7.2.1 Desk studies and field reconnaissance

The Association of Geotechnical Specialists advises that ‘Desk Studies’ are the most cost-effective element of a site investigation (AGS 2006). Online resources available through the internet, when used in conjunction with hard-copy data sources found in libraries and local authority records, enable a comprehensive compilation of existing data about a site to be quickly and cheaply assembled. Table 7.1 lists the types of data that should be collected during the desk study, with examples of likely sources for such information (after Shilston et al. 2012). Such studies mean much more to the person making them if they have been able to visit the site. Accordingly, Table 7.1 includes an initial walkover survey, undertaken by personnel experienced in the type of conditions anticipated at the site (note how prior knowledge of a most general kind is being used from the very start of the process), as part of the desk study.

There are two important outputs from such desk studies. First and foremost should be a preliminary conceptual model of the ground in 3D (Fookes 1997; Parry et al. 2014) that accommodates all the available data on ground conditions at the site under investigation. This ground model must draw on the conditions expected within the glacial and periglacial landsystems presented in Chapters 3–5. The ground model must be considered as a ‘living’ concept whose correctness and totality will continue to improve as more data become available during the progress of the site investigation (Culshaw 2005; Royse et al. 2009). The ground model must take into account all the data collected before the ground investigation commences, namely the expected geology based on existing maps and reports, the interpretation of the remote sensing data (Section 7.2.2), compilation of that data in a GIS (Section 7.2.2), detailed engineering geomorphological and geological mapping (Section 7.2.3) and use of archaeological investigations (Section 7.2.4); further, it should synthesize these data into one or more documents that put on paper a visualization of what the ground could contain. This stage of ground investigation is not complete until the data collected has come out of the files into which it has been stored and been used to create a 3D depiction of the ground, including its landform, sediments, structures, bedrock and the anticipated fluctuations in water table and piezometric levels.

The second set of outputs from this preliminary stage of the ground investigation should be the inclusion of those data into the risk registers for the project. Project risk registers include all elements of project risk: those naturally occurring (Maund & Eddleston 1998); those associated with normal construction; and those of a financial and management nature (Baynes 2010). However, the major risk by far for every project is the nature of the ground (SISG 2013) because it is not a product manufactured by man and no one can know exactly what it contains; investigations probe only a tiny fraction of the volume of ground involved with and affected by engineering work (de Freitas 2012a). Information on ground conditions must therefore form part of a risk register that is updated throughout the project as ground conditions and ground response become known (Clayton et al. 1995; Simons et al. 2002).

7.2.2 Remote sensing (RS) and geographical information systems (GIS)

The term remote sensing (RS) covers a wide variety of techniques for acquiring data, normally images, from satellites, aircraft and ground-based instrumentation, including various parts of the spectrum (e.g. optical, infrared, radar) as well as laser and underwater bathymetric imaging. Google Earth and other web-portals have provided the public at large with familiarity of high-definition optical satellite imagery, but the resolution of RS data can range from 0.5 m on the ground, for optical images and digital elevation models (DEMs) produced by light detection and ranging systems (LiDAR) using lasers from aircraft, to millimetre accuracy from radar satellite interferometry and ground-based instruments. In addition, there is an overlap between remote sensing and some geophysical techniques as the former can provide information about the physical and chemical properties of the surface materials being imaged. However, when considering the use of RS data for ground investigation it is the ability of these systems to image, that is, to distinguish and capture, that is their key asset. Geographical information systems (GIS) is an all-embracing term to describe computer-based methods for capturing data sources such as imagery, mapping and geophysics in digital form, organizing their various elements, spatially co-registering them and providing tools to visualize, compare and analyse the data in three-dimensional space. GIS also ties together conventional mapping (geological, geomorphological or topographic), RS data, geophysics, geochemistry and hydrology. GIS is now routinely used in the preparation and publishing of geological maps and much of the world's geological mapping is available in GIS formats, the most popular being Arc-GIS from Environmental Systems Research Institute (ESRI) (see http://www.esriuk.com).

Viewing GIS data, either in Arc-View or some of the other viewing tools freely available via the internet, is fairly straightforward; however, importing and registering multiple datasets from disparate sources and then performing complex comparisons and analysis on them is much more challenging. GIS technicians, trained for the task, will often be able to achieve this in a more time- and cost-effective way than the non-specialist. A good summary of GIS tools and techniques is freely available at http://www.MapAction.org/resources (‘A field guide to humanitarian mapping’). Although both RS and GIS disciplines continue to evolve rapidly, the review of both provided in Shilston et al. (2012) (albeit oriented towards engineering studies in arid deserts) continues to serve the reader well. This section summarizes aspects of remote sensing of relevance to ground investigation in glaciated and periglaciated ground, notes potential pitfalls with its use for this purpose, highlights successful applications and indicates where further development may be expected.

Almost all the new developments in RS and GIS data acquisition, storage and manipulation either result from, or have been facilitated by, the exponentially increasing power and memory capacity of microcomputers. The latest computers and mobile phones are able to download satellite imagery, gather data in the field with GPS tagging, check published mapping from cloud-servers and even display the results by pointing the phone at a domestic TV screen, although more sophisticated computers and interfaces are required for advanced image processing and GIS manipulation. The driver for these RS and GIS developments has mainly been military intelligence (including, of course, GPS which is so valuable for rectifying and co-locating all the data), but satellites and sensors which have a purely environmental monitoring or sociological purpose are increasing in number, driven by the growing overlap between commercial and consumer-led data acquisition and its applications, such as outdoor leisure. At present, the economic drivers for purely geologically oriented satellites (for example those with ‘hyper-spectral’ sensors that could be very useful for mapping rock types and mineral distributions) are lacking; the geoscientist is therefore mainly reliant on the technical spin-off from satellites designed primarily for other applications.

In addition to stimulating and facilitating satellite imaging, the digital revolution has revived the use of all forms of aerial reconnaissance. Aerial photos are routinely acquired at a wide range of scales (especially over and around urban areas), and processing power allows fast and semi-automatic mosaicking and extraction of digital elevation models (DEMs); this can be done at low cost for small areas (typical of much civil engineering work) using hand-launched unmanned aircraft or drones (Groom 2013). This latter development is already used in site surveys, not just for geoscience but also for establishing an environmental baseline inventory. In addition to airborne cameras and scanners sensing in parts of the spectrum that sometimes discriminate surface reflectance differences not be seen by eye, there are also radar systems (on both satellites and aircraft) which can include frequency bands that can map near-surface material changes, penetrate a cover of vegetation and are able to image in permanently cloud-covered areas. In addition, there are aircraft- and helicopter-borne LiDAR systems that provide very detailed DEMs, including bathymetric maps of great value for mapping glacial features now submerged by the postglacial recovery of sea level (Fig. 7.1).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.1.

Combined high-resolution LiDAR topography and multibeam bathymetry for Northern Resurrection Bay, Seward, Alaska (Labay & Haeussler 2008).

A good DEM is as important a source of geological information as an image of the surface cover (Figures 7.2, 7.3, 7.4), often more so, and GIS can be used to enhance subtle surface features with artificial low-angle illumination. Those DEMs derived from satellite images are routinely available worldwide with a ground resolution of 25 m, soon to be improved to 12.5 m and to 1 m for specific areas (Fig. 7.5). The DEMs produced from airborne sensors can have even better resolution and, in some parts of the world (including the UK), archived airborne LiDAR data may be available from government or commercial sources at moderate cost. Nevertheless, data acquired by satellite (whether as pictures of the surface cover or as DEMs) tend to be the most readily available and lowest in cost, especially over remote areas; they are therefore the starting point for most areas of investigation. With so many possible sources of remotely sensed imagery available, the non-specialist wishing to use it is strongly advised to seek advice from an experienced data provider, preferably one not linked too closely to any one satellite operator.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.2.

A detailed zoom into the LiDAR data of Figure 7.1 just north of the town of Seward (Labay & Haeussler 2008). Active fan and talus cone systems are clearly seen at the foot of the mountains on the left-hand side (west) of the image, and the braided nature of the stream systems is evident beneath the conifer forest and sedge grass cover.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.3.

Bare earth LiDAR data from Puget Sound showing smoothed, low elevation hills and slopes characteristic of glaciated terrain. The DTM reports the terrain rather than the tree tops as shown in the image on the right. (DTM courtesy of The Puget Sound Lidar Consortium; Satellite image courtesy of Google Earth.)

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.4.

DEM of Anderson-Robe mapsheet (Alaska) from USGS NED database, with 12.5 m resolution (http://nationalmap.gov/viewer.html). Two north-facing spurs dominate the onshore area of the image near Valdez (Prince William Sound, AK), and a petrochemical plant is clearly seen on the shoreline. The eastern spur is scarred by a major landslide whose back wall is formed by a NW–SE-trending fault (marked ‘Flt’). The western spur is similarly affected by three or four faults also with the same orientation. North-dipping bedding of the deformed metasedimentary Valdez Group lies close to slope angle and may be prone to slab sliding.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.5.

DEM of an area in Armenia generated from fore-, aft- and nadir-viewing stereo Worldview-2 imagery and processed to a 1 m grid posting. Tri-stereo imagery is advantageous over mountainous and hilly terrain or in areas with significant urban centres where building shadows and occlusions are a significant issue. The relative vertical accuracy in open terrain is c. 1–2 m, but in many cases this is exceeded. Ground control points are required for absolute accuracy. Imagery ©DigitalGlobe, DEM processing ©CGG-NPA Satellite Mapping.

It is easy to forget that the purpose of acquiring the data is to interpret and subsequently use such information it provides; the latter is generally outside the range of standard geoscientific training and sometimes very far beyond it. It is also easy to overlook the fact that imagery downloaded from Google Earth and co-registered to ground data does not equate to a full integration and understanding of the ground from remotely sensed data. The basic tasks of drainage mapping, topographic and geomorphic analysis, analyses of subtle soil reflectance or vegetation changes and similar studies still have to be performed with skill and experience and all require time, care and attention to detail. This is before considering the possibility of using spectral mapping to discriminate soil or rock types or moving to analysis and processing of radar data. Such techniques require data not found in Google Earth and processing techniques that are not commonplace; specialist help from experts in their field will generally be required. Imagery from Google Earth and other ‘cloud’ sources, good though it is in many parts of the world, is only a snapshot in time and will almost certainly not be the optimum data source for every individual task. There is often much to be gained by looking at imagery acquired at different times of the year and acquired as a chronology captured to show gradual surface changes occurring with time.

Much of the world's periglacial and glacial terrain lies in environments with extensive superficial and vegetation cover. It will contain the land systems described in Chapters 3 (Section 3.4.2), 4 (Sections 4.1.2 and 4.1.5) and 5 (Section 5.1.5) with all their complexity, requiring recognition and mapping, often beneath tree cover. Separating these is a key objective in assessing the likely terrain in which a site is sitting and hence a definition of the possible character and behaviour of the ground for the purposes of engineering design and construction. Choice of either suitable bands for radar data acquisition or methods of LiDAR acquisition and processing may offer a step-change in the ability to penetrate tree cover (especially P-band airborne radar) and reveal the true terrain surface below the forest canopy instead of just the tree tops, enabling a digital terrain map to be constructed rather than a DEM (Fig. 7.3).

Satellite radar interferometry (InSAR) allows multiple satellite radar images acquired over a period of a few months to many years to be stacked; the archive can extend back over previous years, enabling millimetre-scale changes in surface elevation and hence morphology to be measured. This can reveal ground movement and deformation and display it graphically; further, there is strong evidence that relative surface movements can reflect aspects of geology concealed by superficial materials including urbanization. InSAR is therefore a tool with great potential for use with land system mapping, as for example in glaciated areas, but brings with it the need to understand the many possible drivers for small changes such as temperature, tectonics and groundwater, and their possible interactions. In fact, this technique is enhanced in urban areas due to reflectance from buildings, exactly where site investigation, especially subsurface, or geophysics are most difficult to use. Satellite radar interferometry will undoubtedly play a significant role in monitoring ground stability associated with extraction of shale gas, much of which happens to be in areas once glaciated, and which could have a high economic profile in the coming years. A recent study over the London Basin gives an insight into the range of new geological features that may be revealed (Aldiss et al. 2014).

At present, satellite InSAR requires very complex processing and is only undertaken by a few highly specialized teams; however, as cloud-computing develops, it is likely that archived data stacks of satellite radar will soon be pre-processed and added to continually, so that the user is free to work on their interpretation rather than complex processing. This is increasingly true of other data types that are already automatically accumulated in geo-registered databases. Now the techniques and algorithms have been developed for stacked satellite InSAR processing, it has been realized that they can be applied to ground-based radar sensors; a new generation of equipment is therefore being developed for monitoring stability of slopes, rock-faces, embankments and so forth, based on a spin-off from satellite technology (Thomas et al. 2013).

7.2.3 Engineering geological and geomorphological mapping

At an early stage in the desk study a decision should be made on whether engineering geomorphological and engineering geological mapping would contribute materially to an understanding of the ground and to the definition of its conceptual model. This sort of mapping is inexpensive and can be immensely valuable; the role, importance and methodology of such mapping are well documented (Anon 1972; Dearman 1991; Griffiths 2002; Cole 2010). Field mapping by experienced practitioners, allied to remote sensing interpretation as described above, has a long and proven history of providing critical data in a cost-effective way for all forms of engineering project (Fookes 1997; Griffiths 2014). The typical data to be recorded on such maps are described in Table 7.2. Data compiled on such maps, when accompanied by comprehensive tables, vertical cross-sections and 3D landsystem diagrams of the sort presented in Chapters 4 and 5, establish the first version of the conceptual ground model, the one that enables the subsequent ground investigations to be designed and interpreted.

As noted in Table 7.2, effective mapping must incorporate geomorphological data (Griffiths 2001). Mapping geomorphology requires a slightly different methodology from the geology-based techniques familiar to graduate geologists (Smith et al. 2011). Geomorphological mapping emerged initially as an academic attempt to link the form of the Earth's surface to the natural processes that shape it, and to further link those processes to the properties of the underlying soil and bedrock (Hutchinson 2001; Verstappen 2011). Such mapping seeks to subdivide the landscape into ‘terrain’ units characterized by similar form, materials and processes (Lee 2001; Smith et al. 2011). These units are typically created on essentially homogeneous parent material, defined by underlying geology, water regime and topography. Although the geomorphological units vary in size they are usually mapped at scales between 1:10 000 and 1:500, and the choice of mapping scale is normally dependent upon the type of investigation and the scale of the base maps available (Griffiths et al. 2015). For smaller-scale mapping an approach using terrain systems mapping is often adopted (Chapter 3, Sections 3.2 and 3.3), which splits the terrain into a series of hierarchical units from the ‘land system’ (the largest), to the ‘land unit’ and down to the individual ‘elements’ (Lawrance et al. 1993; Phipps 2001, 2003). For large-scale mapping the individual elements of the landscape are mapped in detail and four different types of output map can be generated (Cooke & Doornkamp 1990).

• Morphological or morphometric map: this is the factual portrayal of the shape of the ground surface, including the shape and extent of the landforms.

• Morphochronological map: this displays the landform units according to their age of initiation with a distinction usually made between recent or current forms and relict forms whose character is related to past processes of landscape development.

• Morphographical map: on this, landforms are identified by a geomorphological name taken from a standard lexicon (e.g. Goudie 2004).

• Morphogenetic or dynamic map: this shows the origin and development of the landscape, including details of the processes that are presently active.

While a complete engineering geomorphological map would show relevant aspects of all these various components, during the initial stages of the geomorphological investigation it is often useful to prepare separate maps to ensure all the data are collected and can be seen in an uncluttered way (Griffiths & Whitworth 2012).

Engineering geological and geomorphological mapping are therefore complimentary techniques that rely on the input from desk studies, the interpretation of remotely sensed data and field mapping to produce an output that is directly useful for engineering design. They provide a scientifically rigorous method for seeing and interpreting evidence on the ground and for collecting field data; further, they deliver a systematic spatial framework for accommodating and interpreting the more localized site-specific data typically collected in a ground investigation. Field mapping, allied to the analyses and interpretation of remotely sensed data, should be regarded as fundamental to the creation of the ground model (Section 7.7) and can be used in the investigation of all types of terrain, including the glacial and periglacial environment.

7.2.4 Use of archaeology

Archaeology has been used in its modern sense since the early nineteenth century, and is primarily concerned with analysing the modifications humanity has made to our environment and the detritus we have left behind. The application of archaeology to geology is particularly relevant in the Quaternary Period where ancestral hominins (our human ancestors) and their highly durable stone tools are found in sediments, a period which archaeologists now refer to as the Palaeolithic (old stone) age (Fig. 7.6).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.6.

Timeline illustrating the climatic, geological and archaeological chronology for the last 1 million years; global sea levels after Lisiecki & Raymo (2005) and details on the geology and archaeology after Lowe & Walker (1997), Funnel (1995), Kukla et al. (2002), Dix & Westley (2004), Gibbard & van Kolfschoten (2004), Lee et al. (2006) and Wymer (1999).

