Monitoring and assessing reclamation settlement in coastal areas with advanced InSAR techniques: Macao city (China) case study

Transcription

1 TRES_A_ International Journal of Remote Sensing Vol. 00, No. 00, Xxxx 2010, 1 24 Monitoring and assessing reclamation settlement in coastal areas with advanced InSAR techniques: Macao city (China) case study LIMING JIANG, HUI LIN* and SHILAI CHENG Institute of Space and Earth Information Science, Chinese University of Hong Kong, China Shatin, Hong Kong, China 5 (Received 22 March 2009; in final form 6 February 2010) In the present study, we have investigated spatial-temporal behaviours of the land subsidence induced by reclamation activities in Macao Special Administrative Region, a coastal city of southern China. An advanced InSAR technique, referred to as Interferometric Point Target Analysis (IPTA), was applied to retrieve the deformation rate and displacement time series during the period from July 2006 to March Validated by levelling survey measurements, the InSAR-derived results showed a fairly stable and homogeneous pattern within the land of Macao before 1912, which consists mainly of the three granitic islands of Macao Peninsula, Taipa Island and Coloane Island. In contrast, relatively large deformation rates (between -15 and -41 mm year -1 ) and local spatial settlement variability were discovered within the latest reclamation areas. A quantitative comparison analysis of the relationship between the observed settlements and the evolution of land reclamation indicated a time-dependent settlement behaviour with respect to the age of the reclamation. Another key result was that differential settlements were detected over short distances in reclamation areas, particularly between the ground surfaces and the adjacent buildings, thus providing valuable information not only for early detection and remedial activities of potential settlement of such buildings but also for the design of future facilities adjacent to the buildings, particularly for that of large infrastructure developments Introduction To meet the growing demand for more housing and other land uses, it is common practice to reclaim land from the sea in coastal areas in many countries, including China, Great Britain, Korea, Japan, Malaysia, Saudi Arabia, Italy, the Netherlands and the USA (Glaser et al. 1991, Stuyfzand 1995, Kim et al. 2005). Ground settlement has been a significant problem in reclaimed areas as the reclamation is usually carried out by dumping uncompacted fill materials onto a seabed of unconsolidated marine sediment (Pickles and Tosen 1998, Kim and Won 2003, Kim et al. 2005, Jiang and Lin 2010). In particular, the settlement variability is crucial to performance assessment of the reclamation development because this differential settlement, rather than total settlement, can lead to damage of ground constructions (such as buildings, bridges, runways and highways) and underground facilities, with possible consequences in terms of human and economic losses (Tosen et al. 1998). *Corresponding author. huilin@cuhk.edu.hk International Journal of Remote Sensing ISSN print/issn online # 2010 Taylor & Francis DOI: /

2 2 L. Jiang et al. Over past decades, the space-borne Synthetic Aperture Radar Interferometry (InSAR) technique has been proven as an effective remote sensing tool to map ground deformation from natural causes or human activity, such as volcano dynamics (Lu et al. 1997, Fernandez et al. 2005), earthquakes (Fialko et al. 2005, Yen et al. 2008), slope instability (Metternicht et al. 2005, Meisina et al. 2006), groundwater overexploitation (Bawden et al. 2001, Hoffmann 2005), mining (Herrera et al. 2007, Jung et al. 2007) and coastland reclamation (Kim et al. 2005, Teatini et al. 2007, Jiang and Lin 2010). Compared with deformation distribution on sparse points generated by ground-based geodetic methods (e.g. Global Positioning System (GPS) or spirit levelling), this technique can determine a two-dimensional (2-D) deformation field at a large scale and at a reasonable cost, obtained by computing the phase differences (interferograms) between SAR image pairs acquired at different times. Recent remarkable improvements in SAR interferometry have led to innovative approaches based on use of a large data set of SAR images over the same area to overcome intrinsic limitations of conventional InSAR in terms of temporal and geometrical decorrelation as well as atmospheric disturbances (Ferretti et al. 2001, Mora et al. 2003, Werner et al. 2003, Hooper et al. 2004, Lanari et al. 2004, Kampes 2006). The advanced InSAR techniques (herein referred to as Persistent Scatterers Interferometry, or PSI) identify and exploit a subset of image pixels corresponding to phase-stable privileged reflectors (so-called point targets, or PTs) that can be thought of as a high density natural geodetic network, allowing both average rate and temporal evolution of ground surface displacements to be estimated at millimetre level. In the literature a large number of research projects and applications have demonstrated the remarkable capacity of the PSI technique to detect and investigate tiny ground deformation phenomena, particularly in urban areas, for instance that induced by groundwater overexploitation. Nevertheless it is still a challenge to monitor such reclamation settlement due to the scarcity of PT points and the relatively large deformation rate that typically occurs in coastal reclamation areas. Moreover, complicated deformation regimes relevant to underlying geological conditions and building foundation types, in addition to ongoing construction activities, have a significant influence on the interpretation of InSAR-derived results and further on the assessment of reclamation settlement corresponding to buildings at risk. The present study is focused on Macao Special Administrative Region (hereafter referred to as Macao), a coastal city of China mostly extended by reclamation since the seventeenth century, and an advanced InSAR technique (SAR Interferometric Point Target Analysis, or IPTA) (Werner et al. 2003) was applied to a total of 22 ENVISAT ASAR images acquired over the study area between July 2006 and March Average deformation rates and deformation history in the form of time series were retrieved by means of elaborate IPTA processes and were validated by levelling survey measurements. These InSAR results were further combined with geological information and reclamation evolution data available, and accordingly discussions on the relationship between the observed subsidence and reclamation evolutions were provided. Subsequently, a detailed analysis of the reclamation settlement affecting the buildings was carried out, concentrating especially on differential settlement of the buildings and the adjacent ground; indeed, this kind of information can be crucial to early detection and remedial activities of critical displacements for the investigated structures and design of future facilities adjacent to the structures, particularly the large infrastructure developments in a reclamation area Q1 Q2 Q3

