Reflection seismology

The general principle of seismic reflection is to send elastic waves (using an energy source such as dynamite explosion or Vibroseis) into the Earth, where each layer within the Earth reflects a portion of the wave's energy back and allows the rest to refract through. These reflected energy waves are recorded over a predetermined time period (called the record length) by receivers that detect the motion of the ground in which they are placed. On land, the typical receiver used is a small, portable instrument known as a geophone, which converts ground motion into an analogue electrical signal. In water, hydrophones are used, which convert pressure changes into electrical signals. Each receiver's response to a single shot is known as a “trace” and is recorded onto a data storage device, then the shot location is moved along and the process is repeated. Typically, the recorded signals are subjected to significant amounts of signal processing before they are ready to be interpreted and this is an area of significant active research within industry and academia. In general, the more complex the geology of the area under study, the more sophisticated are the techniques required to remove noise and increase resolution. Modern seismic reflection surveys contain large amount of data and so require large amounts of computer processing, often performed on supercomputers or computer clusters.

P-wave reflects off an interface at normal incidence

When a seismic wave encounters a boundary between two materials with different acoustic impedances, some of the energy in the wave will be reflected at the boundary, while some of the energy will be transmitted through the boundary. The amplitude of the reflected wave is predicted by multiplying the amplitude of the incident wave by the seismic reflection coefficient ${\displaystyle R}$, determined by the impedance contrast between the two materials.

For a wave that hits a boundary at normal incidence (head-on), the expression for the reflection coefficient is simply

${\displaystyle R={\frac {Z_{2}-Z_{1}}{Z_{2}+Z_{1}}}}$,

where ${\displaystyle Z_{1}}$ and ${\displaystyle Z_{2}}$ are the impedance of the first and second medium, respectively.

Similarly, the amplitude of the incident wave is multiplied by the transmission coefficient to predict the amplitude of the wave transmitted through the boundary. The formula for the normal-incidence transmission coefficient is

${\displaystyle T=1-R={\frac {2Z_{1}}{(Z_{2}+Z_{1})}}}$.[6]

As the sum of the squares of amplitudes of the reflected and transmitted wave has to be equal to the square of amplitude of the incident wave, it is easy to show that

${\displaystyle Z_{1}(1-R^{2})={\frac {Z_{1}(Z_{2}+Z_{1})^{2}-Z_{1}(Z_{2}-Z_{1})^{2}}{(Z_{2}+Z_{1})^{2}}}={\frac {4Z_{2}Z_{1}^{2}}{(Z_{2}+Z_{1})^{2}}}=Z_{2}T^{2}}$.

By observing changes in the strength of reflectors, seismologists can infer changes in the seismic impedances. In turn, they use this information to infer changes in the properties of the rocks at the interface, such as density and elastic modulus.

Diagram showing the mode conversions that occur when a P-wave reflects off an interface at non-normal incidence

The situation becomes much more complicated in the case of non-normal incidence, due to mode conversion between P-waves and S-waves, and is described by the Zoeppritz equations. In 1919, Karl Zoeppritz derived 4 equations that determine the amplitudes of reflected and refracted waves at a planar interface for an incident P-wave as a function of the angle of incidence and six independent elastic parameters.[5] These equations have 4 unknowns and can be solved but they do not give an intuitive understanding for how the reflection amplitudes vary with the rock properties involved.[7]

The reflection and transmission coefficients, which govern the amplitude of each reflection, vary with angle of incidence and can be used to obtain information about (among many other things) the fluid content of the rock. Practical use of non-normal incidence phenomena, known as AVO (see amplitude versus offset) has been facilitated by theoretical work to derive workable approximations to the Zoeppritz equations and by advances in computer processing capacity. AVO studies attempt with some success to predict the fluid content (oil, gas, or water) of potential reservoirs, to lower the risk of drilling unproductive wells and to identify new petroleum reservoirs. The 3-term simplification of the Zoeppritz equations that is most commonly used was developed in 1985 and is known as the "Shuey equation". A further 2-term simplification is known as the "Shuey approximation", is valid for angles of incidence less than 30 degrees (usually the case in seismic surveys) and is given below:[8]