Geologist Clement Reid in his book Submerged Forests (Reid 1913) provides a summary of some of the most interesting archaeological discoveries and, most importantly, how they relate to and provide information for interpreting sedimentary sequences in which they occur. The recognition that archaeological remains are disturbed by construction has made archaeology, in some countries, a legal requirement within planning applications; for example in England and Wales this is now governed by Planning Policy Statement 5 (2010), and in England by the National Planning Policy Framework (2012). Offshore developments in the UK are subject to the United Nations Educational Scientific and Cultural Organisation convention, governing underwater cultural heritage (UNESCO 2001), and English Heritage (2002) have taken responsibility for planning-related heritage control from the low tide mark up to the 12 nautical mile limit in England.

7.2.4.2 Interpretation

Of particular interest to archaeological studies is the role of natural processes in the formation and preservation of archaeological sites, and this is where the utilization of geotechnical data by archaeologists can be beneficial. Site investigation data can be reviewed and interpreted by a geo-archaeologist before construction projects commence to indicate where potential archaeological remains might occur. With collaboration between geotechnical engineers and archaeologists, specific horizons of archaeological interest can be targeted and samples taken suitable for geo-archaeological recording (Fig. 7.7), scientific dating and palaeoenvironmental analyses.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.7.

Geoarchaeological recording of vibrocore samples.

Fluctuations in global temperature and sea levels have influenced the patterns of colonization by humans over the Quaternary period. The hominin occupation of the British Isles is a good example of a discontinuous occupation related to climate, possibly from as early as c. 970 ka BP (Parfitt et al. 2010) (Fig. 7.6). Surrounded by shallow seas at this time, the British Isles were probably colonized on foot when periods of lower sea level provided a land connection to the continent (Bridgland 2002; Jacobi & Higham 2009), as colonization was not possible during periods of higher sea levels and before the invention of sea-going water craft (around 8–10 ka ago; McGrail 2004). On land, some of the most useful records of these fluctuations, of interest to archaeologists and geologists alike, are river terraces; the best preserved and probably the most studied are those of the Thames (Bridgland 1994, 2002; Wymer 1999; Spurrell 2014).

7.2.4.3 Case studies

Case Study 7.1:

Dogger Bank

Archaeologists have been working closely with the geotechnical engineers since 2009 to understand the prehistoric archaeology of the Dogger Bank in the North Sea, the proposed site of an extensive offshore windfarm (Fig. 7.8). The majority of the Dogger Bank comprises complex glacial sediments of little archaeological interest; however, much of the area is relatively shallow c. 18 –35 m and has long been known as an area potentially inhabited by our ancestors after the last glacial retreat and prior to its final submergence during the Holocene (Reid 1913; Coles 1998). By interpreting and correlating geophysical and geotechnical datasets (Fig. 7.8), past environments have been reconstructed including lakes, rivers, estuaries and coastlines. Core samples obtained from these features contained organic material, and the plant remains (Figs 7.9 & 7.10) indicate a range of past environments (woodland, marsh, coastal and estuarine), the study of which is interesting from an archaeological perspective as it was during these periods that the North Sea floor was occupied by humans (Louwe Kooijmans 1970–71; Long et al. 1986; Coles 1998) who first began to modify the environment on a large scale (woodland clearance). The archaeological investigations and geo-archaeological and palaeoenvironmental data have been integrated to augment the geological model of the area, and aid the interpretation of anomalous data within the geotechnical datasets (Cotterill et al. 2012).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.8.

Dogger Bank, North Sea borehole BH1282, preliminary geotechnical log palaeoenvironmental assessment samples, sub-bottom geophysical data example and radiocarbon dating.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.9.

Betula cf. pubescens (Downy birch) and Menyanthes trifoliata (bogbean) from the Dogger Bank, North Sea.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.10.

Dogger Bank radiocarbon dates Holocene sea levels (sea-level curve after Ward et al. 2006).

Case Study 7.2:

Olympic Park

Geo-archaeological work involved a site wide assessment of the 246 hectare Olympic Park in London, the interpretation of over 5000 borehole logs and the analysis of archaeological remains from 32 deep trenches, was undertaken between 2009 and 2012 (Powell 2012). The borehole and trench data were used to recreate the prehistoric and historic landscape of the Lea Valley, including former watercourses, wetlands and areas of raised ground suitable for human occupation, all of which could then be investigated (Fig. 7.11). Radiocarbon dating and environmental analyses were also undertaken which, coupled with the archaeological and geo-archaeological data, revealed palaeoenvironmental and anthropogenic changes of the site over the last 14 ka years (summarized schematically in Fig. 7.12). This model was used to refine the site investigations and, as a result, revealed archaeological remains including a Neolithic pottery, a flaked flint axe and a timber structure along the edge of one of the buried channels. Radiocarbon dating of the timber structure suggests that it was constructed during the Early Neolithic.

• Download figure
• Open in new tab
• Download powerpoint

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.11.

Topography of the Lea Valley, London showing channel areas existing by the Late Pleistocene–Early Holocene and the location of the trenches from which environmental sequences were analysed.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.12.

General development of the landscape of the Olympic Park site through time, showing the main features of sedimentation, vegetation and land-use occurring during the: (a) Late Pleistocene; (b) Mesolithic; (c) Bronze Age; (d) Saxon; and (e) post-medieval periods.

7.2.5 Additional requirements for planning offshore and nearshore SI

Marine over-water investigations typically make the distinction between offshore (water depths greater than 25 m) and nearshore operations (water depths less than 25 m) due to typical plant and vessels used for access, although significant overlap can occur. The development of offshore wind power schemes around the world over the last decade, especially in NW Europe, has significantly increased the need for marine geotechnical investigations. These had previously been dominated by geotechnical support to the offshore oil and gas industry, cable or pipeline route planning or nearshore coastal structures. The anticipated global growth in wave and tidal power generation schemes is likely to provide further demand for well-planned and executed marine ground investigation.

The growth of nearshore and offshore wind farm projects, particularly in NW Europe and the Americas, invariably means that the principal soils considered for foundation design are of a glacial origin. While intrusive and non-intrusive investigation techniques used in the assessment of offshore glacial deposits are fundamentally the same as those onshore (e.g. use of seismic geophysics, rotary cored drilling, occasional cable percussion and cone penetration testing), the delivery of these techniques and scope of coverage is fundamentally different from onshore operations; significant additional constraints are imposed by location, water depth, weather, wave height and the nature of the glacial (and interglacial) Quaternary deposits that are typically encountered as the margins of continental shelves are approached. The following is not provided as a definitive description of all marine ground investigation practice in glacial deposits, but to highlight some of the key delivery systems and constraints present.

7.3.1 Geophysical techniques and the physical properties of glacial and periglacial materials

The composition, internal structure and bulk engineering properties of geological materials, including their fluid content in pores and fissures, can be estimated by surface and borehole geophysical techniques. These properties are deduced from the response of the materials to: a known stimulus (e.g. a hammer blow or similar acoustic stimulus); a known potential gradient (e.g. a field of electrical or magnetic or heat flow); or to detectable rates of change of composition (e.g. flux of gamma rays). The physical properties used include density, elastic moduli, electrical conductivity and magnetic susceptibility, among others.

Geophysical methods can be passive, monitoring ambient potential fields (e.g. gravity), or active, when deductions are based on the response of the ground to energy propagated into it (e.g. as acoustic waves). The following list of techniques comprises those commonly used in studies of glacial and periglacial terrains, but is not exhaustive.

• Electromagnetic methods (EM) measure the response of materials with contrasting electrical conductivity/dielectric constant and include EM induction techniques and ground-penetrating radar (GPR). Electrical conductivity, or its inverse, resistivity, can also be estimated by electrical techniques including electrical resistivity tomography (ERT), and complex resistivity by induced polarization (IP).

• Seismic methods monitor seismic wave velocities through geological materials, and include active methods such as reflection, critical refraction and multi-channel analysis of surface waves (MASW) and the passive microtremor array measurements (MAM) method.

• Potential field methods include microgravity methods that enable mapping of density contrasts, and magnetic methods influenced by the magnetization and magnetic susceptibility of materials.

The geophysical signature of a deposit is intrinsically related to its physical characteristics including its mineral composition, grain size, bulk density, internal structure, pore space and the distribution and composition of its associated pore fluid. Different geophysical signatures can be used to characterize geological materials, enabling discrimination and identification of discrete glacigenic and periglacial facies.

7.3.2 Physical properties of glacial and periglacial materials

Near-surface geophysics is used to constrain key geological features within a range of glacial and periglacial deposits and landforms, including: moraines (Harris 1991; Harris et al. 1997; Williams et al. 2001; Bakker 2004; Bennett et al. 2004; Sadura et al. 2006; Benediktsson et al. 2010; Parkes et al. 2013); drumlins (Kulessa et al. 2007; Hiemstra et al. 2008, 2011); eskers (Burke et al. 2009); outwash fans and sandurs (Cassidy et al. 2003; Sadura et al. 2006); kame and kettle topography (Busby & Merritt 1999); meltwater channels (Sénéchal et al. 2000), tunnel and palaeo-valleys (Ahmad et al. 2009; Stumpf & Ismail 2013); glacially associated lacustrine deposits (van Dam 2012); talus slopes (Sass 2008); rock glaciers (Musil et al. 2006; Monnier et al. 2009); polygonal/patterned ground (van Dam 2012); and for facies characterization (Kilner et al. 2005; Ismail et al. 2014).

Although the detailed geophysical signatures of glacial and periglacial facies vary greatly, there are general responses that are associated with specific physical characteristics of a facies. Particle size is perhaps one of the most important characteristics that govern the geophysical signature of a deposit, as it will also influence porosity, permeability and associated pore fluid.

7.3.2.1 Fines (clays and silts)

Fines-dominated units (grain size <0.063 mm) are formed by clay and silt-sized particles. These deposits are usually accumulated in low-energy fluvial, lacustrine and aeolian or hillslope environments. Fines enhance the attenuation of electromagnetic signals and will influence, for example, the response of GPR (Fig. 7.13). When electromagnetic energy is applied to clay, the particle charges migrate, converting energy first to kinetic and then to thermal forms, so attenuating the EM signal. Fine sediments are generally conductive materials because they typically have high moisture contents compared with coarser-grained facies. The mineral structure of clays, in particular, can accommodate ions dissolved in surrounding porewater. These ions can absorb an EM signal, and at higher EM frequencies the water itself absorbs the signal. Standard electrical methods are unable to differentiate between clay and silts; however, the chargeability of silts and clays differs and induced polarization (IP), which measures chargeability, can discriminate between them. Fines-dominated facies when stiff typically have high seismic velocities (P-wave velocities of 1000–2500 m s⁻¹ and S-wave velocities of 400–1000 m s⁻¹).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.13.

Geophysical evidence from multiple techniques of a glaciotectonic terminal moraine's internal structure, Cheshire Basin, UK, including (a) electrical resistivity tomography; (b) ground-penetrating radar; (c) seismic refraction; and (d) associated interpretation of the combined data. Geophysical anomalies, where possible, were verified by shallow excavation. Interpretation of ERT profiles was made more robust by the collection of in situ micro-resistivity data.

7.3.2.2 Sands

Sand deposits are those formed of particles of size between 0.063 and 2 mm. Within glacial and periglacial environments, sands are usually deposited within glaciofluvial systems with moderate to high energy (e.g. in meltwater channels and over outwash fans and plains) or by wind (as coversand). Sand grains are generally less conductive and more electrically resistant than fines, although this is strongly influenced in sand deposits by composition of their pore fluid, with air and fresh water being resistive and saline fluids highly conductive. GPR is suited for investigations of coarse-grained deposits due to their low attenuation, unless pore fluids are saline. In coarse materials EM energy is capable of differentiating between beds of fine-grained sands and coarser deposits (Fig. 7.14), as detectable reflections occur at interfaces between such beds. Geological structures such as faults can be identified by offset and truncation of reflectors (Harris et al. 1997; Bennett et al. 2004). Sand-dominated facies are generally associated with P-wave velocities of 200–1000 m s⁻¹ and S-wave velocities of 80–400 m s⁻¹, that is, lower than those of materials dominated by fines, but these velocities can be significantly influenced by pore fluid. When saturated they display higher than usual seismic velocities (P-wave 800–2200 m s⁻¹ and S-wave 320–880 m s⁻¹), which overlap with the measured velocities for fine material.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.14.

Seismic reflection profile of a glaciotectonic landform at Dinas Dinlle, Wales, showing four large thrust structures (image from: Harris et al. 1997)

7.3.3 Geophysical techniques

Although a considerable number of geophysical techniques are available, the most commonly employed within glacial and periglacial environments have been GPR, ERT and seismic methods. A typical GPR array comprises two antennae, a transmitter in the range of 0.025–1 GHz and a receiver (Van Dam 2012). Transmitted EM energy is reflected back from interfaces with different dielectric properties and recorded to produce a cross-section of the subsurface (Bennett et al. 2004). Surveys are usually conducted by common offset profiling with survey optimization, such as step size and separation, determined by the frequency of the antennae used (Schrott & Sass 2008). Antennae can be either shielded or unshielded, and while unshielded provide greater depth penetration they are prone to interference from surface objects (Milsom & Eriksen 2011). Recent advances have led to the development of multiple-receiver systems, which provide improved resolution and resolvability of features by increasing the ratio of signal to noise (Van Dam 2012).

GPR works best in coarse-grained materials such as sands and gravels (Monnier et al. 2011), and provides a graphical output that eases data interpretation. It has been used to investigate numerous landforms and deposits including: moraines (Bakker 2004); glaciofluvial deposits (Cassidy et al. 2003; Heinz & Aigner 2003); rock glaciers (Ahmad et al. 2009); and talus slopes (Sass 2008). GPR gridded surveys enable 3D models to be constructed (Van Dam 2012). The main disadvantages of the technique are signal attenuation across deposits dominated by fines and/or saline pore fluid. Generally, lowering EM frequencies will increase depth penetration; however, this results in a lowering of both the resolution of data and the resolvability of features. The optimal frequency can be selected by trialling various frequencies across test areas whose subsurface characters are known.

In ERT a current is passed between two current electrodes and the potential difference is measured between two potential electrodes, the depth of detection being a function of the electrode spacing. Modern systems use an array of electrodes and a computer to switch between electrode pairs, selecting incremental changes in spacing and position along the array. The data obtained are inverted to produce 2D profiles, showing the distribution of resistance within the subsurface (Loke 2004). Gridded arrays can be used for generating 3D surveys of the ground.

ERT has been used to map the distribution of subsurface facies for a range of landforms including moraines (Hambrey et al. 2008; Parkes et al. 2009, 2013), drumlins (Kulessa et al. 2007; Hiemstra et al. 2008, 2011), ramparted depressions (e.g. Ross et al. 2011) and glaciolacustrine environments (Pellicer & Gibson 2011). These have required different methods of interpretation including comparisons with published values from ERT (Palacky 1987), data from boreholes (Kulessa et al. 2007) and in situ micro-resistivity measurements of key facies (Parkes et al. 2013).

One of the key advantages of ERT is the wide range of array configurations that can be used to optimize a survey for specific purposes, such as mapping different structures (Figs 7.13, 7.14). Different arrays can change the depth penetration and the horizontal extent of mapping (pole–pole), improve the resolvability of horizontal interfaces (Wenner–Schlumberger arrays), enable mapping of vertical and subvertical features to be achieved (dipole–dipole) and vary the sensitivity of a survey to both horizontal and vertical structures (Wenner–Schlumberger array). The asymmetrical pole–dipole array provides good horizontal and vertical coverage with less noise than a pole–pole array. However, upon completion the survey should be reversed and repeated and the results combined to avoid biasing of the inversion model. The induced polarization technique is capable of differentiating between clays and silts due to clay minerals having chargeability, where as other resistivity methods have difficulties because both clays and silts are typified by low electrical resistivity. IP is seeing increased use as data can now be collected using ERT equipment (Van Dam 2012).

Another key advantage of ERT is that it is largely automated and presents its results in real time, allowing the surveyor to interpret results while in the field and to respond to these results. ERT can also be used for long-term monitoring and characterization of a site through time-lapse surveying (Pellicer et al. 2012). The disadvantages of ERT are that its results are susceptible to interference from buried objects such as pipes and cables, it requires good electrical contact with the ground (which may not be possible in landforms such as talus slopes or rock glaciers) and its interpretations are not unique.

Seismic reflection surveys record the two-way travel time of seismic waves to and from interfaces between layers with different acoustic impedance, the product of density and seismic velocity. Both P- and S-waves can be recorded with suitable equipment. Seismic reflection provides graphical sections similar to, but with lower resolution than, those produced by GPR; it is therefore an alternative to GPR when GPR may not be appropriate, such as in fines-dominated deposits or those with saline pore fluids. However, the acquisition and processing of reflection data is more time-consuming and labour-intensive than other methods. Further, ground conditions must be suitable for producing reflections. Seismic reflection was used by Harris et al. (1997) to map the depth of thrust structures observed within a coastal cross-section through the Dinas Dinlle moraine, Wales (Fig. 7.14).

