Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast

Published by the Victorian Government Department of Environment and Primary Industries Melbourne, April 2014 © The State of Victoria Department of Environment and Primary Industries Melbourne 2014 This work is licensed under a Creative Commons Attribution 3.0 Australia licence. You are free to re-use the work under that licence, on the condition that you credit the State of Victoria as author. The licence does not apply to any images, photographs or branding, including the Victorian Coat of Arms, the Victorian Government logo and the Department of Environment and Primary Industries logo. To view a copy of this licence, visit http://creativecommons.org/licenses/by/3.0/au/deed.en Printed by Impact Digital – Brunswick ISBN 978-1-74326-841-4 (Online/pdf) Front Cover Image: Bunarong Coast, West Gippsland (Rebecca McNaught, DSE Colac) Accessibility If you would like to receive this publication in an alternative format, please telephone DEPI Customer Service Centre136186, email [email protected] (or relevant address), via the National Relay Service on 133 677 www.relayservice.com.au. This document is also available in on the internet at www.depi.vic.gov.au Disclaimer: This publication may be of assistance to you but the State of Victoria and its employees do not guarantee that the publication is without flaw of any kind or is wholly appropriate for your particular purposes and therefore disclaims all liability for any error, loss or other consequence which may arise from you relying on any information in this publication.

Contents Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast

2

Key Outcomes from the UNSW studies Preliminary review of inland subsidence Future Surveys

3 3 7

Summary of technical findings - UNSW

8

Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast 1

Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast The Gippsland Region Sustainable Water Strategy noted the ongoing concern that declining groundwater levels due to off-shore oil and gas extraction may contribute to the subsidence of the Gippsland coast. Subsidence is when the ground sinks due to compression of the layers beneath. One cause of compression can be the lowering of groundwater levels or pressures. In order to provide for a long term, cost effective monitoring program for subsidence, the Gippsland Region Sustainable Water Strategy identified the potential of satellilte based techniques. Specifically, Action 3.18 required: “The Department of Sustainability and Environment (now Department of Environment and Primary Industry) will undertake a pilot study using satellite imagery for monitoring subsidence along the Gippsland coast. If successful, the pilot study will be used to design a longer term monitoring program.” The Department of Environment and Primary Industry commissioned the University of NSW to undertake a study using InSAR (Interferometric Synthetic Aperture Radar) technology and three different analysis techniques to identify and measure subsidence in the Gippsland Basin. The project was independently reviewed by Monash University. The Gippsland Basin lies in south-eastern Victoria. While the study was undertaken primarily to assess subsidence along the Gippsland coast, the technique allows for analysis to extend inland to the foothills of the great dividing range (Figure 1).

Figure 1 – extent of analysis for the Gippsland Basin This report comprises: •

A summary of the technical findings of three methods applied, including recommendations for application of the technique

•

A summary of the analysis undertaken to assess possible mechanisms of inland subsidence in the Holey Plains/Stradbroke area (area shown as Path 380 in Figure 1)

There are also three technical reports describing the different methods and results which are presented in a combined report on the InSAR trial and are not included in this summary report. These are: •

Gippsland Subsidence Technical Report DInSAR (Differential Interferometric Synthetic Aperture Radar, which measures subsidence to within approximately 5mm)

•

Gippsland Subsidence Technical Report PSI (Persistent Scatterer Interferometry, which measures subsidence to within 1-2mm); and Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast 2

•

Gippsland Subsidence Technical Report APSI (Advanced Persistent Scatterer Interferometry, which measures subsidence to within 1-2mm)

Key Outcomes from the UNSW studies This study undertook extensive InSAR analyses on a large number of satellite radar images acquired over the Gippsland Basin. Different remote sensing techniques were assessed for accuracy, with the following outcomes: • • •

• •

DInSAR analyses can achieve measures of (typically) 5 mm accuracy. The PSI/APSI technique uses a ‘stack’ of SAR images to achieve (typically) subsidence rates in the order of of 1–2 mm per year. Accurate assessment of land subsidence is dependent on the recurrent collection of satellite radar images. It is recommended that radar imagery is collect regularly and conduct periodic InSAR analysis once every 3–5 years. Radar images have multiple uses (e.g. for flood mapping) and can be of benefit to multiple stakeholders. This investment should therefore be cost-effective to the state. Further improvement of InSAR results could be obtained if some high-quality geodetic measurements from GPS or levelling, for example, are available to assist InSAR analysis. These measurements will assist in reducing errors from orbit and atmospheric interference. Improvements could be achieved by deploying three to five (3– 5) permanent GPS stations along the Gippsland coast.

