3D imaging of GPR data has been around and in use for a long time. Until relatively recently collecting a lot of 2D GPR profiles either as a uni- or bi-directional survey grid was the only way of collecting a 3D survey (and remains an effective way in some circumstances). Typically the time required to locate the 2D profiles in 3D space can be costly in the post processing workflow. This is especially true if the grid is not a perfect square which is the case for the vast majority of surveys. That being said, the availability of sub-centimetre accurate GPS which can spatially tag GPR data in real-time has made things much easier.
3D imaging allows us to present the data such that patterns become obvious to the human eye.Without 3D imaging we would then have to interpret 2D GPR profiles individually whilst considering data from adjacent profiles to look for patterns. 3D imaging allows us to present the data such that patterns become obvious to the human eye. This is particularly true for human-made structures which tend to consist of shapes rarely found in nature and geology, in particular regular shapes and straight lines. Tomographical imaging of the data allows the analyst to view the survey in a series of slices or cut-aways at a given depth, similar to medical imaging techniques such as CT scans using X-ray.
Making things easier still has been the relatively recent development of 3D GPR technology. This technology utilises an array of antenna elements very densely, but evenly, spaced in a transverse beam arrangement. In most configurations they collect a lot of parallel 2D GPR profiles in the direction the array is pushed or pulled. For tomographic imaging to be most effective, the longitudinal scan spacing should be the same as the transverse antenna element spacing, but it can be varied for different applications.
3D GPR technology has been around for a while, early adopters and influencers were those in the archaeological and utility sectors as it has clear advantages when identifying human made structures such as pipes and wall foundations. However, is 3D GPR technology of any use in highways applications?
Here at TRL we first considered 3D GPR technology as viable technique almost 5 years ago when working on a project involving a tunnel in London where we were looking for defects in the tunnel deck. This turned out to be a success but the size of the area was small enough that 3D imaging using 2D technology would have also been a viable methodology.
The next opportunity came in 2014 with one of the first tenders for asset data collection to support one of the early Smart Motorway schemes here in the UK. The scope called for longitudinal GPR in each lane including the hard shoulder along the nearside wheelpath reporting pavement material layer thickness, this is a fairly standard deliverable for a traffic-speed GPR survey. What was less typical was a requirement for full carriageway width transverse GPR sections every 250m along the carriageway.
The solution we came up was to carry out 3D GPR surveys of each lane and hard shoulder at highway speeds so no traffic management was required.Could we have achieved this with 2D technology? Certainly the longitudinal wheelpath surveys could have been accomplished using a vehicle based traffic speed 2D GPR configuration. The transverse would have been a different matter entirely. The only way to achieve this would be to use 2no. lane closures per 4km and walk the GPR across the lanes and for slip roads we would probably need total slip road closures. In actual fact TRL were already carrying out other surveys which required a similar traffic management plan so there were lane closures we could have used. However, by removing the GPR works from the closures it allowed us to collect it early in the project to aid planning of other works and minimise the risk of delays to the programme due to called off shifts.
So the solution we came up was to carry out 3D GPR surveys of each lane and hard shoulder at highway speeds without traffic management. At the time the antenna array available had a survey width of only ~1.6m, therefore we would carry out three passes per lane. We would mount the antenna such that it favoured one side of the vehicle and then the other so we could obtain as much data up to the road marking as possible without crossing over them. In post processing we would stitch the pass data together to form a full 4m lane width data set and even stitch the lanes data together to form a full carriageway width data set. Later use of a ~1.9m array allowed us to reduce the number of passes per lane by 1, increasing productivity in both data collection and processing.
This method allows us to view the data as a series of longitudinal, transverse and horizontal cut-away slices (tomography cuts) at almost any chainage, transverse offset or depth (within the effective depth range of the antenna). We can interpret and deliver the wheelpath data by choosing a transverse offset position for the longitudinal slice coinciding with the wheelpath. The analysts can then interpret the layers which they see using a set of automatic and manual interface analysis tools. Not only can we deliver the construction of the pavement along the current wheelpath, part of a Smart Motorway upgrade will very likely include lane realignment, so we could also deliver a profile along what the designers determine to be the new wheelpath for each lane. Similarly for the transverse data we can choose a slice through all of the data passes and interpret it. This means we are not limited to choosing the position of our transverse profiles before data collection; we can position them to make best use of them.
A more recent development takes advantage of the near total coverage we get with a 3D GPR survey. A transverse profile every 250m might not be often enough to capture all the construction variations going on in the scheme, in other places it might be too many as there might be long lengths of contiguous construction. We can now use existing construction data to target where the transverse should be carried out, an algorithm was developed to query the construction database for major changes and insert a transverse each time one occurred. This meant that we knew in advance where transverse profiles were located in the data so we could concentrate efforts on these locations improving the detail and accuracy of our analysis. The alternative is to analyse all data equally and then carry out the transverse selection.
3D GPR technology represents a major paradigm shift in GPR for highway applications. TRL have demonstrated the benefits of full carriageway width GPR data for Smart Motorway design but it is just as applicable in smaller schemes or could even be applicable on the network level. Other applications are also benefiting from the technology such as void and moisture assessment as well as bridge and tunnel deck assessment where data collection is much more rapid and defects are much easier to visualise and contextualise.