Michael Riedmann and Adam Thomas, Fugro NPA Ltd. Edenbridge, Kent, UK;

Richard A. Sims, EBA Engineering Consultants Ltd., Vancouver, British Columbia, Canada;

Caroline Rogg, Definiens AG, Munich, Germany;

and Oliver Schleider, Intergraph, Ismaning, Germany

A three-year market development activity sponsored by the European Space Agency as part of its Earth Observation Market Development (EOMD) program demonstrated that it is possible to develop effective pipeline monitoring systems based on recent progress in satellite-based remote sensing and context-oriented image processing software.

The prime mover of the PIPEMON project was Fugro NPA Ltd., in collaboration with six business partner companies. The long-term goal is development of integrated services for pipeline operators using earth observation data. Services would include:

  • Pipeline-related ground and structure motion monitoring: integration of InSAR — radar interferometry as measured with satellite-based synthetic aperture radar (SAR) reflectors or transponders — into newly built pipelines; and measure motion along pipelines in relation to landslip, river crossing, or tectonic hazard zones using the persistent scatterer interferometry (PSI) technique.
  • Pipeline route planning: use of earth observation (EO) data and services, i.e., both optical and radar data, in combination with spatial data such as digital elevation models (DEMs) and terrain, land cover, or other mapping; and of interferometry methods such as PSI for detection and characterization of ground hazards, as part of pipeline route planning.

To start, the project’s seven partners already had built strong working relationships during previous projects. Fugro NPA, Definiens, and Tele-Rilevamento Europa (TRE) had experience in the EO industry, while EBA Engineering Consultants Ltd., Advantica, and Intergraph had experience in the pipeline consultancy industry. The British Geological Survey was the project’s science and technical reviewer.

The first project phase defined and implemented two EO service areas for pipeline operators, while the second applied the services at test-site locations. The project team worked with pipeline companies to ensure an appropriate, integrated, and easily accessible geo-information system.

The pre-commercial trials were undertaken at locations requested by pipeline companies, and each used established EO techniques, and in particular, radar interferometry and very high resolution optical data. Commercial implications taken from the trials were studied to ensure subsequent technical requirements could be met cost-effectively, and fall within price-sensitive limits for the pipeline industry.

Ground motion monitoring

The ideal ground motion monitoring tool provides a reliable basis for pipeline integrity risk prediction, such that subsidence events are managed and pipeline structural incidents or failures avoided.

Natural gas underground storage areas are frequently composed of leached salt domes or depleted natural gas reservoirs. Where governed by mining law, storage monitoring typically requires ongoing documentation of any soil subsidence and upheaval. Typically, ground movements of more than 1 cm above a closed, usually circular area with a diameter of 500 m to 5 km need to be detected and monitored.

In parts of Europe, natural gas pipelines may run through coal mining regions affected by considerable differential subsidence. The conventional requirement is that subsidence of more than 5 cm in usually single — but not necessarily regularly shaped — areas, sized 0.25 km2 to 100 km2 must be detected and monitored.

For landslide areas — including local landslip zones, unstable locations associated with river crossings or other features, or pipelines constructed through regions characterized by inherently unstable terrain — requirements for ground-motion detection and ongoing monitoring are typically similar to those for coal mining areas. Landslide areas have to be regularly checked to measure possible drift or movement of buried or surface pipelines, and over time to understand characteristics and rates of movement — for example, predictable seasonal patterns and extents of ground motion.

Route planning

One objective in planning a proposed pipeline is to iteratively test and refine routing options, evaluating constraints associated with a proposed corridor, including environmental, engineering, physical, socioeconomic, land ownership, legal, and spatial considerations. It is a complex exercise involving scenario testing and options evaluation.

While each pipeline is unique, spatial data management and iterative constraint assessment is similar across the board. The project under discussion focused on developing tools to expedite iterative assessments using EO and other corridor-related data.

In densely populated, heavy infrastructure areas, spatial and spatial-related information describing the physical environment is often already available. In such cases, planning a pipeline route entails collation and assembly of required information, as well as spatial analyses that focus on constraint mapping, with particular emphasis on existing infrastructure.