The following conclusions could be made from the analysis: • No subsidence was found to be occurring along the coastal zone. PSI analyses show that the Gippsland coast is stable over the period of analysis (from 1992 to 2011, especially between 2006 and 2011). This finding is consistent with previous ground-based surveys which showed no subsidence. • Subsidence around coal mines was identified in the PSI analysis with a maximum rate of up to 30 mm per year near Morwell. This subsidence was expected and is monitored and managed as part of mine operations. • One of the three techniques identified patchy and localised areas of subsidence of up to 30 mm per year in the Stradbroke and Holey Plains area. This region is part of the onshore extension of the Central Deep subdivision of the Gippsland Basin, where the offshore oil and gas fields are situated. A possible reason for the analysis method identifying potential subsidence is discussed further in the following section. As a result of the success of the pilot study, the Department of Environment and Primary Industry will use this information to design an ongoing monitoring program using InSAR.

Preliminary review of inland subsidence This short analysis has been undertaken to identify if there are other potential contributing factors to the calculated higher rates of subsidence (>15mm/yr) by one of the three analysis methods in the on-shore Gippsland Basin in the Holey Plains / Stradbroke area. Analysis of this area is warranted as it is the on-shore extension of the basin where oil and gas is extracted off-shore, and at first glance there occurance is unexplained. The highest resolution analysis technique, APSI (Advanced Persistent Scatterer Interferometry), can calculate subsidence rates in the Gippsland Basin of as little as two millimetres per year. The technique identified patches of subsidence >15mm/yr in the Stradbroke / Holey Plains area of Gippsland (Figure 2, Figure 3). The five largest sites were investigated further. The investigation comprised review of aerial photography over the survey period, and a review of the “goodness of fit” analysis as part of the InSAR report. The investigation identified that the specific parcels of land (paddocks) where the APSI InSAR analysis technique identified potential subsidence has undergone a change in land use from dryland pasture to plantation forestry over the survey period. The reported “goodness of fit” (Figure 5) of the subsidence data also indicated the areas were generally at the lower reliability end (although not unique to these areas). Figure 4 demonstrates the observed pattern of subsidence has a good visual fit to these parcels of land specifically. One of the areas assessed, west of Giffard, is shown in Figure 3. A summary of this preliminary analysis is provided in Table 1.

Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast 3

Table 1: Summary of observed features of five areas of potential subsidence greater than 15mm/yr. Area (shown in Figure 2) 3.7km N of Darriman 3km W of Giffard (shown in Figure 3) 3.7km NW of Stradbroke 5.5km SE of Darriman 3km NE of Napier

Reported data reliability Each area has reported poorer “goodness of fit” to the data (lower reliability). It is not unique to these areas. (Figure 5)

Land-use changes In all cases, there has been a land use change from dryland pasture to plantation forestry over the period of analysis.

Relationship of subsidence to other man-made features In all cases, the areas of higher subsidence has a general visual fit to specific parcels of land (paddocks) – the paddocks that have had a land-use change from pasture to plantation. Likewise the “goodness of fit” data demonstrates a similar pattern.

It is noted that there are some areas of potential subsidence that occur within areas that were forested over the entire survey period, particularly to the south and north of Stradbroke West. There are also some areas where plantations have been developed, but the analysis method has not identified subsidence. This investigation suggests, however, that the APSI InSAR method does respond to some aspects of ground surface change that are interpreted as subsidence. Rather than further specific site based studies, this report supports that the survey be repeated at a regular interval using multiple InSAR methods to inform the potential for subsidence.

Figure 2: Reported Rates of Subsidence 2006 to 2011 (mm/yr). Illustrates subsidence around the Latrobe Valley coal mines, and identifies localised pateches of higher potential subsidence in the Holey Plains / Stradbroke area.

Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast 4

Figure 3: Reported rates of subsidence approximately 3km west of Giffard. Aerial imagery shows forested areas in January 2000, and then in April 2011.

Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast 5

Figure 4: Comparison of InSAR results with Changes in land use. Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast 6

Future Surveys University of New South Wales recommended a repeat of the survey every 3 to 5 years. The need to repeat the surveys is supported. The UNSW also recommends the use of 3 to 5 ground-based survey points to assist in improving overall accuracy. The recommendations above will be considered in designing an ongoing program to monitor subsidence in Gippsland.

Goodness of fit (lower the value better the fit)

Figure 5: Reported reliability of the data as a “goodness of fit”. The best “fit” to the analysis technique represents a higher reliability of the analysis.

Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast 7

Summary of technical findings - UNSW

Trial of satellite radar interferometry (InSAR) to monitor subsidence along the Gippsland Coast 8

Satellite monitoring of subsidence in the Gippsland Basin

By Linlin GE and Xiaojing LI Geodesy and Earth Observing Systems Group (GEOS) School of Civil and Environmental Engineering (CVEN) The University of New South Wales Sydney, NSW 2052, AUSTRALIA

Contact A/Professor Linlin GE Phone: +61-2-9385 4177 Fax: +61-2-9313 7493 Mobile: 0423 287 219 Email: [email protected]

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW9 | P a g e

Copyright statement and disclaimer:

© State of Victoria, Department of Sustainability and Environment 2013

This publication is copyright. Apart from fair dealing for the purposes of private study, research, criticism or review as permitted under the Copyright Act 1968, no part may be reproduced, copied, transmitted in any form or by any means (electronic, mechanical or graphic) without the prior written permission of the State of Victoria, Department of Sustainability and Environment. All requests and enquiries should be directed to the Customer Service Centre, 136 186 or email [email protected]

Disclaimer: This publication may be of assistance to you but the State of Victoria and its employees do not guarantee that the publication is without flaw of any kind or is wholly appropriate for your particular purposes and therefore disclaims all liability for any error, loss or other consequence which may arise from you relying on any information in this publication.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW10 | P a g e

Table of Contents Executive Summary ................................................................................................................................................. 12 1

Background ...................................................................................................................................................... 16

2

Methodology ................................................................................................................................................... 17 2.1

SAR – Synthetic Aperture Radar .............................................................................................................. 17

2.2

InSAR – Interferometric Synthetic Aperture Radar ................................................................................. 17

2.3

DInSAR – Differential Interferometric Synthetic Aperture Radar ........................................................... 18

2.4

PSI – Persistent Scatterer Interferometry ............................................................................................... 19

2.5

APSI – Advanced Persistent Scatterer Interferometry ............................................................................ 19

3

Data sets .......................................................................................................................................................... 19

4

Key findings...................................................................................................................................................... 20 4.1

Findings from DInSAR – Differential Interferometric Synthetic Aperture Radar .................................... 20

4.2

Findings from PSI – Persistent Scatterer Interferometry ........................................................................ 22

4.3

Findings from APSI – Advanced Persistent Scatterer Interferometry ..................................................... 23

5

Concluding remarks and recommended future work ..................................................................................... 25

6

Glossary ........................................................................................................................................................... 26

7

Key references ................................................................................................................................................. 27

Appendices .............................................................................................................................................................. 28

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW11 | P a g e

Executive Summary As part of the Gippsland Region Sustainable Water Strategy, the Department of Sustainability and Environment contracted the School of Civil and Environmental Engineering (CVEN) at the University of New South Wales (UNSW) to undertake a pilot study into the use of radar satellite imagery to monitor subsidence in the Gippsland Basin. The pilot study produced three technical reports and this report summarises the findings of these technical reports (see Appendices). Groundwater levels are declining in parts of Gippsland due to off-shore oil and gas production, coal mine dewatering and groundwater use for irrigation. As groundwater levels decline, aquifers may "compress", causing subsidence of the land surface. This research was undertaken to improve understanding of land movement over time to inform if subsidence is occurring in the Gippsland Basin, particularly along the coast. The pilot study used satellite remote sensing to improve the accuracy of assessments of subsidence and to complement earlier studies in the area, which were based on field surveying and modelling. The project measured land subsidence along the coastline of the Gippsland Basin, from Lakes Entrance to Corner Inlet (Figure 1), between 1992 and 2011 using satellite imagery from three satellites.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW12 | P a g e

Figure 1 Project area from Lakes Entrance to Corner Inlet, Victoria, on Google Maps.

Previous studies on subsidence along the Gippsland coast used field survey techniques, such as GPS and line levelling, and collected data on a point-by-point basis. These techniques are time-consuming and generally cover a limited area. In some studies subsidence risks were assessed using computer models, but this technique is less accurate for assessing risks on the coast. This is because only limited data is available for offshore areas and, as a result, many assumptions have to be made (Freij-Ayoub et al., 2007).