3 Monitoring reclamation settlement with InSAR 3 2. Description of the study area Macao is located to the western side of the Pearl River estuary, about 60 km west of Hong Kong, and is immediately adjacent to the Zhuhai Special Economic Zone in Guangdong Province (see figure 1). It comprises four parts: the Macao Peninsula, the islands of Coloane and Taipa, and a newly reclaimed area known as Cotai (Coloane- Taipa). As one of the most densely populated areas in the world, Macao in late 2007 had a population of , living in a land area of 29.2 km 2. Due to an acute lack of useable land in Macao, land reclamation from the sea has been common practice, to expand the land area to satisfy the demand for rapid industrial and population growth since the seventeenth century. The evolution of reclamation in Macao as shown in figure 2 is highly related to the historical urban development. Initial land reclamation before the early twentieth century made Macao into a peninsula from an original island of only 2.78 km 2, its geological setting being dominated by bedrock of fine-grain and fine- to medium-grain granite of the Jurassic to Cretaceous age (Liang and Gao 2000). The growth of land reclamation accelerated during the last quarter of the twentieth century, the reclaimed area increasing from 15 km 2 in 1972 to 16.1 km 2 in Towards the end of the 1980s, new land-reclamations were mainly associated with large civil infrastructure projects, including the new Figure 1. The geographic location of the study area.

4 4 L. Jiang et al. COLOUR FIGURE Figure 2. Reclamation evolution of Macao during the period from 1936 to 2007 and the spatial distribution of levelling survey benchmarks. Macao international airport, the deep-water harbour at Ka Ho and the new ferry terminal to other main ports in the Pearl River region. The latest large-scale reclamation in Macao has mainly been conducted in the Cotai region since the 1990s. It created a 5.2 km 2 piece of newly reclaimed land from the sea between Taipa and Coloane islands in order to provide Macao with a new gambling and tourism area. Numerous large-scale urban developments have been conducted in the Cotai reclamation area including the Macao University of Science and Technology, Macao East Asian Games Dome, Cotai Strip, Galaxy World Resort, and so on. The Cotai Strip is one of the largest casino and tourism projects in the world, an investment by the US gaming giant Las Vegas Sands Corp, where dozens of casinos and hotels are currently being built