${\displaystyle R(\theta )=R(0)+G\sin ^{2}\theta }$

where ${\displaystyle R(0)}$ = reflection coefficient at zero-offset (normal incidence); ${\displaystyle G}$ = AVO gradient, describing reflection behaviour at intermediate offsets and ${\displaystyle (\theta )}$ = angle of incidence. This equation reduces to that of normal incidence at ${\displaystyle (\theta )}$=0.

The time it takes for a reflection from a particular boundary to arrive at the geophone is called the travel time. If the seismic wave velocity in the rock is known, then the travel time may be used to estimate the depth to the reflector. For a simple vertically traveling wave, the travel time ${\displaystyle t}$ from the surface to the reflector and back is called the Two-Way Time (TWT) and is given by the formula

${\displaystyle t=2{\frac {d}{V}}}$,

where ${\displaystyle d}$ is the depth of the reflector and ${\displaystyle V}$ is the wave velocity in the rock.

A series of apparently related reflections on several seismograms is often referred to as a reflection event. By correlating reflection events, a seismologist can create an estimated cross-section of the geologic structure that generated the reflections. Interpretation of large surveys is usually performed with programs using high-end three-dimensional computer graphics.

Sources of noise on a seismic record. Top-left: air wave; top-right: head wave; bottom-left: surface wave; bottom-right: multiple.

In addition to reflections off interfaces within the subsurface, there are a number of other seismic responses detected by receivers and are either unwanted or unneeded:

The airwave travels directly from the source to the receiver and is an example of coherent noise. It is easily recognizable because it travels at a speed of 330 m/s, the speed of sound in air.

A Rayleigh wave typically propagates along a free surface of a solid, but the elastic constants and density of air are very low compared to those of rocks so the surface of the Earth is approximately a free surface. Low velocity, low frequency and high amplitude Rayleigh waves are frequently present on a seismic record and can obscure signal, degrading overall data quality. They are known within the industry as ‘Ground Roll’ and are an example of coherent noise that can be attenuated with a carefully designed seismic survey.[9] The Scholte wave is similar to ground roll but occurs at the sea-floor (fluid/solid interface) and it can possibly obscure and mask deep reflections in marine seismic records.[10] The velocity of these waves varies with wavelength, so they are said to be dispersive and the shape of the wavetrain varies with distance.[11]

A head wave refracts at an interface, travelling along it, within the lower medium and produces oscillatory motion parallel to the interface. This motion causes a disturbance in the upper medium that is detected on the surface.[5] The same phenomenon is utilised in seismic refraction.

An event on the seismic record that has incurred more than one reflection is called a multiple. Multiples can be either short-path (peg-leg) or long-path, depending upon whether they interfere with primary reflections or not.[12][13]

Multiples from the bottom of a body of water and the air-water interface are common in marine seismic data, and are suppressed by seismic processing.

Cultural noise includes noise from weather effects, planes, helicopters, electrical pylons, and ships (in the case of marine surveys), all of which can be detected by the receivers.

Reflection seismology, more commonly referred to as "seismic reflection" or abbreviated to "seismic" within the hydrocarbon industry, is used by petroleum geologists and geophysicists to map and interpret potential petroleum reservoirs. The size and scale of seismic surveys has increased alongside the significant increases in computer power since the late 20th century. This led the seismic industry from laboriously – and therefore rarely – acquiring small 3D surveys in the 1980s to routinely acquiring large-scale high resolution 3D surveys. The goals and basic principles have remained the same, but the methods have slightly changed over the years.

The primary environments for seismic hydrocarbon exploration are land, the transition zone and marine:

Land – The land environment covers almost every type of terrain that exists on Earth, each bringing its own logistical problems. Examples of this environment are jungle, desert, arctic tundra, forest, urban settings, mountain regions and savannah.