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW13 | P a g e

Satellite remote sensing uses radar satellites that send out radar beams and collect the return beams reflected by objects on the ground surface. They are not affected by weather conditions and can operate day or night. Satellite radar provides consistent, long-term and full coverage of the Gippsland Basin. By combining two synthetic aperture radar (SAR) images, interferometric SAR (InSAR) generates a new image called an interferogram. The resulting image shows topography and changes in ground levels, as well as any interference from orbit error, weather or background noise. By isolating the topography and changes in ground levels, and minimising the interference, an accurate differential InSAR (DInSAR) result identifies the land subsidence only. DInSAR analysis is accurate to about 5 mm. DInSAR can be improved by using another technique called persistent scatterer interferometry (PSI). PSI uses a ‘stack’ of SAR images (typically more than 20) over the same area to measure the rate of change over the period of the stack. These additional measurements increase precision significantly and can be accurate to 1–2 mm. Both DInSAR and PSI/APSI (advanced PSI) together are also known as InSAR. InSAR images from three different satellites were used in the Gippsland Basin: (1) the European ENVISAT satellite – images from 2003 to 2009 (2) the Japanese ALOS satellite – images from 2006 to 2011 (3) the European ERS-1/2 satellites – images from 1992 to 2001. Because of its much longer wavelength, the Japanese satellite (ALOS PALSAR) was significantly more accurate than either of the European satellites (ENVISAT ASAR and ERS SAR). InSAR with radar systems of longer wavelength is less susceptible to interferences due to vegetation for example. DInSAR detected ground elevation change from open-pit mining and associated dewatering, as well as some changes that may be linked to agriculture. Subsidence along the Gippsland coast is, however, not large enough to be detected with DInSAR. This finding is consistent with previous ground-based surveys. Using PSI, the findings show that the Gippsland coast is generally very stable. Land subsidence can be seen, however, near Morwell, Traralgon and Stradbroke, with land surface changes of more than 30 mm per year. In addition, the area predicted to be subject to subsidence in the models (Freij-Ayoub et al., 2007) is in fact stable over the period of assessment. In the Yanakie area, which had not been picked up by the PSI technique as a land subsidence area, has been identified from the APSI analyses. In summary, space-borne radar interferometry (InSAR) has demonstrated uses in measuring regional land subsidence and accurately supports field survey and modelling. DInSAR can detect subsidence as small as 5mm and PSI can measure even smaller subsidence at 1–2 mm. No detectable subsidence was found along the Gippsland coast over the period of assessment. However, significant subsidence was found around the open-cut coal mines, which is consistent with previous ground-based surveys. While accuracy could be improved by further measurement and better calibration, InSAR has demonstrated its capacity for ongoing, long-term mapping for a number of stakeholders in the Gippsland Basin.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW14 | P a g e

Recommendations (1) Use independent geodetic measurements (e.g. GPS or levelling) to calibrate InSAR analysis in order to minimise the orbit error in the subsidence result. Improvements could be achieved by deploying three to five (3–5) permanent GPS stations along the Gippsland coast. (2) While no significant subsidence was detected along the coast with InSAR, it is clearly a costeffective technique to fill in the gap between ground-based surveys and subsidence modelling. Taking into account InSAR detectability and other factors, it is recommended to conduct periodic InSAR analysis for the region once every 3–5 years. This timeframe will facilitate the early detection of subsidence based on prediction of ground displacement for the region. That is, it is not so short that the accumulated subsidence is not large enough to be measured by InSAR. But it is not too long to miss the opportunity to detect subsidence as soon as it is detectable by InSAR with confidence. It also agrees well with requirements on monitoring impact from groundwater extraction in Gippsland region. The suggested timeframe is also well within the technical capability of current radar satellites whose design life is typically 5-7 years. (3) Optical satellites such as Landsat rely on sunlight and collect images at daytime. They collect images globally all the time so that a comprehensive archive of images is available for retrospective assessments. By contrast, radar satellites have to send out powerful radar pulses to illuminate the area of interest so they cannot afford to leave the sensor on and collect images all the time because of the high power needed for the satellite to generate these radar beams. Unless the satellite operator has a well-planned comprehensive global coverage program (as in the case of the Japanese ALOS satellite), or there is a satellite programming request from the local authority (e.g. DSE or UNSW) for an area of interest, it is impossible to conduct retrospective assessments based on satellite radar data due to insufficient images available. This has been seen in our analysis with the three European radar satellites. Therefore, satellite radar images should be collected regularly in the Gippsland Basin to monitor subsidence but also to serve multiple purposes for the benefit of stakeholders – for example, the Victorian Government, offshore oil and gas licence holders, Gippsland irrigators, Catchment Management Authorities, Gippsland urban and rural water authorities, and local community members.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW15 | P a g e

