Radar is a system that uses electromagnetic waves to identify the range and direction of objects. A radar system has a transmitter that emits electromagnetic waves that are reflected by the target and detected by a receiver. The time difference from transmission to reception of the echo is proportional to the range.
The radar range resolution is the ability of the radar to distinguish two targets with similar ranges. The resolution is determined by the signal bandwidth, BW. The smallest time interval that the radar can resolve is the inverse of the bandwidth: τ=1/BW which gives a range resolution of cτ/2 = c/(2BW), where c is the signal velocity (30 cm / ns in air). For example, a typical navigation radar will have a maximum bandwidth of approximately 15 MHz (corresponding to a pulse width of 70 ns), giving a resolution of 10 m. There are many ways to generate high bandwidth. One method is to use short pulses of very high power, like most navigation radars do. Another alternative is to sweep through a wide frequency range over time. The figure below shows the resulting signal, (using 3 different bandwidths) when two echoes of different range are received.
Figure 1: Radar range resolution. The ability to separate objects is improved as resolution improves.
Ground Penetrating Radar (GPR) is a technique used for subsurface mapping based on emission of radio waves. The required resolution is very high (in the order of millimeters) compared to other radar types. The bandwidth of ground penetrating radars is typically in the range 200 MHz to several GHz (corresponding to a time resolution from 5 ns to less than 0.5 ns). On the other hand, penetration of radio waves into the ground decays rapidly with increasing frequency, so the center frequency needs to be as low as possible. Impulse based GPR systems usually have a bandwidth equal to the center frequency, that is, ranging from 0.5 fc to 1.5 fc. To achieve higher resolution (high bandwidth) the center frequency must be increased, thus reducing penetration. The 3d-Radar GeoScopeTM can generate waveforms from 100 MHz up to 2 GHz or 3 GHz in less than a millisecond. Therefore, there is no tradeoff between penetration and resolution. The step-frequency radar has a coherent receiver which means that the whole waveform length is used as 100% efficient integration time. By comparison, impulse GPRs use stroboscopic sampling with significant loss of energy.
In GPR, the targets of interest are subsurface layers, utilities or other objects. A two dimensional image is generated as the radar is moved along the ground surface. A subsurface object will generate an echo that is received in the radar. At first, the range is rather long, but as the radar moving towards the object, the range is reduced until the radar is directly on top of the object. Thereafter, the range will increase once again. The resulting echo path will be a hyperbola (much like the hyperbolas observed on fishfinders). If the signal wave speed is known, the resulting image may be “focused” in a process called migration, where the data are summed along the path of the hyperbolas.
Figure 2: Focusing of images by migration removes hyperbolas and sharpens the image.
Since the resultant images are basically vertical slices through the ground, they might seem a little bit confusing to the layman. There is no way of telling the shape of the object in 3 dimensions. For example, there is no way of telling if the hyperbola above (or in the focused image: “dot”) is a pipe or a localized object. To do that, we need 3-dimensional imagery which is the case for many applications.
One way of achieving 3D imagery is of course to perform multiple scans, side by side with a single antenna. In order to achieve good resolution, the spacing between scans might be as low as 5 cm (2″). This is an extremely time consuming task. In some applications, this method is not even an option. For instance, with anti-tank mine detection, going for a second scan on top of the mines is not feasible. The alternative is, of course, to use an array of closely spaced antennas and gather data from a wide swath in one pass. The figure below shows the same targets as previously along with depth slices at two different depths. It is obvious that one of the objects is a round body (in fact a 30 cm AT mine) and the other is a pipe.
Figure 3: The horizontal slice gives the object outline. The two topmost images are horizontal slices on different depths (Blue lines in bottom images). Bottom images are vertical slices.
The need for 3-dimensional imagery is even more evident when covering larger areas. The 60 x 60 m area around the roman temple in Silchester would have been almost impossible to map with comparable resolution using a single channel system. Even if you had the time and managed to perform the scans accurately, the weather would change during the weeks of acquisition. With the 3d-Radar system, the site was scanned in a couple of hours.
Figure 4: Roman temple (Silchester, England).