Transition Zone (TZ) – The transition zone is considered to be the area where the land meets the sea, presenting unique challenges because the water is too shallow for large seismic vessels but too deep for the use of traditional methods of acquisition on land. Examples of this environment are river deltas, swamps and marshes,[20] coral reefs, beach tidal areas and the surf zone. Transition zone seismic crews will often work on land, in the transition zone and in the shallow water marine environment on a single project in order to obtain a complete map of the subsurface.

Diagram of equipment used for marine seismic surveys

Marine – The marine zone is either in shallow water areas (water depths of less than 30 to 40 metres would normally be considered shallow water areas for 3D marine seismic operations) or in the deep water areas normally associated with the seas and oceans (such as the Gulf of Mexico).

Seismic data acquisition is the first of the three distinct stages of seismic exploration, the other two being seismic data processing and seismic interpretation. [21]

Seismic surveys are typically designed by National oil companies and International oil companies who hire service companies such as CGG, Petroleum Geo-Services and WesternGeco to acquire them. Another company is then hired to process the data, although this can often be the same company that acquired the survey. Finally the finished seismic volume is delivered to the oil company so that it can be geologically interpreted.

Desert land seismic camp

Receiver line on a desert land crew with recorder truck

Land seismic surveys tend to be large entities, requiring hundreds of tons of equipment and employing anywhere from a few hundred to a few thousand people, deployed over vast areas for many months.[22] There are a number of options available for a controlled seismic source in a land survey and particularly common choices are Vibroseis and dynamite. Vibroseis is a non-impulsive source that is cheap and efficient but requires flat ground to operate on, making its use more difficult in undeveloped areas. The method comprises one or more heavy, all-terrain vehicles lowering a steel plate onto the ground, which is then vibrated with a specific frequency distribution and amplitude.[23] It produces a low energy density, allowing it to be used in cities and other built-up areas where dynamite would cause significant damage, though the large weight attached to a Vibroseis truck can cause its own environmental damage.[24] Dynamite is an impulsive source that is regarded as the ideal geophysical source due to it producing an almost perfect impulse function but it has obvious environmental drawbacks. For a long time, it was the only seismic source available until weight dropping was introduced around 1954,[25] allowing geophysicists to make a trade-off between image quality and environmental damage. Compared to Vibroseis, dynamite is also operationally inefficient because each source point needs to be drilled and the dynamite placed in the hole.

A land seismic survey requires substantial logistical support. In addition to the day-to-day seismic operation itself, there must also be support for the main camp (for catering, waste management and laundry etc.), smaller camps (for example where the distance is too far to drive back to the main camp with vibrator trucks), vehicle and equipment maintenance, medical personnel and security.

Unlike in marine seismic surveys, land geometries are not limited to narrow paths of acquisition, meaning that a wide range of offsets and azimuths is usually acquired and the largest challenge is increasing the rate of acquisition. The rate of production is obviously controlled by how fast the source (Vibroseis in this case) can be fired and then move on to the next source location. Attempts have been made to use multiple seismic sources at the same time in order to increase survey efficiency and a successful example of this technique is Independent Simultaneous Sweeping (ISS).[26]

Marine seismic survey using a towed streamer

Plan view of NATS and MAZ surveys

Plan view of a WATS/WAZ survey

Seismic data collected by the USGS in the Gulf of Mexico

Seismic support vessel

Traditional marine seismic surveys are conducted using specially-equipped vessels that tow one or more cables containing a series of hydrophones at constant intervals (see diagram). The cables are known as streamers, with 2D surveys using only 1 streamer and 3D surveys employing up to 12 or more (though 6 or 8 is more common). The streamers are deployed just beneath the surface of the water and are at a set distance away from the vessel. The seismic source, usually an airgun or an array of airguns but other sources are available, is also deployed beneath the water surface and is located between the vessel and the first receiver. Two identical sources are often used to achieve a faster rate of shooting. Marine seismic surveys generate a significant quantity of data,[27] each streamer can be up to 6 or even 8 km long, containing hundreds of channels and the seismic source is typically fired every 15 or 20 seconds.

