The sensitivity of seismic refraction velocity models to survey geometry errors, assessed using Monte Carlo analysis

Seismic refraction models should routinely be reported with their associated uncertainty. Tomographic solutions are widespread, but estimating uncertainties in these via Monte Carlo simulation places great demands on computer resource, hence this task is often omitted. By considering the Plus-Minus method of seismic refraction interpretation, we use Monte Carlo simulations to evaluate the uncertainty in seismic refraction results and determine the sources of uncertainty that are most impactful on the reliability of the output model. Our analysis considers the impact of survey mislocation (i.e., geophones misplaced from a planned position) and interpretational problems (i.e., misidentification of first-break picks and uncertainty in identifying crossover distances) on the overall uncertainty in inferred unit thicknesses and seismic velocities. These are considered for synthetic data with varying subsurface velocity structure, and for field data collected at a shallow (< 50 m) bedrock site in north Wales (UK). Analysis of synthetic data shows that the impact of the aforementioned errors on thickness estimates is ∼1000 times that on velocity estimates. Of all permutations tested, the most significant impact on thickness uncertainty was the accuracy of first-break picks, with the variance in target thickness estimates increasing roughly exponentially with first-break pick uncertainty. It is therefore prudent to minimise such uncertainty through appropriate survey practice (e.g., maximising source energy, taking multiple shots for stacking) and to properly define the resultant uncertainty in unit thickness and velocity estimates. The simplicity of the Plus-Minus method makes it an effective tool for highlighting the errors that would impact more sophisticated interpretation approaches, such as tomography or Full Waveform Inversion. The results from such analysis can be directly applied in straightforward environmental or engineering investigations and can be used to inform more advanced refraction methods. As such, the practice we highlight should be considered for any refraction interpretation.