5 Monitoring reclamation settlement with InSAR 5 The reclamation in Macao used fill materials from both land and marine sources, and was founded over offshore Quaternary superficial deposits after dredging the marine mud. The offshore deposits brought down the Pearl River are characterized by highly compressible alluvial sediments comprising a layer of silt and clay typically between 20 and 50 m thick below the seabed (Liang and Gao 2000). As a result, the reclaimed land of Macao has experienced ground settlement that has been principally induced by the overburden of some external loads on the alluvial deposits. These loads are basically due to the imposition of man-made fills and the construction of infrastructure and buildings, whose contribution depends on the type of foundation and their corresponding depth. Similar to that of other reclaimed coastal areas, the land settlement in Macao is characterized by a large subsidence rate, commonly of the order of m year -1 during the initial stage after reclamation completion, and the rate generally decreases with time. In particular, the buildings in Macao were constructed and founded on the alluvial deposits with different techniques in terms of foundation types and structures of buildings during different reclamation periods; consequently, a complicated subsidence pattern might be observed by the InSAR method due to multiple bounce effects (i.e. double- and triple-bounce) relevant to the complex interaction between the ground level and buildings. Moreover, most of the buildings in the new Cotai reclamation area were still under construction during the investigation period between 2006 and The construction activities, as well as the rapid subsidence, lead to a low coherence in this area. These features of subsidence in the reclaimed land increase the difficulty of InSAR analysis and interpretation due to high temporal decorrelation, complicated subsidence patterns and scarcity of PT points InSAR data and methodology In the study, we applied interferometric processes to a data set of 22 ENVISAT ASAR images at VV polarization, acquired between July 2006 and March 2009 along descending track 175. Initial estimates of the interferometric baselines were calculated from DORIS precise orbit state vectors provided by the ESRIN help desk of the European Space Agency (ESA). A local DEM of the study area, with a regular grid spacing of 5 m and a vertical resolution of 2 m, was used to simulate and remove the topographic phase contribution within the interferogram generation process, as well as for geocoding InSAR results from range-doppler coordinates into map geometry corresponding to the WGS84 coordinate system. We considered in this work an advanced InSAR technique referred to as Interferometric Point Target Analysis (IPTA) developed by GAMMA Remote Sensing, which allows exploitation of the temporal and spatial characteristics of interferometric signatures collected from point targets to accurately map surface deformation histories, terrain heights and relative atmospheric path delays (Werner et al. 2003). The IPTA processing sequences involve SAR SLC (Single Look Complex) image co-registration to common geometry, PT determination based on SLC spectral and radiometric characteristics, initial height estimation using an external DEM, calculation of initial baselines, interferometric point analysis and model refinement (Wegmüet al. 2004, Stramondo et al. 2008). A detailed discussion on the IPTA method is outside the scope of this work and can be found in Werner et al. (2003) and Wegmüller et al. (2004). Accordingly, we will focus on the following special and key processing steps within the IPTA scheme in order to get a solution for Q4 Q5 Q6 Q7

6 6 L. Jiang et al. monitoring and assessing the land subsidence in the reclaimed coastal areas of Macao, which is characterized by low coherence, large subsidence rate and complex deformation regimes Formation of small-baseline interferograms A key step of the interferometric processes in this study is the selection of InSAR pairs used to generate the interferograms. These InSAR pairs should be properly chosen with the main objective of mitigating decorrelation effects, due to high subsidence rates and frequent construction activities in the case of Macao. In particular, we limit interferometric pairs to those having temporal baselines lower than 140 days and perpendicular baselines of less than 300 m. Moreover, it is noted that an important modification related to the interferogram formation scheme is that no single master SAR image is used as reference for all interferometric pairs, but multiple masters are considered, to increase the number of generated interferograms and therefore to maximize the sum coherence of pixels. As a result, we built a total of 60 interferograms from the co-registered 22 ASAR SLC images that were used in all further computations (see figure 3). 3.2 Generation of candidate point targets Identification and selection of PT points is regarded as one of the most critical steps in the advanced InSAR techniques based on the point-like target analysis scheme. Two different PT identification approaches were applied. The first one involved initial identification of PT candidates based on the low temporal variability of the back Figure 3. Baseline-time plots relevant to the ASAR images and interferograms used in this study. The points represent the 22 image acquisition dates and the solid lines represent the 60 interferograms with duration, 140 days and perpendicular baseline, 300 m.