The seismic refraction method records P-waves that are critically refracted at interfaces on either side of which significantly different seismic velocities occur. Processing of arrival times enables the seismic velocities, thickness and depths to be derived for up to three detected layers using the plus-minus method (Hagedoorn 1959) or other inversion methods. Seismic refraction is a useful technique for distinguishing between unconsolidated glacial and periglacial sediments and their consolidated equivalents, including bedrock (e.g. Kulessa et al. 2007), as well as the depth to water. The disadvantages of the technique are that it lacks resolution and is unable to image strongly dipping units or image layers where the velocity decreases with depth.

The multi-channel analysis of artificially generated surface waves (MASW) and microtremor array measurements (MAM) of naturally occurring waves both use techniques that generate dispersion curves from surface waves to reveal the variation of S-wave velocity with depth (Park et al. 1999). A shear wave velocity profile of the ground is produced by both methods and multiple measurements can be combined to produce 2D and 3D profiles for the ground. Importantly, the shear wave velocity can be directly related to the stiffness of a material, as it is a function of the material elasticity and its undrained strength. Consequently, both MASW and MAM are useful tools for ground engineering purposes; however, the application of MASW to glacial sediments is currently limited (Ismail & Sargent 2006).

Geophysical methods, including potential field methods and other propagating approaches that might be useful for investigating ground affected by glacial and periglacial processes, are summarized in Table 7.3.

7.3.4 Marine geophysical investigation: additional considerations

Marine geophysical investigations face similar challenges to those of onshore environments in terms of the ground conditions associated with glacial and periglacial terrains. However, geophysical investigations offshore have to be performed from vessels at the sea surface (not withstanding downhole geophysics), therefore making the task of acquiring accurate and useful data different from onshore. Offshore geophysical survey planning should be predicated on objectives set out in preliminary desk studies with a thorough understanding of the geological domain where data are to be acquired. This may prove difficult where data are either not available or limited and often results in the cyclic nature of marine SI works where information is collected, assessed, SIs are re-planed and new, better-quality data are acquired over several phases of work.

Three primary geophysical sensors are used for offshore surveys: multi- and single-beam bathymetry, side-scan sonar and sub-bottom profiling (seismic reflection). Bathymetry and side-scan sonar are acquired to investigate depth to seafloor, seabed sediment composition and the size/shape of any objects on the seafloor (e.g. glacially derived boulders). Equipment can be deployed from vessels of varying size, usually dependent on depth to seabed and sea state. Small craft can deliver sensors into the nearshore environment.

For investigation of the subsurface (below seafloor), sub-bottom profiling is most frequently used. The type of equipment deployed is dependent on the purpose of the investigation and the required penetration and resolution of the data required. For example, high-resolution, shallow-penetration geophysical profiling equipment needs to be used to accurately image features between 5 m and 10 m below seafloor, such as infilled iceberg scour-marks and remnant ice-rafted debris. Deeper penetration (therefore lower resolution) sub-bottom profiling may be required to image subsurface features such as glaciotectonic structures and other buried ice-contact features, erosional channels (including large-scale tunnel valleys), and to image multiple glacial tills either overlying bedrock or underlain by interglacial deposits.

An initial geophysical survey in the marine environment provides the opportunity for detecting potential shallow-marine gas as this produces a ‘blanking’ effect on seismic profiles. While this may significantly diminish the quality of the seismic dataset and will mask underlying geologies, it is imperative that such findings and the associated risks of drilling into gas-bearing soils in the marine environment are communicated to design and logistics teams to allow suitable planning for subsequent intrusive investigation works.

In continental shelves affected by ice cover and grounded ice sheets, such as the North Sea (<150 m water depth), surface-towed or vessel-mounted geophysical equipment is typically used. In deeper-water environments of greater than 150 m depth, deep-towed remotely operated vehicles (ROV) and autonomous underwater vehicles (AUV) can be used with the aim of getting the seismic sources and sensors as close as possible to the seabed being imaged.

In the marine environment, additional consideration must be given to the following.

• Understanding of the logistical challenges, vessel capabilities and sensitivity of equipment (e.g. downtime or poor data acquisition due to periods of bad weather).

• Integration between marine and onshore seismic profiling techniques. This is particularly important for planning investigations for cable/pipeline routes that will likely require data collection in both the offshore, nearshore, beach/tidal flat and terrestrial environments.

• Integration between seismic profiling data of varying resolutions such as shallow (<10 m below seafloor) and deep (>10 m below seafloor) data types.

• Differences in the physical properties between glacial and periglacial deposits and Holocene–Recent deposits, leading to differential penetration of seismic profiling systems in the marine realm. For example, as with stiff to very hard glacially loaded clay deposits overlain by more recent marine sands and gravels.

Ground-truthing of geophysical data acquired offshore is logistically and financially much more challenging than doing so onshore. This is further described in Sections 7.4 and 7.5.

7.3.5 Conclusions for best practices with geophysical surveying

• Near-surface geophysical methods can prove suitable for use on land when glacial and periglacial features and their associated sediments lack exposure, but no single technique is applicable for all situations; indeed, some methods are very specific in their performance. Methods must be selected based upon site conditions and survey requirements.

• The geophysical survey forms a critical component of marine SI works, particularly for large offshore and nearshore developments such as wind farms where the size of some sites (>100 km²) necessitates additional data capture between borehole locations and an integrated approach to ground model development and geohazard identification.

• Appreciating the likely physical properties of glacial and periglacial materials enables the best combination of geophysical methods to be selected for their investigation (Tables 7.3 & 7.4). Facies have different geophysical signatures due to their physical properties, including their grain size, porewater and degree of compaction. Fine-grained facies generally have low resistivity values and cause attenuation of electromagnetic energy. Coarser-grained facies generally have higher resistivity values, generating reflectors within GPR data due to grain size differences. Seek the advice of an experienced geophysicist before commissioning a method.

• The main issue with all geophysical techniques is the acceptable compromise between resolution and depth of penetration. Surveys should be optimized to provide the required data; greater penetration and lower resolution for surveying large-scale structures, but shallower penetration and higher-resolution data for surveying small-scale complexities. Multiple geophysical techniques should be employed to avoid misinterpretation of a single data source, as advocated by Schrott & Sass (2008).

• A range of geophysical techniques exist that are suitable for use within both active and relict glacial and periglacial environments. Each method is associated with advantages and disadvantages, so certain techniques will be more suitable for use than others in certain conditions and some cannot be used under water (e.g. offshore). The most commonly employed are GPR, ERT and seismic methods.

• Reconnaissance surveying is useful to determine ground conditions before extensive surveying begins. Profiling is a commonly used approach, but gridded surveys enable modelling and constraint of structures in three dimensions. For more detailed surveying it is useful to collect in situ measurements such as from micro-resistivity, so enabling more robust interpretation and mapping of a facies.

• Land-based reconnaissance surveying using ERT enables likely facies to be identified from published data and therefore enables an informed decision on the applicability of other techniques (e.g. in fines-prone environments, seismic methods would be the optimal approach rather than GPR).

• GPR and ERT provide complimentary data that enable both sedimentary structures and their likely composition to be identified (e.g. Hambrey et al. 2008; Parkes et al. 2013). GPR should be used for identification of near-surface structure, but is prone to attenuation in fines-dominated facies. ERT enables mapping of facies and identification of large-scale structures; it is therefore a useful technique for reconnaissance surveying and for assessing the applicability of other techniques.

• Seismic methods enable identification of structure and are a suitable alternative technique should GPR be precluded due to unfavourable ground conditions (e.g. fines).

• It is critical to ‘ground truth’ geophysical data; it should be interpreted with the aid of geological evidence from shallow trenching and other exposures. Deeper features identified on geophysical surveys provide suitable targets for probing and drilling, enabling these to be directly related to geological reality.

7.4.1 Introduction

The description of soils and rocks in practice is a fundamental component of ground investigations, as it provides the written evidence that forms the field log. Whether this log is being prepared for engineering or sedimentological purposes makes no difference in terms of the process used for describing, but there are differences in some of the terminology such as detailed below. Field logs should record the materials and strata seen in a borehole, trial pit or any other type of investigation hole, or in an exposure such as a cliff or quarry face. The record presented as the field log includes descriptions of the strata encountered and a range of other field observations that provide additional information to that presented within the description alone. The descriptions within the field log are a basic element of the factual information that underpins the entire understanding and interpretation of the ground conditions at the site. The field log is therefore a basic building block in the compilation of the ground model.

It is important to remember that the field log is essentially a factual report of observations; classification as to the origin or type of soil or rock seen is a step that follows using the observations reported (i.e. the facts). Classification is not something that needs to be carried out in the field, although the logger can be thinking about these matters as part of the logging process; a good description enables a classification to be made from it, but the reverse need not be true. However, because of the complexity and variability of depositional assemblages in glacial and periglacial soils, sedimentologists have developed a shorthand set of facies classification terms to enable more rapid communication with other practitioners. Two sets of such classification terms are given in Table 7.5 and Figure 7.15.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.15.

A classification of poorly sorted sedimentary rocks (Moncrieff 1989).

There is nothing wrong with this, and new terms will keep being added into this set, but note that those in use are not universally accepted. Although initially useful, it is apparent that these shorthand notations cannot, and should never, replace a full and detailed description of either a sample or an exposure.

The descriptions contained within a field log provide:

• the reader with information to interpret the ground conditions:

  • for designers seeking information to guide the scheduling of field and laboratory tests, and/or

  • for co-workers who are interested in the sequence and conditions under which the ground was formed;

• a framework for the interpretation of the sequence;

• a framework for interpretation of the test results; and

• a model for the lateral and vertical interpolation and extrapolation of those results to the whole site.

In many cases it is the description that is relied upon to provide information on likely mechanical and hydraulic properties, especially when test results are either not available or cannot be made available because representative samples were absent. This is particularly the case in many glacial soils.

In civil engineering, many investigations commissioned by design teams concentrate on one or more aspects of the engineering problem, but most commonly the focus is on foundations. However, most investigations actually require information on a number of other problems, in particular the geological setting. Without this geological understanding, any predictions of ground conditions between investigation points are likely to be erroneous.

The descriptions are a key element in the preparation of the permanent works design, providing information that helps towards understanding how the ground will behave and interact with the structure. The field logs and descriptions also provide information that is critical to the contractors who have to carry out the works. The aspects that a contractor will be interested in may centre on the site preparation or the temporary works; their interest will include aspects of how the ground will respond to excavation, whether the materials are suitable for reuse and what support needs to be provided to the construction works.

The importance of a full field record, whether of a bag sample, a core or a trial pit is that the exposure is normally only seen by a single individual. The recording of all the information available from that exposure may or not be straightforward, but communication of the findings to those other parties who will use the information is a critical part of the exercise. The field log of the exposure is what the client who has commissioned the investigation is paying for, namely the collection of essentially factual data of ground conditions to inform the design process. The field log is also all that remains after the investigation has taken place, and after the trial pits have been backfilled and the borehole samples and cores discarded. The field log therefore has a life well beyond that of the investigation.

For all these reasons it can be seen that complete, accurate and systematic description of the soils and rocks is fundamental to all investigations; the compilation of the descriptions that appear on a field log must therefore be carried out by a field logging geologist or suitably qualified engineer. The system that should be followed in compiling these descriptions is set out at three levels:

• in the UK Standards BS EN ISO 14688-1, 14689-2 and 14689-1 based on the European standards;

• in the European standards EN ISO 14688-1, 14688-2 and 14689-1 (British Standards Institution 2003, 2013a, b), and

• in an explanatory text on soil and rock description (Norbury 2015).

Field logs prepared by a logger in the field should utilize a defined terminology in a standard word order as part of the systematic process of description. The purpose of a standardized word order is twofold. First, it enables the logger to work systematically and in an orderly fashion, and therefore to include all relevant aspects within the description. Second, the presentation of descriptions in a consistent word order is important to enable the reader to access the information. The familiar word order enables the key information and changes between strata to be more readily recognized. For example, the reader worried about the grading, the strength or the fissuring will know where to look for the information and to obtain the specific information that they are seeking. No preferred word order is included in EN ISO 14688 or 14689. The standard word order used in UK practice is in accordance with BS 5930:2015 as given in Table 7.6.

UK descriptive practice since 1999 has been to allow the division of a full description into more than one sentence with the intention of avoiding unduly complicated descriptions. If a description is complicated because of a long string of adjectival descriptors, it is likely that it will include an element of ambiguity.

The first sentence of a multiple-sentence description must name the principal soil type or name the rock and provide the quantification of the secondary constituents. The first sentence will also usually provide the overall information on the material. As many additional sentences as necessary are then used to provide any additional adjectival qualifiers on either the principal or the secondary constituents where required. This is particularly relevant in many glacial soils and the additional information can include description of the shape, grading, lithology or strength of particles. The additional sentences will also include details on the mass characteristics including the weathering and full descriptive information on the character of discontinuities and/or fabric features. Examples of multiple-sentence descriptions are given in Norbury (2015).

7.4.2 Grain size

In broad terms, the engineering geologists and sedimentologists use a similar grain size notation with the four main size fractions similarly defined. It is in the subdivisions of these size fractions that there are differences. The two sets of terms are given in Table 7.7.

7.4.3 Consistency

The consistency of soils is to be described according to EN ISO 14688-1 and is explained in Norbury (2015). Consistency is a descriptor applied in the field and is without reference to any tests to measure the strength of the soil. The terms for consistency of soils abut to the terms for the strength of rocks as shown in Table 7.8. The term hard should only be used for transported soils, such as many glacial soils, but for in situ material such as bedrock the rock strength terms should be used.

7.4.4 Widely graded soils

Glacial soils tend to have a wider range of particle sizes present than many laid down by other agents such as water or wind. Soils containing similar proportions of clay, silt, sand, gravel and even cobbles pose particular problems in soil description. The conventional split in describing the fine fraction (<63 µm) and the coarse fraction (63 µm–63 mm) separately is difficult to apply as the sand fraction (63 µm–2 mm) is usually bound in to the fine soil matrix. If the soil is dry, that is, stiff or very stiff in consistency, wetting the soil in order to properly assess the sand fraction is not easy, and is not often successfully made in the field. The importance of scheduling particle size analyses in these cases should be noted.

A solution to this problem was put forward by Spink & Norbury (1993) in describing the matrix and the clasts, and the proportions of each, separately; the matrix in this case would include the clay, silt and sand fractions. This approach has not been widely adopted, but offers a descriptive approach that is sympathetic to the engineering behaviour of the soils in question.

7.4.5 Coarse and very coarse fractions

The transport capacity of ice and related agents means that glacial soils include significant proportions of coarse and very coarse particles (2–63 mm and >63 mm, respectively). These need to be described in some detail in order to record information pertinent to their geological origin and to their engineering behaviour.

A common problem in the description of coarse and very coarse soils is accommodating the difference between volume and weight percentages. The human eye sees the proportions of sand and gravel by area percentage. The heap of sand and the pieces of gravel occupying the same space are seen as the same area proportions. However, the correct visual description is according to the weight percentage where the piece of gravel will weigh twice that of the same area of sand.

The description of coarse soils (gravels and sands) should be straightforward as representative samples are recoverable from most types of investigation hole in sufficient quantity for most tests; the particles can be seen by the naked eye and so are readily identified, measured and described.

Very coarse soils are not common in nature, which is fortunate as we are not able to sample or test them with confidence. Very coarse soil particles (cobbles and boulders) can be individually picked up and examined as hand specimens. Although the large particle size makes the description of these particles just like that of a hand specimen, implying that this task of description is straightforward, it must be remembered that most investigation methods cannot recover such particles intact. This raises two problems. First, the presence of very coarse particles needs to be inferred from indirect evidence, such as evidence of particle breakage, angularity and remarks by the driller. Second, the sample presented for logging will probably not be large enough to be representative of the in situ soil, in particular when the particles include those much larger than c. 100 mm.

Reference to the size of samples required by Standards for testing in the laboratory (e.g. for particle size grading) puts the minimum quantity to be sieved to obtain a representative grading curve at 200 kg if a cobble is present in the soil, or 1000 kg if a boulder is present. Although sieving of such samples is not practical, strictly speaking a similar quantity should be available to the logger for the description to be representative. It is not possible to recover very coarse particles from a standard borehole however, and the volume recovered may be considerably less than that theoretically required.

A common problem in very coarse soils is being clear as to which dimension it is that is measured. In the laboratory particle size analysis, the technician is required to manually offer the particles to the sieve aperture to see if it will fall through. The relevant dimension is the size of the square sieve aperture through which the particle would pass; the particle size therefore defined is not necessarily any of the maximum, average or minimum dimensions.

A range of details of the coarse and very coarse particles can usefully be included in a description. A scheme was suggested by Norbury (2015) and is presented in Table 7.9.

