Quality checks and common mistakes¶
A checklist to go through before considering the processing complete. Almost all wrong results in seismic refraction arise from a few recurring causes, nearly always upstream of the calculation.
Quick checklist¶
| # | Check | Where |
|---|---|---|
| 1 | Number, position and spacing of the geophones match the actual spread | Data input → Geophone positions |
| 2 | The Z elevations of the geophones have been entered (if the ground slopes) | Project → Geophone positions |
| 3 | Every shot has its own source position | Data input → Source position |
| 4 | The first arrivals increase regularly with distance from the source | Traces |
| 5 | The number of layers is justified by the straight segments of the traveltime curve | Traveltime curves |
| 6 | The curve chosen for the optimum XY is the straightest | Interpretation → Velocity function |
| 7 | The final section is geologically plausible | Morphology of refractors |
| 8 | The views are set the way you want them in the report | before Generate report |
The most frequent mistakes¶
Imprecise picking¶
By far the main cause of wrong sections. A first arrival read a few milliseconds too late distorts the traveltime curve and, downstream, the morphology of the refractor.
How to recognise it: a point that on the traveltime curve "steps out" of the alignment of its neighbours; anomalous steps or cusps in the final section. Remedy: reopen Show zoom on separate window and check again trace by trace. Do not rely on Automatic picking.
Spread geometry that does not match¶
Spacing or position of the first geophone left at the default values. The distances are the X axis of every traveltime curve: if they are wrong, all the velocities are proportionally wrong.
Too many layers¶
Increasing the number of layers always improves the apparent quality of the fit, but it introduces refractors that the data do not support. Add a layer only if there is a recognisable straight segment to justify it.
Excessive filtering¶
A filter that is too aggressive rounds off the wavefront and shifts forward the first arrival. Filter the minimum necessary to make the onset legible.
Lithological name inferred from velocity alone¶
Velocity also depends on compaction, fracturing and porosity: on its own it does not identify a material. Where possible, calibrate against a borehole.
Limitations of the method: when the data are not there¶
Some situations are not processing errors but physical limitations: the software cannot flag them and the section will look "clean" all the same.
Situations in which refraction cannot see
- Velocity inversion — a slower layer beneath a faster one does not generate total refraction: it is invisible to the method.
- Hidden layer — a layer with an intermediate velocity may not produce a segment of its own on the traveltime curve.
- Thicknesses that are too small — they do not give a recognisable segment.
- Non-uniqueness — identical sets of times can be explained by different models.
For this reason, when the stakes are high, the interpretation must be calibrated against a borehole or cross-checked with other surveys (for example MASW for the Vs profile).
How to choose the spread at the design stage¶
- Small spacing → greater detail of the refractor, smaller investigated depth.
- Large spacing → greater depth, but lower accuracy on the calculated depths.
- Always provide for external shots beyond the two ends of the spread: they serve to constrain the refractors beneath the head and tail geophones.
Return to the Tutorial index or review the fundamentals in Understanding seismic refraction.