Limitations come in the form of physical barriers, including buildings; existing land ownership (e.g., financial limitations related to property transfer or access); land dedications; or “no-go” environmental constraints. Routing options are therefore often highly restricted.

In sparsely populated areas, with less infrastructure or development, other challenges come into play — including limited spatial data upon which to base constraint investigations; regional land use, ownership, and management issues; regulatory and environmental issues; and physical barriers associated with rugged or challenging landscapes.

For sparsely populated areas, available information is sometimes incomplete, and coverage of available spatial information is scattered or piecemeal. In these cases, exploitation of new satellite platforms and more robust analytical tools further assist pipeline planners in filling the “spatial data gap,” so as to enable iterative studies of corridor alternatives.

Project services: motion monitoring

In 1993, synthetic aperture radar interferometry (InSAR) was introduced to the wider remote sensing community with publication in Nature of the interferogram depicting ground deformation caused by the Landers earthquake. Although interferometry’s power was demonstrated, the conventional technique has not always been applicable in operational scenarios. The last few years, however, technical developments emerged that provide more precise motion rates, extraction of specific motion histories, and precise targeting.

A recently developed satellite radar technique called Persistent Scatterer Interferometry (PSI) can be applied to measure ground movement over underground storage areas. The technique measures motion of individual structures and ground features to millimetric precision over entire regions. PSI’s power resides in an archive of radar data that stretches back to 1992, allowing for up-to-date motion history for every measurement point. The technique assumes buildings or other man-made infrastructure on the ground, as these will provide measurement point locations.

Conventional differential interferometry (DifSAR) can be used to measure ground subsidence over coal mining areas. Ground movements are typically very fast during the removal of a mining panel (for example, 50 cm within two months). In contrast to PSI, which cannot measure velocities of more than 10 cm per year, DifSAR can measure faster movements during or after the mining activity event.

For landslide areas with slowly creeping soil, PSI and DifSAR may be used under some circumstances, or alternatively, artificial corner reflectors can be employed in an array to precisely measure the sliding ground at specific locations. Corner reflector interferometry (CRInSAR) allows ground displacement measurements at centimetric precision, carried out remotely. CRInSAR cannot be used for real-time measurements or to measure fast-moving landslides.

Project services: route planning

Route planning requires constraint mapping along corridors being considered for a proposed pipeline, such that various scenarios or alternatives can be compared objectively and — as new information may become available — rapidly updated and re-analyzed with new information, weighting scenarios, logic, or constraints. Geographic Information System (GIS)-based constraint mapping, in particular, is well suited to support route planning, as it automatically provides spatial data interpretation and weighting. Moreover, it can readily integrate spatially based EO data.

An automated GIS-based method for route-selection constraint mapping is inherently reproducible, even when there are needs to integrate vague models, weighting matrices, or linguistic information. The automatic reasoning process associated with a GIS-based constraint mapping process, if properly incorporated, will be transparent and available for iterative options testing. It incorporates weighting and ranking protocols, so individual features can be emphasized in constraint mapping. GIS-based analysis also helps avoid time-consuming reinvestigations and recalculations as the iterative process proceeds.

The procedure, as developed within the project, incorporates EO data as part of an overall information base for constraint mapping, and consists of four main steps:

  1. Coarse land-cover analyses, updating any available land-use GIS, and incorporating overall constraint features; and an initial ranking of routing options, in overview, to exclude difficult terrain and pre-select general route corridors. This process can usually be conducted using more generalized land cover and use information, including lower-resolution EO data.
  2. Identification of viable route alternatives within the corridors, pre-selecting routing options and spatially delineating critical corridor and route segments by constraint mapping of input data, including higher-resolution EO data or digital airborne imagery. Constraint evaluations at this stage may also incorporate a combination of ground-based features or measures. The result is a set of routing alternatives, such that each corridor and route segment alternative is associated with different numbers and different weightings of constraint features.
  3. Selection of route alternatives associated with fewer or lower-weighted constraint features, as well as further analyses. This involves further combination of EO and other spatial and constraint data using GIS modeling and expert knowledge tools. At this point, the analyses should be driven by experts familiar with the routing selection criteria. As the investigation proceeds, iterative refinements that emerge as constraint features are re-weighted and remaining information content requirements determined. This step is iteratively repeated, based upon expert feedback or new or updated information.
  4. A final round of analyses incorporates all GIS modeling rule-sets, appropriate spatial information, and expert knowledge to create final constraint maps. The final outputs identifying one or a few preferred routing options are then checked and further considered by pipeline planning experts.