A seismic vessel with 2 sources and towing a single streamer is known as a Narrow-Azimuth Towed Streamer (or NAZ or NATS). By the early 2000s, it was accepted that this type of acquisition was useful for initial exploration but inadequate for development and production,[28] in which wells had to be accurately positioned. This led to the development of the Multi-Azimuth Towed Streamer (MAZ) which tried to break the limitations of the linear acquisition pattern of a NATS survey by acquiring a combination of NATS surveys at different azimuths (see diagram).[29] This successfully delivered increased illumination of the subsurface and a better signal to noise ratio.

The seismic properties of salt poses an additional problem for marine seismic surveys, it attenuates seismic waves and its structure contains overhangs that are difficult to image. This led to another variation on the NATS survey type, the wide-azimuth towed streamer (or WAZ or WATS) and was first tested on the Mad Dog field in 2004.[30] This type of survey involved 1 vessel solely towing a set of 8 streamers and 2 separate vessels towing seismic sources that were located at the start and end of the last receiver line (see diagram). This configuration was "tiled" 4 times, with the receiver vessel moving further away from the source vessels each time and eventually creating the effect of a survey with 4 times the number of streamers. The end result was a seismic dataset with a larger range of wider azimuths, delivering a breakthrough in seismic imaging.[28] These are now the three common types of marine towed streamer seismic surveys.

Marine survey acquisition is not just limited to seismic vessels; it is also possible to lay cables of geophones and hydrophones on the sea bed in a similar way to how cables are used in a land seismic survey, and use a separate source vessel. This method was originally developed out of operational necessity in order to enable seismic surveys to be conducted in areas with obstructions, such as production platforms, without having the compromise the resultant image quality.[31] Ocean bottom cables (OBC) are also extensively used in other areas that a seismic vessel cannot be used, for example in shallow marine (water depth <300m) and transition zone environments, and can be deployed by remotely operated underwater vehicles (ROVs) in deep water when repeatability is valued (see 4D, below). Conventional OBC surveys use dual-component receivers, combining a pressure sensor (hydrophone) and a vertical particle velocity sensor (vertical geophone), but more recent developments have expanded the method to use four-component sensors i.e. a hydrophone and three orthogonal geophones. Four-component sensors have the advantage of being able to also record shear waves,[32] which do not travel through water but can still contain valuable information.

In addition to the operational advantages, OBC also has geophysical advantages over a conventional NATS survey that arise from the increased fold and wider range of azimuths associated with the survey geometry.[33] However, much like a land survey, the wider azimuths and increased fold come at a cost and the ability for large-scale OBC surveys is severely limited.

In 2005, ocean bottom nodes (OBN) – an extension of the OBC method that uses battery-powered cableless receivers placed in deep water – was first trialled over the Atlantis Oil Field in a partnership between BP and Fairfield Geotechnologies.[34] The placement of these nodes can be more flexible than the cables in OBC and they are easier to store and deploy due to their smaller size and lower weight.

Time lapse or 4D surveys are 3D seismic surveys repeated after a period of time. The 4D refers to the fourth dimension which in this case is time. Time lapse surveys are acquired in order to observe reservoir changes during production and identify areas where there are barriers to flow that may not be detectable in conventional seismic. Time lapse surveys consist out of a baseline survey and a monitor or repeat survey, acquired after the field was under production. Most of these surveys have been repeated NATS surveys as they are cheaper to acquire and most fields historically already had a NATS baseline survey. Some of these surveys are collected using ocean-bottom cables because the cables can be accurately placed in their previous location after being removed. Better repetition of the exact source and receiver location leads to improved repeatability and better signal to noise ratios. A number of 4D surveys have also been set up over fields in which ocean bottom cables have been purchased and permanently deployed. This method can be known as life of field seismic (LoFS) or permanent reservoir monitoring (PRM).[35]

