Sediment Profile Imagery - History and Application

History and Application

Conventional diving is limited to shallow waters. Remotely sampling deeper sediments of high water content is often unreliable due to sampler bow waves, compaction upon impact, or variably disrupted surface sediment features (Somerfield and Clarke 1997). In 1971, Rhoads and Cande described an instrument to address the problems of adequately observing and collecting silty sediments. Their remote sampling equipment introduced the field of in situ vertical sediment profile imagery and what are now commonly called SPI cameras. The device mainly consists of a wedge-shaped box mounted in a frame. The box has an oblique face made of transparent acrylic and a downward-looking camera (Figure 1). Weights force the wedge and its internal mirror into the sediments. The mirror, at 45° to the transparent section, reflects an image of the pierced sediment-water interface to the underwater camera, like a periscope. In order to remain rigid at depth, the wedge is filled with distilled water.

Figure 1. Schematic drawing of the profile camera in partial cross section showing the cradle in the down position intersecting the bottom. A- slack winch-wire; B- oil-filled cylinder; C- piston rod; D- piston containing a small diameter hole; E- battery housing with magnetic reed switch, F- lead weights, G- camera (oriented vertically); H- light; I- Plexiglas guillotine filled with distilled water; J- sediment-water interface; K- 45° angle mirror reflecting the sediment-water interface profile 90° to the camera lens. Taken from Rhoads and Cande (1971).

Their device returned images such as that shown in Figure 2. At first glance SP images may appear unremarkable, but analysis of dozens of images allows the breadth of information they contain to become apparent. In Figure 2 the gross texture and water content of the sediment is immediately apparent. Since resolution allows imaging of individual sand grains, the classic textural parameters (percentage of gravel, sand, and mud) can be assessed and a mean grain size estimated. The sediment-water interface is clear. If the image was taken immediately upon insertion, this observation indicates that the device entered the seabed with little disturbance. Furthermore, the interface is distinct. While seemingly straightforward, some seabeds have, instead, a boundary layer of suspended sediments with a broad density gradient instead of a discrete transition point. This condition has a fundamental importance to many benthic organisms. Biological activity is readily apparent as well. When calibrated using traditional grab samples or cores coupled with a few SP images, resolution allows identification of some infauna including the tubicolous sabellid polychaetes, a bisected nereid, and the mound produced by a sea cucumber seen in Figure 2.

Figure 2. Sediment profile photograph of a mud bottom 35 m deep in Cape Cod Bay, Massachusetts. The place of the photograph passes through a fecal mound produced by Molpadia oolitica (holothurian). The apex of the cone is populated by the sabellid polychaete Euchone incolor (A). An errant polychaete has been cut by the guillotine (B). Void spaces at depth are produced by the feeding activities of M. oolitica (C). Light-coloured oxidized (sulfide-poor) sediment extends about 3 cm below the sediment surface. Taken from Rhoads and Cande (1971).

Another significant feature of Figure 2 is the distinct colour change between surface sediments and those deeper. This gradient of colour change, though continuous, is known as the apparent redox potential discontinuity depth (ARPD) when reduced to an average transition point. When properly considered in conjunction with local geology and bioturbation levels, the depth and character of the ARPD can provide profound insights into the interactions between sediment geochemistry and biologic activity. Graf’s review (1992) supports the early observations of Jorgensen & Fenchel (1970) that sediments can be divided into oxic, suboxic, and anoxic levels with fundamental consequences for biota. They defined these boundaries as occurring at the >300 mV (oxidation reduction potential) level for oxic and less than 100 mV for anoxic chemoclines (with suboxic in between) as presented in Figure 3. The vertical position of these boundaries can vary seasonally and locally in response to detrital supply and mixing (due to bioturbation or physically mediated mixing) as fast as 1 cm d-1. Anoxic sediments tend to be toxic to most animals because of free H₂S and low pH. In this reducing environment, heavy metals can also precipitate. Some heavy metals, like cadmium and copper, are stabilised as sulphides and do not readily dissolve, but can be remobilised quickly and pollute boundary layer water if oxic conditions are restored (Graf 1992). The sediment penetration of chemical species from overlying waters to these layers will depend heavily upon the size and shape of sediment grains. Using a fluid bromide tracer, Dicke (in Graf 1992) found molecular diffusion alone to penetrate soft sediments to 4 cm in one day and 8 cm after 4 days. Bioturbation can accelerate this process by up to a factor of ten. Thus, the chemoclines affect and are, in turn, affected by benthic organisms. Besides exclusion and bioturbation effects of aerobic organisms, Fenchel and Riedl (1970) pioneered investigations into an unusual fauna inhabiting the suboxic regions of sediment. Clearly, SPI tools have much to offer in investigations of this sort.

