faults and faulting

faults and faulting A fault is a fracture, or a zone of several fractures, across which movement has taken place, as shown by the fact that a reference marker of some kind has been offset. As such, faults differ from joints, which exhibit no measurable displacement. In the upper crust, above a depth of about 15 km, fault displacement is the principal means by which energy generated at plate margins is dissipated. Understanding the geometrical expression of faults, and the mechanisms by which they develop, is crucial to a number of key needs of modern society, such as the exploitation of fault-related energy and mineral reserves, and the prediction of earthquake hazards.

Classification of faults

Two main categories of fault geometry are recognized: dip-slip and strike-slip. Dip-slip faults are those faults on which the slip vector is parallel to the dip of the fault, whereas strike-slip faults have a near-horizontal slip vector. Dip-slip faults fall into two further subdivisions (Fig. 1). In normal faults the rock overlying a dip-slip fault plane, the hanging-wall, moves down relative to the underlying footwall. Conversely, reverse faults push hanging-wall rocks up, relative to the footwall. Normal faults are the principal structure by which the upper crust undergoes an overall lengthening, or extension. By placing relatively older, deeper rocks on younger footwall rocks, however, reverse faults cause shortening. A vertical prfe, such as a borehole, through a reverse fault will therefore encounter the same vertical succession of rocks twice, in both the hanging-wall and the footwall, whereas a borehole through a normal fault will record apparently thinned, or even ‘missing’, rock units across the fault plane (Fig. 1).

Strike-slip faults are also subdivided according to the relative displacement between each side of the fault. If we imagine ourselves standing on one side of a strike-slip fault that cuts a linear marker, the fault is termed dextral or sinistral, according to whether the marker is offset to our right or left, respectively.

Linked fault systems

In practice, individual exposures of faults seldom display perfect dip-slip or strike-slip geometry. Slip-parallel striae, termed slickenside lineations, usually indicate oblique slip, in which displacement is sub-vertical or sub-horizontal. Furthermore, dip-slip and strike-slip faults commonly occur in the same setting, forming part of a linked fault system. In cross-section, these consist of a staircase-like profile of sub-horizontal flats, and steeper ramps, which cut through the footwall, detaching a relatively thin-skinned slice of hanging-wall rocks. The idea that disparate faults with different senses of slip often belong to a single, linked fault system spawned a major new research theme in the 1970s and 1980s. Such studies explored the range of patterns of linked fault systems, and the way in which a hanging-wall succession deforms internally in response to its movement over variously oriented ramps and along flats in the underlying fault system. These studies are based on data obtained from fieldwork, especially from mapping reverse fault systems in the Canadian Rockies, the Appalachians, and the European Alps, and from seismic reflection profiles imaging normal fault systems in the sub-surface.

A more-or-less universal observation from linked fault systems is that the constituent faults have a listric, or curvi-planar cross-sectional profile (Fig. 1). Listric fault systems are particularly widespread at fairly shallow depths within sedimentary basins. They form most commonly in well-layered rock sequences, comprising a variety of strong and weak horizons, when gravity is important as a driving force. At the surface, the ramps of listric dip-slip faults will commonly attain angles of between 30° and 60° to the horizontal, but their dip will become gentler with depth, before they become detached along a flat.

Flats commonly consist of a particularly weak rock-type, such as halite (rock salt) or a mud containing trapped, high-pressure water. The combination of the relative weakness of these low-strength, décollement (detachment) horizons and the entrapment of high-pressure fluid makes them an ideal décollement zone because they minimize the frictional resistance to movement of the hanging wall over the footwall. Off the Caribbean island of Barbados, where a major reverse fault system is actively deforming the sea floor, mud volcanoes record the advance of a flat up to 10 km in front of the ridge marking the line of most intensive hanging-wall deformation. Provided the frictional resistance to movement along a flat is minimal, the hanging-wall will slide passively over the footwall, and little or no internal deformation will occur.

Faulting and folding

Two main fold structures are associated with deformation above ramps, rollover anticlines and ramp anticlines (Fig. 1). Rollover anticlines form where a relatively steeply inclined ramp in a listric normal fault system flattens with increasing depth to connect with a sub-horizontal flat. Gradual displacement along the fault system will, theoretically, open an expanding, unfilled gap between the footwall and the hanging-wall. In practice, however, such gaps are never observed as the hanging wall responds by collapsing into the gap. The result is a monoclinal rollover anticline, its tightness increasing progressively with continued fault displacement. Excellent examples of rollover anticlines are associated with listric fault systems affecting oil- and gas-bearing sandstones in Tertiary sedimentary sequences in the Gulf of Mexico and the Niger delta.

