What is InSAR measurement precision?
InSAR can measure displacement with millimeter precision. The precision typically increases with the number of processed images and the length of the period of the analysis. Acquisition gaps, strong atmospheric disturbances but mostly surface variations in the period of analysis reduce the Signal to Noise Ratio, i.e. negatively impact precision. Two quality parameters accompany each time series of displacement measurements:
- Displacement rate standard deviation [V_STDEV], which indicates the precision of the annual displacement rate.
- Coherence is an estimation of the Signal to Noise Ratio for each measurement point. Time series characterized by smooth variations of the displacement values in time and low levels of noise correspond to MP exhibiting coherence values close to1. Users should be cautious in interpreting individual TS associated with low coherence values, since they are more prone to errors. Coherence is only calculated for the LOS measurements, which constitute the input of the 2D (vertical and east-west) measurements.
Displacement (LOS) | Displ.t rate standard deviation | Time series error bar |
---|---|---|
Precision | ± 1 mm/year | ± 5 mm |
Typical displacement precision values for a measurement point less than 1 km from the reference point and a dataset of at least 30 images.
With what accuracy is the position of individual InSAR measurement points known?
While the precision of the displacement time series is millimetric, the position of individual measurement points is known with metric accuracy, and depends on the satellite system being used (see table below).
Satellite | Band | Wavelength [cm] | Resolution RGxAZ [mxm] | North-South [m] | East-West [m] | Elevation [m] |
---|---|---|---|---|---|---|
SNT | C-band | 5.6 | ~5x20 | ±8 | ±12 | ±8 |
TSX (Stripmap) | X-band | 3.11 | ~3x3 | ±1 | ±3 | ±1.5 |
Typical location precision values associated with the UTM coordinates of a measurement point at mid-latitudes.
1-D (Line-Of-Sight) Measurements
InSAR measures the projection of the true vector of displacement onto the satellite incidence angle, known as the Line of Sight (LOS), which is inclined with respect to the vertical. The measurement is 1-D (i.e. away or towards the satellite) and its sign and value depend on the orientation of the displacement vector with respect to the LOS. Negative values (from green to red) indicate movement away from the satellite, while positive values (from green to blue) indicate movement towards the satellite. Because the SAR antenna is right looking, the LOS in the ascending orbit (flying S-N) is looking towards the east, while the LOS in the descending orbit (flying N-S) is looking towards the west.
SqueeSAR® measures the projection of real displacement (Dreal) onto the LOS. Ascending orbit (south to north flight direction) measurements will produce different values compared to descending orbit (north to south flight direction) readings.
2-D (Vertical and East-West) Measurements
As all SAR satellites orbit in a quasi polar orbit (travelling across the poles) and have right-looking antennae, it is possible to acquire data stacks both when the satellite is flying towards the south (descending dataset) and towards the north (ascending dataset). The two datastacks (ascending and descending) will produce completely independent displacement maps, each from a different perspective (Line-Of-Sight, abbreviated to LOS): the same displacement may produce different readings when viewed from different LOS directions.
Ascending and descending acquisition geometries. The SAR antenna is right-looking, which means it is looking eastward in the ascending part of the orbit (S-N) and westwards in the descending part of the orbit (N-S).
The combination of 1-D (LOS) InSAR measurements obtained from ascending and descending orbits over a same area and overlapping period, produces 2-D (vertical and east-west) measurements. As satellites identify different measurement points from ascending and descending orbits, to produce 2-D results the measurement points are averaged using a spatial grid and their time series resampled. As such, 2-D cells do not represent radar targets on the ground but rather areas, which are represented by synthetic points located at the centre of the cells. Note that north-south movement cannot be measured, because SAR satellites are not sensitive to movement parallel to their travel direction!
The dimension of the 2-D measurement grid is chosen according to the image resolution, for high resolution data it is generally 10mx10m; for low resolution data it is generally 50mx50m.
The figure below demonstrates that 2-D measurements can only be calculated in areas where LOS measurements from both the ascending and descending dataset are present - note that the 2-D measurement points only appear in the areas where the two 1-D datasets are overlapping.
2-D measurements are estimated by subsampling ascending and descending data on a common spatial grid. The measurements of all MPs contained within the same cell are averaged to produce 2-D measurement points located at the centre of the cell. The 2-D procedure only produces readings for cells containing MP from both orbits (white cells)
For the 2-D analysis, both displacement rate and acceleration are displayed. In the displacement rate maps, each point is color-coded according to the magnitude of the displacement rate. In a vertical data set, negative values (red) indicate downward surface displacement, while positive values (blue) indicate upward surface displacement. In an east-west data set, negative values (red) indicate westward motion, while positive values (blue) indicate eastward motion. In the acceleration maps, the measurement points' colour indicates whether the time series is linear (light blue) or nonlinear (dark blue).
Ascending and descending LOS measurements correspond to the full resolution network of natural reflectors identified on the ground and provide the projection of the real movement to the specific LOS. The combination of ascending and descending data produces a regular grid of vertical and east-west measurements.