Background Land subsidence contributes to increased risk of flooding in coastal areas; damage to buildings, pipes, roads and railways; groundwater system changes; increases in seawater intrusion; and can reactivate faults. Subsidence of the land surface along the Gippsland coast may occur in response to a drop in groundwater levels and pressures caused by offshore oil and gas production, mine dewatering and excessive groundwater use. Although no significant subsidence has been observed in Gippsland, there are concerns that the coastal land surface will start to subside in the future. This concern is based on borehole measurements that have shown that groundwater levels in the underlying Latrobe Group Aquifer have been falling since 1975. Some estimates suggest that parts of the coast could subside by up to 1 m at some point this century, but these results are highly uncertain (McInnes et al., 2005). In 2007, the CSIRO assessed coastal subsidence (Freij-Ayoub et al., 2007) and determined that: • • •

Coastal land subsidence is predicted to occur along almost the entire length of Ninety Mile Beach Greatest subsidence is predicted in an area centred on the coast at Golden Beach Maximum land subsidence was predicted to be 0.51 m by 2031 (about 21 mm per year, assuming subsidence is linear between 2007 and 2031) and 0.48 m by 2056 (about 10 mm per year) • Worst-case-scenario subsidence was predicted to be 0.87 m by 2031 (about 36 mm per year) and 1.2 m by 2056 (about 24 mm per year) • Because only limited data is available for the offshore area, these predications were not specific or locally accurate. • To date, ground surveys have not detected any land subsidence, other than at the Latrobe Valley open-pit coal mines (AAMHatch, 2006 & 2007). Coastal flood risk along the Gippsland coast may be made worse by future land subsidence. The range of predicted land subsidence is significant: for example, it is estimated that the coast could subside by between 13 and 977 mm at Alberton by 2070 (McInnes et al., 2005).

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW16 | P a g e

Methodology Two slightly different InSAR techniques, namely DInSAR and PSI, are used to analyse satellite imagery. DInSAR and PSI are explained briefly in the following sections, preceded with background on SAR and InSAR.

SAR – Synthetic Aperture Radar Radar is used widely in weather forecasting and navigation. Advanced imaging radar — the synthetic aperture radar (SAR) — creates an image at the same resolution, regardless of the distance between the object and the sensor. This means that a SAR sensor installed on a satellite can be used from space. Optical satellites use cameras to photograph the ground surface and can only operate during the day when the sky is clear. Radar satellites use much longer wavelengths and therefore can penetrate thick clouds. Satellite imaging radar can operate day or night, during any weather conditions. SAR has been used increasingly in emergency responses, such as flood mapping (Hu et al., 2012) and is also an ideal system for long-term monitoring of subsidence. The SAR system works in three different microwave radiation bands, that is, 3 cm X-band, 6 cm C-band and 24 cm L-band (See Appendix 1 – DInSAR Analysis). Shorter wavelengths (for example, X-band) are more sensitive to subsidence and, in addition, the new generation SAR satellites operate at three different resolutions (1 m, 10–30 m and 100 m), with images covering an area of 10 km in length and 10 km in width (that is, 10 km by 10 km), 70 km by 70 km and 500 km by 500 km. Because of these developments, a large number of high-resolution images across all three wavelength bands can be used in order to balance resolution and coverage.

InSAR – Interferometric Synthetic Aperture Radar Recent significant technological advances have created interferometric SAR (InSAR, also known as IfSAR). (See Appendix 1 – DInSAR Analysis.) By combining two images, a new image can be generated. The resulting image shows topography and changes in ground levels, as well as any interference from orbit error, weather or background noise. When two SAR images are collected at the same time, terrain topography becomes the most dominant signal in the interferogram. The Shuttle Radar Topography Mission (SRTM) collected InSAR images over 80 per cent of the global land mass in only 11 days (NASA JPL, 2000). These images were used to create the first, detailed global digital elevation model (DEM). Figure 2 shows the SRTM DEM for the Gippsland Basin.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW17 | P a g e

Figure 2 SRTM DEM for Gippsland Basin

DInSAR – Differential Interferometric Synthetic Aperture Radar Differential InSAR (DInSAR) is an extension of the InSAR technique. By removing the topographic component and reducing interference from orbit, weather and radar noise, it is possible to isolate the land subsidence component. To resolve inconsistencies between the two images, various assumptions have to be made in DInSAR analysis given only two images are available, which in turn limits its accuracy (Ge et al., 2004, Ge et al., 2007). Despite the limitations, DInSAR has found wide applications in monitoring land subsidence due to mining (ERSDAC 2008, Ng et al., 2010 and Ng et al., 2012a), groundwater extraction (Ge et al., 2010) and earthquakes (Ge et al., 2009).