Ramp anticlines record a comparable process to the formation of rollovers, but here applied to listric reverse, or thrust, faults. In the case of thrusts, an anticline will always form above an underlying ramp: the empirical rule states that the angle between a ramp and the geological marker horizons it cuts must be approximately maintained. Thus, if thrust ramps attain angles of around 30° to the horizontal (and hence 30° to a marker, such as sedimentary bedding), then subsequent movement of the hangingwall up the ramp and on to a flat must be accompanied by folding as the bedding flexes to maintain its original angle with the flat. Continued displacement along the thrust will result in the increasing separation of the ramp anticline from the ramp above which it originally formed. Alternatively, a new fault may develop which cuts into the former footwall, thereby generating a closely spaced, or imbricate, array of faults. According to whether or not they connect at both the foot and the top of the ramp, such structures are termed, respectively, duplex or imbricate fans (Fig. 2). Studies of the detailed history of internal deformation of hanging-wall rocks that have experienced movement over multiple ramps reveals complex sequences of repeated bending and unbending. For this reason, ramp anticlines are commonly referred to as fault-bend folds. A drive across any of the world's currently or recently active thrust belts—the foothills of the Himalaya, or the Jura Mountains of the European Alps—demonstrates spectacularly the role of fault-bend folding in generating their topography. These regions are characterized by linear ranges of hills and valleys, the hills marking the ramp anticlines, the valleys and plateaux expressing the flats between thrust ramps. Ramp anticlines in thrust belts in Colombia, Papua New Guinea, and particularly the Zagros Range in Iran, have yielded some of the largest single oil accumulations in the world.

Strike-slip faulting

So far, discussion of linked fault systems has concentrated exclusively on dip-slip settings. However, many of the characteristic features of linked fault systems—their staircase profile, comprising variously orientated ramps and flats, the interconnection between disparate fault types, and the intimate relation between faulting and folding—are seen also in areas undergoing strike-slip faulting—with the exception, that is, that the ramps and flats are turned on end so that the staircase profile is seen, not in cross-section, but in plan view (Fig. 3). Strike-slip fault systems are therefore dominated by long straight fault segments along which the adjacent sides of the fault move passively past one another. Deformation is largely confined to the offsets of these straight segments, at bends along the fault (Fig. 3). Two types of bend are recognized: transtensional bends (Fig. 3), where the stresses are locally tensional, and compressional transpressional bends. Seismic reflection profiles through transtensile and transpressive bends reveal dendritic arrangements of normal and reverse dip-slip faults respectively, termed flower structures. Rocks adjacent to a major strand of an active strike-slip fault commonly display evidence of repeated episodes of reverse faulting, and normal faulting, as they pass through multiple zones of transpression and transtension.

Strike-slip faults are the most long-lived of all fault systems, sometimes remaining active for 100 Ma, over which time they can accumulate as much as 5000 km displacement. The only structures capable of absorbing such enormous displacements are the plate margins. Among the best-known examples of these trans-lithospheric, or transform, faults are the dextral San Andreas and Alpine faults in California and New Zealand respectively.

Growth faulting

One of the most exciting developments in the study of linked fault systems has been the realization that faulting and sedimentation are commonly synchronous processes. In the case of displacement along a listric normal fault, the development of a rollover anticline will cause a topographic depression, itself a focus of further sediment accumulation. Consequently, sediment will accumulate preferentially in the hanging-wall of an active normal fault, thereby generating a thicker sedimentary succession. This is growth faulting and it can lead to dramatic changes in sedimentary thicknesses on each side of a fault, the thicker sedimentary succession sitting always on the downthrown side of a growth faults. Furthermore, growth faulting will lead to variable inclination of growth strata; the older, stratigraphically deeper beds will dip more steeply into a listric normal growth fault because they have experienced a longer period of fault displacement and bed rotation.

Growth faulting is also an important process in active thrust belts, in which thicker successions are predicted to accumulate in the footwalls, where extra space is continuously being created by hanging-wall uplift. By the same token, actively amplifying ramp anticlines experience syntectonic erosion of their crests, thereby providing a ready source of sediment. Using seismic reflection profiles, analysis of growth strata associated with modern thrust faults in the Philippines, Venezuela, California, and Oklahoma provides a promising means of assessing earthquake hazard. Here, dating of the growth strata yields a direct estimate of rates of fault-bend folding during the past 8 Ma. or so, the most rapid rates corresponding to the more earthquake-prone fault segments.

