Complete the form to request product details, pricing, schedule a demo, or to discuss an upcoming project.
Terra Insights’ Pierre Choquet and 3vGeomatics’ Jon Leighton presented a live webinar, The Future of InSAR for Geotechnical Monitoring. They discussed:
We’ve highlighted some of the attendees’ questions:
Question: How do you define vegetation? Does low vegetation mean only grass-covered to a depth of a few centimetres and does high vegetation mean trees and forests?
Answer: As important as the height of vegetation is, how arid the ground surface is equally important. In Arizona, for instance, you can have decent-sized bushes covering a site and yet you can still get good coherence. In areas in Brazil, with just short grass, but high-water content in the ground, we have worse coherence there. For X-Band data, as a rule of thumb, usually, once you get beyond the grass, and unless the ground is very arid, you are not going to get much of a result there. C-Band data would do a bit better. Whereas, L-Band data would do a lot better.
Q: Are you using in-house software or commercial software for processing the data?
A: We use solely in-house software, software that we’ve been developing since 2007. That’s a good question to ask about vendors because the off-the-shelf solutions are going to have less flexibility and probably won’t have well-built products. Do you remember when we said about the importance of products and that realization of what geotechs wanted? That’s really what separates vendors—how good their products are. We all really know how to do InSAR these days. It’s not a completely solved science, but what differentiates us, I think is the products and the compute power.
Q: How defensible is InSAR data for forensic work and in legal matters? For example, looking at when in the past settlement occurred when there were allegations of construction-related settlement. How defensible is InSAR data?
A: We have been involved in legal cases. One that I am allowed to talk about is the Millenium Tower. In San Francisco, there is a large skyscraper that was known to be tilted. People were rolling pool balls across the top floor and could see that. InSAR had a very good view of it. We could not only see the displacement but see the direction it was going and the rate it was going. That was used in a class action law suit on behalf of the class action plaintiffs. I believe successfully. InSAR data is as defensible as any other good data source that can be shown to be a good data source that can be validated.
Q: Could you explain a little more about the non-PSI (non-persistent scatters) methodology for processing InSAR data? Is it related to the small baseline approach (SBAS)?
A: Small Baseline Subset (SBAS) is a way of using—it’s effectively very similar—you create a network of interferograms based on their obit geometry. The small baseline bit (SBAS) is about the orbit geometry. It’s related to that, yes, but it would take me too long to explain. But it is also related to the way that targets get chosen. With persistent scatter geometry, the classic way to choose a measurable target is to use the amplitude information. The problem with that is we are using it as a proxy for the real measurement. The real measurement is to take the phase. 3vG, a long time ago, figured out a way to use the phase directly and see how that behaves and use that as a way of selecting targets. When you do that, you run into far fewer problems when you get to issues—that targets come and go. When targets come and go, with the amplitude method, it is, hard to incorporate—reincorporate—them. You have to do a kind of a reset every couple of months or years. With the phase method you can keep coming and going. Targets can come and go as they please.
Q: How InSAR performance gets affected by adverse weather, like clouds, snow, rain, etc.?
A: We pass through the weather layer, but we still can see it in the data. If there is, for instance, a thunderstorm right at the time of acquisition, then we could get some serious disturbances in the data. But you’d have to be pretty unlucky to have a thunderstorm consistently every 11 days. It’s very unlikely. What you end up with in the end is water vapour. Water vapour is the main thing that we see in the data most of the time. Clumps of water vapour, which they differ from displacement because they are random over time and they’re a different scale. They’re at the mesoscale of tens of kilometres wide. You can filter them out by using those assumptions, because for displacement we expect it to continue over time and it usually is smaller.
We also see weather effects from barometric pressure and temperature. If your site is at vastly different heights—you’ve got mountains and valleys or the bottom of a pit—than these pressure differences will be visible in the SAR data because you are staring through less atmosphere at the top of the mountain then you are at the bottom of the pit. There’s a very good way to model that to almost 95% of the effect and it can be removed.
Q: Can you talk about InSAR for railways? Can you monitor the deformation of the rail, specifically the track.
A: We wrote a paper with the Canadian Geological Survey about the Ashcroft Valley in British Colombia, which has a railway—a north/south railway there—and detected all kinds of geohazards along the rail. You have to distinguish between the railway itself and the embankments surrounding it. That approach was detecting the geohazards in the embankments in the rock slope cuttings. You can also measure the railway directly because it shows up as a bright target, consistent and unchanging. A persistent scatterer, if you like. It pokes right out of the data. You can tune your processing approach, so you only see the bright, high-quality railway line points. But then you don’t see the embankments and the cutting slopes. The best approach is to do both of these things. When there is a lot of vegetation on the embankments and cuttings, as there often is, then you are tied to using L-Band data. That will give you a good result. There’s several projects with CNRL, for instance, not just looking at railway displacement but also the effect of beavers undermining railway tracks as well. You can see those kinds of changes, too. InSAR is getting there for railway lines. Engineers are starting to consider it a useful tool, but we’ve got some way to go before it becomes a big deal.
Q: Can LiDAR and InSAR be integrated successfully?
A: Yes. It depends what you mean, but we commonly use LiDAR with InSAR processing. One of the requirements for InSAR—one of the things to get going—is an elevation model of the ground. You want to map the topography because InSAR is itself sensitive to topography. You are trying to subtract that element. You don’t care about topography. You are trying to measure displacement. If you’ve got a LiDAR elevation model, that’s as good as you can get for elevation. You really do a great job of subtracting that signal. That’s one way we integrate it. We’re modelling out the signals.
In terms of integrating them both as measurements, you could do that as well. But they have some significant differences. If you compare LiDAR scans then you’re going to see height changes that are meaningful at the decimetre level, which for most geotechs is way too late. Whereas InSAR will detect millimetric level displacement. The two have some very different sensitivities. But LiDAR has some advantages. It can measure the absolute level of the ground, but InSAR can measure the much more subtle changes. I can imagine ways that you could pair those things together in a product.
This transcript has been edited for clarity. Watch the webinar on-demand to learn about trending developments in InSAR monitoring.