The Gold Standard of Archaeology | Multichannel Analysis of Surface Waves, Dispersion Curves and Depth Resolution

1. Begin with dispersion, not a coloured section

Back to contents

Rayleigh waves of different frequency sample different effective depths. In a layered ground, their phase velocities vary with frequency, forming a dispersion curve. MASW estimates that curve from multichannel records and inverts it for a shear-wave velocity profile. The measured object is dispersive wave energy; the subsurface model is an inference.

State the target before collecting data. A broad compacted platform may alter stiffness over wavelengths the array can sample. A thin trench fill with velocity close to surrounding soil may not. Archaeological importance does not guarantee geophysical visibility.

2. Design source, receivers and offsets as one system

Back to contents

Receiver spacing influences aliasing and the shortest wavelengths that can be represented. Spread length, source offset, source bandwidth, geophone response and record length shape the usable frequency range. Too-short offsets can create near-field effects; excessive offsets can attenuate high-frequency energy and reduce shallow detail.

Record geometry and ground coupling at every shot. Test several offsets and sources before production. A dense acquisition interval cannot undo spatial smearing caused by a long receiver spread, and a long line does not guarantee useful low-frequency energy. Field design must serve the desired depth and lateral scale.

3. Build and inspect the dispersion image

Back to contents

Processing transforms a shot gather into frequency–phase-velocity space so coherent energy appears as ridges. Analysts then pick a dispersion branch. Noise, body waves, higher modes and spatial aliasing can create competing ridges. Automatic picks should be checked against wavefields and adjacent gathers.

Publish representative gathers and images with the selected curve. Retain rejected or uncertain picks. A smooth curve can be wrong when the analyst follows the strongest mode at one frequency and another mode elsewhere. Mode identity is part of interpretation, not a hidden software detail.

4. Treat higher modes as information and risk

Back to contents

The fundamental Rayleigh mode is often targeted, but higher modes can dominate in strongly layered ground. Including correctly identified modes can improve depth information; mixing modes can generate a convincing but false velocity structure. Archaeological deposits with sharp contrasts may produce complex modal behaviour.

Test multimode interpretations and compare synthetic dispersion from candidate models with the observed ridges. If mode identity remains uncertain, propagate that uncertainty through several inversions. One preferred curve should not erase other energy supported by the data.

5. Understand why inversion is non-unique

Back to contents

Many combinations of layer thickness and shear-wave velocity can fit a finite, noisy dispersion curve. Inversion therefore uses starting models, bounds, layer parameterisation and regularisation. A low misfit shows compatibility with picked data; it does not prove that the recovered layering is unique.

Run families of plausible models and report their spread. Examine sensitivity to density or compressional-velocity assumptions where relevant. Use smooth and layered parameterisations that match the research question. A model with more thin layers can fit better simply because it has more freedom.

6. Worked example: a supposed buried road

Back to contents

In an invented survey, a ridge runs from 180 metres per second at 25 hertz to 420 metres per second at 8 hertz. A simple inversion places a faster layer near three metres depth. If the useful wavelength at the low-frequency end is roughly 420 divided by 8, or 52.5 metres, that energy samples a broad depth range rather than a precise three-metre boundary.

A second inversion with a gradual velocity increase fits nearly as well. A cross-line shows the fast zone extending beyond the expected road alignment. The defensible result is a broad stiffness increase; a buried road remains one hypothesis until resistance, coring or excavation supplies geometry and material evidence.

7. Translate spread length into lateral resolution honestly

Back to contents

A 2-D MASW section commonly assembles many 1-D profiles located near spread midpoints. Each profile represents energy sampled across much of the receiver spread. Features narrower than that footprint can be blurred, shifted or averaged with their surroundings. Smaller roll-along steps add redundancy but do not restore detail already smeared.

Report the effective footprint alongside pixel size. Test synthetic targets close to expected ditch, wall or palaeochannel widths. Interpolation between profile centres is display, not new measurement. A fine colour grid should never be described as centimetre-scale archaeological resolution.

8. Changed-context transfer check: open field to stone courtyard

Back to contents

Move from an open field with consistent coupling to a paved courtyard inside a monument. In the field, long spreads and low-frequency sources may support deeper profiling. On paving, coupling, reverberation, limited aperture and irregular boundaries alter the wavefield, while the archaeological target may be shallower.

The transferable rule is to rebuild the acquisition test for the new wavefield. Shorter arrays may improve access but reduce depth and dispersion resolution. Complementary ground-penetrating radar or resistance data can constrain geometry without pretending that one method repairs another automatically.

9. Know the limits and common misconceptions

Back to contents

Shear-wave velocity reflects stiffness and density-related structure, not archaeological identity. High velocity is not automatically stone and low velocity is not automatically a void. Saturation, compaction, weathering, roots and modern fill can change the model. Resolution worsens where usable wavelength coverage is narrow.

Depth of investigation is not a single fraction of maximum wavelength valid everywhere. It depends on signal, mode, parameterisation and acceptable uncertainty. MASW sections are not photographs, and a good curve fit cannot validate a wall label.

10. Audit field record, mode pick and inversion together

Back to contents

Archive raw gathers, geometry, elevations, source and receiver settings, coupling notes, preprocessing, dispersion images, all picks with uncertainty, mode decisions, starting models, bounds, regularisation, misfit and accepted model ensembles. Preserve software and code versions.

Repeat inversion across plausible picks and parameterisations. Compare adjacent gathers and crossing lines. Validate with independent observations positioned to test the proposed feature. Readers should see which structures are required by the data, which are regularised and which come from archaeological expectation.

11. Frequently asked questions

Back to contents

Does MASW measure depth directly? No; depth comes from inversion of dispersion. Is the strongest ridge always the fundamental mode? No. Does a smaller acquisition step give finer resolution? It adds redundancy but cannot beat the spread footprint by itself. Can MASW find walls? It can detect stiffness contrasts compatible with walls, not wall identity.

Why do different models fit the same curve? The inverse problem is non-unique and data cover a limited wavelength range. Are cross-lines useful? Yes, because they test lateral continuity and orientation. Must MASW stand alone? No; integration is often essential.

12. The gold standard reports the recoverable velocity structure

Back to contents

MASW is valuable because surface waves carry depth-dependent information across an efficient multichannel survey. Its strength becomes credible only when dispersion energy, modal interpretation, resolution and model families stay visible.

The gold standard designs the spread for the archaeological scale, publishes uncertain picks, tests non-uniqueness and describes velocity before naming a feature. That turns a colourful section into a bounded inference that excavation and complementary evidence can genuinely test.