Trial results, motion monitoring

The project undertook pre-commercial trials at a number of test sites, in conjunction with pipeline operators, to test the applications described above. The PSI technique was applied over a salt cavern field in Germany used for crude oil and natural gas storage. Subsidence of the cavern field has been monitored since 1975 with a yearly ground-leveling survey over the entire field area. PSI processing was carried out using a dataset of 70 descending ERS-SAR images, covering a time span of 12 years, from May 9, 1992 to January 25, 2005. An extract of the final estimated velocity field in the direction of the satellite line-of-sight is shown in Figure 1.

Measurement points were obtained from pipelines and related infrastructure as well as farmhouses and outbuildings on the site, which individually strongly reflected the radar signal back to the satellite. A close-up of the point coverage is shown in Figure 2.

Sampling involved a non-invasive monitoring technique that measures ground movement over a wide area with sub-millimeter precision. For each point mapped in Figure 1, a time-series motion history dating back to 1992 can be derived.

The network of ground-leveling measurements was compared to the PSI processing result. To do so, the leveling measurements had to be interpolated to the PSI results, both spatially and temporally.

A coal-mining subsidence test site associated with a gas pipeline in the UK was processed in a previous project and its utility further studied in the PIPEMON project. Figure 3 shows a prominent displacement contour amounting to at least 12 cm within a two-month period, centered on an elongated subsidence pattern of 1.5 km. Subsidence is directly attributable to mining activity along the block highlighted in red. The area of subsidence extends in width over a previously mined block (in yellow) and onto a section of the pipeline, shown in blue.

Regarding landslides, a cluster array of six metallic corner reflectors (CR) was placed within the pipeline right-of-way to measure ground-movement rates over a known landslip area associated with a pipeline river crossing in north-central Canada. CRs guarantee a clear, strong, and time-persistent target response to the satellite radar sensor, especially necessary in vegetated areas where few or no natural reflectors are present in the target area. CRs were deployed in March 2006 at the Canadian test site, and data collection continued for two years.

The underground pipeline is subjected to deep seated and slow land creep on a slope just before the river crossing. The slope drops 100m in height over a distance of 600m, as shown in Figure 4. Five satellite images revealed differential motions. All reflectors were orientated correctly towards the satellite and returned a sufficiently strong signal.

The investigation showed a centimetric downhill movement of the slope; for the 2006 data, for example, this movement was in the order of ~3.5mm from April to October. Inclinometer measurements confirm ongoing ground movement at this site, in the order of 5mm to 10mm per year since 1997. Further CR measurements, over subsequent years, are required to further validate the results.

Trial results, route planning

Route planning test sites one and two, as outlined below, demonstrate that hard data, such as land cover and GIS-based positions of existing pipelines, can be appropriately assessed and weighted along with manually input “soft information” — for example, interest expressed by adjacent land owners and local planners to undertake specific activities if certain conditions are met.

For a test site in Northern Germany, Ikonos imagery and other spatial data incorporated within a land-use GIS helped construct constraint mapping for a section of a proposed pipeline route. Also contained within the project GIS for the northern German test site was spatial information about existing pipeline networks, as well as interpreted patterns of ground movement over time (derived from PSI analyses) within the vicinity of the proposed pipeline corridor.

Constraint mapping was undertaken using object-oriented image-analysis software from Definiens Software that intuitively fuses EO data with thematic data like ESRI shapefiles, to perform derivative analyses. In a first step of the route-planning constraint mapping, image objects are created containing all EO data, GIS information, and their relationships. This enables the planning expert to later access information for every area in the test site.

