Satellite based ground motion surveys
for environmental safety and security
Satellite based ground motion surveys
for environmental safety and security
Land Motion Surveys for Environmental Safety and Security
Frequently Asked Questions:
Germany Deformation Map
Satellite Data:
What satellite data was used?
SAR images from the Sentinel-1 satellite mission, which is part of the European Union’s Copernicus Earth Observation Programme: http://www.copernicus.eu
What sensor was used?
Synthetic Aperture Radar (SAR)
What Sentinel-1 product was used?
The Interferometric Wide product
How many Sentinel-1 image frames were used?
25 image frames from 5 satellite tracks were used to cover the whole of Germany
How many images were processed per frame?
Around 85 images were processed per frame
How many images were used in total?
More than 2200 Sentinel-1 images were processed, more than 8 Terabytes of data
Where did you get the satellite images from?
All images were downloaded for free over a standard internet connection from the Copernicus Open Access Hub. https://scihub.copernicus.eu
What is the period covered by the images?
March 2016 – March 2018
Data Processing:
How was the data processed?
The data was processed using GVL’s in-house interferometric SAR (InSAR) processing software, PUNNET. Using this software, we can generate values of average motion from a large stack of satellite SAR acquisitions
Is the entire set of Sentinel-1 data processed all-at-once?
No. The data is processed frame-by-frame and the results are mosaicked together at the end of the process
Why do you need so many images per frame?
To achieve the best accuracy and to minimise the effects of errors, such as atmospheric delay, as many images as possible should be used
Can fewer images be used?
In theory, we can use as few as thirty images but the quality and coverage of the results, especially in vegetated areas, will be affected
Are enough images available over my country/region/site for an ISBAS survey?
Yes. Sentinel-1 data is being gathered world-wide and there is now a sufficient archive to survey just about anywhere in every continent. This may not be the case with other satellite missions
What hardware was used for the processing?
A multi-core Windows PC with 128GB RAM and using 6x1TB SSD drives
Is the processing manually intensive?
No, it is almost wholly automated. The processing was performed by a single operator
How long did the processing take?
The processing was done during company downtime and was completed over a period of seven weeks
What orbital and temporal baseline thresholds were applied to the InSAR processing?
The perpendicular orbital baseline threshold was 75m and the temporal baseline threshold was 365 days
How many differential interferograms were generated?
In total, around 1750 differential interferograms were generated per frame, meaning that approximately 45,000 interferograms were analysed in total across the UK
Can I get a copy of the PUNNET software?
No
Is the ISBAS method published?
No. It is Patent Pending
Who owns the IP to the ISBAS method?
The University of Nottingham, who have granted an exclusive license to GVL
How can I get an ISBAS survey done of my area?
Please contact GVL who would be happy to provide details: contact@geomaticventures.com
What is done about layover and shadow areas?
Layover and shadow masks are calculated from the image and satellite geometry and are excluded from the survey
I thought InSAR only worked over urban and rocky areas. How do you get results over vegetated and rural terrain?
We use the unique Intermittent Small Baseline Subset (ISBAS) method which can provide results over such areas through the use of an advanced coherence analysis technique
There appear to be ISBAS results in forested areas. How is this possible?
It is well-documented that C-band radar backscatter contains a component from the forest floor and that this component has high coherence, as opposed to the backscatter from the forest canopy which has low coherence. Using a coherence analysis, the ISBAS method can identify those pixels dominated by backscatter from the forest floor and accumulate these results to obtain near-continuous coverage of ground motion in these areas
Has the ISBAS method been validated?
Yes. See our list of peer-reviewed publications
Can I read some peer-reviewed papers on the ISBAS method?
Yes. See our list of peer-reviewed publications
Do you use an absolute reference, such as a GPS station or a surveyed corner reflector?
No reference points are used
About the Product:
What do the pixels mean?
Each pixel contains measurements of relative average motion over the period covered by the data, in millimetres per year. Values are projected in the vertical (up/down) direction. Negative values are down; positive values are up
What do you mean ‘projected in the vertical (up/down) direction’?
SAR sensors look from the side, not from the vertical, direction. Therefore, vertical motion is often underestimated, so we have corrected for this
Is all motion in the vertical (up/down) direction?
No. Landslides are good example of this. Landslides on west-facing slopes will appear to be going up and those on east-facing slopes will appear to be going down (see figure below). In fact, any slippage in the east-west direction, rather than up-down, will have the same effect. This is an unfortunate consequence of the geometry
What do you mean by ‘average motion’?