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW18 | P a g e

PSI – Persistent Scatterer Interferometry In order to increase the number of measurements over an area, persistent scatterer interferometry (PSI) uses a ‘stack’ of SAR images (usually more than 20) collected when the satellite revisits the same area (See Appendix 2 – PSI Analysis.) As a result, there are more measurements and PSI can achieve an accuracy of millimetre per year in rate of subsidence when compared with continuous GPS measurements (Ng et al., 2012 b & c; Zhang et al., 2012). However, because PSI is dependent on persistent scatterers such as buildings, it is easier to use in cities but more difficult in mostly rural regions such as the Gippsland Basin.

APSI – Advanced Persistent Scatterer Interferometry The Advanced PSI (APSI) technique combines measurements from both persistent point scatterers and distributed radar scatterers so that the density of ground deformation measurements can be greatly increased (See Appendix 3 – APSI Analysis.) Consequently, APSI is more reliable in rural regions such as the Gippsland Basin.

Data sets The area of interest (AOI) for this project is along the coastline in the Gippsland Basin, from Lakes Entrance to Corner Inlet. In order to detect potential deformation in the area, all images collected with the three different satellite sensors are used (Figure 3): •

•

•

Dataset 1: The ASAR sensor aboard the European ENVISAT satellite. It is a C-band system with a wavelength of about 5.6 cm and a revisit time / cycle of 35 days. There are 13 ASAR images available for the AOI acquired between 9 December 2003 and 13 October 2009. Dataset 2: The PALSAR sensor aboard the Japanese ALOS satellite. It is an L-band system with a wavelength of about 24 cm and a revisit time / cycle of 46 days. There are 139 PALSAR images available for the AOI acquired between 30 December 2006 and 9 March 2011. Dataset 3: The SAR sensor aboard the European ERS-1/2 satellites. It is a C-band system with a wavelength of about 5.6 cm and a revisit time / cycle of 35 days. There are 48 ERS images available for the AOI acquired between 20 June 1992 and 2 July 2001.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW19 | P a g e

Figure 3 Coverage of ALOS PALSAR (yellow), ENVISAT ASAR (blue) and ERS 1/2 SAR (red) in the Gippsland Basin

Key findings Findings from DInSAR – Differential Interferometric Synthetic Aperture Radar Given the small subsidence anticipated in Gippsland, the time over which the images are collected has to be as large as possible so that any land subsidence is detectable with DInSAR. Using the largest subsidence rate predicted in Gippsland (36 mm per year), the subsidence anticipated over the ALOS satellite’s 46-day cycle would be only 36 mm per year or 4.5 mm in a cycle. This is well below the limit DInSAR is able to detect. To detect land subsidence, DInSAR requires images over a long time period and this leads to a loss of quality in the resulting image. The Japanese ALOS PALSAR satellite performs much better than either of the European satellites (ENVISAT ASAR and ERS SAR) because of its much longer wavelength. InSAR with radar systems of longer wavelength is less susceptible to interferences due to vegetation for example. DInSAR did identify subsidence caused by open-pit mining and associated dewatering around Yallourn, Loy Yang and Hazelwood (Figure 4). DInSAR also detected smaller changes related to agricultural use, however subsidence along the Gippsland coast is

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW20 | P a g e

generally not large enough (i.e. up to centimetre level) to be detected with DInSAR. This is consistent with the findings of ground-based surveys (AAMHatch 2006 & 2007). The fact that many DInSAR results are affected by atmospheric changes suggests the use of more advanced InSAR techniques, such as the persistent scatterer InSAR (PSI), will improve the accuracy of measurement. For example, changes in the atmosphere such as water vapour concentration can create patterns similar to subsidence in DInSAR results although this does not present a problem for general radar remote sensing. These atmospheric changes can be estimated with PSI.

Figure 4 DInSAR derived from two PALSAR images collected on 24/12/2010 and 08/02/2011 with a temporal separation of 46 days

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW21 | P a g e

Findings from PSI – Persistent Scatterer Interferometry PSI analysis used images collected by the Japanese ALOS satellite, however some images were excluded because of poor quality. The PSI analysis shows that the Gippsland coast is very stable (Figure 5a). Several significant subsidence patterns can be observed near Morwell, Traralgon and Stradbroke, with land surface changes of more than 30 mm per year. In addition, the area predicted to be subject to subsidence in the models (Freij-Ayoub et al., 2007) is in fact stable over the period of this assessment (Figure 5b).