Selection preferences and priorities as identified by pipeline planning experts were translated into sets of rules, which the image-analysis software could then use, in a fuzzy logic way, to classify spatial extents into five constraint classes.

For a second test site in Germany, coarse and detailed analyses were performed. The coarse level-one analysis used publicly available data such as Landsat imagery and broad DEM data to generate a generalized overview over the test site. In the second step, areas of interest were analyzed using more detailed input data, such as aerial photo-imagery and spatially based land use and ownership data.

Analysis and evaluation

To evaluate suitability of EO data and services for the pipeline industry, operators participating in the project were asked to evaluate the service results. Regarding ground movement monitoring, the operator of the German gas storage facility indicated that the PSI measurements corresponded to the leveling survey results, despite some differences in the center of the bowl. The subsidence bowl was clearly defined by the PSI data.

A strict direct comparison between ground survey data and PSI data was not possible, as measurements were interpolated spatially and temporally. The operator indicated that additional measurement points might further improve results; in such instances, where natural scatterers for PSI are insufficient or lacking, CRs can be installed. The operator sees InSAR as being particularly useful for providing important information on areas, such as wetlands, that are typically inaccessible for ground survey teams.

The PSI technology has the advantage of exploiting a satellite data archive going back to 1992, providing historical ground movement data not possible to replicate with conventional methods. Especially slow ground movements can be detected and spatially defined, in particular movements that might be overlooked using conventional ground-based methodologies. PSI’s high-point density over urban areas allows the identification of unstable areas at a glance.

DifSAR has proven useful in measuring fast ground movement over short intervals. While both DifSAR and PSI cover wide areas, CR arrays within relatively localized areas can monitor the movement of specific sites or structures.

In general, satellite-based ground motion monitoring is attractive compared to conventional approaches. It can be conducted remotely (following CR installation, if required) and cost-effectively in remote geographical areas. It can be used to non-invasively, accurately, and effectively survey large remote areas. It can confirm provisional findings regarding ground movement as measured by ground-based measurement tools, or establish spatial limits around areas suspected to be moving over time.

Feedback from operators indicated that there are at least two service issues that need to be addressed. A proposed pipeline network configuration is sensitive, strategic data that must, however, be considered as part of the route selection process. What’s needed are ways to protect corporate information while making portions available for constraint mapping and route selection activities; certain services already use network data to generate third-party interference information.

Production costs, especially for EO data and services, need to be reduced before mass deployment will be achieved. This requires low data costs, clearly defined costs for population of GIS analytical tools, and iterative development of constraints mapping components. Service and data pooling, for example, should be further considered.

Final words

Pipemon was undertaken to better define the potential for geo-information services for pipeline operators, in particular in relation to ground motion monitoring and route planning. The pre-commercial trials demonstrated the potential for EO data and services to the pipeline industry. Based on the project’s findings, it is clear that there is considerable interest by pipeline operators in the technology and potential for EO data and services to assist them in their operational requirements.

Satellite-based ground motion monitoring can be very attractive compared to conventional approaches. The process can be conducted remotely and may be particularly cost-effective in remote geographical areas. It can be used in a non-invasive manner to accurately and effectively survey large remote areas. It can also confirm provisional findings regarding ground movement as measured by ground-based measurement tools, or establish spatial limits around areas suspected to be moving over time.

More specifically, regarding use of PSI over gas storage facilities, findings show the PSI approach to be very useful. The high point density over urban areas allows identification of unstable areas at a glance; the methodology makes use of archival satellite data that cannot be accessed or generated using other methodologies; and, the approach can detect ground movements overlooked using conventional methods.

With respect to route selection, the project identified that EO data and services can be combined with analytical tools to assist route selection. EO data and services can be used, in particular, as a key component of spatially based constraint mapping, conducted in an iterative manner to refine routing and to help select final corridor options for a proposed pipeline route.

When integrated into service chains, EO services can contribute to providing EO-based information products for pipeline operators. SDI (spatial data infrastructure) platforms already host service-chain components (e.g. data provision, EO service, business logic) and provide information products for pipeline operators and other industries.

Acknowledgment

Based on a paper presented at the 6th International Pipeline Conference, Calgary, Alberta, Canada.