Detailed information such as lithology and shape is helpful in identifying the origin of the material included within the glacial deposits, in terms of material source and degree of transport. In addition, information on such aspects as striations, faceting and any preferred orientation (or absence of preference) of the particles can help in assessing which type of deposit is present.

In engineering geological practice, the description of coarse particles is normally limited to their roundness although, in some cases (e.g. where a particular shape dominates) this too can be given; these terms have been borrowed from sedimentological practice so there are no discrepancies here. In sedimentological practice, it is normal to describe the particle shape, roundness and sphericity.

7.4.6 Bedding

The thickness and spacing of beds should be recorded on the field log whenever seen. Terms for the description of these features are given in Norbury (2015). Bedding features are less common in glacial soils than many other types of soil but, where bedding is present such as in glaciomarine or glaciolacustrine deposits, it is important that it is recorded. The best-known forms of bedding are the laminated and varved deposits. The features that need to be recorded are the thickness of the layers and their variation in thickness; these provide important information on the rates of deposition. The varved nature is picked out either by different colours of material or by different materials; in the latter case it is essential to record the range of different materials present. The standard terms thickly laminated and thinly laminated (6–20 mm and <6 mm respectively) do not provide sufficient detail in describing such soils; the thickness of individual laminations should be measured and reported in millimetres. The terms used to describe the thickness of beds in sedimentological practice differ from those used in engineering geological practice. The two sets of terms are given in Table 7.10.

In engineering practice it is not usual to provide details of the form of any bedding, although any discernible geometry can be described. On the other hand, sedimentologists tend to provide considerable detail on the vertical and lateral variation of these features, together with the size and shape of any bedding forms present and contacts between the strata in order to provide the basis for interpretation of the depositional facies (Reading 1996; Collinson et al. 2006; Stow 2006). Descriptions for either engineering or sedimentological purposes would tend to use the same lexicon of terms so there is no conflict, but the level of detail and the terms used should be clearly defined and agreed before the fieldwork commences.

7.4.7 Discontinuities

The logging and recording of discontinuities in soils and rocks is covered at length in Norbury (2015, chapter 11). The discontinuities that may be seen and logged include bedding fractures, joints or fissures and shear surfaces. The information that should be recorded where available includes orientation, spacing, persistence, termination, roughness, wall strength, aperture, infilling and seepage. The amount of information that can be recorded depends on the size and quality of the exposure (sample, core, trial pit or cliff face).

Many glacial soils have been subjected to high stresses in their past, including those from freezing and desiccation, and higher than they are under now so are ‘over-consolidated’. As a result, many soils and rocks can include the tensile cracks termed joints; the synonymous term ‘fissure’ is often used in soils. These can be at any angle and of any form, and can include evidence of shearing. Low-angle shears are particularly important in glacial and periglacial soils. There are many references in the literature to slope failures related to such features; classical early works include Early & Skempton (1972) and Skempton et al. (1991). These features are often related to downslope movement taking place in the supersaturated soils at the base of the active layer. Finding such features is not easy; at Carsington (Chapter 1, Section 1.2, Case History 1.5) they could only be found after the soil in the trial pit walls had dried in the summer sun for several weeks (Skempton et al. 1991).

7.4.8 The soil–rock boundary

The key issue at the boundary between soil and rock materials is the identification of rockhead. However, rockhead can assume many forms and what it means depends on the purpose for which it was defined. Geological rockhead is here defined as the upper surface of in situ materials, that is, those without transport from their original position. Engineering rockhead is the boundary between materials that will behave as a soil and those that will behave as a rock. A number of rockhead levels may be possible depending on the criteria being adopted. The particular environments operating under glacial and periglacial conditions can give rise to much variation in rockhead conditions as follows.

• Advances of glacial ice can cause in situ deformation and disruption of bedrock beneath a till or glaciotectonite, removing many of the rock-like characters from the rock (such as bedding and cohesion) and even locally remoulding the disaggregated rock, so reducing it, in appearance and strength, to a soil.

• Rafts of bedrock can be transported by glaciers and now found embedded in ‘drift’ with clearly identifiable till beneath them.

• Outwash moraines can include angular cobbles and boulders of local bedrock, so underlain by the parent material.

• Periglacial soils can be weathered in situ by a combination of chemical and physical processes to provide a gradational weathered profile (e.g. as in brecciated bedrocks; see Chapter 5, Section 5.2.2.1).

In each of these situations, the actual identification of geological rockhead will not be straightforward, especially in sequences that are gradational. A rational basis for identifying decisions may be possible, given decent exposure, but identification from borehole evidence alone will be difficult and at times can be impossible; to complicate matters further, it is likely that the borehole recovery will be incomplete in both quantity and quality. The identification of engineering rockhead may be more straightforward given adequate exposure, but can be very difficult where exposure or core recovery is limited. Sampling for laboratory testing is likely to be difficult, if not impossible. Field observation and recording are therefore left as the usual means for identifying rockhead. It is clear that selection of the most appropriate method(s) of investigation is critical, and important to realize that more than one method may be required, including geophysics (Chroston et al. 1999; Coswill et al. 2016).

7.4.9 Expectations from the conceptual model

Derivation of the initial conceptual model was introduced in Sections 7.2.1, 7.2.3 and 7.2.5 and its evolution is described in detail in Sections 7.4.2.2 and 7.7. It is imperative that the field logger is aware of this model before going into the field, so that the logging becomes part of the continuous process of checking the model for agreement with reality. Any findings that do not agree with the model should prompt a review of the model itself and the field investigation programme, providing potential for updating both. Key features of the ground, relevant to its response to engineering, are unlikely to be recorded if logging is conducted outside the context of the ground model. With the benefit of a set of field logs that correctly and completely describe the ground exposed, the classification of the materials encountered is possible and the conceptual model can be updated.

7.5.1 Introduction: issues with the investigation of glacial and periglacial deposits

The complexity of deposits and spatial variability that can be found in glaciated or periglacial terrain warrants a holistic approach towards planning, managing and ultimately executing a ground investigation so that its desired objectives are achieved. This includes the appropriate balance of geophysical and intrusive investigation techniques along with geomorphological mapping and phased approaches to investigation. Within this, the importance of selecting appropriate intrusive technique(s) employed to obtain suitable samples, and the suitability and penetrative capability of in situ testing techniques, needs to be considered (Clayton et al. 1995).

The following section considers the range of intrusive ground investigation techniques that can be employed to study glacial and periglacial deposits, in both the onshore and offshore environment, and the selection of the best technique for the type of deposit being investigated and parameters to be established; design of the investigation, including suitability of sampling regimes, will also be considered. The section will also briefly refer to the benefits of using desk studies and developing conceptual site models for the site being investigated (see Section 7.2.1), as well as the possible need for expert knowledge to interpret the depositional environments and glacial/periglacial processes that may have been at work (Chapters 3–5); an understanding of these may well be needed from an early stage to ensure the ground investigation is a success.

7.5.2 Planning investigations

The inherent variability of glacial and periglacial deposits dictates the need for detailed planning and selection of intrusive investigation techniques. The scale, variability and nature of glacial deposits will be intrinsically linked to latitude, bedrock geology, history of environmental change and the resulting glacial processes that have affected the terrain under consideration. Previous chapters of this book explain in detail the role of these factors in determining the palaeoglacial and periglacial environment of the deposits seen today, and the reader is directed to the concept of the ‘landsystem’ approach (Chapters 3–5) for characterizing and predicting the deposits and landforms that may be encountered during an investigation. Additional constraints related to marine site investigations, including the nature of deposits that may be encountered, are provided in Section 7.2.5. Such foresight in planning an investigation may be achieved through a desk study and the development of a conceptual site model; the experience of specialist advice should be considered (see Sections 7.2 and 7.3). Examples of the conceptual models that should be considered are illustrated in Chapters 3 and 4. Figures 7.16 and 7.17 illustrate the variability of ground to be expected at the scale of a borehole and a local site.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.16.

Example of glacial lithostratigraphy (Szuman et al. 2013).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.17.

Stratigraphic architecture of channel gravels (Yorke et al. 2012).

7.5.3 Choice of ground investigation techniques and plant applicable to glacial and periglacial deposits

The suitable range of investigation techniques available can only be considered once the objectives of the investigation have been set and the investigation designer has understood the ground conditions and working environment likely to be encountered. This section briefly outlines the various intrusive investigation methods available to the investigation designer. Each method of intrusive investigation has specific benefits and limitations based on accessibility, achievable depth, the suitability for penetrating and sampling the types of deposits typical of glacial and periglacial terrains, and cost. Some techniques may be more suitable than others depending on the stage of investigation (preliminary, main or detailed) under consideration. An example of the difficulties that can be encountered with shell and auger drilling in glacial soils is recorded in the account from a driller (Fig. 7.18).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.18.

A cable percussion driller's experience of glacial deposits.

Table 7.11 compares, in general terms, the range of intrusive investigation methods available to the ground conditions that might be present based on the CSM, with specific interest in key glacial and periglacial deposits that might be encountered.

Tables 7.12, 7.13, 7.14, 7.15, 7.16, 7.17, 7.18, 7.19, 7.20, 7.21, 7.22, 7.23 provide a description of the method of each investigation technique, with its general and specific application and limitations to ground investigation, starting with the least sophisticated. An indication of the cost of each technique is provided, but it should be remembered that the cost of using each technique would also be determined by access restriction, the type of terrain that must be traversed to access the positions and commercial considerations (see Tables 7.24, 7.25, 7.26).

Tables 7.12, 7.13, 7.14, 7.15, 7.16, 7.17, 7.18, 7.19, 7.20, 7.21, 7.22 also discuss the use of each of the techniques against the specific constraints and geohazards of intrusive investigation in glacial and periglacial deposits. This includes the suitability of the samples that can be obtained by the technique together with the textural and structural details that can be observed in those samples for their use in detailed investigations of their formation, which may need to be considered for engineering purposes. Table 7.23 provides an indication of the plant and intrusive investigation techniques that may be appropriate for accessing sites in different terrains, from marsh land to offshore and in constrained positions within urban areas.

7.5.4 Offshore and nearshore intrusive investigations

Intrusive investigations in the marine environment will utilize many of the principal drilling and in situ test techniques detailed in Section 7.5.3, but will be principally limited to rotary coring, occasionally cable percussion, CPT and vibrocores for near-seabed deposits. The principal differences between onshore and marine SI equipment are mainly related to the delivery system and the size and depth capabilities of the drilling rigs used.

7.5.5 Resolution of data

The spatial frequency of points of intrusive investigation will be determined by the investigation's objectives, the available plant, the terrain on and around the site, and cost. The frequency of investigation points, and the sampling and testing regime chosen, will dictate the level of certainty in the data collected and the interpretations that can be derived from this. Greater levels of certainty would of course be preferred but are generally moderated by cost, and in part this should be accounted for by the phasing of investigations from preliminary studies to detailed investigations.

Table 7.26 provides examples of typical spacing of investigation points across a site as suggested by various codes of practice and guidance documents. The selection of intrusive investigation spacing should however be tailored to the specific ground conditions that may be encountered and the nature of the proposed development.

The designer of an investigation may plan the borehole locations based on the aims of the investigation and this may typically involve either an initial survey at regular spacing along a line, as in the case of a road, pipeline or cable route, or as a grid for developments covering a large area, offshore wind farms, etc. The investigation points may also be based on specific aspects of the engineering design, that is, the desired location of building foundations, escalators, underground junctions, depths of cut-off and so forth. However, because the specifics of the development design may change and because ground associated with glacial and periglacial processes can be so variable, deriving geotechnical parameters at point locations and then expanding this conceptually across the site, in the hope that this approach will predict the geology over the area and depth of the site, is extremely risky and rashly optimistic of success unless a detailed understanding of the depositional processes and spatial variation of the geology is known. The need for a CSM and expert knowledge when assessing the geology and geotechnical parameters of a whole site in glacial and periglacial ground must be remembered.

The investigation designer should utilize both expert knowledge, including the CSM as it develops, and the phased approach to site investigation when considering the location of investigation points. A grid or linear model is suitable where insufficient previous knowledge on the site is known and should be augmented by more localized and detailed investigation of points of interest as the investigation progresses. Expert knowledge can advise on the location of boreholes given the placing of likely landform-sediment associations in the CSM. While such an approach may be considered ‘intelligent’, it may tempt the designer into reducing the overall number of investigation points and should not be considered a complete substitution for a suitable number of exploratory hole positions specifically located to provide a degree of certainty in the dataset.

Where variable geology is suspected, the investigator has a number of further techniques available to decrease the level of uncertainty in the investigation. This can include the use of CPT for rapid assessment of soil types and the characterization of basic geotechnical parameters. Alternatively, geophysical assessment, including the use of electromagnetic resistance surveys of seismic reflection profiles, can be used to study the ground between borehole or CPT locations. Such approaches are conducive to the phased approach of investigation described above, and the benefit of their use in an integrated investigation is discussed further below.

The suggested spacing regime of intrusive investigation in glacial and periglacial terrains should ideally be at the tighter end of the ranges given in Table 7.26. Glacial and periglacial deposits and landscapes are among the most varied in the world in terms of composition and engineering behaviour, comparable in their variability with volcanic rocks. It is the responsibility of the investigation designer to ensure that the variability of the deposits on site is understood and to design the investigation appropriately in order to achieve the aims and objectives of the specific project with a level of certainty the project requires.

7.5.6 Choice of sampling and laboratory testing

Selection of appropriate sampling and testing regimes for geotechnical assessment of glacial and periglacial soils will ultimately be determined by the nature of the proposed development and parameters required for design and construction. The above sections outline the investigation plant required and discuss the merits of each technique in relation to the ground conditions that might be encountered. To these concerns should be added those of selecting the type and quality of the samples that can be recovered by each technique within any given ground conditions. Many suitable texts are available elsewhere for this purpose (e.g. Head 2006; Head & Epps 2011, 2013; Clayton & Smith 2013); however, in designing an investigation consideration must be given to the range and nature of samples that can be taken for geotechnical testing. In selecting a sampling regime, the investigation designer must consider the difference between recovering disturbed rather than intact or ‘undisturbed’ samples during the construction of exploratory holes.

7.5.6.2 Intact and undisturbed sampling

Total stress and porewater pressure at the base of a hole change during any drilling, almost regardless of technique used, and so the sample released and extracted from the ground is disturbed from its in situ state. Consequently, no samples are truly undisturbed. However, where detailed investigation requires a better than usual understanding of both the depositional and strain history of the soils, greater emphasis on recovering as ‘undisturbed’ a sample as possible is needed. This may be accomplished by increasing the number of driven open-tube-type samples with cable percussion (Table 7.17) or the selection of rotary-type drilling techniques from which intact specimens can be removed from the recovered core (Tables 7.18, 7.19). The sampling equipment used should, of course, be considered against the anticipated geology with a view to recovering as intact a sample as is possible within the budgetary constraints of the investigation. Due to the inherent variability of glacial deposits, local knowledge and experience in selecting appropriate sampling techniques to minimize disturbance should be used when available.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.19.

Example of detail that can be gained from an exposure. See Figure 7.34 for examples of microstructure in samples.

A classification of quality is used for undisturbed samples and the testing programme selected should consider this in terms of the ground conditions present and the plant used. Table 7.27 shows the ‘classes’ of undisturbed sample that can be achieved from cable percussion and rotary coring, and advice on the geotechnical analysis that can be undertaken on the recovered sample.

View this table:

• View inline
• View popup

Table 7.27.

Sampling quality

Class 1 samples are suitable for specialist investigation techniques that can aid the interpretation of till provenance and depositional process, and may prove important to the aims and objectives of a specialist investigation in glaciated terrain (see Section 7.5.7). Microscopic analysis of tills, including microfabric and micromorphological analysis, can be undertaken either on samples recovered from class 1 piston samples or good-quality rotary core. Macro-clast fabric techniques could also be applied to such samples. Orientation of the sample should be considered during the drilling technique wherever possible (a requirement which, when present, may restrict collection of samples to exposed faces; Fig. 7.19). Class 1 samples also provide a better medium than disturbed samples from which to extract clasts or smaller particles for visual and geochemical lithological analysis, particularly in thinly bedded or otherwise heterogeneous deposits where sediment provenance is unclear.

7.5.6.3 Sample screening

Class 1 undisturbed samples are necessary for determining properties in terms of effective stress and for more advanced testing including cyclic testing, something which is becoming routine in site investigations for the design of foundations for wind farms. Such advanced laboratory testing is both expensive and time-consuming.

Experience with glacial deposits has shown that, having carefully collected undisturbed samples and prepared them for testing and analysis, and having waited for the results, the final failure of the sample sometimes reveals internal structures and textures that are thought to be uncharacteristic of the majority of the deposit from which the sample was taken. It is then considered that the failure mechanism inferred from the test will differ from the wider failure mechanism of the sediment unit as a whole and the testing must be completed on other samples, at extra cost. In order to save costs, samples of till are now regularly screened using x-rays and scans before testing to detect the presence of small-scale textural and structural variations which can occur in a deposit that would have otherwise been described as a massive diamicton.

