Mapping Tunnel-related Ground Deformation using Satellite

InSAR: From Pre-construction Baseline to Post-construction

Monitoring
Adam Thomas, Rachel Holley, Harry McCormack & David Hamersley
NPA Satellite Mapping, CGG, [email protected]
Keywords – Interferometry, Subsidence, Persistent Scatterer Interferometry,
tunnelling.

Abstract
Since the early 1990s a number of major tunnelling projects have taken place
across London, UK, including the Jubilee Line Extension (completed in 1999),
Channel Tunnel Rail Link (also known as High Speed 1 - completed in 2005, and
Crossrail which is due for phased completion in 2018/2019. All of these projects
have one thing in common: they are underground excavations that took place in
the heart of central London. They had to be carefully weaved through existing
underground infrastructure, and beneath buildings of historical importance.
Coincidentally, it was in the early 1990s that the first operational satellite
equipped with an interferometry-enabled Synthetic Aperture Radar (SAR)
sensor was launched by the European Space Agency. Since then, a further 10
operational SAR satellites have been launched, all of which have contributed to
the satellite data archive across London.
Using a satellite remote sensing technique called Interferometric Synthetic
Aperture Radar (InSAR), precise ground deformation measurements are derived
from SAR data to reveal the impact of unground tunnelling and excavations. It
can reveal ground conditions prior to engineering works, the impact of
excavation activities (including dewatering), and post-construction deformation
many years after completion.
This paper will highlight the use of InSAR across London, showing the evolution
of ground deformation from the early 1990s to the present day, and correlating
these changes with the wide range of engineering activities that have taken place
across the city. The presentation will highlight the benefits of using InSAR to
better understanding tunnel-related ground deformation and will advise on how
this information could be used in future projects within the Middle East.

1. Introduction

Underground tunnelling projects have the potential to be impacted by a range of

geotechnical and geological hazards. Subsidence, clay shrink-swell, karst cavities,
sabkha, faults, seismic activity and liquefaction, landslides, shallow groundwater,
and pre-existing buried utilities, all have the potential to derail a project during
its planning, construction and operational phases.
Our increasing need to develop new infrastructure in an already congested
environments inevitably leads to designers and engineers to push the design
envelope. This has the potential to exacerbate pre-existing hazards, create new
hazards and/or increase the likelihood of infrastructure being impacted due to
their proximity to hazard prone regions. For instance, the UK’s rich mining
history has resulted in wide-scale subsidence in central England. More recently
there is evidence that this subsidence regime is now being compensated by
groundwater recharge, resulting in land levels rising. These ‘silent hazards’ are
often overlooked as they can manifest over many years, sometimes decades. In
the long run these hazards can have a significant impact on infrastructure,
particularly where infrastructure crosses regions of differential movement.
The impact of natural and man-made deformation hazards are no more obvious
and complex than in a major city such as London. With underground space at a
premium due to the network of rail, power and sewerage tunnels that already
criss-cross the city, the importance of understanding historical ground
deformation, the status of current deformation hazards, and the long-term
impact of engineering works, has never been greater.
In the last 35 years London has been subject to a many big infrastructure
projects; not least those relating to the 2012 Olympics, and more recently
tunnelling efforts associated with Crossrail – currently one of the biggest
engineering projects in Europe. Monitoring of these works has been critical to
ensuring the integrity of existing underground infrastructure, and buildings of
national heritage and importance.
This paper focusses on the impact of underground tunnelling on ground
deformation across London. It will demonstrate the importance of satellite
interferometric synthetic aperture radar (InSAR) for remotely mapping ground
deformation associated with underground tunnelling, and will highlight the
importance of understanding the status of ground deformation before, during
and after major tunnelling projects.