OBN has proven to be another very good way to accurately repeat a seismic acquisition. The world's first 4D survey using nodes was acquired over the Atlantis Oil Field in 2009, with the nodes being placed by a ROV in a water depth of 1300–2200 m to within a few meters of where they were previously placed in 2005.[36]

There are three main processes in seismic data processing: deconvolution, common-midpoint (CMP) stacking and migration.[37]

Deconvolution is a process that tries to extract the reflectivity series of the Earth, under the assumption that a seismic trace is just the reflectivity series of the Earth convolved with distorting filters.[38] This process improves temporal resolution by collapsing the seismic wavelet, but it is nonunique unless further information is available such as well logs, or further assumptions are made. Deconvolution operations can be cascaded, with each individual deconvolution designed to remove a particular type of distortion.

CMP stacking is a robust process that uses the fact that a particular location in the subsurface will have been sampled numerous times and at different offsets. This allows a geophysicist to construct a group of traces with a range of offsets that all sample the same subsurface location, known as a Common Midpoint Gather.[39] The average amplitude is then calculated along a time sample, resulting in significantly lowering the random noise but also losing all valuable information about the relationship between seismic amplitude and offset. Less significant processes that are applied shortly before the CMP stack are Normal moveout correction and statics correction. Unlike marine seismic data, land seismic data has to be corrected for the elevation differences between the shot and receiver locations. This correction is in the form of a vertical time shift to a flat datum and is known as a statics correction, but will need further correcting later in the processing sequence because the velocity of the near-surface is not accurately known. This further correction is known as a residual statics correction.

Seismic migration is the process by which seismic events are geometrically re-located in either space or time to the location the event occurred in the subsurface rather than the location that it was recorded at the surface, thereby creating a more accurate image of the subsurface.

The goal of seismic interpretation is to obtain a coherent geological story from the map of processed seismic reflections.[40] At its most simple level, seismic interpretation involves tracing and correlating along continuous reflectors throughout the 2D or 3D dataset and using these as the basis for the geological interpretation. The aim of this is to produce structural maps that reflect the spatial variation in depth of certain geological layers. Using these maps hydrocarbon traps can be identified and models of the subsurface can be created that allow volume calculations to be made. However, a seismic dataset rarely gives a picture clear enough to do this. This is mainly because of the vertical and horizontal seismic resolution[41] but often noise and processing difficulties also result in a lower quality picture. Due to this, there is always a degree of uncertainty in a seismic interpretation and a particular dataset could have more than one solution that fits the data. In such a case, more data will be needed to constrain the solution, for example in the form of further seismic acquisition, borehole logging or gravity and magnetic survey data. Similarly to the mentality of a seismic processor, a seismic interpreter is generally encouraged to be optimistic in order encourage further work rather than the abandonment of the survey area.[42] Seismic interpretation is completed by both geologists and geophysicists, with most seismic interpreters having an understanding of both fields.

In hydrocarbon exploration, the features that the interpreter is particularly trying to delineate are the parts that make up a petroleum reservoir – the source rock, the reservoir rock, the seal and trap.

Seismic attribute analysis involves extracting or deriving a quantity from seismic data that can be analysed in order to enhance information that might be more subtle in a traditional seismic image, leading to a better geological or geophysical interpretation of the data.[43] Examples of attributes that can be analysed include mean amplitude, which can lead to the delineation of bright spots and dim spots, coherency and amplitude versus offset. Attributes that can show the presence of hydrocarbons are called direct hydrocarbon indicators.