Figure 5 (a) Linear displacement rate map generated from ALOS PALSAR data for Gippsland Basin

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW22 | P a g e

Figure 5 (b) Linear displacement rate map from ALOS PALSAR data overlaid on the predicted 2031 subsidence contours as blue lines (Freij-Ayoub et al., 2007)

Findings from APSI – Advanced Persistent Scatterer Interferometry Use of the PSI technique for mapping land subsidence in Gippsland Basin is limited by the lack of PS points available in rural areas. APSI is a newly developed technique that is an extension of the PSI technique and can overcome the limitations of PSI in rural areas. When APSI was used to map land subsidence in Gippsland, several areas of subsidence were identified and these matched well with the PSI results (Figure 6). APSI also identified a further area around Yanakie that had not been picked up from the PSI result.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW23 | P a g e

Figure 6 APSI measured linear displacement rate map generated from ALOS PALSAR data for Gippsland Basin.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW24 | P a g e

Concluding remarks and recommended future work This study undertook extensive InSAR analyses on a large number of satellite radar images acquired over the Gippsland Basin. Different remote sensing techniques were assessed for accuracy, with the following outcomes: 1. DInSAR analyses can achieve measures of (typically) 5 mm accuracy. The DInSAR technique successfully detected ground elevation changes caused by open-pit mining and associated dewatering, as well as land surface changes that may be linked to agriculture. However, the subsidence along the Gippsland coast is generally not large enough to be detectable with DInSAR. This finding is consistent with previous ground-based surveys. 2. The PSI/APSI technique uses a ‘stack’ of SAR images to achieve (typically) 1–2 mm accuracy. Extending PSI using the APSI technique makes it more applicable to rural regions such as the Gippsland Basin. PSI analyses show that the Gippsland coast is stable over the period of analysis (from 1992 to 2011, especially between 2006 and 2011). However, several areas of land subsidence have been identified in the PSI results near Morwell, Traralgon and Stradbroke, with a subsidence rate of 30 mm per year. In addition, the area over the predicted subsidence identified in computer modelling (Freij-Ayoub et al., 2007) is in fact stable over the period of assessment. Previously unknown subsidence around Yanakie has been identified using APSI. Regular and consistent collection of images from the Japanese satellite ALOS has enabled both DInSAR and PSI/APSI to accurately measure land subsidence across the Gippsland region. Satellite InSAR is clearly useful in measuring regional-scale land subsidence and can support field surveying and computer modelling. Because there was no regular and consistent collection of images from the European satellites, PSI/APSI analyses cannot be done on these images, although the wavelength used by these satellites should be more sensitive to subsidence. Accurate assessment of land subsidence is dependent on the recurrent collection of satellite radar images and therefore it is recommended that DSE collect radar imagery regularly and conduct periodic InSAR analysis once every 3–5 years. Radar images have multiple uses (e.g. for flood mapping) and can be of benefit to multiple stakeholders. This investment should therefore be cost-effective to the state. Further improvement of InSAR results could be obtained if some high-quality geodetic measurements from GPS or levelling, for example, are available to assist InSAR analysis. These measurements will assist in reducing errors from orbit and atmospheric interference. Improvements could be achieved by deploying three to five (3–5) permanent GPS stations along the Gippsland coast.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW25 | P a g e

Glossary ALOS: the Japanese Advanced Land Observation Satellite. AOI: area of interest. APSI: advanced PSI. It is an advanced technique to expand the application of PSI from urban to rural areas. ASAR: the Advanced SAR, an image radar sensor on-board Envisat. DInSAR: differential interferometric synthetic aperture radar. DS: distributed scatterers, the distributed radar targets used together with PS in APSI analysis. Envisat: the European Environmental monitoring Satellite. ERS: the European Earth Resources Satellites. Two of them were launched, namely ERS-1 and ERS-2. GEOS: the Geodesy and Earth Observing Systems Group, one of the leading InSAR research groups, collaborating closely with leading global researchers from, e.g. Stanford University, NASA Jet Propulsion Laboratory (JPL), the German Aerospace Centre (DLR) and the Japan Aerospace Exploration Agency (JAXA). GPS: Global Positioning System, one of the GNSS (Global Navigation Satellite Systems). A GPS receiver can be set up over a ground survey mark and collect data for more than two hours in order to measure its coordinates. Such repeat GPS measurements over the same marks (for example, once a year) can be used to monitor the displacement at the marks accurate to around 1 cm in the vertical direction and 3–5 mm in the horizontal direction. InSAR: interferometric synthetic aperture radar (also known as IfSAR). Line levelling: a technique to measure relative heights of ground survey marks accurate to 1 mm with surveying instruments such as a digital level. LoS: line-of-sight, also known as radar looking direction. PALSAR: the Phased Array type L-band Synthetic Aperture Radar, an image radar sensor on-board ALOS. PS: persistent scatterers, the stable radar point targets used in PSI analysis. PSI: persistent scatterer InSAR, also known as PSInSAR. Sometimes DInSAR and PSI are also referred to as InSAR in its broad sense. SAR: synthetic aperture radar. Satellite SAR sensors can be operated day and night, in any weather conditions, unlike the optical counterparts, which can only be used at day time when the area of interest is free of cloud.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW26 | P a g e

CVEN: School of Civil and Environmental Engineering, the largest of the nine schools in the UNSW Faculty of Engineering. UNSW: The University of New South Wales, one of the eight research-intensive universities in Australia.