2. Tunnel monitoring

Building damage resulting from tunnelling-induced ground movements is one of

the largest risks to an underground engineering project in an urban area. It is
widely recognised that notable ground movements can occur as a result of either
poorly designed or managed tunnelling works, or the presence of unforeseen
and adverse ground conditions (Millis et al. 2008).
Traditional monitoring techniques involve deploying a network of monitoring
points along the tunnel alignment. This may include the use of Robotic Total
Stations (and associated prisms), GNSS, and tilt and crack sensors. Due to the
cost and effort involved in installing and managing these systems, monitoring
networks are often distributed along the alignment and concentrated in those
sections of the route that are deemed to be most critical, for instance areas
where the tunnel will pass under sensitive structures. In densely populated
cities, such as London, this can make tunnel monitoring challenging.
Since the early 1990s the science around InSAR (Section 3) has evolved in line
with the increasing availability of satellite Synthetic Aperture Radar (SAR) data
(Section 4). The engineering community has realised that InSAR has a vital role
to play in mapping and monitoring ground deformation throughout the lifecycle
of a tunnelling project, from planning and route selection, through to
construction and operation. This latter point i.e. the ability to provide historical
and contemporary measurements throughout a project, acts to serve as a
uniquely defining aspect of InSAR. The increased use of InSAR in tunnelling
projects has resulted in it being included in the ITAtech Guidelines for Remote
Measurements Monitoring Systems (Schneider et al. 2015), a guide to the
application of InSAR and other technologies to the monitoring of tunnels.
The archive of SAR data that exists back to 1991 (Section 4) enables InSAR
specialists to retrospectively survey a proposed tunnel alignment in the early
planning stages of the project. Notwithstanding the fact that InSAR requires no
ground-based intervention (thus removing safety concerns and minimising
additional costs) the following operational benefits can be realised:
• Pre-construction baseline:
o Determine whether the tunnel alignment is optimal by inferring
whether it will pass through areas of historical ground
deformation hazards, which may increase risks and costs
throughout the entire project
o Due diligence during the early stages of a project ensures that
engineers embark on the work in full knowledge of hazards that
may be encountered.
o ‘Legal security’ if future claims are sought relating to structural
damage of buildings and properties. With a definitive baseline and
monitoring data (during and after construction) it is possible to
determine whether damage can be attributed to construction
activities, or whether they are indicative of unrelated pre-existing
activity.
• Construction monitoring:
o High density InSAR measurements along the tunnel alignment can
be used to ‘densify’ existing survey points, and provide an
important validatory check on other monitoring measurements.
o Measurements can be extended out beyond the main tunnel
alignment, allowing engineers to understand the wider impact of
tunnelling and related activities, such as dewatering.
• Post-construction monitoring:
o Long-term monitoring of the tunnel alignment (and surrounding
areas) to ensure settlements and other deformations have been
correctly compensated for and/or dissipate according to modelled
predictions. InSAR also provides a low-cost alternative to
maintaining a conventional monitoring system years or decades
after tunnel construction.

All of these factors listed above have the potential to reduce risk and financial
exposure during a tunnelling project.

3. Satellite InSAR

Persistent Scatterer InSAR (PSI) is the InSAR processing technique most

commonly used to map high-precision, multi-year measurements, and is
therefore particularly advantageous for urban tunnelling projects.
PSI identifies common geographical locations in each SAR image acquired during
the time period of interest. It looks for pixels that reliably and persistently reflect
the radar signal back to the satellite during each orbit. Referred to as Persistent
Scatterers (PS), these locations typically correlate to fixed man-made structures
(e.g. buildings, bridges) and some natural features (e.g. rocky outcrops). The
exact location and densities of PS are serendipitous and cannot be predicted in
advance of processing. In urban environments it is certain that PS densities will
be high, and it is possible to guarantee the inclusion of PS at known location by
installing artificial radar reflectors.
PSI compares differences in radar phase for each PS across all SAR images in a
data stack to derive their individual deformation measurements. PSI generally
requires a SAR data stack of at least 20 to 25 images (depending on the
parameters of the application) in order to generate an accurate result. The large
data stack allows for statistical estimation of an ‘atmospheric phase screen’ to
minimise the contribution of atmospheric effects that would otherwise be
retained during more simplistic InSAR processing techniques such as Differential
InSAR (DifSAR). A large data stack also allows errors associated with the satellite
orbits and topography to be minimised. Given this, PSI is capable of mapping
millimetric-to-decimetric annual deformation rates, with millimetric precision.
PSI measurements generally provide mean displacement rates as well as multi-
year deformation time series for individual PS. These measurements can be
compared with other geo-monitoring data and used as part of a holistic network-
level monitoring solution that has inherent redundancy. The rate at which
measurements are generated and fed into the tunnel monitoring system depends
on the type of SAR data used. Section 4 discusses this in more detail.