Crustal-scale faulting and earthquakes

Despite the undoubted importance of listric faults, deep geophysical surveys have demonstrated that the largest-scale faults which control the overall shape, position, and long-term development of sedimentary basins are, in fact, more commonly planar. Some planar faults have been shown to extend as deeply as 30–40 km, to the base of the crust, sometimes actually offsetting the Moho itself. In the northern North Sea, planar normal faults define many of the largest oilfields, such as the Brent field. Field-based studies of planar normal fault systems liken them to the behaviour of books on a bookshelf. With increasing extension, or shear, vertically stacked books will progressively rotate, attaining ever-decreasing angles to the horizontal. Examples mapped in the highly extended Basin and Range province of Nevada show early formed, highly rotated planar normal faults now orientated at near-horizontal attitudes.

Earthquakes occur during episodes of sudden fault movement and they express the catastrophic release of energy which builds up within a volume of rock immediately adjacent to a geologically stressed fault plane. The existence of earthquakes is perhaps the most graphic evidence we have for the ability of the upper crust to store energy elastically. Analysis of global earthquake data shows that the overwhelming majority of earthquakes of small to medium magnitude have their epicentres within the top 10–15 km of the crust. Below 15 km, where temperatures begin to exceed 250–300 °C, the brittle upper crust changes its physical properties and begins to deform in a more ductile fashion. This is the brittle–ductile transition, and it explains why few earthquakes are recorded from deeper than this seemingly arbitrary depth: beneath the brittle–ductile transition, only small quanta of elastic energy can accumulate along an active fault segment before it deforms plastically, and harmlessly. Studies of magnificently exposed planar normal fault systems in the Aegean region of Greece reveal a remarkable consistency in the dimensions of continental fault systems. Active faults are inclined at angles of between 40°–60° to the horizontal, and they are spaced 20–100 km apart, with individual faults attaining lengths of between 15–20 km. Such relative consistency has led researchers to suggest that it is the fairly uniform deformational characteristics of the brittle upper crust that exert the chief control over the dimensions of continental fault systems.

Why in general do large-magnitude earthquakes not occur above the brittle–ductile transition? This is due to the way in which the upper crust is criss-crossed by billions of pre-existing faults and fractures. From the scale of millimetres to hundreds of kilometres, all these fractures represent planes of weakness, each with the potential to be reactivated time and time again. In the upper crust, therefore, elastic energy will not usually build up for long before the magnitude of stress is sufficient to overcome the frictional resistance to reactivation of a pre-existing fracture, and only a small earthquake will result.

Slipping and creeping

Cumulative displacement on large fault systems can be of the order of thousands of kilometres in the case of strike-slip fault systems and hundreds of kilometres for dip-slip fault systems. However, measurement of fault scarps, the surface expression of recently ruptured fault planes, indicates that the amount of slip that accompanies even a large-magnitude earthquake is usually less than a metre. This introduces the notion of co-seismic slip and aseismic creep along faults, defined as fault displacement occurring respectively during and after an earthquake. Sporadic episodes of co-seismic slip take place at rates of between 100 mm and a metre per second, whereas more-or-less continuous aseismic creep, as the name suggests, occurs as slowly as 1 to 100 mm a year. Despite the considerably more dramatic and hazardous manifestation of faulting during co-seismic slip, it is actually responsible for generating as little as a millionth of the total length of a fault. Aseismic creep also has the effect of smoothing the spatial variation in displacement over the area of a fault plane. Most faults exhibit a smooth decrease in cumulative displacement, from a maximum at the centre of the fault to zero at the fault tip.

Faults and plate tectonics

Is there a systematic distribution of the different fault types according to plate-tectonic setting? Using earthquake data to map the global distribution of faulting, it has long been appreciated that most faulting is restricted to linear zones of intense activity, with few faults in between. These are, respectively, the plate margins and intra-plate regions. The dynamics of plate movement exerts the principal control over the global distribution of horizontal stress. Three main stress domains can be discriminated, each with characteristic types of faulting. Normal faults occur where the least principal, or tensile, stress is horizontal, and the greatest principal, or compressive, stress is vertical. Reverse and thrust faults are found in regions where the least (but not tensile) stress is vertical and compressive stress is horizontal. Lastly, strike-slip faults occur where both the compressive and tensile stresses are horizontal, and the intermediate stress direction, which is compressive is vertical. It is easy to see, therefore, why the main plate tectonic settings of reverse faults are the world's major mountain belts, where strong horizontal compression is generated in response to the collision of adjacent plates. Conversely, normal faults are commonest in proto-oceanic environments of continental splitting, where plate divergence generates strong horizontal tension. Strike-slip faults, however, characterize those plate margins where adjacent plates are moving past one another, neither colliding nor splitting apart.

Jonathan P. Turner

Bibliography

Bolt, B. A. (1993) Earthquakes. W. H. Freeman, New York.
Hancock, P. L. (ed.) (1994) Continental deformation. Pergamon Press, Oxford.
Twiss, R. J. and and Moores, E. M. (1992) Structural geology. W. H. Freeman, New York.