Screening undisturbed samples and whole core runs, particularly for tills, should include as a minimum the selection of subsamples for splitting and description to ascertain the general level of textural and structural variation present. The selection of the sample for advanced testing should then be made with suitable judgement.

Screening techniques include the use of x-ray and MRI scans to identify the internal structure of samples. Such approaches may appear costly, but must be balanced against savings made through the subsequent testing. The level of detail identified in such screening (Fig. 7.20) can be used later in the description of the sequence and may be beneficial to interpretation of depositional history and strain response of the deposit.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.20.

Three-angle X-ray images of a single core of till used to identify suitable samples for cyclic testing. Dark spotting marks clasts. Shading of matrix demonstrates density changes in the deposit (darker relates to greater sediment density). Note detailed bedding of fine-grained deposits at base, and zones of higher and lower density. Core length displayed is approximately 700 mm. Sample selection in such variable deposits may be critical to ensure accurate representation of the ground conditions.

If a screening regime is selected, care must be taken to assess the bias it creates in sample selection. While it allows large clasts, zones of textural variation and bedding structures in the sample to be avoided (as seen in the example in Fig. 7.20), the engineering geologist must be satisfied that the engineering properties of those zones are not governing the engineering characteristics of the deposit as a whole.

7.5.7 Specialist investigation

Glacial and periglacial terrains can record a complex arrangement of multiple erosional and depositional landforms and deposits reflecting spatially and temporally variable processes. It may be the case that, to understand these terrains for engineering purposes, the engineering geologist must understand the history of landscape-forming processes that have operated on site (see for example Ross 2013). In certain cases, it is necessary to use specialist knowledge and techniques more akin to those used by the academic glacial geologist or Quaternary scientist than those used by the ground engineer. Table 7.28 provides a brief overview of some of the techniques available and highlights the added value they can provide to a project.

Each of the techniques described may provide suitable added benefit to the investigation of glacial and periglacial terrains. However, the designer of the investigation must measure the cost of such approaches to the benefits they might bring to the investigation. Many techniques may be suitable for answering specific questions about deposits beneath a site that typical assessment by an engineering geologist or geotechnical engineer alone cannot answer. Such problems may only come to light after the preliminary or main phase of investigation has taken place, and such approaches may be considered as part of a detailed phase of targeted investigation.

When planning the design of an investigation in glaciated ground, the question must be asked whether the technical resources available are capable of answering the complicated engineering problems that complex glacial and periglacial deposits can pose. Figures 7.14, 7.16, 7.17 and 7.21 provide examples of the level of detail that may be required in describing glacial deposits to ensure that their engineering properties are fully understood. The designer of the investigation must assess whether an engineering assessment of deposits to BS5930 (2015) and Eurocode alone (and as described in Section 7.4 on soil and rock descriptions) will be sufficient for the level of detail required, or whether the use of specialist resources is warranted.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.21.

Stratigraphic profile from Usk valley, South Wales, demonstrating the typical level of detail required when visualizing glacial sequences. Example demonstrates the complexity of a glaciotectonite, multi-layered till sequence reflecting a major fluctuation in an ice margin during a single glacial episode. The sequence reflects an upwards transition of strain response in the sediment which has imparted distinctively different engineering properties to the different till units (Coleman 2007) (c, clay; z, silt; s, sand; g, gravel; c, cobble; vertical scale in centimetres).

7.5.8 Integration of site, geophysical and laboratory data

The most important fact regarding ground that has been glaciated and/or subject to periglaciation is that it is infinitely complex, challenging and frequently surprising. Many different intrusive site investigation techniques have been described together with their benefits and limitations. No one technique by itself is likely to be suitable for assessing a site and providing the understanding of geology and engineering parameters needed for any large engineering project.

Large-scale site investigations in glacial and periglacial deposits and terrains need to follow a flexible and integrated approach; Figure 7.22 illustrates one such cycle. Decisions on the location of exploratory holes were based on this geophysics and the preferred initial location of foundations. Initial exploratory holes identified a multi-till sequence over a very variable rockhead of chalk, potentially much more complex than had been foreseen. Downhole geophysics confirmed the variability in the till layers. A specialist in glacial geology was consulted and postulated the potential presence of buried tunnel valleys eroded into the bedrock which preserved much older glacial deposits than expected, all subsequently buried by younger till. The engineering parameters of these separate till units were obtained and a programme of CPT survey identified the extent of these deposits and erosional features across the site. It was recognized that the use of boreholes alone would not have identified the presence of the variable ground conditions, and that the integrated approach was appropriate and assisted with the selection of advanced testing. Another part of this chapter provides a review of geophysical investigations (Section 7.3) and the detailed logging of glacial and periglacial deposits (Section 7.4), and the approaches mentioned there should be incorporated into the scope of intrusive ground investigations and kept under review.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.22.

Conceptual approach and process pathway for assessment of geotechnical parameters and development of a detailed ground model for a major construction project in complex glacial terrain, incorporating an initial surface geophysical assessment, boreholes, CPT and laboratory testing programme. Based on work in the Irish Sea.

7.6.1 Introduction

Glacial and periglacial deposits are located at, or close to, the ground surface. They can be located within the built environment, and their associated groundwater is in contact with surface waters. From a hydraulic perspective, glacial and periglacial deposits may differ significantly from bedrock aquifers as a result of their potential to be strongly heterogeneous at all spatial scales. The impact of poorly permeable materials and their heterogeneity on heads and water flows, and on the approach adopted for the investigation of a site and the interpretation of its results, all need to be taken into account. Consequently, glacial and periglacial deposits should not be thought of in terms of ‘aquifer hydrogeology’ (see Section 7.6.4.5 as an example), a point considered further later. The approach described in this section addresses these issues, but note that it is not possible to describe a rigid approach applicable to all situations. The specific approach, spatial scale of the assessment and/or model, and level of detail required should be commensurate with the ‘problem to be solved’. While the general approach described in this section covers the activities that might be undertaken for the largest scale and most complex sites and situations, these activities may not be relevant for smaller-scale sites and local situations. This is illustrated in Table 7.29, although it should be noted that this does not constitute specific recommendations. Some characteristics and approaches are relevant to all sites and spatial scales.

This section explains these key behaviours and observations and describes practical approaches that can be used to measure, interpret and model the overall (upscaled) hydraulic properties and behaviour of heterogeneous glacial and periglacial deposits.

7.6.2 Approach

A ‘whole-system’ approach starts at the broadest scale, considering the site within its wider geological and geomorphological setting then focusing in on the site and adjacent areas of interest. The objective is to progressively describe the water balance, that is, where does water enter the system, where does it flow through, where does it discharge and what are its flow rates? The focus is therefore on describing how the system behaves. For some problems, such as the example in Table 7.29 of infiltration (seepage) into an excavation, it may be concluded that only the local conditions are important and the wider system does not need to be considered further, although this need not always be the case.

The whole-system approach starts with an initial desk study (Section 7.2.1) to develop a preliminary conceptual model (Section 7.6.3) that can be used to undertake initial analyses, underpin testing of broad concepts using numerical modelling, and identify and specify the site investigation requirements. Analysis of existing data may form an important part of the desk study. Similarly, an understanding of hydrogeology and relevant analytical approaches in glacial and periglacial deposits is important for the design and execution of a site investigation. Data analysis is described in Section 7.6.4, and is relevant to all sites and spatial scales.

The specific approach to site investigation and then modelling activities will depend on the objectives of the investigation and the site conditions, in particular whether groundwater flow is of interest (e.g. for water management and engineering) or whether groundwater quality is also important (as in contaminant transport). The key difference between flow and transport problems is that transport problems typically require a greater spatial density of borehole and monitoring well data than a study of flow, due to the effects of heterogeneity on local flow paths and hence the direction and speed of contaminant migration.

Once site investigation activities are complete the results can be used to update the preliminary conceptual model of the site, although there may still be significant uncertainties at this stage. The objective of subsequent numerical modelling and analysis is two-fold: firstly, to test the conceptual model in order to explore and reduce uncertainties through an iterative process of numerical modelling and updates; and secondly, to apply the numerical models to answer the questions of interest and so achieve the project objectives.

7.6.3 Desk study and preliminary conceptual model

The objective of the desk study here is to develop at an early stage in the investigation a preliminary conceptual model for the hydrogeology that can be used to inform the subsequent design of a site investigation and any preliminary groundwater modelling that is appropriate, for example, to test broad concepts and define the likely boundaries to the system. Relevant sources of information available to the desk study will include photos and site plans, maps (land use and topographical, soil type, vegetation and geological and potentially also hydrogeological); existing borehole and potentially also river flow and catchment hydrological data; and evidence from site walkovers and geophysics. Some sense of the way in which glacial and periglacial processes can govern the distribution and patterns of groundwater flow can be gained from the following cases: drainage where relict periglacial shears are present (Davies et al. 2003); groundwater associated with the origin of ramparted depressions, interpreted as relict frost mounds (Ross et al. 2011; Ross 2013); flow in frost-patterned ground where a polygonal pattern of sand-filled ice-wedge pseudomorphs penetrating till directed water into a nearby pipeline trench (Morgan 1971); and in brecciated bedrocks, which are widespread in chalk and other fine-grained limestones and in other rocks, where the brecciated weathered profile has the potential to strongly affect groundwater flow.

The first stage in developing the preliminary conceptual model is to consider the regional or wider setting of the site (e.g. is it located on high ground that is likely to be an area of groundwater recharge, or low-lying ground such as the base of a valley that may be an area of groundwater discharge), and to identify its natural hydraulic boundaries (e.g. topographic divides or ridge lines, rivers, streams, ditches and the sea). All these will help in identifying the direction of groundwater flow to establish where recharge enters and exits the system as groundwater. Areas of discharge such as springs, streams, a river or the sea, are more obvious to identify than areas of recharge; maps of streams and ditches, when used in conjunction with geological knowledge, are invaluable for this purpose.

Figure 7.23 illustrates how information of surface hydrology can guide an understanding of a hydrogeological system. If ‘spring A’ (Fig. 7.23) is present, then it suggests the deposit labelled glacial ‘A’ is less permeable than the more extensive formation on which it sits and upwards flow towards ground level can be expected. If deposit ‘A’ is glaciofluvial, it may be sufficiently permeable to be hydrogeologically indistinguishable from the underlying formation, but the presence or absence of spring ‘A’ will be a guide.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.23.

Schematic cross-sections of some typical glacial-on-bedrock configurations.

It is reasonable to assume that the direction of flow in glacial deposit ‘B’ (Fig. 7.23) will be downwards. The presence or absence of spring ‘B’ and its location on the slope are good guides to the permeability of deposit ‘B’; the higher up the slope spring ‘B’ emerges, the less permeable is deposit ‘B’. The presence of springs on the slope between glacial deposits ‘A’ and ‘B’ will also indicate the likely location of the water table in the more extensive formation on which ‘A’ and ‘B’ sit.

Streams and rivers incised into ‘sheet-like’ glacial deposits (see Chapters 3, Section 3.8, and Chapter 4, Sections 4.1.1 and 4.1.2 for examples of the ground to be expected in various glacial landscapes) are relatively common in Britain, and often result in configurations such as shown in Figure 7.23. This situation is complex because there must be interaction between a regional groundwater system in the underlying aquifer and a local system in the glacial drift. The aquifer will be confined and will have a piezometric surface within the glacial drift. The location of the piezometric surface is not readily deduced from the available information. How much of the glacial drift experiences downwards flow (from surface recharge) and how much of it experiences upwards flow (from the underlying aquifer) is therefore unknown, as is the elevation at which downflow and upflow are balanced.

The configuration in Figure 7.23c is a multiple version of Figure 7.23a & b but includes a coastline where the amount of discharge is unmeasureable, and comments on how the multiple groundwater systems might interact. Note that the direction of groundwater flow may vary with depth, with shallow flows reflecting the local topography and hydraulic boundaries and deeper flows reflecting the larger-scale topography and associated hydraulic boundaries.

A preliminary conceptual model of the types shown in Figure 7.23 is essential for assessing the potential value to an investigation of surface geophysics, boreholes at any location and depth, surface flow gauging and the possibility of achieving a meaningful water balance. The preliminary conceptual model is also the basis for identifying and managing uncertainties, including: significant uncertainties that need to be reduced through site investigation and/or modelling; uncertainties that are not significant; and uncertainties that cannot be removed and whose potential implications are such that they will need to be managed.

In trying to make an initial estimate of how a groundwater system in glacial sediments might function, it is useful to bear in mind the following basic concepts.

1. The effective permeability of layered sediments is directional (Fig. 7.24). Where gravels are inter-layered with clays, the difference is extreme. Where sediments and their stratification are ‘disorganized’, the geometric mean of the individual components is the most appropriate large-scale value to use (Fig. 7.24b).

2. The maximum rate at which rain can infiltrate into soil or rock is equal to the value of its fully saturated hydraulic conductivity in the vertical direction. Average rainfall in northern Britain is about 1 m a⁻¹ (equivalent to 3 × 10⁻⁸ m s⁻¹) so soils and rocks with a value of hydraulic conductivity below about 1 × 10⁻⁸ m s⁻¹ are likely to have water either permanently on or close to their surface (taking account of topographic expression and recent weather at the time of observation). It is therefore reasonable to consider formations with a permeability less than 1 × 10⁻⁸ m s⁻¹ as having a ‘low permeability’.

3. Where groundwater crosses formation boundaries, it is refracted towards the normal in the low-permeability formation (cf. the law of refraction). In glacial sediments, the difference in values of permeability between adjacent sediments can be so large that it is reasonably accurate to assume that groundwater flows essentially parallel to formation boundaries in high-permeability formations and almost directly across low-permeability formations. Curvilinear flow lines should not occur in conceptual models of low-permeability formations, except immediately adjacent to discharge boundaries such as cuttings and deep foundations.

4. The lower the permeability of a formation, the closer to the ground surface is the likely location of the water table.

5. Perched water tables are particular features of permeable deposits in locations such as the tops of hills and crests of inter-fluves. In Britain, the high-permeability elements of glacial deposits in which perched water could develop are seldom extensive, thick or high enough to enable perched water tables to form to any noticeable extent. Claims of their existence should be treated with scepticism. In lower ground where more permeable glaciofluvial deposits can accumulate to a greater extent, the larger-scale (regional) water table (or piezometric surface) is likely to be close to at or above ground level and unsaturated conditions are unlikely unless abstraction is lowering the water table artificially.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.24.

The dependence of large-scale permeability on direction of flow and the degree of sedimentary organization.

The approach described above can be applied to problems at all spatial scales, for example ranging from a field where water drains to perimeter ditches to much larger areas such as with landfill or industrial sites, where the site and areas downstream are of interest.

For larger-scale sites and situations where long-term management and monitoring of groundwater is required, a useful and often neglected source of information for establishing the site water balance is hydrological data. Typically, data will not be available at the site scale (e.g. site streams will not have been gauged). However, flow and catchment water balance information may be available for major rivers, and this can provide some important insights into the regional hydrogeology and site hydrogeological setting. For example, a river hydrograph exhibiting rapid response to a rainfall event (i.e. ‘flashy’ behaviour) suggests that the glacial and periglacial deposits in the catchment are of low permeability and possibly saturated, while a slower runoff response suggests water can infiltrate, be stored and slowly released through the glacial and periglacial deposits. However, care is required when analysing river hydrographs because the rate at which rainfall drains to river systems is also influenced by other factors, including the average flow distance to enter the river system.

7.6.4 Data analysis

Existing site investigation data and other reports and publications may be available at the desk study stage; if so, they should be used to plan any site investigation that is to follow. The following discussion applies to problems at all spatial scales.

7.6.4.1 Boreholes and completions

It is important to know the type of materials that are intersected by the well screen and piezometer. In glacial deposits that have rapidly varying characteristics, and therefore significant heterogeneity, this can be uncertain. The well screen or piezometer could have been installed in a clay, silt, gravel or some combination of these. In the field, certainty about the nature of the formation being penetrated by drilling is a function of the method used, but may also be influenced by the experience of the driller and any specification the driller has been asked to follow (see Section 7.5). For desk studies using existing data, it should be noted that a variety of drilling methods and specifications may have been used at different times. Field records may also be of variable quality due to the subjective interpretation of different operators. This particularly applies to lithological logs (see Section 7.4).

The key to a reliable groundwater investigation is the integrity of borehole completions. There are guidelines on installing monitoring zones in boreholes (e.g. Aller et al. 1991; Fretwell et al. 2006) (Fig. 7.25). It is most important to know in which horizon the monitoring zone is located; this should ideally be within a single lithology, particularly if it is to be used for hydrogeological testing or a monitoring zone or test interval. If it contains layers of silt and gravel, the results obtained from testing will only yield hydrogeological information about the gravel.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.25.

Borehole completions involving monitoring zones. (a) Typical recommended completion. (b) Some common causes of leakage. (c) Monitoring zone altering local flow.