4. SAR Data

In 1991 the first operational InSAR-compatible SAR satellite, ERS-1, was

launched by the European Space Agency (ESA). Its C-band SAR sensor started
routinely collecting data around the world, with the ability to capture data across
the same location at 35-day intervals. Each image captured at the same location
on Earth provides a record of the position of Earth’s surface at that time.
In 1995 ERS-2 was launched as a continuity mission to ERS-1, further increasing
the collection of SAR data across the world. This pattern has continued ever
since. Since ERS-1 & -2, a further 9 InSAR-compatible missions have been
launched by national space agencies and commercial satellite operators, with six
currently in operation (Table 1).
In recent years there has been a shift towards SAR constellations as this
increases the frequency of data collection, down from e.g. 35-days to e.g. 1-day at
some locations. Constellations also provide redundancy in the case of satellite
malfunctions or failures. There have also been technical advancements with the
SAR sensors themselves, with 1m high resolution options available at a radar
wavelengths ranging from X-band through to L-band. The parameters for each
satellite can be found in Table 1.
 Radar Incidence Spatial Revisit
Image size
SAR Satellite wavelengt Lifespan angle range resolution time
range (km)
h (°) range (m) (days)

ERS-1 5.6 cm 1991 - 2000 20 - 26 30 35 100 x 100

ERS-2 5.6 cm 1995 - 2011 20 - 26 30 35 100 x 100

100 x 100 -
Envisat ASAR 5.6 cm 2002 - 2012 15 - 45 30 - 150 35
400 x 400
50 x 50 - 500
Radarsat-1 5.6 cm 1995 - 2013 20 - 59 8 - 100 24
x 500
2007 - 20 x 20 - 500
Radarsat-2 5.6 cm 20 - 59 3 - 100 24
present x 500
30 x 30 - 100
ALOS PALSAR 23.6 cm 2006 - 2010 8 - 60 10 - 100 46
x 100

2007 - 10 x 5 - 100 x
TerraSAR-X 3.1 cm 20 - 55 1 - 16 11
present 1500

2007 - 10 x 10 - 200
COSMO-SkyMed 3.1 cm 25 - 50 1 - 100 1 - 16
present x 2000

2014 -
Sentinel-1A 5.55 cm 20 - 46 5 - 100 12 80 - 400
present

2014 - 30 x 30 - 350
ALOS 2 23.6 cm 8 - 70 1 - 100 14
present x 350

2016 -
Sentinel-1B 5.55 cm 20 - 46 5 - 100 12* 80 - 400
present

Table 1. Characteristics of SAR satellites. *denotes that orbit is offset by 6 days

from Sentinel-1A

5. London tunnelling projects

Since the 1980s there have been approximately 28 major tunnelling projects in
London, starting with the London Falconbrook Flood Relief scheme (1979 and
1983) (Francis., et al. 2013).
Of key relevance to this paper are the 17 projects that have taken place since
1990 (and therefore span the period of the launch of ERS-1 and subsequent SAR
satellites), and of those the following five major tunnelling projects that are
discussed in more detail in Section 6 (Table 2). These are: Pimlico and
Wandsworth to Wimbledon cable tunnel (1992-1995), Jubilee Line Extension
(1993 to 1999), Channel Tunnel Rail Link (also known as High Speed 1 – 2001 to
2005), National Grid power tunnels (2011 to present) and Crossrail (started in
2012 and due for phased completion in 2018/2019).
 Excavated diameter of
Construction
Project name Geology maximum dimension
period
(m)
Pimlico and Wandsworth to London Clay under sand
1992 – 1995 2.97
Wimbledon cable tunnel and gravel
Barking Reach power station Chalk 1993 – 1994 3.9
Superficial deposits
Heathrow Express – bored tunnels 1993 – 1997 6.1
overlying London Clay
Jubilee Line Extension – running London Clay, Lambeth
1993 – 1999 4.9
tunnels group, Thanet sands
St Johns Wood to Pakenham Street
London Clay 1996 2.9
cable tunnel
London Woolwich and
DLR extension to Lewisham Reading beds (Lambeth 1996 – 1998 5.9
group) and Thanet sands
London Clay and
West Ham to North Greenwich cable Woolwich and Reading
1997 – 1998 2.9
tunnel London Beds (Lambeth Group)

