Description

AVA 1D Decon L2 performs an amplitude-versus-angle (AVA) inversion of a single pre-stack angle (or offset) gather - such as a common image gather from a CDP or migrated point - to estimate relative changes in P-wave velocity, S-wave velocity and density with time. It links the observed amplitude behaviour across angles to the underlying elastic-parameter reflectivity using a linearized Aki-Richards approximation, and "deconvolves" the effect of the supplied wavelet using a least-squares (L2-norm) solver, so that the estimated properties are corrected for wavelet shape rather than just raw amplitude changes. Use it to turn an angle gather into quantitative AVA attributes (relative Vp, Vs and density reflectivity) for lithology and fluid discrimination, or to QC how well a simple elastic model can reproduce the observed amplitude-versus-angle behaviour at a given location.

Input data

Input gather

The pre-stack angle or offset gather to be inverted - typically a single CDP or common-image-point gather where each trace corresponds to a different incidence angle (or offset used as an angle proxy). The trace offset (or angle) value stored in each trace header is used directly by the inversion, and every trace must share the same sample interval as the input wavelet.

Input wavelet

A single-trace wavelet (for example, one produced by a wavelet extraction or estimation module) that represents the source signature imprinted on the input gather. It is used to build the convolution operator that links the underlying reflectivity to the observed amplitudes, so the estimated elastic-parameter changes are effectively deconvolved from the wavelet shape. The wavelet trace must have exactly the same sample interval as the input gather - if it does not, the module stops with an error asking you to supply a matching wavelet.

Parameters

Noise

Sets the amount of damping (regularization) applied to the least-squares inversion, expressed as a fraction/percentage. Raising this value stabilizes the solution and reduces sensitivity to noisy or inconsistent amplitude-versus-angle behaviour, at the cost of resolving power; lowering it lets the solver fit the observed amplitudes more tightly, which can amplify noise if the gather is not well conditioned. This control only affects the "each sample" decon type described below - when "all trace" is selected, the module applies its own fixed internal regularization instead and this value is ignored.

Default: 0.01 (about 1%). Increase it if the inverted Vp/Vs/density traces look erratic or overly sensitive to noisy far-angle traces.

Number of Iteration

Sets the maximum number of iterations the least-squares solver performs while converging on a solution. In "each sample" mode this is the iteration limit used for every individual time sample. In "all trace" mode the full-profile inversion runs an internal five-pass reweighting scheme (to sharpen boundaries), and this value controls how many solver iterations are allowed within each of those five passes. Higher values allow the solver to converge more fully and can improve accuracy, but increase processing time; lower values run faster but may stop before fully converging, especially on noisy or poorly conditioned data.

Default: 100 (minimum allowed value is 1). The default is usually sufficient; increase it only if the result still looks unstable or you suspect the solver has not converged.

Scale Window

Defines the half-length (radius) of a symmetric time window used to compute a running RMS amplitude level around each time sample, across all traces of the gather. Before inversion, every sample is normalized by this local RMS value, which compensates for overall amplitude trends with time (for example, wavelet strength or gain differences) so that the amplitude-versus-angle behaviour analyzed by the inversion reflects true angle-dependent reflectivity rather than time-varying gain. The actual averaging window spans roughly twice this value (this many units before and after each sample).

Default: 0.1 s (100 ms as shown in the interface). Increase it for a smoother, more stable amplitude-normalization trend on noisy data; decrease it if strong, closely spaced reflectors are being over-smoothed by the normalization.

Decon type

Chooses the inversion strategy used to estimate the elastic-parameter changes. "each sample" solves an independent, small least-squares problem at every time sample using only the amplitudes recorded at that sample across all angles - it is fast and simple, but treats every sample separately with no vertical continuity constraint. "all trace" instead builds a single operator for the entire 1D profile that combines the angle-dependent Aki-Richards relationship, the wavelet convolution, and a vertical derivative term, and solves it with an iteratively reweighted least-squares scheme that favours a blocky (step-like) solution - this produces sharper, more geologically continuous boundaries in the output logs and also returns a reconstructed synthetic gather for comparison against the input.

Default: "each sample". Switch to "all trace" when you want a blockier, more interpretable elastic-log style result and a QC synthetic gather, at the cost of longer processing time (it repeats the solve five times internally to sharpen the result).

Output data

Output gather

A QC gather whose content depends on the selected decon type. With "each sample", it carries the per-sample residual/error level reported by the solver, useful for spotting time intervals where the fit to the observed amplitudes was poor. With "all trace", it instead carries a synthetic gather rebuilt by forward-modelling the inverted Vp, Vs and density changes through the same wavelet and angle operator - compare it visually with the input gather to judge how well the inversion explains the observed amplitude-versus-angle behaviour.

delta VP

A single-trace log of the estimated relative change (reflectivity-style perturbation) in P-wave velocity at every time sample of the input gather, as recovered by the AVA inversion. Peaks and troughs mark time levels where the amplitude-versus-angle response indicates a P-wave velocity contrast.

delta Vs

A single-trace log of the estimated relative change in S-wave velocity at every time sample, recovered alongside delta VP and delta Density from the same inversion. Because S-wave behaviour largely controls how amplitude varies with angle, this output is particularly diagnostic for AVO/AVA-based lithology and fluid analysis.

delta Density

A single-trace log of the estimated relative change in bulk density at every time sample. Density changes are generally the weakest and hardest-to-resolve component of the Aki-Richards amplitude response, so treat this output as lower-confidence than delta VP and delta Vs, especially when the input gather has a narrow angle range or a low signal-to-noise ratio.

ERR

A reserved output channel intended to carry a per-sample inversion error/residual curve as a QC companion to the Vp/Vs/density outputs. For general QC of the inversion quality, rely primarily on the Output gather described above, which is populated with the solver's residual (in "each sample" mode) or a reconstructed synthetic gather (in "all trace" mode).

References

Aki, K. and Richards, P. G., 1980, Quantitative Seismology: Theory and Methods: W. H. Freeman and Co.