Monitoring zones are commonly sealed with bentonite pellets; however, experience indicates that these seals commonly leak (Fig. 7.25), particularly if the zone is more than 10 m below ground or the formation is of very low permeability. If the seals leak, hydrogeological tests may simply become a test of the seal permeability rather than of the permeability of the ground. The piezometer will not provide a reliable measure for either groundwater head or quality when vertical hydraulic gradients are present; as previously mentioned, flow through low-permeability deposits is dominantly vertical. There is no simple test for seal integrity except that alluded to by Black (2010).

A particular problem of monitoring zones in glacial formations is their ability to participate in the regime of local vertical flow (Fig. 7.25). Where a subhorizontal glacial formation is of low permeability, the natural groundwater flow regime will be vertically across it and usually downwards unless directed otherwise by an excavation operating as a sink. The gravel pack around a screen is of very high permeability relative to the ground, and will act as a local conduit gathering flow from its upper end and discharging it at its lower end. Some sort of water level will be monitored based on an averaging of the total heads involved and dependent on the adjacent values of transmissivity; the upper part of the gravel pack may possibly be unsaturated, which is a circumstance easily misinterpreted as a water table. Such through-flow may also have the effect of locally changing water chemistry. These problems are partially avoided by placing short screens towards the bottom of short gravel packs and attempting to locate monitoring zones within the more permeable parts of a heterogeneous glacial sequence.

Even though multiple completions within a single borehole are useful, because vertical groundwater flow is common leakage from one zone to another is also very common. Single completions are preferred, unless budgets allow extreme care to be taken when installing multiple completions and tests to be undertaken to check their integrity.

7.6.4.2 Permeability testing

Glacial deposits are not well suited to large-scale pumping tests since they are often not of sufficient thickness or do not have significantly extensive transmissive formations. Further, it is difficult to ensure that monitoring zones are within the same formation as that pumped since unconnected, underlying confined formations will, when saturated, respond equally well to pumping, thus imitating a connected zone.

Single borehole tests are the general rule, conducted ideally in a monitoring zone located within a single known lithology. Testing during drilling is not recommended unless the test zone is specifically isolated using a packer. Cable-tool casing through a clay or silt formation does not form a seal sufficient for reliable testing.

The most common single borehole test type is the slug test (also known as the falling/rising head test) but it has limits of application that need to be adhered to (Fig. 7.26). Constant rate and constant head (usually injection) tests can be used in order to test a larger region, but are more complicated to perform than variable head tests where gravity is the dominant (and constant) driving force. If applied to horizons near ground level, constant head or rate tests may affect the water table and unsaturated zone significantly and yield results that cannot be reproduced.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.26.

Range of values of hydraulic conductivity that can be measured by slug tests.

Analytical methods vary but are summarized in: Butler (1997); BSI 5930 (British Standards Institution 2015); and ISO 22282-2: (International Organisation for Standardisation 2012). They are of two types: one allowing for near-well storage of water (e.g. Cooper et al. 1967); and the other without (e.g. Hvorslev 1951; Bouwer & Rice 1976). The advantage of the analyses that include storage is that the region of influence of the test can be estimated.

For pumping tests, standard hydrogeological texts propose that the region of influence is the nearest distance from the tested borehole at which there is negligible head change. This is not realistic or practicable for slug tests, since any change of properties must be both recorded and recognized in data acquired at the source borehole. In order to calculate a realistic ‘region of influence’, Guyonnet et al. (1993) posed the question, ‘How close would a planar barrier boundary have to be in order to be recognised in a slug test?’ Recognition is the key, in that the divergence from theoretical behaviour must be maintained for a sufficient length of time and be large enough to be distinguished from noise. The idea and the calculation are shown in Figure 7.27 for a 150 mm diameter standpipe in a 200 mm diameter borehole in a water table formation. It can be seen that the region is extremely small and is virtually entirely dependent on the estimate of formation storage. Figure 7.27 reveals nothing about how long it would take for the deviation to take place, but timing is directly proportional to formation permeability.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.27.

An approach to estimating the ‘range’ of a slug test (based on Guyonnet et al. 1993). Note that r_t is the radius of the tubing in which the water level fluctuates; S: storage coefficient.

In unconsolidated homogeneous glacial sediments, bag samples can be collected and reconstituted in the laboratory to densities similar to those in the field in order to provide reasonable results for placing the results from field tests in context. Naturally, it would be exceedingly difficult to recreate from a bag sample the heterogeneity of an in situ sediment.

7.6.4.3 Interpreting permeability measurements

Most projects investigating glacial deposits rely on many small-scale, single borehole and laboratory permeability tests that yield results that vary over a wide range yet need to be consolidated into single values or distributions for modelling purposes. A simple procedure for doing this, based on plotting the cumulative rather than frequency distribution of results, is outlined in Figure 7.28a. All the results for a given formation (Formation A) are assembled in terms of test length and permeability and sorted according to permeability (Column B in Fig. 7.28a). The total length is calculated together with the cumulative length in terms of a fraction of the total length (Column D). The key manipulation is to use the standard Excel function of the inverse of the standard normal cumulative distribution to arrange the results according to six equal-area standard deviations (i.e. using ‘+3’ to plot 6 standard deviations). The resulting plot (Fig. 7.28b) is valuable for a number of reasons. Firstly, a pure log-normal distribution will plot as a straight line; the steeper the line the narrower the distribution. Secondly, for a normal distribution the median equals the geometric mean (note, as previously described, in ‘disorganized’ deposits the average permeability is equal to the geometric mean; Fig. 7.24b). Thirdly, the harmonic and arithmetic means are equal to the median plus and minus one standard deviation (Fig. 7.28b).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.28.

Assembling and presenting permeability results. (a) Spreadsheet to produce (b) an equal area standard normal cumulative distribution plot. Note that the use of the ‘+3’ term in Column E produces a plot showing 6 standard deviations.

Using the above method, it is possible to assess whether or not the glacial deposit is likely to be a single formation for hydrogeological purposes. It is also easy to decide on a permeability value to apply to the deposit where flow is either parallel to bedding or across it. Finally, it is possible to incorporate tests where no result was possible because the test zone was below the measurement limit of the equipment used (i.e. Formation B in Fig. 7.28b).

7.6.4.5 Interpreting head measurements

Hydrogeology evolved as a method for quantifying water supply and, as a result, focused on understanding aquifers rather than their bounding low-permeability formations of the sort found in many glacial deposits. It is often assumed that an aquifer is a formation containing negligible vertical gradients of head, so that a monitoring zone placed anywhere within a single vertical borehole will register the same water level. Unfortunately, confusion results when this idea is applied to poorly permeable deposits where vertical head gradients are almost always present.

Figure 7.29a illustrates the form of a typical result from many ground investigations in glacial deposits that is often misinterpreted as a system of perched water tables (Fig. 7.29b). Given the nature of many glacial deposits, an equally plausible and possibly more likely interpretation of the observed water levels that should be considered is the fully saturated concept. This is explained in Figure 7.29c & d, which shows how the discrete evidence of ‘perched’ water can be placed into a continuous field of downwards flow across a low-permeability deposit; as it flows down it uses up the potential energy it possesses by virtue of its elevation. Hence, it has only the potential of its elevation at any elevation where it is measured, so the water level measured at any elevation will equal its elevation head. Here driller's records can be of help, especially if the material recovered from BH2 and 3 (Fig. 7.29a) is wet and ‘seepages’ are noted as more permeable seams are penetrated.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.29.

Alternative concepts to explain the variation in water level measurement at a site involving glacial deposits of unknown properties. (a) Schematic of water level measurements; (b) and (c) alternative concepts; and (d) explanation of head variation in fully saturated concept (i.e. (c).

In theory, perched water systems should register heads above the elevation of the base of the saturated section from which they are measured and register no heads in the unsaturated zone between the perched levels. In practice, the generally low permeability of many glacial deposits means that monitoring zones can remain ‘dry’ for long periods of time (especially when the clayey deposits expand into the void created by drilling, locally reducing their porewater pressures and effectively ‘sucking’ water into their de-stressed zone), and it is easy to believe that they are located within unsaturated zones. In addition, the impact of vertical flow on water levels within borehole monitoring zones (Fig. 7.25) also gives the impression of an open water surface, again causing confusion.

As a first approach to interpretation, the following should be borne in mind when working with glacial sediments.

1. The concept of perched water is one associated with ground of high permeability and not the permeability of clay-rich sediments.

2. Lenses of cobbles or gravel within a silt or sand do not constitute highly permeable ground because they are lenses, and their configuration does not produce the continuity required for ‘highly permeable ground’.

3. The upper surface of a perched water level is a water table as the pressure of its water there is atmospheric; above it the water pressure is less than atmospheric. The water in ground below the base of the saturated body of a perched water level is also at a pressure less than atmospheric until atmospheric pressure enters the ground at that elevation.

4. Groundwater systems are continuous as is their field of elevation and pressure head, and the pore pressures within that field can vary from less than atmospheric to more than atmospheric.

5. Perched water tables of any appreciable extent require ‘headroom’ within the highly permeable material in which they sit in order to absorb recharge and not end up coalescing to form the alternative concept, a zone of fully saturated vertical flow. This also means that the perched water concept is normally inappropriate in landscapes of low relief.

There are two methods to distinguish between the two concepts of perched or continuous.

The first method is directly related to Figure 7.29, in that the potential at any point is equal to the elevation of that point. A plot of head (as water level) against elevation of the measurement point should yield a 1:1 relationship. Figure 7.30 shows results from a limited investigation of an old landfill in northern England. The local configuration is similar to that shown in Figure 7.23a with the landfill (an old brick-pit) located on the top of the hill.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.30.

Plot of water level variation as head v. the elevation of the screened interval.

The blue ‘boxes’ show the average water level plotted v. water level range, and thus the range in elevation of the recorded water level in the screened intervals. The key aspect of this representation is that all the water levels, barring that above the Magnesian Limestone interval, occur within the screened interval. This implies a saturated system of downwards-flowing groundwater that consumes all its excess potential energy in flowing from the water table near the surface to the elevation of the screened interval (as in Fig. 7.29d). If this was a series of perched water levels, then at least one could be expected to occur above the screened interval. The Magnesian Limestone is a confined formation with groundwater at the indicated value of head. There is a river c. 1 km away at an elevation of c. 45 m OD to which the whole system would appear to be draining. The position of the water level within the screened interval is a function of the rate at which water comes into the more permeable gravel packed interval v. the rate it exits. If the formation at the top of the interval is more permeable than that at the bottom, then the water level will settle towards the top of the interval (e.g. the highest glacial till (sand) interval) and vice versa (e.g. the lowest glacial till (sand) interval). The long open interval in the landfill with a water level towards its base is interpreted as meaning that the landfill was more permeable than its glacial drift surroundings and was draining into the underlying glacial sand. In general, Figure 7.30 reveals that interpretation of water levels is seldom clear-cut; knowledge of the wider regime often provides clarity.

The second method of distinguishing between these concepts depends on having time-series data of water-level fluctuations. It depends on the concept that, when there is no recharge, a water table will decline at a rate proportional to its height above its discharge point and that this decline will be an exponential decay whose slope can be extrapolated (Olin & Svensson 1992). The problem with applying this concept traditionally lay in being unable to wait long enough to observe a sufficiently dry period for a recession curve to develop from which a ‘discharge point or base level’ could be extrapolated. That problem has been partly solved by the advent of accurate pressure transducers and large-capacity data storage. The method is summarized in Figure 7.31 and relies on the fact that the water level will attempt to decline at the highest rate possible all the time and will do so during any period of zero recharge, however short. By using frequent measurements, the rate of change can therefore be calculated for every two data points and generate a dataset of ‘spot rates’ v. elevation. This is plotted (as a cloud) in Figure 7.31c. Only the ‘rate-of-fall’ data are of interest, since the ‘rate-of-rise’ data is entirely dependent on the rate of rainfall/recharge.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.31.

Using detailed short-term hydrograph data to predict long-term base level.

As shown in Figure 7.31, the water level can decline at any rate but cannot exceed some maximum rate that is related to its base level. This maximum rate should form a reasonably linear envelope that can be projected to zero rate, which is the base level of the groundwater system under scrutiny. In the case of perched water levels the base level indicated should coincide with some identifiable low-permeability layer. Note that the period shown in Figure 7.31 was followed by a long period of near drought and water-level decline, which confirmed the projection shown in Figure 7.31.

A further benefit of the ‘spot rate’ method is that water table conditions can be separated from confined conditions on the basis of how they behave. In the case of water tables, the envelope to their rate-of-fall is roughly linear and projects to the base level (as in Fig. 7.31c). On the rate-of-rise side of the zero line (i.e. the positive side), virtually any rate is possible and the ‘spot rate’ cloud is distinctly asymmetric (Fig. 7.32b). In contrast, where ground water is confined, the rate of change of pressure is a function of the elasticity of the formation and its water, and the way both distort under the difference between overburden load and change in pore pressure. This results in a symmetrical ‘spot rate’ cloud (Fig. 7.32a). Clearly, the groundwater hydrographs yield no such definitive information.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.32.

Characteristic responses of water tables and confined conditions in terms of ‘spot rates’.

7.6.5 Site investigation

The design of a site investigation obviously depends on the amount of existing information and the purpose of the project. The spatial and temporal scale of the investigation and the required data types should be consistent with the problem to be solved.

Site investigation requirements are most exacting in situations where contaminant transport is of interest, because groundwater velocities are important. These are difficult to evaluate in mixed glacial deposits unless the contaminant has already spread with the groundwater and the input point (or source) is known together with the likely date of the contaminant's release. Even then, for the ground investigation a high density of monitoring points may be required to delineate contaminant plumes in three dimensions and identify the influence of heterogeneity on transport paths. Pre-investigation modelling to assess the sensitivity of outcomes to inputs based on existing field measurements, and to establish useable boundary conditions in terms of the extent and depth of an optimum investigation, is valuable.

The bulk of any hydrogeological investigation will focus on boreholes and their completions. The following specific requirements are relevant to glacial and periglacial materials:

• piezometers should be installed at different elevations, typically including installation in bedrock and/or below the elevation of the major hydraulic boundary(ies) to which groundwater is thought to be draining;

• the lateral distance between piezometers at different elevations should be sufficiently small to ensure that subhorizontal and vertical gradients can be differentiated;

• short monitoring zones (piezometers) should be used, located within a single lithology; and

• single borehole hydraulic tests should be commissioned.

In addition, consideration should be given to having two phases of ground investigation. Although two phases of ground investigation will result in additional mobilization costs, and potentially also additional time to complete the ground investigation, the overall costs might be reduced considerably because the conceptual model can be developed more rapidly and the investigation can be better focused on what might be a complex environment. Phase 1 piezometers should be used to identify the dominant flow directions, vertical gradients and establish the outline conceptual model. The conceptual model can then be refined and used to focus Phase 2 of the ground investigation, which will be targeted on providing detailed information in the area for construction, or down-gradient groundwater quality monitoring. Two phases of ground investigation will be particularly beneficial for situations where contaminant transport is of interest, and will provide the opportunity for an iteration of modelling between phases.

Choosing a borehole diameter raises a fundamental problem in formations of low permeability: a large diameter allows easy groundwater sampling and the possibility of installing multiple measuring points, but mitigates against a highly responsive groundwater hydrograph and may involve long times to gain samples. This problem further supports an approach involving two phases of ground investigation: Phase 1 boreholes have single installations in narrow diameter boreholes; and Phase 2 boreholes are similar or of larger diameter with multiple installations, depending on the Phase 1 findings.

In poorly permeable deposits the seasonal variation in heads and the water table may be significant, for example of the order 1–2 m, except where the summer water table is very close to the ground surface. The magnitude of this variation and the specific change in the elevation of the water table may be important for engineering design. The requirement for time series monitoring may therefore have to be factored into the site investigation plan.

The limitations of single borehole tests, principally slug tests, have been described above. The implications of uncertainty in the site water balance are likely to be smaller than the implications of uncertainty in the hydrogeological tests. This shows the value of defining the site water balance and, for situations where groundwater flow through the site and beyond is of interest, it may be appropriate to collect weather and surface water flow data to support the water balance analysis. The components of the site water balance are illustrated in Figure 7.33.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.33.

Components of the site water balance.

Groundwater recharge can be estimated in a number of ways. The simplest method is to estimate hydrologically effective rainfall (HER) from weather station data, taking into account the nature of the land use. However, for low-permeability deposits and topographic lows, where the water table is close to the ground surface, recharge may be significantly less than HER may predict.