7 Monitoring reclamation settlement with InSAR 7 scatter derived from time series of SLC data. High quality radiometric calibration and image co-registration with accurate interpolation are important technical requirements for this approach. It works well for large data stacks. The second one, mainly of interest for small stacks, utilizes the spectral properties of each individual SLC datum. It is performed by employing identification criteria of high back-scattering and low spectral phase diversity, since the point-like targets are expected not to exhibit the speckle observed for extended targets, and their back-scattering intensity remains more or less the same when processing different sub-looks with fractional azimuth and range bandwidth. In this work, it is noted that point targets determined by the two approaches described above were combined into a single preliminary point list Selection of reference points The deformation rates and histories derived by the IPTA method are relative to the selected reference point. Consequently, the selection of a high quality reference point is important to avoid introducing unnecessary phase noise in the two-dimensional bilinear regression model used for a solution to relative terrain height corrections and relative linear deformation rates between the reference point and the second point, especially for points which are far away from the reference point. Some criteria for the selection of the reference point are dictated by the method applied, such as the high quality of the point in terms of phase stability over time. In the case of Macao, a specific processing difficulty is the inability to accurately unwrap the entire scene using a single reference point. The presence of a water body of more than 3 km in width between Macao Peninsula and Taipa Island (see figure 2) leads to no PT points being available for identification, and therefore hampers the unwrapping procedure propagating from one side of the land to the other. So we carried out two independent IPTA processes using two different reference points, one corresponding to the Macao Peninsula and the other the areas of Macao including the islands of Coloane and Taipa and the Cotai reclaimed area. In terms of the a-priori knowledge of the deformation regime of the study area, two stable points were chosen within urban areas built on the original land of Macao Peninsula and the Taipa Island (see figure 4) Interferometric point analysis As with similar ideas in other advanced InSAR approaches, the interferometric phases in the IPTA are only interpreted for the selected PTs and the phase model of each PT is the same as that used in conventional interferometry (Wegmüet al. 2004). The unwrapped interferometric phase f unw is expressed as the sum of topographic phase f top, deformation phase (linear and non-linear contribution) f def, differential path delay phase (also called atmospheric phase) f atm and phase noise (or decorrelation) f noise terms: f unw ¼ f top þ f def þ f atm þ f noise (1) The interferometric phase model indicates a linear dependence of the topographic phase on the perpendicular baseline component and a linear time-dependence on the phase contribution to linear deformation. Consequently, a two-dimensional linear regression analysis of the differential interferometric phase can be performed with the dimensions corresponding to the perpendicular baseline of the interferometric pairs

8 8 L. Jiang et al. COLOUR FIGURE Figure 4. Linear deformation rates over the study area during the period from July 2006 to March 2009 obtained by the IPTA technique. The colour tones from green to light green correspond to displacements around zero (stable areas); the dark green indicates slight uplift; the tones from light yellow to red correspond to subsidence areas. This result is superimposed on a multispectral QuickBird image, and the two reference points within the urban areas built on the original land of the Macao Peninsula and Taipa Island were used for the two independent IPTA processes.

9 Monitoring reclamation settlement with InSAR 9 and to the time difference between the two SLCs of the interferometric pairs. The regression analysis is used for phase unwrapping in the temporal domain and for solving both height correction with respect to an external DEM and constant deformation rate of the point target relative to the reference. A standard deviation of the phase from the regression is used as a quality measure, permitting the detection and rejection of those points that are not suited for IPTA analysis. Specifically, points with a phase standard deviation larger than the indicated threshold value (. 1.2) were rejected in the case of Macao. Generally, for large stacks of SAR images and a small phase gradient, the two-dimensional linear regression by using the wrapped phase data of complex valued differential interferograms can find the correct phase ambiguities (Wegmüet al. 2004). However, the spatial separation of point-like scatterers is an important factor affecting the capability of the time-dependent phase unwrapping technique to resolve high phase gradients over large distances, particularly in areas of low point target density. So spatial phase unwrapping was carried out prior to the regression to mitigate the phase ambiguity errors. It was accomplished in this work using minimum cost flow optimization techniques applied to a triangular irregular network (Costantini and Rosen 1999). Once the reliable unwrapped phase is available for all interferometric pairs and the baselines are refined with a least-squares approach, the atmospheric phase, nonlinear deformation, and error terms of the residual phase could be discriminated based on their differing spatial and temporal dependencies. The spatial and temporal filtering processes in IPTA for this discrimination of residual phase terms are similar to other PSI approaches (Ferretti et al. 2001, Mora et al. 2003, Hooper et al. 2004, Lanari et al. 2004, Kampes 2006). An important aspect of the IPTA technique is the possibility of a step-wise, iterative improvement of the model parameters after the discrimination of residual phase terms. The main improvements include the consideration of a height correction, a deformation rate, a baseline refinement, atmospheric phase terms and the PT list. The final results obtained by interferometric point analysis consist of height corrections, linear deformation rates, atmospheric phase, refined baselines, phase quality information (temporal coherence) and non-linear deformation histories for each PT point. 4. Results and interpretation 4.1 InSAR-derived results and validation A total of PT points were ultimately identified and interpreted, of which more than (approximately 490 PT km -2 ) were located within the Macao region. The spatial distribution of the PT points and corresponding linear deformation velocity detected by the IPTA are illustrated in figure 4. Most of these PT points were situated in urban and suburban areas, but a few were in rural environments, such as in sparsely vegetated areas on Coloane Island. It is worth noting that a limited number of PT points were found in the most recently reclaimed areas, such as Cotai, New Port in the southern tip of Macao Peninsula and the Macao International Airport. This was mainly due to construction taking place during the investigation period and an absence of man-made angular structures available in these areas. Generally, moderate (light green and yellow, between -3 and -10 mm year -1 )to strong (red,, -25 mm year -1 ) settlement was observed in the reclaimed land of Macao, particularly in the Cotai reclaimed region and in New Port on Macao Peninsula. It should be noted that an exceptional subsiding area, belonging to the

