Multiples in GPR data are not very common except in very specific scenarios such as ice profiling and underground mining but do occur in a couple everyday situations that may surprise you.
The vast majority of GPR data is created when GPR signals reflect once after travelling from the GPR transmitter before arriving at the GPR receiver (Figure 1).
Figure 1
Most GPR cross-sections (right) show GPR signals that have reflected once from a subface object or boundary.
However, in some situations, it is possible for GPR signals to reflect two, three or more times from the same object or boundary before arriving at the GPR receiver. These types of reflections are called “multiples” . Identifying multiples in your GPR data is one of the most difficult interpretations.
In this TIPS, we discuss the conditions necessary for multiples to occur in GPR data and show some examples, including an everyday example and others in very special circumstances.
Ice
The most common example of multiples in GPR data is when the GPR signal enters a layer with large contrasts in dielectric permittivity at both the top and bottom of the layer. This results in large reflectivity values at both interfaces and much of the GPR energy becomes essentially “trapped” in the layer and reflects up and down multiple times. A good example of this can be seen in ice thickness data (Figure 2).
Figure 2
The high reflectivity at both boundaries above and below ice produces the perfect conditions for GPR signals to reflect multiple times.
Ice has a dielectric permittivity of 3.2 while the air above the ice has a dielectric permittivity of 1, and the water below the ice has a dielectric permittivity of 80. When the GPR signal travels to the bottom of the ice, it encounters the ice-water interface with a reflectivity of 67%, which means about 2/3 of the energy reflects back up into the ice. Then, when the GPR signal reaches the ice-air interface at the top of the ice, the reflectivity is 28%, resulting in a significant amount of signal reflecting back down into the ice, where the process can repeat multiple times until the signal is attenuated.
While the GPR data looks like there are several layers (Figure 3), there is really one layer that the GPR signal has reflected from several times at longer and longer travel times.
The fact that the layers perfectly mimic one another (Figure 3) is a characteristic of a multiple to look for to identify multiples.
Figure 3
Multiples in ice thickness data. The second and third reflections mimic the first reflection in time so a difference of 1ns in the first reflection is 2 ns in the second reflection and 3ns in the third reflection. This means that thickness variations such as the V-shaped area in the box gets more and more exaggerated as the number of multiple reflections increase.
Water Puddles
Another, more common place to see a similar response is multiples from a water puddle (Figure 4). When a water puddle is deep enough to submerge both GPR antennas in the water, multiples can occur.
Figure 4
Water puddles, with high reflectivity at both boundaries above and below the water produce another scenario for GPR signals to reflect multiple times.
Water has a dielectric permittivity of 80 while the air above the puddle has a dielectric permittivity of 1 and the asphalt below the puddle has a dielectric permittivity of 6. The water to asphalt reflectivity at the bottom of the puddle is 57% while the water to air reflectivity at the top of the puddle is 80%, resulting in the conditions to produce multiples.
Figure 5
GPR signal multiples in a water puddle. Note that the water depth is greatly exaggerated to show the path of the GPR waves.
Multiples when crossing a puddle with GPR tend to produce much more complex responses compared to ice (Figure 6) because puddles are often small enough that the GPR collects data from the edges of the puddle, where the water depth goes to zero (Figure 5).
Figure 6
GPR signal multiples in a water puddle. While it looks like penetration depth has increased under the puddle, the GPR signal is mostly trapped in the water layer, reflecting up and down multiple times. The varying water depth across the puddle and the edges of the puddle together generate a complex pattern of responses. Note the velocity pull down in the direct ground arrival due to the very slow velocity of water (0.033 m/ns) compared to asphalt (0.13 m/ns).
The danger of water puddle multiples for GPR operators is two-fold; 1) misinterpretation that the multiples represent a real subsurface target and 2) that these signals mask the reflections from real subsurface objects below the puddle.
Understand that this multiple response will NOT occur when the ground is simply wet. The water depth must be enough to submerge both the transmitting and receiving antennas in the water.
We discussed this phenomenon in more detail in the Subsurface Views newsletter from April 2011:
The same principles described here can cause non-metallic pipe multiples (Figure 7). Since the pipe is non-metallic, GPR signals can enter the pipe and reflect from the bottom. Some of that energy will then encounter the high reflectivity interface at the top of the pipe and reflect downwards again. GPR energy can reflect one or more times within the pipe, producing hyperbolas that mimic the hyperbolic responses from the top and bottom of the pipe . This effect is very pronounced in water-filled pipes (as GPR signals travel slowly in water and therefore the travel-times in water are longer).
Figure 7
GPR signal multiples in a non-metallic pipe. The GPR cross-section on the right shows the hyperbolic responses from the top and bottom of the pipe followed by the first and second multiples. The hyperbolas are equally spaced vertically because the travel time increases by exactly two pipe diameters for each reflection.
The difference in travel time between the reflections can be used to determine the approximate pipe diameter (but only when the material in the pipe is known). We discussed this in a story in our January 2020 newsletter (https://www.sensoft.ca/blog/tips-determining-pipe-diameter-from-gpr-data/).
Underground Mines
Our last example comes from data collected in the tunnel of an underground mine. A customer, Compass Minerals, sent us a terrific example of low frequency 100 MHz data collected in a salt mine penetrating about 15 meters (50+ feet, Figure 8).
Figure 8
100 MHz center frequency data collected down a long tunnel in an underground salt mine. Notice the higher frequency content, slightly undulating reflectors at about 50 and 100 ns (2.5 and 5 m depth). These are multiples from the ceiling of the mine tunnel
The reflectors at about 50 and 100 ns are multiples from the ceiling of the tunnel.
Figure 9
Two multiples, reflections from the tunnel ceiling are annotated in the GPR cross-sections in Figure 8 and 10.
One reason why these are likely multiples is because they cut across geological structures (red box on Figure 10), which would be unlikely if the reflection was caused by a geological structure.
Figure 10
Same data as Figure 8 but with annotations to highlight the details. The first multiple is about 50 ns and the second at exactly twice the time, 100 ns. One clue that these are multiples is that they cut across geological features (red box). Note the weak reflector indicated at the bottom of the section, mimics the strong reflector 50 ns above it.
Since the speed of GPR signal in air is known (speed of light – 0.30 m/ns), like the multiples in the non-metallic pipe example above, we can calculate the height of the tunnel:
An interesting interpretational exercise is to determine if the deeper, weaker, undulating reflector at 12-to-15-meter depths, that mimics the stronger reflector at 9-to-12-meter depths, is a real reflector or a multiple. If you notice that the time difference is 50 ns, exactly the same as the multiples higher in the section, you quickly come to the conclusion that it is caused by the GPR energy that reflected from the ceiling and penetrated the subsurface to reflect from the real reflector at 9-to-12-meters, with a delay of 50 ns – the time it took to reflect from the ceiling. This path is animated by the black arrows in Figure 11.
Figure 11
The strong, undulating reflector at 9-to-12 meters in Figure 10 is created by the GPR energy following the path of the blue arrows in this animation. The deeper, weaker reflector indicated at the bottom of Figure 10 is likely not a real reflector, but one caused by the GPR signal reflecting from the ceiling of the tunnel before entering the subsurface and reflecting from the strong, undulating reflector at a depth of 9-to-12-meters (black arrows). This animation shows the paths of the GPR signals that combine to generate the GPR cross-section in Figures 8 and 10.
Multiples are not often visible in GPR data, but in situations with a high reflectivity boundary, like these examples, watch out for them and be careful not to interpret them as real, subsurface reflectors.
