Abnormal Events

In reflection seismic acquisition and processing, abnormal events are coherent or incoherent signals recorded on seismic traces that are not primary reflections from subsurface horizons. The term covers diffractions (from point scatterers like faults or boulders), multiples (reflections that bounce two or more times between reflectors before reaching the receiver), refractions (energy that travels along a high-velocity boundary), surface waves (ground roll, Rayleigh waves), guided waves, direct arrivals (energy traveling through the air or along the surface), air-coupled waves from the seismic source, and various forms of ambient noise from wind, traffic, or industrial activity. Despite the label suggesting these signals are exceptions, abnormal events often dominate raw seismic shot records and represent one of the main challenges in producing a clean, interpretable seismic image. Each type of abnormal event has a characteristic pattern of arrival times (move-out) across a receiver array, and understanding these patterns allows the processor to design filters that attenuate them without damaging the primary reflections.

Key Takeaways

Multiples are the most damaging abnormal event for deep exploration targets. A primary reflection bounces once off a reflector and returns to the surface. A multiple bounces at least twice: it might reflect off the seafloor, travel back up to the surface, reflect downward again, bounce off the target reflector, and return to the receiver. The multiple arrives at almost the same time as a primary from a deeper horizon but carries no new geological information. In marine seismic data, the sea surface is a nearly perfect reflector, generating strong multiples from every shallow horizon. Demultiple methods (surface-related multiple elimination, SRME; parabolic Radon filtering; predictive deconvolution) are among the most computationally intensive steps in seismic data processing.
Diffractions are generated by any abrupt discontinuity in the subsurface: a fault plane, a channel edge, a salt flank, a buried boulder, or a cavern. When a seismic wave hits a point-like scatterer, energy radiates in all directions from that point. On a seismic section, diffractions appear as hyperbolic tails emanating from the diffractor location. Migration (the computational process of repositioning seismic energy to its true subsurface location) collapses diffractions to their point of origin. Diffractions are actually useful: their shape on unmigrated data encodes the seismic velocity at that depth, and diffraction-based velocity analysis is a growing area of seismic processing research.
Surface waves (ground roll) are a persistent problem in land seismic surveys. Ground roll is a low-frequency, high-amplitude Rayleigh wave that travels along the earth's surface at speeds of 200 to 600 metres per second, much slower than P-wave reflections (2,000 to 5,000 metres per second). On a seismic record, ground roll appears as a cone of low-frequency energy dominating the short-offset traces. It masks shallow reflections and must be attenuated using geophone arrays (spatially averaging out the coherent slow-velocity noise), high-pass filters, or specialized ground roll removal algorithms (f-k filtering, polarization filters). Ground roll is particularly severe in areas with thick near-surface soil: prairie acquisitions in Saskatchewan and Alberta are affected when moisture content raises the Rayleigh wave amplitude.
Direct arrivals are energy from the seismic source that reaches the receivers by traveling through the shallowest layer without reflecting off any deeper boundary. They arrive first among all recorded events (hence the term "first breaks") and are used to build a near-surface velocity model by first-break tomography or refraction statics analysis. Once used for near-surface velocity estimation, direct arrivals are muted (blanked out) from the data before subsequent processing steps because they interfere with shallow reflections.
Ambient noise (wind noise, traffic vibration, power line interference at 50 or 60 Hz) is a time-variable, spatially incoherent abnormal event. Unlike multiples or surface waves, ambient noise does not have a predictable move-out pattern that can be exploited for filtering. It is attenuated by stacking many traces that share the same subsurface reflection point (CDP stacking), which reduces incoherent noise by the square root of the fold (number of contributing traces). In areas of high ambient noise (near industrial facilities, highways, or windy open plains), high fold is the main defence, requiring dense receiver and shot patterns that increase acquisition cost.

Why Abnormal Events Matter for Exploration

A geologist looking at a seismic section wants to see reflections from rock boundaries. An abnormal event is noise in that context: it obscures the reflections, creates false structural highs or lows, and may be misinterpreted as a real geological feature. The history of seismic exploration includes embarrassing cases where bright amplitude anomalies that looked like gas sands turned out to be multiple reflections from a shallower boundary, or where apparent anticlines were created by velocity pull-up below a high-velocity body such as a salt dome.

Proper identification and attenuation of abnormal events during processing is therefore not just a technical quality step but a direct safeguard against geological misinterpretation. The processor's job is to deliver data where what the interpreter sees is as close as possible to the earth's true reflectivity, free of the artifacts introduced by multiple bounces, surface waves, diffractions, and noise.

The tricky part is that some abnormal events carry geological information. Diffractions from a fault reveal the fault location. Guided waves in a low-velocity channel can illuminate the geometry of that channel. The line between "noise to suppress" and "signal to use" is blurry, and the best processors and interpreters think carefully about what they are attenuating before applying aggressive filters.