10 10 L. Jiang et al. original land of the Macao Peninsula, was identified in the urban areas near the Lou Lim Ieoc Garden. In contrast, the sectors corresponding to Macao centre and the islands of Coloane and Taipa appeared quite stable (see figure 4). In the majority of observed deformation patterns, mean deformation velocities were homogeneously distributed over most of the investigated area; however, IPTA-derived deformation velocity maps also pointed out a relative spatial variability even over a short distance, particularly in the reclaimed land adjacent to buildings such as the East Asian Games Dome and the Venetian Macao Resort Hotel within the Cotai area. A discussion on the cause of this differential settlement will be presented in the next sub-section. An accuracy assessment was carried out by comparing the InSAR-retrieved deformation results against in situ measurements collected repeatedly by first-order spirit levelling surveys. The locations of a total of 69 levelling benchmarks are shown in figure 2. To enable the comparison, time series of levelling measurements have been projected along radar LOS (line of sight) direction and interpolated via a linear regression within the interval common to the SAR data used. The basic idea is to compare all PTs identified in a circle defined around the benchmark position. In particular, the starting radius of the circle was defined as 50 m, but when no PTs were found it was increased to 100 m and 150 m. Subsequently, for each benchmark, we calculated the average and standard deviation of the deformation rate difference between the selected PTs and the corresponding benchmark. Finally, for the whole set of 69 benchmarks, the mean values of the two validation parameters (average and standard deviation of the velocity difference) are m ¼ 1.1 mm year -1 and ¼ 2.4 mm year -1. These validation results are comparable with those in the most recent validation experiments available in the scientific literature (Ferretti et al. 2001, Teatini et al. 2005, Casu et al. 2006, Crosetto et al. 2008, Raucoules et al. in press), and demonstrate a reliable achievement in this study. It should be noted that if we only consider the benchmarks with a deformation rate less than -3 mmyear -1, mostly situated within reclaimed areas, these values are m ¼ 2.8 mm year -1 and ¼ 3.2 mm year -1 (see table 1). It is mainly due to the underestimation of deformation rate measurements retrieved by the IPTA in some cases that benchmarks are located in the most recently reclaimed land, in particular the benchmark CS8 and CS13 (see table 1). This underestimation might result from the interpretation that radar double-bounce reflections record a complex interaction process between the adjacent subsiding ground and stable buildings, namely, structures which have been constructed on piled foundations in the reclaimed land Observed subsidence vs reclamation evolution In order to investigate the relationship between observed settlement and the evolution of land reclamation, we divided the whole Macao region into three settlement effect zones, which correspond to the initial land of Macao before 1912 (referred to as zone 1), the reclamation area obtained before the 1990s ( ) (referred to as zone 2) and that reclaimed after the 1990s ( ) (referred to as zone 3). The InSAR deformation velocity maps corresponding to the three zones are illustrated in (a) (c). First of all, we focus on zone 1, the initial land area of Macao before 1912, mainly composed of the three original islands of Macao, Taipa and Coloane. With a high spatial density of PT points, particularly in Macao Peninsula, the InSAR results give evidence of a fairly stable and homogeneous pattern within the majority of this zone (see figure 5(a)). It can be explained by the interpretation that the geological conditions of the three original islands are dominated by bedrock of fine-grained and

11 Monitoring reclamation settlement with InSAR 11 Table 1. Results of the comparison of the mean deformation velocity between all the PTs identified in a circle around the position of the levelling benchmark. Absolute difference Benchmark Deformation in LOS (mm year -1 ) Maximum distance to selected PT (m) Number of PTs in the circle m (mm year -1 ) (mm year -1 ) TMA TMA TCA NGP223A NGP NGP051A NGP MS MS MS MS CS CS CS CS CS CS CS CS Average fine-to medium-grained granite. We remark that an exceptional localized subsiding phenomenon has been detected in the urban area near the Lou Lim Ieoc Garden (see the black box in figure 5(a)). In this case, a significant subsidence with a deformation rate of -10 mm year -1 to -16 mm year -1 is highlighted in the zoomed-in deformation velocity map in figure 6(a), and is further represented by the deformation time series relevant to the pixels labelled in figure 6(a) as PT_A, PT_B and PT_C (see the plots in figure 6(b)). This subsiding area is actually situated on an area of old reclamation land (reclaimed before the 1880s) on the northern margin of the original Macao Island, and is characterized by a small thickness of the recent alluvial deposits (Lin et al. 2008). A possible explanation for this anomalous phenomenon might rely on a possible natural consolidation process of the compressible soil. However, such a hypothesis is ruled out by the observation that most of the buildings on the nearest old reclamation land within this zone, such as the northern tip of the zone, are not subjected to obvious subsidence (see figure 5(a)). Therefore, the natural process of soil consolidation should not be taken as the reason for this localized subsidence. The most likely cause for the deformation lies in irregular and localized lowering of the groundwater table. Unfortunately, there are insufficient monitoring data of groundwater levels available to validate the effect in relation to the observed displacements, but an inference relevant to this cause can be made from the interpretation that a continuous drop of water level in the pond in Lou Lim Ieoc Garden has been recorded, and might have been induced by large-scale construction nearby since 2006 (Lin et al. 2008) Q8