Figure 3. The redox potential discontinuity (RPD) - layer concept of Fenchel & Reidel (1970). The sediment is divided into anoxic, suboxic, and oxic layers. Along the walls of tubes and burrows of animals the redox isolines are depressed (cf. Jorgensen & Revsbech, 1985). According to micro-electrode measurements of oxygen, the so-called oxic layer does not really contain free oxygen over the entire depth. Figure taken from Graf (1992).

Rhoads and Germano (1982) developed a list of parameters taken from SPI in an effort to reduce and quantify specific environmental attributes and make them amenable to traditional statistical analysis. Their list has been modified and qualified throughout the literature, but is summarised in Table 1. A few of these parameters can be calibrated and are reproducible in a variety of habitats. Gross sediment texture is probably the least controvertible and most immediately informative parameter for producing benthic habitat maps and identifying sediment-modifying impacts. The apparent redox potential discontinuity (ARPD) can also be a powerful assessment parameter. For example, one of the reported effects of sustained aquaculture activity on coastal environments is the deposition and accumulation of organic-rich sediments near the production site whether from the faeces and pseudofaeces of shellfish or uneaten food and excretion of fin fish. This can result in an increase in oxygen consumption by the sediment, formation of anoxic sediments, and the production and release of harmful gases such as methane, H₂S, and CO₂ which can affect the water column, benthic macrofauna (Pocklington et al. 1994), and meiofauna (Mazzola et al. 1999). The relationships between infauna, suboxic sediments, and organic enrichment are well documented (Weston 1990; Rees et al. 1992; Hargrave et al. 1997). This system is much like that described by Pearson and Rosenberg (1978) as presented in Figure 4. Rhoads and Germano (1982) took this concept one step further by assigning categories to the various successional stages in an attempt to integrate the biotic and geochemical responses to organic enrichment. To be used reliably, successional stage determinations must be made within the biological and physical context of each study, are necessarily subjective, and are unlikely to be more than broadly informative between analysts. Similarly, the majority of parameters presented in Table 1 are site- and study-specific. Acting in a similar manner to a cone penetrometer, the SPI wedge penetration depth into soft sediments may be generally useful as a proxy for sediment fabric if calibrated, but results will be sensitive to differences in equipment and deployment.

Table 1

Figure 4. Diagram of changes in fauna and sediment structure along a gradient of organic enrichment (Pearson and Rosenberg 1978).

Even with these limitations SPI can be an extremely powerful analytical, reconnaissance, and monitoring tool. Sediment-type maps have often been constructed by retrieving grab or core samples followed by days or weeks of laboratory-based processing. After an SPI device is lowered into the sediment and the image recorded, it can be hauled up and lowered repetitively without fully recovering the device. Such a vessel ‘stitching’ an SPI device along a prescribed route can survey an area with unprecedented economy compared to physical sample recovery. There is, of course, a trade-off between sampling data quality and quantity. SPI allows much greater spatial coverage for a given amount of field time at the cost of the detailed sediment descriptors typically produced from physical cores (half phi interval texture analysis, carbon content, etc.). Managing this balance is the essence of good SPI use and highlights its strengths. For example, Hewitt et al. (2002), Thrush et al. (1999), and Zajac (1999) call attention to the value of integrating macrofaunal community observations collected at different scales and their application in describing processes occurring at different scales within a heterogeneous benthic landscape. When evaluating landscape-scale questions it is rarely feasible to simply and comprehensively sample the total spatial extent with dense, equivalently detailed sampling points. The researcher must compromise between data collection grain, dimensions of the actual sampling unit (typically 0.1 m2 grab or similar), and lag- distance between sample units over which results will be interpolated (often tens to hundreds of metres for grab samples). Sediment profile imagery can be an efficient monitoring tool when coupled with more detailed sampling techniques such as macrofaunal core sampling, or continuous sediment survey transects (Gowing et al. 1997). It offers point data that can be economically collected at sufficient frequency to connect more resource-intensive samples in an ecologically meaningful way. A study can therefore operate at nested spatio-temporal scales with SPI providing overall maps and connectivity while other sampling techniques are used to characterise assemblages and variability within habitat types. This type of integration is necessary for developing our understanding and predictability of soft-sediment processes (Thrush et al. 1999; Noda 2004).