London – West Ham – Greenwich

London Clay 1997 – 1998 3.0
Peninsula
London – Surbiton Hogsmill London Clay 1997 – 1998 3
London Clay and
Nunhead to Deptford water main Woolwich and Reading 1998 – 2000 1.4
Beds (Lambeth Group)
City of London London Clay 1999 – 2000 2.9
London Clay, Woolwich &
Reading beds
(Lambeth Group), Thanet
Channel Tunnel Rail Link sands, Harwich 2001 – 2004 8.11
formation, Bullhead beds,
lower Chalk and
Upnor formation
Channel Tunnel Rail Link London Clay 2001 – 2005 6.84
London Clay, Lambeth
Lower Lea Valley cable tunnels 2006 – 2009 4.0
group, Thanet sands
London Clay, Lambeth
National Grid power tunnels 2011 – present 4.75
group
Lee tunnel Chalk 2011 – present 8.8
Crossrail – running tunnels London Clay, Chalk 2012 – present 7.1
Table 2. Tunnelling projects in the UK post 1980 (Francis., et al. 2013)

6. Results

Data spanning 1992 to 2015 have been processed using PSI. Data from ERS-1 & -
2, Envisat and Radarsat-2 were processed as separate epochs to highlight key
changes during 1992 to 2001; 2005 to 2010 and 2010 to 2015 respectively.
The data presented in Figure 1 show the mean displacement rate maps for each
epoch, where green indicates stability of ±1.5mm/year; red indicates
deformation away from the satellite (predominantly subsidence) of >5mm/year;
blue indicates deformation towards the satellite (predominantly uplift) of
>5mm/year.
 CTRL Power Crossrai

Jubilee Line

Cable Shrink- Dewatering

A B C

Figure 1. Multi-epoch PSI mean displacement rate measurements across central

London. Each image is 15 km across. Image A shows ERS-1 & -2 measurements
spanning 1992 to 2001; Image B shows Envisat measurements spanning 2005 to
2010; Image C shows Radarsat-2 measurements spanning 2010 to 2015. © ESA.
RADARSAT-2 (Data and Products) © MacDonald, Dettwiler and Associates Ltd.
(2010-16) – All Rights Reserved. Image © CGG 2017.

Figure 1A shows relative stability across central London, however there are
notable localised areas of deformation relating to (as labelled):
• Pimlico-Wandsworth cable tunnel works (1992 to 1995) relating to the
underground installation of a 2.97m diameter tunnel through London
Clay under sand and gravel. The PSI results presented in Figure 1 show a
rate of 5mm/year settlement during construction which slows to
2mm/year after construction. The total amount of tunnelling-related
settlement is approximately 20mm.
• Jubilee Line Extension running tunnel works (1993 to 1999). This
involved the installation of a 4.9 m diameter tunnel through London Clay,
Lambeth group and Thanet sands. The PSI results presented in Figure 1
show a rate of 3mm/year during construction with the total amount of
tunnelling-related settlement at approximately 20 mm. Francis et al. state
that “the JLE tunnels potentially influenced a large number of properties
and sensitive structures, notably buildings around Westminster Square
and Network Rail viaducts east of Waterloo Station, but no significant
impacts were experienced.

Image B shows relative stability across central London, however there are
notable localised areas of deformation relating to (as labelled):
 • Channel Tunnel Rail Link (CTRL) tunnel works (2001 to 2004) relating to
the underground installation of up to 8.11m diameter tunnels through a
variety of geological units, including London Clay, Woolwich & Reading
beds (Lambeth Group), Thanet sands, Harwich formation, Bullhead beds,
lower Chalk and Upnor formation. The majority of tunnelling was carried
out through the Lambeth Group and Thanet Sand, occasionally dipping
into the Upper Chalk. These formations are water bearing, and extensive
de-watering of the project area was carried out before construction
(Francis et al.).

There is also a large area of apparent subsidence, with a rate of 2mm/year, in the
region of Pimlico and Lambeth that may relate to natural shrink-swell cycles
across London. It may be linked to the cable tunnel works discussed above, but
this would require further investigation.
Image C also shows relative stability across central London. There are notable
localised areas of deformation relating to (as labelled):
• National Grid power tunnels (2011 onwards) relating to the underground
installation of 4.75m diameter tunnels through London Clay and Lambeth
group units. The PSI results presented in Figure 1 indicated general
levels of settlement of the order of 5 mm in many locations along the
alignment, with isolated areas settlement of approximately 10mm.
National Grid used 1.5% volume loss to assess ground movements, which
equates to 4 to 5mm of maximum settlement (Francis et al.).
• Crossrail running tunnels (2012 onwards) relating to the underground
installation of 7.1m diameter tunnels through London Clay and Chalk
units. Along the alignment levels of settlement vary from 5mm to 20mm,
with the highest levels of settlement correlated to the rail stations. PSI
measurements associated with Crossrail tunnelling are discussed in
greater detail below and in Figure 2.