12 12 L. Jiang et al. COLOUR FIGURE Figure 5. Linear deformation rates within the three zones of Macao: (a) zone 1 corresponding to the initial land of Macao before 1912; (b) zone 2 corresponding to the reclamation land obtained before the 1990s ( ); (c) zone 3 corresponding to the reclamation land obtained after the 1990s ( ). The black box in figure 5(a) is the urban area near the Lou Lim Ieoc Garden which is highlighted zoomed-in in figure 6(a), while the example areas (black boxes) in figure 5(c) labelled 1 and 2 are investigated in figures 7 and 8.

13 Monitoring reclamation settlement with InSAR 13 COLOUR FIGURE Figure 5. (Continued.)

14 14 L. Jiang et al. Q18 COLOUR FIGURE Figure 5. (Continued.)

15 Monitoring reclamation settlement with InSAR 15 Q19 COLOUR FIGURE Figure 6. IPTA-retrieved results near the Lou Lim Ieoc Garden, the location of which is highlighted in figure 5(a) (black box). (a) Linear deformation rate superimposed on QuickBird imagery (the colour scale of the velocity as given in figure 4) and (b) displacement time series relevant to the PT points labeled as PT_A, PT_B and PT_C. Note that the blue polygon in Figure 6(a) represents the pond boundaries within the garden. Compared with the homogeneous and stable behaviours of the majority of land present in Macao before 1912 (figure 5(a)), an overall settlement pattern with a relatively large deformation rate and local spatial variability was detected in the two reclamation zones corresponding to reclamation land obtained before the 1990s ( ) and after the 1990s ( ) (see figures 5(b) and(c)). These new reclamation zones are characterized by highly compressible recent alluvial deposits 345

16 16 L. Jiang et al. Table 2. Statistical results of the InSAR-derived reclamation settlement corresponding to the three effect zones of Macao. Original land of Macao (zone 1) (before 1912) Reclamation area before the 1990s (zone 2) ( ) Reclamation area after the 1990s (zone 3) ( ) Statistics For all PTs For PTs on buildings For PTs on the ground For all PTs For PTs on buildings For PTs on the ground For all PTs For PTs on buildings For PTs on the ground Number of PTs Deformation rate (mm year -1 ) Max. value Min. value Mean value Standard deviation