Where regional groundwater that flows into and out of a surface water catchment can be estimated, groundwater recharge can be approximated as the difference between groundwater inflow and outflow, plus the slowflow (i.e. baseflow) component of the catchment hydrograph, assuming there is no significant change in storage. If a more accurate estimate of recharge is necessary, a soil moisture balance model can be developed, where recharge is the residual when evapotranspiration, surface water runoff and storage in the unsaturated zone are subtracted from precipitation. Storage of water in the unsaturated zone and travel times of water to the water table are the most difficult parts of the water balance to quantify, particularly in heterogeneous deposits where preferential flow paths (e.g. fractures in till) or lateral flow paths are important. This level of detailed estimation will rarely be necessary, but methods for calculating recharge in low permeability and layered systems are available (Cuthbert et al. 2009, 2010, 2013).

In some ground investigations, it may be useful to collect data on the physical properties of the subsurface using non-invasive geophysical methods (Section 7.3) to aid interpolation of geological features between boreholes. There is a strong contrast in electrical resistivity between clays and sands, which correlates with the low- and high-permeability layers in glacial deposits. Geophysical methods that measure the electrical properties of the subsurface, such as electrical resistivity tomography (ERT) and electromagnetic surveys and ground-penetrating radar (GPR) are therefore particularly useful for investigations in heterogeneous glacial materials; ERT is most useful where distinct clay and sand-gravel deposits are present, but may be less conclusive where mixed deposits such as silty sands are present. Results may be affected by the extent to which deposits have been over-consolidated. The depth penetration of GPR is low in saturated, clay-rich soils. Surveys can be designed to collect information from the 1 m to 100 m scale, depending on the amount of information required. For example, ERT with a horizontal resolution of 5 m and a vertical resolution of 0.5 m to a depth of 5 m has been used in the UK to map the extent of glacial till sheets. This level of detail would not be appropriate in many site investigations, but could be useful if a detailed knowledge of flow paths was necessary.

Geophysical data often have to undergo an amount of post-processing to interpret the measurements in terms of geological units (Section 7.3). The most useful information is often the contrast in properties rather than the actual values, and it is therefore important to use some borehole or auger data to ground-truth the measurements and assign properties to geological units. Data should be carefully studied to understand any possible anomalies that could arise from the presence of man-made structures and metal objects, particularly on brown-field sites.

7.6.6 Numerical model

The site investigation results will enable the preliminary conceptual model to be updated and, if the site investigation data suggest that the preliminary conceptual model is wrong, then careful consideration and possibly further investigation will be required to understand where the problem lies and how the conceptual model can be corrected. The conceptual model can be tested by numerical modelling and further refined by additional ground investigations and/or iterations of modelling and investigation. The extent to which uncertainties need to be reduced will depend on the objectives of the modelling and the potential implications of the uncertainties. Future development of the conceptual model will therefore focus on the key uncertainties that need to be addressed and on scoping the potential implications of the uncertainties that remain.

7.6.7 Applying the model

Once the model has been demonstrated to adequately reproduce observations, it can be applied to the purposes for which it was designed. As previously described, the end-uses can be divided into groundwater flow problems (water resources and engineering), and flow and transport problems (groundwater quality and contaminant transport).

7.6.8 Summary

Perhaps the most important and possibly surprising message to convey is that glacial and periglacial deposits should not be thought of in the usual ‘aquifer terms’ as, in most cases, they will behave en masse as materials of low permeability. They are much more complex than the aquifers normally exploited for water supply, hence the note of caution concerning the application of ‘water supply experience’ to these materials. As a consequence, their overall key behaviour should be identified as early as possible in any study of their hydrogeology; the observations required for this have been explained, and practical approaches described. With the aid of these concepts, hydrological data from these materials can be used to measure, interpret, develop and apply models appropriate for their glacial and periglacial origins. Previous chapters have graphically illustrated how variable these origins can be and how the product of one glaciation can be affected by the glaciations that follow; this great variability means that the specific approach appropriate for a given situation will depend on the size and type of hydrological ‘problem’ being addressed.

7.7.1 A ground model

With so many inputs possible (see Sections 7.5.2.1, 7.6.2 and 7.6.3), it is prudent to recall that the purpose of a ground model is four-fold:

• To depict for those involved with design, construction and payment of the work the content, volumes and surfaces of the volume of ground affected by the engineering works, and the risks these pose to the successful completion of the work. The ground model will extend beyond the footprint of the structure and beyond the depth to which the physical work (e.g. footings, cut-offs, wells) penetrate, especially if water quality is of concern.

• To enable the various characteristic values for the ground, most notably those for strength, stiffness and permeability, from which design values for these are derived, to be considered within the geological context from which they have been generated, and thus root them in the reality of the ground as some may have to be deduced statistically and others by numerical modelling.

• To enable the natural relationships within the ground, including those for its ground water, to guide the prediction of the response of the ground and its water to the changes engineering will impose upon it.

• To do all this in a logical way that starts with a hypothesis (a concept) probably based on almost no data other than a map and experience, and evolve by observing nature and obeying the laws of science in order to be able to express itself as a ground model that captures the summation of knowledge for that site.

The model will initially take the form of ideas, that is, it will be conceptual. These ideas have to be committed to paper in order to be able to show what is on site and how it is likely to respond to an engineering process. This requires those ideas to be refined through observation to form an ‘observational model’ (Parry et al. 2014) that eventually supports an ‘analytical model’ (Parry et al. 2014).

The first refinement and the first pieces of paper are likely to be 2D vertical cross-sections based on existing data; these can benefit from experience gained from other sites in similar geological situations, assisted by 3D cartoons of likely engineering geomorphology and underlying geology (e.g. Fookes et al. 2015). These are nothing more than ideas; if they prove to be wrong they can be discarded, but their value at this stage in the mental process of establishing a conceptual model for the ground is that they trigger a scientific approach to studying the ground as a whole. Without these first thoughts, which can be directed to the movement of water and gas on site just as well as to the nature of the solid materials on site, initial ideas cannot be given shape and substance which can be tested by ground investigation. It is that need to test an idea that should govern the design of a ground investigation, for example the limits of deposits, the character of the materials within those limits or the hydraulic heads through the deposits. Ground investigations designed simply to gather data provide an insufficient basis for defining a ground model; it is the ideas and concepts of the ground model that need first to be proved and that is the basis for the ground investigation and the observational model. Such models are therefore likely to be essentially geological in character to start with, that is, geological models (e.g. the lithologies present, associated sedimentary and tectonic structures and their ground water) as distinct from geotechnical models, which come later.

The process of testing ideas is crucial for developing a geological ground model that can be said to be ‘justified’; version 1 of the ground model (i.e. the ideas based on existing data and experience) has to be used to make predictions that can then be tested by observation. The predictions will almost certainly be wrong, at least in part, and require the model to be adjusted, so producing version 2 and so on. That approach has already been alluded to in the previous sections in this chapter dealing with geomorphological, sedimentary and hydrogeological models and the various investigations (geophysical, remote sensing, hydrological and geological) they require. Where this whole process of prediction and testing ends depends on the precision required for the model; in theory it is a never-ending task but in practice commercial constraints will impose a limit on what can be achieved at this stage. The precision required will depend on and define the hazards associated with using the geological model, that is, with the correctness of the depiction presented and with the demands of the analytical model required. The first set of risks to consider are therefore those associated with the correctness of the ground model itself; for example, a boundary may be known to ±0.5 m in a vertical borehole but have no basis for its extension radially from the borehole, and that can present quantifiable risks to particular actions such as piling, settlement calculations, groundwater control, etc. Likewise, the direction and nature of groundwater flow could be completely misunderstood, exposing the model to risks not even appreciated by either design or construction. The second set of risks are those associated with the engineering work itself, and are those normally considered in risk assessments where they invariably assume the geological model is correct. It is therefore most important to first assess the completeness of the ground model and the risks associated with using it at any stage in its development.

Generating a model of the ground therefore depends on having an idea to start with of what the geological ground may look like and of what it may contain. That is obviously something that improves with experience, and the purpose of this chapter is to contribute to that experience with an approach that will guide its users through major aspects of the task. However, the geological model is only a starting point; from there a geotechnical ground model has to be developed. The route map for this was established in the UK in the 1970s by Knill & Price (Price 2009) and subsequently revised by Fookes (1997), Knill (2003), de Freitas (2009) and Parry et al. (2014). The next step is to transform a geological model into a geotechnical model, and this is briefly described below.

The solid components of the geological model (e.g. beds, boundaries, structures, etc.) and the fluids they entrain (liquid and gas) constitute the fabric of the ground seen at a macroscopic scale. Each component of this fabric will possess material properties that collectively interact to provide strength, stiffness, permeability and durability, and do this in conjunction with the in situ stresses within it, both total and effective (i.e. total minus pore pressure). Note that only a few of these material properties are visible (e.g. grain size, colour, texture, wetness); those needed for analyses are invisible (e.g. strength, stiffness, stress, permeability) and have to be either measured or assessed from the response of the material to change (unloading, loading, drying, wetting, rainfall, etc.). When these properties are combined with the geological components and fabric to which they belong, a sense of the mass properties of the ground is obtained. These mass properties exist within a larger geological environment defined by stress, pressure, temperature and chemistry of the larger body of ground of which the site is a part. The site may therefore be on a slope and its mass properties can only be appreciated once they are placed within the geological and hydrogeological environment of the slope; the same can be said of sites in valleys and on hill tops, in regions that have been recently unloaded by erosion or excavation or loaded by previous construction or, in the case of glaciations, been loaded and unloaded by nature more than once. By combining mass properties with the geological environment, the ground regime or condition of a site may be quantified. It is this regime that is affected by engineering work, so creating an engineering change that causes the ground to respond, thus generating a ground response. Predicting and explaining the response of the ground to engineering change is the purpose of much engineering geology and hydrogeology. Putting this simply, using the terms described above,

• strength, stiffness, permeability, durability and in situ stress = material properties

• material properties + fabric of the mass = mass properties

• mass properties + environment = ground regime or condition

• ground regime or condition + engineering change = ground response to engineering change.

In the long term it is evident that an account of changes in the natural environment should be considered, hence:

• ground response to engineering change + environmental change = long-term situation.

7.7.2 A basic idea

The extent to which geology (i.e. ground regime or conditions) varies from place to place, is closely related to the work involved in its formation. A simple physical measure of work is force × displacement and this can be used to review, in a qualitative way, the work that has gone into forming a volume of ground and thus gauge its likely complexity. For example, a 100 × 100 × 100 m volume of horizontally bedded over-consolidated stiff and fissured clay-rich till accumulated by sedimentation can be compared with a similar volume of ground containing the same till, but this time cut and displaced by glacial tectonics developing at the same time the till was being deposited, all unconformably overlain by glaciofluvial sands and gravels. It is obvious which volume of ground has been affected by the greatest work and which is likely to be the more complex site; what makes this obvious is the number of steps or stages in the geological development of that volume, that is, its geological history.

The above example is based on the ‘basic’ geology of a site, its ‘solid’ geology and covers of ‘drift’ or superficial deposits; to this has to be added the effects of subsequent glaciations and periglaciation. Immediately it is seen that this extra ‘work’ must add to the complexity of a site and the models to describe it. The message is clear: the actions of glaciation and periglaciation can seriously increase the geological complexity of a site. Add to this anthropogenic activity, which might not involve contamination issues but could easily do so (investigation would be required to ascertain their presence or absence), and it is obvious that any ground model of glaciated and periglacially affected ground, especially in an urban environment, is likely to be very complex. Such thoughts provide a prudent basis for starting a ground model; the model for ‘version 1’ may be simple but knowledge of the glacial and periglacial processes described in the chapters so far should alert the user to the fact that the ground is probably more complex than that. Such thoughts should also alert the user to the fact that the ground investigation techniques used to validate a ground model need to be sympathetic to the scale, delicacy and nature of the data required.

In any ground investigation a balance has to be struck between the investigation needed and the hazards arising from those aspects of the ground, which remain unknown. Those hazards that remain should be removed by design so that they do not materialize as risks; if they are not removed this way, because the ground model being employed contains unidentified errors and omissions (as can easily happen in hydrogeology), unexpected trouble during the construction and working life of the engineering structure can easily arise. Morgenstern (2000) discusses this in detail and how incorrect geological models sow the seed for trouble later.

When designing a ground investigation, it is necessary to define the physical limits of the ground to be investigated and the data required. The physical limits to an investigation are guided by the footprint of the structure, its length, width and depth and by the direction of ground water flow, but are not limited by these. For example, when studying a site on a slope the ground above and below the limits of the site also needs to be considered. Likewise, the depth of investigation can be sensibly gauged from the depth to which a change in vertical stress from the engineering work does not exceed some percent (5–10%) of the in situ stress at that depth prior to construction.

Defining the data required is not so easily done; the data needed will be dictated by what has to be done (which creates the engineering change) and the response this will generate from the ground. Strength, stiffness and permeability of the ground are almost certainly going to be involved unless the ground is to be a source of materials for aggregate and fill, when volumes available, composition and durability often assume greater importance.

A validated or justified conceptual model of the ground is therefore made from layers of information, updated as the model evolves, and the old method of using tracing paper to draw individual layers of information, seen either in plan as with maps or in section, still has much to recommend it (not least the fact that data plotted this way and its relationships have been consciously considered and not treated simply as an exercise in downloading data to the right scale). The first layer provides a base to the work by recording the limits of the intended works, including its anthropogenic elements (e.g. drains, tunnels, sewers) and the ground investigation holes from which any ground data have been derived. Subsequent layers provide, at the same scale as the first layer, the geological model based on these data and any other sources, then the hydrogeological model (including water levels and their range of fluctations, water returns and losses, locations of in situ tests of permeability, locations of piezometers and their respective heads), then the geomechanical data (SPTs, locations and results of in situ tests and laboratory tests), then geophysical results, and finally geochemical data on the quality of solid, liquid and gaseous phases. These layers can be viewed and updated as the investigation continues and during construction, and superimposed to obtain evidence of relationships (e.g. between water levels, water returns and geological material); an example is given in de Freitas (2009).

7.7.3 Adding the glacial and periglacial component

The glacial and periglacial component to be added to the model presented in Section 7.7.2 is similar in some respects to adding the effects of weathering in that it will be related to ground level but, unlike weathering where material is essentially transported as solutions and suspensions, the cold-climate component can contain a substantial element of material transported as solids, as illustrated and explained in Chapters 3 and 5 (Harris 1991; Ballantyne & Harris 1994; Boulton 1996, 2006; Arber & Ber 2007; Martini et al. 2011; Phillips et al. 2013a).

Assessments of the change to the ground created by glacial and periglacial processes are much more difficult to make than those for the change that follows loading and unloading from engineering work. Changes associated with mass movement, as may occur on slopes, in the vicinity of a moving glacier or by freezing and thawing, may be most easily detected using geomorphology, remote sensing and shallow investigations (see Section 7.2) if they have not been covered by more recent drift and anthropological activity. However, changes to ground conditions created by glacial and periglacial processes may be concealed beneath drift deposited as the glacial conditions receded, making such changes much harder to identify as they have to be interpreted from core recovered from ground damaged by them. For example, patterned ground, ice-wedge pseudo-morphs, brecciated bedrock and subglacial tectonic structures including shear zones, and palaeo-shears associated with solifluction all damage the ground in subtle ways making them difficult to identify from core alone. These are discrete features, their location is unknown and, if they are encountered, it will tend to be on a hit-or-miss basis (i.e. randomly). Not finding these features does not rule out their presence on site, but finding them indicates that there are probably more to be found. That is why the use of geophysical surveys is so valuable and can contribute so much to the confidence in such models built with the aid of their use, especially in the marine environment.

This means that the geological model for ground known to have a Quaternary cold-climate history should be considered to contain ground damaged by glacial and/or periglacial processes until excavation demonstrates such features are absent. Note that the combined effects of glacial and periglacial processes should be expected in ground investigations located to the north of the Pleistocene glacial limit, which extends in the UK to the northern fringe of London (see Chapter 2, Figs 2.7 and 2.9 and Chapter 4, Fig. 4.72a–h). Any ground investigations south of this limit should ordinarily involve purely periglacial features. Each ground investigation should be interpreted with cold-climate processes in mind, for example, could the poor state of the core or its poor recovery be the result of glacial and cold-regions processes? Likewise, could the water levels and their change with rainfall or discharge be reflecting a heterogeneity that points to the presence of glacially generated characters? And so on.

To this geological model should now be added the geotechnical characters of the ground – strength, stiffness, permeability and so on – to produce the geotechnical ground model. The outcome from this is a series of ‘what if?’ questions for the designer to consider. This has implications for the types of staff involved in design and the time allocated in the design process for considering the ground. Evidence in previous chapters and in the sections of this chapter demonstrate that, when working in glaciated and periglaciated terrain, the use of specialists in geomorphological mapping, remote sensing, glacial sedimentology and hydrogeology should be balanced against the costs of delay from unforeseen ground conditions and, where this balance is favourable, the involvement of specialists should be considered the norm.

