and its interpretation is enhanced with additional DO and redox sediment porewater profiles to map the aerobic, suboxic, and anaerobic zones within the surficial sediments. For the James Matthews site, decreasing KR values with decreasing DoB for the timber reflectors indicates a higher risk of degradation and material loss due to the more aggressive biological degradation processes found in the shallower aerobic and suboxic sediments. This supposition is consistent with sediment chemistry profiling and the results from timber analyses conducted by WAM. For isolated reflectors around the site, low KR values associated with higher (>0.5 m) DoB estimates are more likely to indicate weak reflector surfaces, such as sediment interfaces, as at these depths anaerobic sediment conditions would tend to favor timber preservation. As the magnitudes of reflection coefficients derived from adjacent nonwaterlogged and nondegraded pine, jarrah, and European oak sleepers were all <0.4, any KR values from the James Matthews site exceeding this value are unlikely to be associated with timber, but rather highly reflective buried materials such as stone or wood.

This overall assessment of lateral extent, depth of sediment cover, and potential relationships between acoustically derived reflection coefficients vs. DoB provides a fuller interpretation of the ongoing risks to the sub-seabed material located on the James Matthews site. The interpretation is consistent with, and supports, the in situ management plan and actions undertaken by WAM to date. It potentially provides a means with which to separate the locations of degrading timber from ballast stone and from weaker sediment interfaces. The risk interpretation model identifies those locations and depths where buried materials, preliminarily interpreted as timber, would be at high risk of further degradation. It also identifies isolated sites away from the intact remains for further scientific study.

The value of future SBP monitoring surveys on this site would be to quantify the stability of the seabed surface—hence the constancy of DoB measurements for those timbers at risk of further degradation—and in the longer term, any changes to their KR values, which may indicate further material (density) loss. In addition, a follow-up SBP survey with more closely and consistently spaced transects would permit a finer resolution model of the site, and more automated processes to calculate KR values would enable a much larger range of reflector targets to be included in the preliminary risk assessment model. Similar SBP surveys and analyses at other sites with different sediment characteristics, water column vegetation, sub-seabed material types, and lower material densities would significantly add to this body of knowledge and broaden its practitioner applicability.

Conclusion

The application of parametric sub-bottom profilers for responsible in situ management purposes of shallow-buried UCH materials has been demonstrated by this study. Performance characteristics, such as high noninvasive survey speeds and rapid data-capture timeframes, detection reliability, accuracy and resolution of sub-seabed burial depth estimates, together with the potential to assess material degradation risk, all align with the principles associated with binding international conventions and current best practice. While these results are based on experimentally controlled and verified measurements at just two adjoining sites, they should encourage practitioners to use parametric sub-bottom profilers to undertake initial baseline surveys of sites for which they are responsible. The data derived from these surveys form input into a site risk assessment and the scope of an in situ management plan. At the same time, these future surveys will broaden the knowledge database and improve interpretation confidence.