17 Monitoring reclamation settlement with InSAR 17 typically between 20 m and 50 m thick below the sea bed. We further remark that the land reclaimed after the 1990s subsides faster than that before the 1990s when we compare visually the InSAR results of the two reclamation zones. In particular, the largest settlements (such as red spots, -25 to -40 mm year -1 ) were detected in the latest reclamation area since 2000, such as the Cotai reclaimed region. Accordingly, we can infer from the comparisons of the InSAR results that the observed subsidence is time-dependent with respect to the age of the reclamation if we consider an overall interpretation. This relationship is further demonstrated by a quantitative statistical analysis in terms of the mean and the standard deviation of the deformation rate corresponding to each of the three zones (see table 2). For instance, the resulting mean values of the deformation rate are -0.2 mm year -1, -1.8 mm year -1 and -4.6 mm year -1 for zones 1, 2 and 3 respectively. Indeed, it can also be explained by the fact that the magnitude of reclamation settlements is principally related to the different developmental stages of the consolidation process of the alluvial deposits, in the presence of overloads induced by the fill materials and constructions (Stramondo et al. 2008, Jiang and Lin 2010) Observed subsidence vs buildings It is worth noting that the observed reclamation settlements can also be influenced by effects relevant to a building s structural features and foundation type. Accordingly, in order to investigate these effects we first selected PT points located on buildings and on the ground with GIS building data at 1:1000 scale, and then carried out a statistical analysis of deformation rates corresponding to two subsets of these PT points. Compared with those located at ground level, the PT points on buildings represent statistically a minor deformation pattern, and in the cases of the reclamation areas zone 2 and zone 3, the mean value of the deformation rate on buildings only is approximately one-third of that on the ground. For example, -0.6 mm year -1 vs -1.9 mm year -1 corresponding to zone 2 and -2.0 mm year -1 vs -5.8 mm year -1 corresponding to zone 3 (See table 2). Further, the stability of most buildings on reclaimed land, in terms of a relatively small deformation rate (green spots), can be inspected visually in the deformation velocity maps with respect to zone 2 and zone 3 (see figures 5 (b)and(c)). The quantitative and qualitative analyses conducted on the buildings settlement effects agree generally well with the interpretation that most of the buildings must be founded on piles reaching the granite bedrock as a rule for reclamation constructions in Macao, in particular for large-scale constructions within recent reclaimed areas, such as Macao East Asian Games Dome and the Cotai Strip in Cotai. Moreover, it should be noted that a local spatial variability of reclamation settlement was observed even over a short distance, particularly in the case of the reclaimed land adjacent to buildings, and the magnitude of differential settlement depends on the distance from the buildings. The double-bounce reflections with respect to PT points that record complex interactions between ground level and buildings might explain the differential settlement behaviour between adjacent subsiding soil and the buildings that have been constructed on piled foundations (Ketelaar and Hanssen 2003). Finally, in order to more clearly highlight the relationship between the observed subsidence and the position of the buildings, some interesting examples are presented here. Let us first focus on the site labelled 1 in figure 5(c) and highlighted in the Q9

18 18 L. Jiang et al. COLOUR FIGURE Figure 7. Results relevant to the area (around the Macao East Asian Games Dome) labelled as 1 in figure 5(c). (a) Linear deformation rate superimposed on topographic data (the colour scale of the velocity is as shown in figure 4); (b) Kriging interpolated results with respect to the PT measurements left; (c) a profile of the interpolated results along the path indicated by the dashed segment A B shown in figure 7(b). Note the sub-segment C D corresponding to the dome boundaries.

19 Monitoring reclamation settlement with InSAR 19 COLOUR FIGURE Figure 8. Results relevant to the area (near New Port) labelled 2 in figure 5(c). (a) Linear deformation rate superimposed on QuickBird multispectral imagery; (b) the site location of LB1 on a building s inner base; (c) comparison between displacement time series relevant to PT1 and of the benchmark LB1; (d) comparison between displacement time series relevant to PT2, PT3 and LB2, note that PT3 is located on a building and PT2 on the adjacent approximately 30 m from PT3; note the differential settlement between PT2 and PT3 corresponding to the site shown in figure 8(e). Q20

20 20 L. Jiang et al. zoomed-in image of figure 7(a). This typical site is characterized by the largest settlement and the presence of the largest buildings up to now, the Macao East Asian Games Dome built in In this case, the deformation rate corresponding to the structure of the dome is very low; in contrast, a very significant settlement is clearly visible in the sector corresponding to the ground adjacent to the buildings (see figure 7(a)). Moreover, larger deformation velocities were the trend in sectors further away from the buildings. This trend is particularly evident when we consider a Kriging interpolated result (figure 7(b)) with respect to the PT measurements shown in figure 7(a). To quantitatively investigate this differential settlement, we present in figure 7(c) a profile of the interpolated deformation velocities along the path indicated by the dashed segment A B shown in figure 7(b). It is noted that the obvious discontinuity occurring in correspondence to the sub-segment C D with respect to the dome boundaries reveals a significant settlement spatial gradient between adjacent soils and the building. As an additional example, we have considered the reclaimed site corresponding to New Port of Macao Peninsula, labelled 2 in figure 5(c) and highlighted in the zoomed-in image of figure 8(a). It was selected particularly because of the availability of the time series of levelling measurements of the two benchmarks marked in figure 8(a) as LB1 and LB2 (violet triangles). We remark that the benchmark LB1 was mounted on the inner base of a building on deep foundations (see figure 8(b)) and represents a fairly stable behaviour. In figure 8(c), the InSAR deformation time series of the point target PT1 on the building, which is closest to the benchmark LB1, exhibits almost no deformation signal, confirmed by the levelling measurements. With regard to the site near the benchmark LB2, two PT points were selected whose distances were approximately 20 m (see figure 8(a)); the one marked PT2 is closest to the benchmark, the other (marked PT3) was selected for a comparison analysis and is located on an adjacent building constructed on piled foundations. By considering the deformation time series of PT2 as shown in the plot of figure 8(d), we measured a linear trend of moderate settlement relevant to the point that agrees well with the levelling measurements of the benchmark LB2. Moreover, the deformation time series of PT3 is also plotted in figure 8(d), and it hints at a rather stable behaviour in correspondence with the ground target, such as the building itself. Note that the differential settlement between PT2 and PT3 is evident by superposing comparisons of their deformation time series, and implies that the point PT2 might be dominated by a double-bounce reflection (soil structure). The evidence of this differential deformation phenomenon is again revealed by in situ inspection, as testified by the picture of the building shown in figure 8(e) Q Q11 5. Conclusions In this case study, a complete analysis of spatial and temporal behaviours of reclamation settlements within the Macao region was carried out by applying the IPTA technique to a total of 22 ENVISAT ASAR images acquired between July 2006 and March The IPTA results have been validated by levelling survey measurements with a mean difference of 2.8 mm year -1 and a standard deviation of 3.2 mm year -1, demonstrating that elaborate IPTA processing can effectively monitor and detect the complicated reclamation settlements in the investigated area. This could be attributed to a couple of special and key IPTA processing steps, including formation of smallbaseline interferograms, generation of PT candidates by two identification