7.7.4 Staff and time resources

In an ideal world contracts would be agreed with sufficient time within them to complete the geological work required to produce a verifiable ground model. When this does not happen, innovation is required. However, innovation within ground engineering occurs in two quite different camps that are often differently resourced: that for designing and constructing the works; and that for constructing the ground model. Innovation in design and construction can be greatly assisted by the use of well-established computer models for easing repeated calculations of structural stability and providing innovative optimal management of the construction process; the economies these provide can win a contract. Innovation in ground investigation is based on discovery of what nature has assembled and man has dissembled; innovation here lies in understanding the geological processes that have been at work. The ground is not something that can be juggled and programmed in the same way as can design and construction; it was not made by man, it is fixed in its position (with the exception of its fluids) and, because its components and their properties are only partially known, they cannot be combined to form some ‘optimum’ in which to work. Further, some components and properties may have an inherent randomness that can only be dealt with by some ‘back-up’ when encountered (e.g. extra pumps, anchors, arches, etc.); they are aleatory in nature. Other components and properties may allow knowledge of them to be improved by further study; they are epistemic. Unfortunately, there are no rules available to guide the designer of a ground investigation either on the relative percentage of aleatory and epistemic components to expect or on the level of knowledge that is sufficient for the ground engineering proposed. The experience of both the client and their technical adviser is crucial; however, the inevitable pressure to ‘save money’ means that many ground models often start with the disadvantage of the work they require being under-assessed and the financial resources allocated to them being inadequate.

The effects of glaciation and periglaciation aggravate this disadvantage by adding complexity to the ground model; glaciated and frost-disturbed ground requires additional resources for its evaluation, just as a contaminated site requires additional resources for its investigation (Nathanail 2013). Various ground models have been described in this publication and reference to them may help a client to appreciate the need for such insights. Helpful though these may be, there is no substitute for education and any geotechnical work conducted in glaciated and periglaciated terrains is served well by staff educated and trained in this subject. Innovation here is directly linked to the discovery of facts, their analyses and their interpretation; these are not activities that can be left to computers.

In order to be understood, complex ground requires the allocation and expenditure of time and resources. Some clients believe the fine details of work in such ground can be left to the contractor provided the contractor is given additional resources for further ground investigation. As a result, the contractor may be asked to carry all risks. There is no guarantee that this further ground investigation, after the award of a contract, will improve either the geological model or the geotechnical model, and thus reduce risks inherent in a poorly understood volume of ground. The principal reason for this is that, by the time the contractor is prepared to accept all risks, the design and construction methods have been decided. Here continuity of involvement by those developing the ground model is significant. Delays in obtaining either permission or funding for work can result in the appointment of contractors who have had no prior involvement with the ground on site. The contractor will presume by this time that the ground model is fit for purpose and further investigation is being used by the contractor to focus on fine-tuning and dimensioning decisions that have already been made in principle.

There are natural links between the complexity of the ground, the ability of those who study it, the ground model that emerges, the risks that accompany the ground model and the contract that governs payment of the work. When those links are strained by contracts that seek to shift responsibility from client to contractor, claims tend to result (Clayton 2001; SISG 2014).

7.7.5 Assumptions, classifications and baseline values

The general accuracy of a ground model is only known when the predictions made for its response to engineering change, especially from excavation (i.e. unloading), loading and change of water levels, are compared with reality. It is optimistic to imagine any model of the ground is complete and correct away from its points of direct validation, for example as at boreholes. Much of any model will be assumed and, if the assumptions are soundly based, the model will be representative; the likelihood of the presence or absence of features can be semi-quantified when data are sufficient for that purpose. Geostatistics has been established now for many decades and concepts of krigging and nugget values have been employed in applied geology, mainly for mineral evaluation (Matheron 1963). Such statistical methods have not been developed in geotechnics to their full potential (Houlsby & Houlsby 2013), and it is easy to see why (Bilodeau et al. 2005) as the raw data available in most cases are scanty.

Instead of using geostatistics, geotechnics has tended to achieve semi-quantification by using schemes of classification and characterization, as these provide an idealization sufficient (although not necessarily satisfactory) for numerical modelling; their great strength is that they have been used for decades and much experience is now attached to them (Barton et al. 1974; Bieniawski 1974; Harrison & Hudson 2000). Systems such as rock quality designation (RQD), rock mass rating (RMR) and Barton’s ‘Q’ ratings provide classifications and characterizations for rock masses (Marinos & Hoek 2000), while grain size, strength and plasticity continue to provide a basis for classifications of bodies of sediment (ASTM 2006; BS 5930:2015; Section 7.4). However, a classification does not remove assumptions; grouping material on the basis of selected similarities such as grain size or unconfined compressive strength disguises differences in other parameters by covering them with a common name (e.g. ‘sand’ or ‘clay’ or ‘moderately strong’). Classification also adds to problems of how to understand materials that occur on and close to the boundaries within a classification, namely ‘marginal materials’, which could almost be classified as something else. The silt/clay boundary in soils is a common source of problems that result in claims. Problems that are a function of a classification are man-made problems that can be made worse by inappropriate contracts; to resolve them it is necessary to consider how the contract and the design can cope the nature of the material. Ground models should therefore be based on the nature of materials; they can include classifications but should not be based on classifications alone.

Given the potential for geological complexity in glaciated or frost-disturbed ground it is good practice first to describe what is present, then to consider how what is present will respond to engineering change, and then the effects it might have on engineering works, before classifying it. Classification is not an end in itself but a tool, and the unthinking use of classifications in rock and soil can result in claims.

Considering the nature of the ground and how it might respond to engineering change provides a sound basis for establishing baseline values (American Society of Civil Engineers 2007); without such values it will be almost impossible to compare the real ground response to engineering change with that predicted from the ground model (Mellors & Southcott 2004). Typical examples of baseline values come from the measurement of water levels and water quality prior to construction, and the position of ground level at time zero. Glaciated or frost-disturbed ground can be in a delicate state of stability and able to change easily in response to engineering change; it is therefore prudent to include baseline values within a ground model and confirm them early in a ground investigation.

7.7.6 Risk

Risk is the probability that unwanted events will occur. The actual number of events divided by the possible number of events equals the probability of an event occurring; this is the definition of probability. The probability of death from playing Russian roulette using a fully loaded gun holding six bullets is therefore 6/6 or 1.0 or 100%. Likewise, should the gun contain no bullets, the risk of death is 0/6 or 0.0 or 0%. Probability therefore ranges between 0 and 1.0. In ground engineering, probability is used to cope with two forms of uncertainty that occur in natural systems, namely ‘epistemic’ (i.e. uncertainty which can be reduced by learning more of a site, e.g. as accompanies every assumption about a ground model) and ‘aleatory’ (i.e. uncertainty which cannot be reduced but can be managed). Methods for doing this tend to fall into two approaches as outlined below.

Approach 1 considers the probability of many aspects of a subject, all of which are considered (either by logic, experience or both) to be interrelated in some undefined way. Using the example above: does the gun come from a reputable supplier and likewise the bullets (i.e. should it and they be expected to work)? Has the gun been maintained (i.e. will it work now)? Will its operator be competent (is he/she likely to be drunk)? Has he/she been trained and has he/she done this before; does he/she know the risk (has he/she completed a risk assessment)? These questions can be expanded as necessary. In Approach 1 the task is either to remove or to significantly reduce the likelihood of those probabilities that will result in an accident, for example find and confiscate the gun or the bullets rather than calculate the probability of their presence and successful use, and prohibit the playing of Russian roulette.

Approach 2 can consider more narrowly defined issues of occurrence and their probability. It uses data that quantifies the number of times a specific and well-defined event has occurred; this could be the number of times a value from a cone penetration test (CPT) has been exceeded in a CPT profile or a particular lithology has been intercepted. It can also ask questions about the relationships of the data (e.g. number of times a CPT value is associated with a lithology). From these datasets, the probabilities of combinations of circumstances (or occurrence of events) can be calculated. In Approach 2 the task is to enhance data obtained at one location with related knowledge obtained from elsewhere (very much akin to what a geologist does in a non-statistical way when mapping); this has been used in attempts to improve the definition of a ‘characteristic value’ of a deposit for an analyses when few values for the deposit exist from the site in question, but many more values exist for the same deposit but elsewhere.

Both approaches are used: Approach 1 because risk has to be considered for every project and Approach 2 because it makes use of data which are known to be relevant (Duncan 2000; Ho et al. 2000; Whitman 2000; Fenton & Griffiths 2008). Each approach will now be briefly reviewed for their application in glacial and periglacial engineering geology.

7.7.6.1 Approach 1

‘Risk’ is the probability of occurrence of an event (E) multiplied by the consequences (C) of that event should it occur; E × C defines the hazard. All the events which are considered to be present within a process and their consequences are assembled to form a matrix of (E) against (C), called a hazard index, where each multiple (E × C) defines a grade of hazard within the index. This general approach and the way it can be used in management is developed and illustrated more fully by van Staveren (2006). To identify the risks of a ground model being wrong, it is necessary to review not only the certainties of the model but also the assumptions made in the model and their consequences for engineering if they are wrong. Each section in this chapter can be used to guide such a review of the risks associated with a ground model.

Boundaries present a particular problem in ground that has been glaciated or disturbed by freeze–thaw processes. Surfaces of shear and tensile failure can occur within such ground from a variety of causes, but especially from gravitationally driven mass movement on slopes and from glacially deformed ground both in front of and beneath a glacier; examples of these can be found in Hutchinson (1991), Burke et al. (2008), Phillips et al. (2013a, b) and Phillips & Hughes (2014). Ground models contain boundaries between materials, but the very concept of a boundary requires the materials on either side of it to be separated by some criterion (e.g. colour, strength, cohesion, stiffness, fossil content, chemical character, etc.) and this is not always easy to achieve. Some natural boundaries, especially those separating material in situ from its transported, deformed and/or remoulded equivalent immediately above it, may not be identified by ground investigation.

Ground damaged by freezing and thawing has been disrupted and it is typical that this disruption will be more in some places than others, especially if cryogenic structures are well developed (Van Vliet-Lanoë 1988, 1991), making the extent of such damage almost impossible to see using boreholes alone. Trial pits, trenches and natural exposures are by far the best way of establishing the presence, scale and repeatability of these features. Where detailed understanding of ground conditions is needed, micromorphological analysis of thin-sections of frost-disturbed soils can reveal features such as micro-scale shears, vesicles, platy structures and textural variation in periglacial deposits (e.g. Harris 1998; Chapter 3, Section 3.6.6).

Subglacial shear will generate discrete boundaries on which failure has occurred and, when they occur within a body of uniform sediment and are parallel to its bedding (e.g. as in a laminated clay), they are difficult to reveal by borehole cores. Such shearing can be almost unnoticed, unlike classical examples of rafting and sub- or proglacial deformation where substantial and very visible disturbance is usually present over large areas (Bleumle & Clayton 1984; Martini et al. 2011).

On a much smaller (microscopic) scale are features generated from the release of porewater under high pressure, often described as water-escape structures and associated with the undrained shearing of deformed ground within the proximity of a glacial snout. Porewater pressure can reach a magnitude that creates hydro-fracturing within weakened bedrock as well as in subglacial till (Chapter 9, Section 9.2.1, Case Study 9.5). Field scale and microscopic evidence often indicates shear displacement has accompanied these episodes of failure (van der Meer et al. 2003, 2008; Phillips et al. 2013b); examples are illustrated in Figure 7.34.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.34.

Example of micro-fabrics found in till (from van der Meer 1993).

Special note should therefore be taken of the description of materials (see Section 7.4), particularly when these are not easily recovered from ground that has been disturbed, because the ground contains an extreme range of particle sizes and shapes together with structures such as shear surfaces, hydro-fractures and their weak infilling (e.g. Norris & Spieker 1966; Menzies 2000). Here core loss is of great concern as it could represent so many things. Indeed, core loss in these materials can easily warrant the use of borehole geophysics if mechanical and/or hydraulic issues are of concern as in the design and construction of foundations, cut-offs, caissons, shafts and situations sensitive to ground movement (Coswill et al. 2016). If an assessment cannot be made of the accuracy of the descriptions for such material, the ground model might contain not only ‘unknowns’ but also ‘unreality’. Material descriptions are also necessary for interpreting the data obtained from surface geophysics and groundwater levels, each of which can provide a ground model in terms of geophysics and hydrogeology.

The response of ground to controlled changes in head (as in tests where water levels in boreholes or wells are changed in a controlled way) can be most revealing as the response of volumes of ground, far larger than any alternative in situ test, and can be considered both in plan and with depth. Too often a drawdown test undertaken to confirm the feasibility of dewatering is only used to obtain a global value for hydraulic conductivity that is of indicative rather than of particular use. Yet the same data often contain sufficient information to detect the presence of hydrogeological boundaries and differences in compressibility of the ground around the site (see Section 7.6), the nature of which, in glaciated ground, can be exceedingly complex (Kruseman & de Ridder 1994; McMillan et al. 2000). Quantification of boundaries and compressibility can be immensely useful in assessing the presence of physical features inherited from a glacial past and probably unseen by vertical boreholes, and possibly uninterpretable from surface geophysics alone. Such groundwater data can be used to check a ground model derived from borehole and geophysical data alone (see Section 7.6). Checks of this sort can reveal the presence of features which might be present within a ground model and could be a hazard, and hence the risks associated with using the ground model as it stands.

Assessing the unknowns that could be present within a ground model suggests that such models should be accompanied by the brief governing the purpose of the investigation from which they are derived, because the design of the investigation and the investigative methods used may be insufficient for supporting the ground model proposed. Judgement of the value of a ground model can best be made when the resources that enabled it to be made are also known. The establishment of ground reference conditions (American Society of Civil Engineers 2007) based on the ground investigation can be a good start to reducing risk, but they are only part of the solution of risk in highly variable ground. They should describe the materials and conditions on site, cater for the range of conditions expected and, very importantly, provide a basis for resolving differences when those conditions are found not to apply.

7.7.7 End note

The lack of certainty in a model of complex ground has tempted some to suggest that, by using construction techniques that can accommodate unknowns in the ground such as groundwater and weak materials, there is no need for a ground model at all; this is a very narrow and dangerous view to adopt. Great advances have been made in designing construction techniques that can accommodate a wide range of ground conditions, for example the earth pressure balanced tunnel boring machine, or a propped box excavation within diaphragm walls or a diaphragm wall shaft. However, the cost of using these sophisticated techniques cannot be sensibly priced if the ground model is rudimentary; much of the economic advantage in using them may therefore be lost. Similarly, those working with the material excavated, the arisings, will be unsure of what might be available for use elsewhere or of its condition. Further, without a ground model there is no basis for predicting the ground movements that should happen so that they can be compared with those that do happen, and so facilitate back analyses and hence a knowledge of ground en masse.

The case histories for the Westminster Box and related work (Burland et al. 2001) show that real-time monitoring enables delicate engineering to be safely accomplished through the most complex ground where, in this case, the complexity came not only from the original solid geology and its drift, but anthropological disturbance and the active creation of boundaries and voids while the work was in progress. All such monitoring can be viewed as ‘learning about the ground during excavation’. The ground model at Westminster was constantly being adjusted and updated in terms of mechanical properties. The same should be expected for the geology of a ground model; ground is only known properly at the end of the job. Every section of this chapter has concluded that a geotechnical engineer's understanding of the ground evolves as the work progresses.

The lesson from this experience is that ground models need to evolve during the lifetime of a job; too often they are frozen at the time of contract. It is not unusual for elements of construction to be dictated by trigger levels and limits that initiate well-defined and priced responses, but this should not justify the end of investigation as some unknown or unsuspected aspect of the ground could still delay the works. There is a strong case, supported by the case history evidence in this publication, to warrant a much wider use of this ‘evolutionary’ approach to a ground model within ground affected by glacial or periglacial activity. The ground model should be expected to evolve as the job progresses and the implication of these changes in the model considered for ground response.

The combination of Bayesian statistics and measured values for ground response, whether from in situ tests or instrumented field work (e.g. drilling, tunnelling, prop loads, end-bearing loads, etc.), enables the probability of outcomes based on data at the scale observed during ground investigations to be compared with the probability of outcomes using these data at different scales in analyses. As such, this approach has the potential to provide an objective, quantitative and hence rational basis for deciding the scale at which to design, a problem for which little is known in ground engineering and hydrogeology.

In this way the two areas of innovation described in Section 7.7.6 can both be used; a ground model which is verifiable yet capable of flexibility can contribute to winning contracts just as effectively as computer-based innovation for design and management techniques. Engineering in ground disturbed by glaciers or frost might do well to consider the possible advantages of this approach, and view the use of a ground model that is fixed at time of contract as not only limiting the knowledge that may be required but also as a primary cause of potential delay, overrun and additional costs.

Finally, there is a third form of uncertainty in addition to those of epistemic and aleatory mentioned above, namely ontological (i.e. uncertainty from events of which we have no knowledge). This is the uncertainty that arises so easily at the interface between professions, when the engineer fails to communicate with the engineering geologist, or the engineering geologists fails to communicate with the Quaternary sedimentologist or glacial geomorphologist, and so on. In this respect, the various chapters in this book will be of relevance and assistance to the engineering investigation and assessment of glaciated and periglaciated ground.