Why do some areas not contain any points?
The density and coverage of InSAR measurement points mainly depends on the ground surface characteristics and the topography of the area.
The point coverage is generally low over:
- Vegetated areas and low reflectivity areas (i.e. areas where the signal backscattered to the satellite is low, such as water).
- Areas affected by temporal decorrelation (i.e. the reflected radar signal changes over time). Decorrelation is generally associated with rapid surface changes, such as active operations areas, water flooding and snow coverage.
- Areas where the satellite is not able to correctly view the ground due to Line of Sight orientation with respect to the local topography (shadowing, overlay, foreshortening).
Why are points present where the ground surface has recently changed (e.g. blasting, dumping, etc)?
The grid of measurement points (MP) is defined with the baseline (setup) analysis and kept consistent during the monitoring period. If the ground surface changes because of active operations, a decrease in signal coherence may occur, resulting in increased noise in the time series (Figure 3). However, the point will remain in the dataset until a new baseline is performed once the surface operations are completed. A new baseline generally requires at least 15 images, and is periodically recalculated, usually every 6 months.
Fig 3: Time series over an area of surface operations. The loss of coherence induced by the surface changes introduces noise in the time series.
Why do my InSAR and ground instrumentation measurements not match?
Discrepancies between InSAR and other ground instrumentation measurements are generally attributed to different acquisition characteristics and/or accuracy of the measurements. When comparing InSAR measurements with other measurement techniques, it is important to take into account the following factors:
- InSAR provides 1-D (LOS) or 2-D (vertical and east-west) measurements and cannot detect north-south movements . Use the same direction of movement for comparisons. This may involve projecting other measurements to the satellite line of sight.
- 2-D InSAR measurements are referring to areas (spatial grid, for high resolution 10mx10m) rather than point targets
- InSAR measurements are referred to a local reference point. Use the same reference point for comparisons or validate the stability of the reference points for all measurement techniques.
- InSAR can measure movements ranging from a few mm/yr up to a few cm/yr. Instrumentation with a focus on faster displacement, or providing frequent data, might be challenging to compare.
In general, when comparing different instrumentation with InSAR it is always worth considering the sensitivity to displacement (the line of sight) and the temporal data sampling of each single instrument. Contact support@tre-altamira.com for further support on this topic.
Is it possible to compare Slope Stability Radar (real or synthetic aperture) results with InSAR results?
It is possible to compare InSAR with ground-based radar systems. However the following factors should be considered:
- InSAR services are designed to gather slow movements, from 1 mm/yr to a few cm per year. Radars are typically targeting displacement in the range of 1 mm/day or 1 mm/month. Slower displacement phenomena are usually under the threshold of sensitivity of radar systems. Therefore a slow moving area detected by InSAR might seem completely stable with ground-based radar data.
- Fast and complex deformation patterns can be followed better with ground-based radars due to the instruments' high acquisition frequency (every few minutes). InSAR, with at best one image per week, is better suited to detecting slow, precursory displacements.
- Radar (both ground and satellite based) measure relative displacement. If the area affected by the deformation is wider than the radar map, the radar will not be able to map the displacement. InSAR data is usually not affected by this problem, because the datasets cover a comparatively large area.
- The capability of both technologies to estimate displacement is highly dependent on the Line-of-Sight of the radar. As ground-based radars typically acquire at a nearly horizontal LoS, the instruments are less sensitive to the vertical component of displacement. Conversely, considering the quasi-polar orbit of SAR satellites, InSAR cannot detect any North-South component of displacement.
In general, GB-SAR systems are used for early warning at the local scale, because they are designed to provide real-time monitoring. Satellite InSAR provides results with a lower update frequency but can cover a significantly larger area.
Is it possible to import InSAR data to our monitoring platform?
The output of an InSAR analysis consists of a point cloud with a database of displacement time series (i.e. history of displacement) for each measurement point. The point cloud and time series database can be generated in all common formats to be ingested into other monitoring and data analysis platforms. TREA InSAR data can be integrated into the most common platforms, including Canary, Vulcan, GeoExplorer, GeoMonitoring Hub, Sensemetrics, Rocscience, K2Fly and more.
Why is the orthophoto not updated underneath?
TREmaps uses Google Maps images as a basemap for displaying the InSAR results. It is possible to use high resolution optical satellites to acquire an updated image. In case an updated background image is not already available, please contact support@tre-altamira.com for guidance on how to proceed.
Is it possible to define alarm thresholds and/or hazard maps based on InSAR results? Can we use InSAR as an early warning system?
Like all monitoring data, InSAR data may be used to create usable and actionable information layers to serve as input for TARPs. However, this is a complex task considering the following:
- That InSAR data are only updated every 11 days (at best 4+7 days), hence do not provide timely data
- That each monitored asset may have different features and possible root failure causes that may require different thresholds.
In order to define thresholds and derive hazard maps, understanding of the geotechnical characteristics of the observed asset is required. TRE ALTAMIRA can recommend partner companies to facilitate this activity, please contact support@tre-altamira.com for guidance on how to proceed.