This is the average rate at which the land surface has moved, over the two-year period of observation
Can this method detect very large rates of motion?
The detection of high rates of motion can be limited by the wavelength of the radar, which in this case is only 5.6cm long. This means that it is excellent at measuring rates of only a few centimetres per year and less. With any InSAR technique, higher rates are only confidently detectable in exceptional circumstances, such as in large urban or arid areas where the phase quality is very high
What can cause the land surface to move?
In Germany, the main causes relate to mining and gas storage. We see evidence of accumulation and depletion at the surface of opencast mines, such as the Lignite mines in North Rhine Westphalia and Lusatia, and the subsidence of underground mines, such as the potash mines in South Harz. Elsewhere, we also see the effects of groundwater recovery after abandonment, such as uplift in the Ruhr Valley and in Saarland.
What do you mean by ‘relative average motion’?
No absolute referencing is used. We have therefore used arbitrary referencing over areas we expect to be stable, such as urban areas, such that rates of motion should be correct. However, the ‘relative’ motion between two adjacent areas is always correct
How big an area is covered by a pixel?
Approximately 90m x 90m
What is the spacing between pixels?
50m
What is the accuracy?
Standard errors are typically of the order of 2mm/year
Why is an absolute reference, such as from an array of GPS stations, not used?
An absolute reference would not alter the relative motion but may add other confusing effects such as continental drift or GIA. These effects are, however, well-known and could be added from third party sources if required
There are a lot of low level patterns of motion across the entire survey, in rural areas. What are they?
We have noticed that most of these patterns are correlated to land use and soil types and cannot simply be dismissed as atmospheric effects or random noise. This is the subject of on-going research
Are there any errors?
Yes, there are many potential errors relating to things like increased noise in rural areas and phase unwrapping errors at the extremities of a frame or shoreline. However, we try to minimize the occurrence of errors by rigorously inspecting every frame and checking that most of the deformation detected makes sense or corresponds with expectations through discussions with experts in geology, mining, environmental science and engineering
Could other products be provided, such as total subsidence over a given period?
Yes. If you are interested in any similar types of product or format, or are interested in surveying a different location, please contact GVL to discuss
The background map appears out-of-date. Why is this?
The background satellite image, roads and labels are provided by the host GIS service provider for display purposes only. These are usually based upon Open Source data (such as Google and Bing maps) and may not be aligned with the dates of our survey. This is entirely out of the control of GVL
Applications:
From this survey, what appears to be the main cause of land motion in Germany?
Active mining but also the continuing effects of mine abandonment. There are also large areas of subsidence relating to gas storage in salt caverns in the north of the country.
What is the cause of the largest rate of land motion?
Subsidence rates exceeding three centimetres per year are seen across many of the mining areas. Some rates may be much higher, but the InSAR method used here is sensitively tuned and may underestimate the value in those locations.
What causes uplift in coal mining areas?
Underground coal mines were pumped of water for decades. When the mines were abandoned and the pumps switched off, the mines have slowly flooded and we often see a commensurate uplift associated with this
Why do opencast mines look like they are subsiding?
The extraction process in opencast mines causes the surface level to go down, which looks like subsidence
Are there any landslides?
Some landslides can be seen in hilly and mountainous terrain, but this map only shows slow-moving ones. Sudden, catastrophic landslides cannot be detected in this map
What causes motion in soils?
A low-level of soil erosion in a naturally occurring process on all land. There are many natural causes such as rainfall and water run-off but human activities such as the tillage of agricultural lands also have a major effect. https://en.wikipedia.org/wiki/Soil_erosion
What causes uplift in soils?
There is some uplift seen in peatland areas which is usually associated with a rising water table. https://goo.gl/4CHHDB
My house appears to be in an area of uplift/subsidence. Should I be worried?
Not necessarily. All the motion detected in this survey is small (mostly sub-centimetre) and slow moving, at a low-resolution (90m) and extending over large areas. This means that it is impossible to tell you if your house is affected by this motion, or not
Who should be interested in maps like this?
Operators, planners and regulators in:
-
Onshore Oil and Gas
-
Carbon Capture and Storage
-
Road and Rail Infrastructure
-
Water and Utilities Infrastructure
-
Environmental and Waste Management
-
Mining, both underground and open cast