Fast Facts

The recognition that seismic records contained significant non-reflection energy dates to the earliest days of exploration seismology. Harry Mayne's invention of common mid-point (CMP) stacking in 1956 (commercialized in the early 1960s) was driven largely by the need to attenuate multiples: different multiples and primary reflections have different move-out velocities, so stacking with the correct primary velocity adds primary reflections coherently while attenuating multiples that arrive at different times on different traces. The development of digital recording (replacing analog film) in the 1960s allowed computer-based filtering of surface waves and multiples that was impossible to do on film. The introduction of high-performance 2-D and 3-D surveys in the 1970s and 1980s provided the higher fold needed to suppress incoherent noise through stacking. Modern processing addresses abnormal events with a suite of specialized algorithms that would not have been computationally feasible before the 2000s.

Identifying Abnormal Events on Seismic Records

On a raw seismic shot record, experienced processors look for characteristic patterns to identify abnormal event types. Ground roll shows as a linear fan of energy centered on the source location, with very slow apparent velocity (short move-out across the receiver array for a given time increment). Primary reflections show as gently curving hyperbolas with high apparent velocity. Multiples show as steeply curving hyperbolas with intermediate velocities between primary reflections and surface waves.

F-k (frequency-wavenumber) domain analysis separates events by their apparent velocity: the slope of a line in the F-k domain corresponds to the apparent velocity of that event in the time-space domain. Ground roll occupies a narrow wedge of low wavenumber (slow velocity) in the F-k plane and can be filtered out with a dip-selective F-k filter that removes the low-velocity content without affecting the primary reflections at higher velocities.

In shallow seismic surveys used for engineering investigations or near-surface hazard assessment, the separation and analysis of abnormal events (especially refractions and surface waves) is actually the goal of the survey, not a problem to be removed. MASW (Multichannel Analysis of Surface Waves) deliberately analyses surface wave dispersion to build a near-surface shear-velocity model, which is used in seismic hazard assessment and ground characterization for construction.

Abnormal events are also called seismic noise, coherent noise (for events with recognizable patterns like multiples and ground roll), or incoherent noise (for random noise). Related terms include multiple (a seismic reflection that has bounced two or more times between reflectors before reaching the receiver; one of the most serious abnormal events in marine seismic data; attenuated by SRME and Radon filtering), diffraction (seismic energy scattered from a point discontinuity such as a fault tip or buried boulder; appears as a hyperbola on a seismic section; collapsed to its true location by seismic migration), ground roll (low-frequency, high-amplitude Rayleigh surface waves that travel along the near-surface earth; a major abnormal event in land seismic data; attenuated by geophone arrays and f-k filtering), seismic migration (the computational process that repositions seismic energy from its recorded position to its true subsurface location; collapses diffractions and corrects dipping reflectors; part of correcting the effects of abnormal events on seismic images), and signal-to-noise ratio (the ratio of the primary reflection energy to the combined energy of abnormal events and random noise; a key measure of seismic data quality; improved by higher fold, better acquisition geometry, and effective noise attenuation processing).

How Misidentified Multiples Caused a Dry Hole in Offshore West Africa

An exploration company was evaluating a deepwater prospect off the coast of Nigeria at approximately 1,800 metres water depth. The 3-D seismic data showed a bright amplitude anomaly at 3.4 seconds two-way travel time (TWT) on a structural high, with a flat spot (horizontal reflection at the base of the anomaly) that the interpretation team identified as a gas-water contact. Amplitude versus offset (AVO) analysis showed a Class IIp (phase reversal with increasing offset) AVO response, consistent with a gas-saturated sand. The prospect was mapped at 120 million barrels of recoverable oil equivalent.

The exploration well was drilled. At the target depth, the drilling team found a 4-metre interval of hard carbonate (chalk stringers), not the porous sand reservoir the seismic suggested. The "bright amplitude" was the impedance contrast between the chalk and the overlying shale, not a gas-sand reflection. The "flat spot" was actually a peg-leg multiple: a reflection that had bounced off the seafloor, traveled back up, reflected off a shallow high-impedance carbonate at 0.9 seconds TWT, and then traveled back down and reflected off the same shallow carbonate again, arriving at the 3.4-second event at exactly the time and moveout expected of a gas-water contact at that depth.

The processing team, when reviewing the data after the dry hole, identified the peg-leg multiple by its characteristic repeat period: the two-way time of the shallow carbonate (0.9 seconds) multiplied by an integer is exactly 3.6 seconds, and the 3.4-second event was within processing uncertainty of that period. In the original processing flow, the multiple attenuation step had used a SRME algorithm that was optimized for the dominant seafloor multiple but had not predicted this lower-amplitude peg-leg path. The dry hole cost USD 85 million. The post-mortem led to a requirement for multiple-model-based attenuation validation before any drill decision on a bright amplitude prospect in deepwater settings.