Complete the form to request product details, pricing, schedule a demo, or to discuss an upcoming project.
Terra Insights’ Savanna Herman and 3vG’s Murray Down presented the live webinar, Advancing Dam and Levee Safety: Using InSAR Remote Sensing for Early Risk Identification. The webinar explored how remote sensing technologies like InSAR data can be used to monitor dams and levees to identify early risks to help asset operators make proactive decisions.
Watch the webinar on-demand to see the entire recording
The webinar covered:
We received several questions during the webinar. We’ve included some of them along with our answers below.
Q: Can InSAR measure changes in seepage in an embankment dam from the downstream service?
A: There are a couple of things to consider here. If the seepage is saturating soils and causing the soils to swell, particularly if they’re clay-rich soils—that displacement from the swelling will be measurable by InSAR. There also are some effects in terms of the dielectric properties of the surface of the soil based on moisture. However, in our experience, it’s extremely difficult to reliably quantify the soil moisture level.
Q: What displacement resolution are you able to capture? And what is the time change between readings?
A: Let’s first clarify the distinction between three metrics you might be referring to, namely spatial resolution vs pixel spacing vs displacement precision, as well as some other terms closely associated with these.
Spatial resolution, typically measured in meters for InSAR, is the size of the smallest features that can be detected or differentiated in an image. This is a bit more complex in radar images than it is in optical images (photographs) because of the significant variation in an object’s radar cross section depending on its shape and orientation, rather than just its size. Spatial resolution is closely related to pixel spacing, and the two are often colloquially treated as being the same but they’re slightly different. Pixel spacing is the distance between the centers of adjacent pixels in an image. As with spatial resolution, pixel spacing has some non-intuitive properties in radar imagery compared to optical imagery. The pixel spacing varies between satellites and beam modes, but also varies within an image based on the ground’s slope and aspect. It is also often anisotropic (differs in value between two axes). The values quoted for each satellite and beam mode are averages. Typically, for higher resolution satellites such as TerraSAR-X or Cosmo Skymed in their most common StripMap mode, raw pixel spacing around 3 m by 3 m. But we can acquire in different beam modes with finer pixel spacing of 2 m, 1 m, or as fine as 25 centimeters. In contrast, Sentinel-1 resolution is much coarser and has more significantly anisotropic spacing at about 5 by 20 m, which is typically resampled to have square pixels in the end products. InSAR results provide a unique displacement time series history of measurements for every single pixel, but the effective resolution and displacement detection contours require groups of several pixels moving together to trigger an alert highlighting a displacement area with confidence above a reasonable threshold.
Finally, there is displacement precision, which refers to the displacement rate being measured for each pixel, typically in millimeters. Precision is often mixed up with accuracy. Accuracy is the difference between the measurement and reality, while precision is the consistency between multiple measurements of the same reality. For InSAR, displacement precision, is the closeness of two or more measurements of the same deformation or displacement of the ground or structures over time. InSAR’s displacement precision is easier to quantify than its displacement accuracy, partly because InSAR inherently involves multiple redundant measurements of the same displacement over time. For example, between three images acquired at times A, B, and C, InSAR can measure displacement between time spans A-B, B-C, and A-C, and the sum of first two spans should theoretically equal the third span. InSAR accuracy is more complicated to quantify but has been quantified many times in studies comparing InSAR with measurements from other sensors such as lidar, inclinometers, extensometers, GNSS, etc. Both precision and accuracy of InSAR vary based on many factors including wavelength, ground conditions (such as vegetation, snow, or digging activity), total number of images (more is better), time between images, and the quality of processing algorithms that process the raw SAR data. Far from being a simple one-size-fits-all commoditized procedure, InSAR processing algorithms vary between InSAR providers, and we are continually innovating better algorithms every year with intensive research and development, as well as tuning special algorithms to suit each application such as mining, dams, oil & gas, pipelines, highways, and urban infrastructure.
There are also interactions between these metrics. For example, in areas with faster displacement rates and/or spatially abrupt displacement rate gradients, displacement precision and accuracy are also affected by pixel spacing and revisit period between consecutive images. The best precision and accuracy are achieved with small pixels and frequent measurements. Precision can also be expressed in different ways. We typically quote the 2-sigma uncertainty. With about 60 images of TerraSAR-X data in good ground conditions, the displacement 2-sigma uncertainty is typically around 2 mm. And in terms of the revisit period or time span between readings, this depends on how frequently this satellite will come back around to the location again, which varies between satellites. Typically, we are looking at around four days to 14 days, with some satellite providers starting to promise some higher frequencies as they become more developed. The combined constellation including TerraSAR-X and PAZ satellites can revisit a site twice per 11-day cycle per look direction, so a total of four times per 11-day cycle period if using dual-look. The ICEYE constellation can revisit multiple times per day. There are currently some orbital considerations that limit how many of those sub-day revisit images are appropriate for InSAR deformation measurement in some areas, but they are improving the InSAR-specific capabilities of sub-day revisit imaging schedules as they add more satellites.
Q: What is the difference between NISAR and InSAR (Interferometric Synthetic Aperture Radar)?
A: InSAR is the name of the overall technology—the technique that we use to do interferometry on synthetic aperture radar data.
NASA-ISRO SAR (NISAR) is a particular satellite that will launch later this year through a collaboration between NASA and ISRO, the Indian Space Agency. What’s exciting about NISAR is that it is both L-band—the long wavelength just like ALOS-2—but like Sentinel, it’s going to have the raw data provided for free. That means that we can use NISAR to produce products at a lower overall price because all you’re paying for is our processing algorithms, which are obviously an integral part of the complete solution, but at least we don’t need to purchase the raw data. We do need to purchase raw data for TerraSAR-X and ALOS-2 and a lot of other satellites, so that will be very exciting to use NISAR for low-cost monitoring of vegetated areas.
NISAR is relatively lower resolution in terms of pixel spacing —it’s coarser—compared to TerraSAR-X and ALOS-2, but it is higher resolution than Sentinel-1. I believe the pixel spacing of NISAR will be about 6 meters. And again, that long-wavelength means you can get through vegetated areas.
We’re excited about starting to offer products with NISAR, depending on when it’s launched. It will take a while to commission and then to build up a stack of images so you can expect the first products for that probably early in 2025.
This transcript has been edited for clarity. Watch the webinar on-demand to learn how InSAR can be used for dam and levee monitoring.