Seismic refraction is a geophysical principle governed by Snell's Law. Used in the fields of engineering geology, geotechnical engineering and exploration geophysics, seismic refraction traverses (seismic lines) are performed using a seismograph(s) and/or geophone(s), in an array and an energy source. The seismic refraction method utilizes the refraction of seismic waves on geologic layers and rock/soil units in order to characterize the subsurface geologic conditions and geologic structure.
Seismic refraction maps contrasts in seismic velocity – the speed at which seismic energy travels through soil and rock. This parameter typically correlates well with rock hardness and density, which in turn tend to correlate with changes in lithology, degree of fracturing, water content, and weathering.
There are two basic approaches to seismic refraction data analysis: layer-cake and tomographic inversion. The former is the more traditional approach, although tomography has become more popular as faster computers have made it much more feasible than in the past.
Especially in the near-surface, it is not always the case that seismic velocities are divided into high-contrast, discrete layers. Nor is it the case that velocities are constant horizontally. Conventional layer-cake inversion techniques, such as the delay-time method, assume both, and require the geophysicist to provide layer assignments before the data inversion can be completed.
Tomography is less constrained in this sense; it does not “think” in terms of layers, and it better accommodates horizontal velocity variations. If discrete layering is not apparent in the raw data, the tomographic approach is generally more appropriate. As such, Geometrics’ SeisImager Refraction Analysis software offers both options.
Estimating rippability prior to excavation
Mapping depth to bedrock/bedrock topography
Mapping depth to ground water
Calculation of elastic moduli/assessment of rock quality
Mapping thickness of landslidesIdentification and mapping of faults
Seismic survey involves deploying shots at the surface and recordings made via a linear array of sensors (geophones or hydrophones). The shots may be a shotgun, accelerated weight drop or explosive. Refracted seismic signals travel laterally through the higher velocity layer (refractor) and generates a ‘head-wave’ that returns to surface. Beyond a certain distance away from the shot, the signal that has been refracted at depth is observed as first-arrival signal at the geophones. Observation of the travel-times of refracted signal from selectively deployed shots enables derivation of the depth profile of the refractor layer. Shots are typically fired at locations at and beyond both ends of the geophone spread and at regular intervals along its length.
The results of the seismic refraction survey are usually presented in the form of seismic velocity boundaries on interpreted cross-sections. Seismic sections represent the measured bulk properties of the subsurface and enable correlation between point source datasets (boreholes/trialpits) where underlying material is variable. Reference to the published seismic velocity tables enables derivation of rippability values.
The data processing is carried out using specialist seismic software. The first stage involves accurate determination of the first-arrival times of the seismic signal (time from the hammer blow to each recording hydrophone) for every shot record, using PICKWIN. Time-distance graphs showing the first-arrival times were then generated for each seismic shot record and analysed using PLOTREFA software to determine the number of seismic velocity layers. Modelled depth profiles for the observed seismic velocity layers are produced by a tomographic inversion procedure that is revised iteratively to develop a best fit-model. The final output of a seismic refraction survey is a velocity model section of the subsurface based on an observed layer sequence with measured velocities that correspond to physical properties such as levels of compaction/ saturation in the case of sediments and strength/rippability in the case of bedrock.
The depth of penetration in a seismic refraction survey is approximately 1/5th of the length of the geophone spread, including offset shots. So if you need to see 10m deep, you will need room to lay out a (minimum) 50m seismic spread, as measured from offset shot to offset shot.
For most engineering refraction work, the best possible source is a 14 or 16lb sledgehammer. A downhole seisgun is not a good refraction source in general, except in cases where the surface is too soft to use a hammer effectively. An accelerated weight drop can be a good source, but is not portable and requires vehicle access to the shot points. Small explosives, such as Kinepak, are ideal when portability is required and the depth of interest is greater than what can be reached with a hammer.
Any Geometrics seismograph can be used for seismic refraction. For 24 channels or less, the SmartSeis ST is ideal. If using a laptop in the field works for you, the ES-3000 is a good alternative to the SmartSeis. For larger surveys, the Geode is recommended. For simple rippability, 12 channels will often suffice. For mapping bedrock topography, at least 24 channels are recommended. In general, the more detail required, the more channels you need.
A hammer-and-plate refraction survey is easily accomplished with two people. Longer lines and/or the use of explosives (which requires digging shot holes) generally require 3-4 people.
Seismic refraction requires that velocities increase with depth. A lower velocity layer beneath a higher velocity layer will not be detected by seismic refraction, and will lead to errors in depth calculations. Fortunately, this is a fairly uncommon occurrence in the shallow subsurface.
The seismic source employed must match the desired depth of penetration. For hammer and plate work, the maximum depth you can expect to explore to is about 15-20m; however, this can vary significantly depending on geology, surface conditions, cultural noise, and the person swinging the hammer.
Refraction is a relatively broad-brush technique – it looks at gross velocity differences, and you should not expect to be able to map more than 3-4 individual velocity layers.
Cultural noise can be a problem – it is more difficult to conduct a seismic survey in an urban environment than in a rural one. Surveying along busy roadways should be avoided when possible. Shooting at night is sometimes necessary in order to achieve acceptable signal-to-noise ratio in busy areas.
Layer velocity (density) must increase with depth; true in most instances. Layers must be of sufficient thickness to be detectable. Data collected directly over loose fill (landfills) or in the presence of excessive cultural noise may result in sub-standard results. In places where compact clay-rich tills and/or shallow water overly weak bedrock an S-wave survey may be used to profile rockhead where insufficient velocity contrast may prevent use of a P-wave survey.
The final product of a refraction survey is a velocity model, such as the layer-cake inversion shown below,