AGC Time Constant: Definition, Seismic Gain, and Amplitude

Automatic Gain Control (AGC) is a signal-processing technique that continuously adjusts the amplification applied to a seismic trace so that the output amplitude stays roughly constant across the record length. The AGC time constant, also called the AGC gate length or operator length, defines the time window over which the algorithm measures the root-mean-square (RMS) or average absolute amplitude of the trace before applying a compensating gain factor. Short time constants, typically 100 to 200 milliseconds, produce tightly levelled traces where nearly every sample looks the same size. Long time constants, typically 500 milliseconds to 1 second or more, allow more relative amplitude variation but still remove the broad trend of amplitude decay with depth. The choice of AGC time constant governs whether the processed seismic data is useful for qualitative structural interpretation, quantitative amplitude analysis, or rock-physics inversion.

Key Takeaways

• AGC equalizes seismic trace amplitudes over a sliding time gate, improving the visual continuity of late-arriving reflections that have been weakened by acoustic attenuation and geometric spreading.
• The time constant (gate length) sets the temporal scale over which amplitude is measured and normalized: short gates (100 to 200 ms) destroy relative amplitude information; long gates (500 ms to 1 s) preserve more relative variation but still suppress true-amplitude contrasts.
• AGC is fundamentally incompatible with quantitative seismic interpretation: Amplitude Versus Offset (AVO) analysis, direct hydrocarbon indicator (DHI) mapping, and 4D time-lapse monitoring all require true-amplitude processing with no AGC applied.
• Physically meaningful alternatives to AGC include spherical-divergence correction (compensates for wavefront spreading) and surface-consistent amplitude corrections (removes acquisition-related amplitude variability without distorting relative reflectivity).
• SEG-Y headers carry scalar fields that record how amplitudes were scaled during processing; these must be checked before any quantitative amplitude work to determine whether AGC was applied and whether it can be reversed.

What Is AGC and Why Is It Used?

A seismic wave radiating from a surface source loses energy continuously as it travels into the earth. Two dominant mechanisms drive this loss. First, geometric spreading (also called spherical divergence) disperses the wavefront energy over an ever-growing spherical surface area, reducing amplitude proportionally to the inverse of the travel distance. For a wave at 2,000 m (6,562 ft) depth, geometric spreading alone reduces amplitude to approximately 0.5 percent of its value at 100 m (328 ft) depth, a factor of 200. Second, intrinsic attenuation (described by the quality factor Q) converts wave energy to heat through friction and fluid flow in the rock matrix. High-frequency energy is attenuated more rapidly than low-frequency energy, causing the wavelet to broaden and lose resolution with depth. Both effects mean that deep reflections arrive at the surface with amplitudes 20 to 60 decibels (dB) weaker than shallow reflections, even when the reflectivity coefficient is the same.

Early seismic recording hardware had limited dynamic range, typically 12 to 24 bits, and could not simultaneously display both the large-amplitude shallow arrivals and the tiny deep arrivals within the same trace window. AGC was introduced to keep the signal within the displayable range of analogue oscillographs and early digital monitors. Even with modern 24-bit analogue-to-digital converters that provide more than 140 dB of instantaneous dynamic range, AGC remains widely used for display and quality-control purposes because it makes reflection continuity visually obvious regardless of depth.

The AGC algorithm, at its core, computes the RMS amplitude within a sliding window centred on the sample of interest, then divides the sample by that RMS value (with a small noise floor added to prevent division by near-zero values). The output sample therefore has an amplitude roughly equal to the ratio of the input sample to the local RMS, standardizing each sample relative to its immediate neighbourhood. If the time constant is 200 ms and the sample rate is 2 ms, the window contains 100 samples, and the normalization removes any amplitude trend that persists over 200 ms or longer.

How the Time Constant Affects the Result

The time constant is the single most important parameter in any AGC implementation, and its effects are directly counterintuitive to interpreters who have not examined trace-level data.

At a short time constant of 100 to 200 ms, the AGC window captures only a few dominant reflection cycles. The algorithm forces every trace segment to have approximately the same RMS amplitude, regardless of whether the underlying geology is a high-impedance carbonate reef, a soft-shale halfspace, or a gas sand with anomalously high reflectivity. Direct hydrocarbon indicators (DHIs), including bright spots (high-amplitude gas reflections), dim spots (amplitude decreases over oil), and polarity reversals, are obliterated by a short AGC because the algorithm interprets the anomalously high or low amplitude as a gain excursion to be corrected rather than geological signal to be preserved. Similarly, fluid contact reflections, which derive their detectability precisely from their relative amplitude contrast, are suppressed relative to surrounding events.

At a long time constant of 500 ms to 2 s, the AGC window spans many reflection cycles and the algorithm removes only the broadest amplitude trends, such as the general increase in reflectivity at a major unconformity or the attenuation effect of a gas chimney above a leaking reservoir. Long-gate AGC preserves more relative amplitude information and is sometimes used for structural interpretation where the goal is to map fault geometry rather than to quantify reflectivity. However, even a 1-second gate removes amplitude information at the frequency of the gate: a reflector whose relative amplitude varies over 1-second windows (depth ranges of approximately 750 to 2,000 m depending on velocity) will still be distorted.

AGC vs. True-Amplitude Processing

True-amplitude processing preserves the proportionality between the recorded trace amplitude and the reflection coefficient of the subsurface interface, after accounting for the physically deterministic processes of geometric spreading and, optionally, Q compensation. The workflow begins by applying a spherical-divergence correction, which multiplies each sample by a gain proportional to the square of the two-way travel time (or by an empirically derived function of the velocity field and time). This single correction removes the largest component of amplitude decay without introducing any inter-sample normalization. Subsequent surface-consistent amplitude corrections equalize the response of individual shot points and receivers, removing hardware differences, near-surface coupling variations, and source-energy variations, while still preserving the relative amplitude of events within any single trace.

The distinction between AGC and true-amplitude processing is not merely academic. Every modern amplitude-variation-with-offset (AVO) workflow, every seismic inversion aimed at porosity or fluid saturation, every simultaneous inversion for P-impedance, S-impedance, and density, and every 4D time-lapse comparison of monitor and base surveys requires that the amplitude of the processed seismic data be a faithful proxy for subsurface reflectivity. If AGC has been applied, the amplitude information is irreversibly corrupted and none of these analyses can produce geologically meaningful results. For this reason, major oil companies including Saudi Aramco, Shell, and Equinor specify in their seismic processing contracts that a true-amplitude preserved version of every processed dataset must be delivered alongside any display-quality AGC version.

Modern full-waveform inversion (FWI) workflows used in acquisition design and processing also require true-amplitude data as input. FWI minimizes the difference between observed and modeled waveforms; any AGC normalization applied to the observed data destroys the amplitude information that drives the elastic parameter updates.