The use of reflection seismology in studies of tectonics and the Earth's crust was pioneered in the 1970s by groups such as the Consortium for Continental Reflection Profiling (COCORP), who inspired deep seismic exploration in other countries such as BIRPS in Great Britain and ECORS in France.[44] The British Institutions Reflection Profiling Syndicate (BIRPS) was started up as a result of oil hydrocarbon exploration in the North Sea. It became clear that there was a lack of understanding of the tectonic processes that had formed the geological structures and sedimentary basins which were being explored. The effort produced some significant results and showed that it is possible to profile features such as thrust faults that penetrate through the crust to the upper mantle with marine seismic surveys.[45]

On land, conducting a seismic survey may require the building of roads, for transporting equipment and personnel, and vegetation may need to be cleared for the deployment of equipment. If the survey is in a relatively undeveloped area, significant habitat disturbance may occur and many governments require seismic companies to follow strict rules regarding destruction of the environment; for example, the use of dynamite as a seismic source may be disallowed. Seismic processing techniques allow for seismic lines to deviate around natural obstacles, or use pre-existing non-straight tracks and trails. With careful planning, this can greatly reduce the environmental impact of a land seismic survey. The more recent use of inertial navigation instruments for land survey instead of theodolites decreased the impact of seismic by allowing the winding of survey lines between trees.

The main environmental concern for marine seismic surveys is the potential for noise associated with the high-energy seismic source to disturb or injure animal life, especially cetaceans such as whales, porpoises, and dolphins, as these mammals use sound as their primary method of communication with one another.[46] High-level and long-duration sound can cause physical damage, such as hearing loss, whereas lower-level noise can cause temporary threshold shifts in hearing, obscuring sounds that are vital to marine life, or behavioural disturbance.[47]

A study has shown[48] that migrating humpback whales will leave a minimum 3 km gap between themselves and an operating seismic vessel, with resting humpback whale pods with cows exhibiting increased sensitivity and leaving an increased gap of 7–12 km. Conversely, the study found that male humpback whales were attracted to a single operating airgun as they were believed to have confused the low-frequency sound with that of whale breaching behaviour. In addition to whales, sea turtles, fish and squid all showed alarm and avoidance behaviour in the presence of an approaching seismic source. It is difficult to compare reports on the effects of seismic survey noise on marine life because methods and units are often inadequately documented.

The gray whale will avoid its regular migratory and feeding grounds by >30 km in areas of seismic testing. Similarly the breathing of gray whales was shown to be more rapid, indicating discomfort and panic in the whale. It is circumstantial evidence such as this that has led researchers to believe that avoidance and panic might be responsible for increased whale beachings although research is ongoing into these questions.

Offering another point of view, a joint paper from the International Association of Geophysical Contractors (IAGC) and the International Association of Oil and Gas Producers (IOGP) argue that the noise created by marine seismic surveys is comparable to natural sources of seismic noise, stating:[49]

"The sound produced during seismic surveys is comparable in magnitude to many naturally occurring and other man-made sound sources. Furthermore, the specific characteristics of seismic sounds and the operational procedures employed during seismic surveys are such that the resulting risks to marine mammals are expected to be exceptionally low. In fact, three decades of world-wide seismic surveying activity and a variety of research projects have shown no evidence which would suggest that sound from E&P seismic activities has resulted in any physical or auditory injury to any marine mammal species."

In 2017, IOGP recommended[50] that, to avoid disturbance whilst surveying:

• Protective measures are employed to address site-specific environmental conditions of each operation to ensure that sound exposure and vessel traffic do not harm marine mammals.
• Surveys planned to avoid known sensitive areas and time periods, such as breeding and feeding areas.
• Exclusion zones are typically established around the seismic source to further protect marine fauna from any potentially detrimental effects of sound. The exclusion zone is typically a circle with a radius of at least 500 meters around the sound source.
• Trained observers and listening devices are used to visually and acoustically monitor that zone for marine mammals and other protected species before any sound-producing operations begin. These observers help ensure adherence to the protective practices during operations and their detailed reports provide information on the biodiversity of the survey area to the local governments.
• Sound production typically begins with a “soft-start” or “ramp-up” that involves a gradual increase of the sound level from the air gun source from a very low level to full operational levels at the beginning of the seismic lines – usually over 20 to 40 minutes. This soft-start procedure is intended to allow time for any animal that may be close to the sound source to move away as the sound grows louder.