The data presented in Figure 2 shows the mean displacement rate map for the
period spanned by the Radarsat-2 data (2010 to 2015). As before, green
indicates stability of ±1.5mm/year; red indicates deformation away from the
satellite (i.e. subsidence) of >5mm/year; blue indicates deformation towards the
satellite (i.e. uplift) of >5mm/year.
The graphic focusses on the Crossrail tunnel alignment:
• Label A and its associated time-series graph relates to tunnelling at Bond
Street station. The red section of the graph shows the period of tunnelling
and associated ground settlement. PSI measured approximately 12 mm of
settlement during the tunnel period, with the rate of settlement
continuing after construction.
• Label B and its associated time-series graph relates to tunnelling at
Whitechapel station. The red section of the graph shows the period of
tunnelling and associated ground settlement. PSI measured
approximately 20 mm of settlement with a slowing in the rate towards
 the end of construction, however unlike Bond Street, there is a clear
levelling off of the displacement rate towards the end of the construction.
• Label C and its associated time-series graph relates to a stable reference
location in Lambeth, close to Waterloo station. During the same period of
time as that shown for Label A and B this location remained
approximately stable.

The large region of subsidence centred on Greenwich relates to dewatering

activities associated with Crossrail. Approximately 10mm of settlement was
recorded between 2013 and 2014, with recovery back to pre-dewatering levels
between 2015 and 2016.

B
A

A B C

Figure 2. Radarsat-2 derived PSI measurements spanning 2010 to 2015. Image is

15 km across. Label A and its associated time-series graph relates to tunnelling at
Bond Street station. The red section of the graph shows the period of tunnelling and
associated ground settlement. Label B and its associated time-series graph relates
to tunnelling at Whitechapel station. The red section of the graph shows the period
of tunnelling and associated ground settlement. Label C and its associated time-
series graph relates to a stable reference location in Lambeth, close to Waterloo
station. RADARSAT-2 (Data and Products) © MacDonald, Dettwiler and Associates
Ltd. (2010-16) – All Rights Reserved. Image © CGG 2017.

7. Conclusions

This paper has demonstrated the capability of InSAR, and specifically PSI, in
mapping and monitoring ground deformation before, during and after
underground tunnelling activities across London, UK. The unique time series of
results that span 1992 to 2015 highlight the benefit of using PSI data throughout
the lifecycle of a tunnelling project, from wide area, synoptic surveying along the
entire alignment, through to discrete measurement monitoring at higher risk
locations.
PSI data ensures that engineers have a much more detailed understanding of
historical and pre-existing ground hazards that might be encountered during
construction, which helps to minimise engineering risks and costs. It also
provides vital information on the post-construction impact of tunnelling, which
when combined with baseline data can provide important legal evidence should
claims be brought against the companies involved in constructing the tunnel.
The measurements recorded by PSI and presented in this paper are comparable
with those cited in the references. The authors of this paper intend to develop
this research further by performing a more detailed analysis of PSI and in-situ
monitoring data to better understand the performance and validatory
capabilities of these independent solutions.
The availability of Sentinel-1A &-B satellite imagery across London, and many
other areas of the world, on a soon to be 6-day frequency, will greatly enhance
InSAR monitoring capabilities, improving comparison with other geo-data, as
well as the overall accuracy of InSAR-derived measurements.

8. References

Millis S.W., Salisbury D., Burren R., Thomas A. 2008, Application of Persistent Scatter
Interferometry to Monitor Tunnelling Induced Settlements in Urban Areas of Hong Kong. The
HKIE Geotechnical Division Annual Seminar 2008.
Schneider O., et al. 2015, ITAtech Guidelines for Remote Measurements Monitoring Systems.
ITAtech REPORT N°3-V2 / May 2015.
Francis M., Hale E., Hitchcock A., Birks J., Mavrommati C. 2013. Impacts of Tunnels in the UK
2013.