Key references AAMHatch, 2006 & 2007. Gippsland Ground Elevation Survey completed for Epochs 2 & 3, June 2004 to May 2007. Prepared for the Department of Primary Industries (Minerals and Petroleum Division). ERSDAC, 2008. http://gds.palsar.ersdac.jspacesystems.or.jp/e/doc/Min_Subsidence_Monitoring_en.htm (last accessed on 3 January 2013). Freij-Ayoub,R., Underschultz, J., Li, F., Trefry, C., Hennig, A., Otto, C., McInnes, K., 2007. Simulation of Coastal Subsidence and Storm Wave Inundation Risk in the Gippsland Basin. CSIRO. Ge, L., A.H.M. Ng, H. Wang and C. Rizos, 2009. Crustal deformation in Australia measured by satellite radar interferometry using ALOS/PALSAR imagery. Journal of Applied Geodesy, 3(1), 47-53. Ge, L., H-C Chang and C Rizos, 2007. Mine Subsidence Monitoring Using Multi-source Satellite SAR Images, Photogrammetric Engineering and Remote Sensing (PE & RS), 73(3), 259-266. Ge, L., X. Li, C. Rizos, and M. Omura, 2004. GPS and GIS Assisted Radar Interferometry. Photogrammetric Engineering and Remote Sensing (PE & RS), 70(10), 1173-1178. Ge, L., X. Li, H.C. Chang, A. Ng, 2010. Impact of ground subsidence on the Beijing Tianjin high-speed railway as mapped by radar interferometry, Annals of GIS, 16, pp. 91 - 102. Hu, Z., L. Ge and X. Li, 2012. Flood Monitoring with Integrated Multi-Source Datasets based on Satellite SAR Coherence, IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2012), Munich, Germany, 22-27 July. McInnes, K.L., I. Macadam, G.D. Hubbert, D.J. Abbs, J. Bathols, 2005. Climate Change in Eastern Victoria, Stage 2 Report: The effect of climate change on storm surges. CSIRO report for the Gippsland Coastal Board, 37pp. NASA JPL, 2000. http://www2.jpl.nasa.gov/srtm/ Ng, A.H.M., L. Ge, K. Zhang, and X. Li, 2012a. Estimating horizontal and vertical movements due to underground mining using ALOS PALSAR, Engineering Geology, Vol 143–144, 18–27. Ng, A.H.M., L. Ge, X. Li, H. Z. Abidin, H. Andreas, K. Zhang, 2012b. Mapping Land Subsidence in the Jakarta city, Indonesia using Persistent Scatterer Interferometry (PSI) Technique with ALOS PALSAR, International Journal of Applied Earth Observation and Geoinformation, 18, 232–242. doi:10.1016/j.jag.2012.01.018. Ng, A.H.M., L. Ge, X. Li, K. Zhang, 2012c. Monitoring ground deformation in Beijing, China with Persistent Scatterer SAR Interferometry, Journal of Geodesy, 86(6): 375 - 392, DOI 10.1007/soo190-011-0525-4.

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW27 | P a g e

Ng, A.H.M., L. Ge, Y. Yan, X. Li, H.C. Chang, K. Zhang, & C. Rizos, 2010. Mapping accumulated mine subsidence using small stack SAR differential interferograms in the Southern Coalfield of New South Wales, Australia. Engineering Geology, 115(1-2), 1-15. Zhang, K., Ge, L., Li, X. and Ng, A. H.-M., 2012. Monitoring ground surface deformation over the North China Plain using coherent ALOS PALSAR differential interferograms, Journal of Geodesy, http://www.springerlink.com/openurl.asp?genre=article&id=doi:10.1007/s00190-012-0595-y.

Appendices 1) Satellite monitoring of subsidence in the Gippsland Basin – DInSAR Analysis 2) Satellite monitoring of subsidence in the Gippsland Basin – PSI Analysis 3) Satellite monitoring of subsidence in the Gippsland Basin – APSI Analysis

Satellite Monitoring of Subsidence in the Gippsland Basin

GEOS-CVEN-UNSW28 | P a g e