21 Monitoring reclamation settlement with InSAR 21 algorithms, selection of two reference points and expansion of the initial PT point list etc. Moreover, it has to be pointed out that phase unwrapping errors are possible in the IPTA two-dimensional linear regression model, particularly in cases of low PT point density and non-linear displacement, also depending on the temporal and spatial sampling available. These errors in the correct estimation of the phase ambiguities, as well as errors of the model-based interpolations or fits, may affect an accurate estimation of non-linear subsidence in this study, particularly in the most recent reclamation areas of Macao. The IPTA-derived results pointed out three distinct trends within the three settlement zones of Macao: l a quite stable and homogeneous pattern within the majority zone1, corresponding to the initial land of Macao before 1912, consisting mainly of the three granitic islands of Macao, Taipa and Coloane, l a general settlement pattern, with a relatively large deformation rate ( 15 to -41 mm year -1 ) and local spatial variability, within the two reclamation zones (zones 2 and 3) corresponding to reclamation land obtained before the 1990s ( ) and after the 1990s ( ) respectively, and l a time-dependent settlement behaviour with respect to the age of the reclamation interpreted by a quantitative comparison analysis of the relationship between the observed settlements and the evolution of land reclamation. In particular, the largest settlement (approximately -40 mm year -1 ) was detected in the area most recently reclaimed since the year 2000, such as the Cotai reclaimed region Moreover, the resulting deformation velocity maps evidenced that an exceptional subsidence ( 10 mm year -1 to -16 mm year -1 ) within zone1 was occurring in the urban area near the Lou Lim Ieoc Garden; the most likely cause for this anomalous settlement phenomenon lies in an irregular and localized lowering of the groundwater table induced by large-scale construction activity nearby. It was also inferred that most buildings in Macao, even within the reclaimed areas, were not subject to significant settlements, these buildings being mostly constructed on piled foundations that extend to the granite bedrock. A key issue of the present analysis of the relationship between the observed subsidence and the buildings is the evidence that the differential settlement was detected over short distances, particularly between the ground surfaces reclaimed and the adjacent buildings; thus providing valuable information not only for early detection and remedial measurement of a building s potential settlement but also for the design of future facilities adjacent to the structures, particularly large infrastructure developments. We further remark that an in-depth knowledge of the geology, hydrology, meteorology, and geotechnical engineering corresponding to a reclaimed area is crucial for understanding spatial-temporal behaviours of such complicated reclamation settlement scenarios. In the near future, ERS SAR archives since 1993, sustained and regular acquisition of ENVISAT ASAR data and more ground observations, especially geological and hydrological material, will be collected, and a joint analysis of multidisciplinary data will be carried out for further studying the reclamation settlement phenomena in this study area. It will allow us to develop a more thorough understanding of the reclamation settlement mechanisms so as to ensure the reliability of the long-term residual settlement predictions, and thereby to assist in performance assessment of reclamation development Q12 Q13

22 22 L. Jiang et al. Acknowledgements The work in this paper was supported by the Competitive Earmarked Research Grant of the Hong Kong Research Grant Council (no. CUHK4665/06H), Innovation and Technology Support Programme, Hong Kong (no. ITS/038/07 and no. ITS/077/08), and was partly founded by the ISEIS-DSCC cooperation project. The authors would like to thank the Macao Cartography and Cadastre Bureau for providing the levelling survey measurements and GIS layers and corresponding data analysis. 490 Q References BAWDEN, G.W., THATCHER, W., STEIN, R.S., HUDNUT, K.W. and PELTZER, G., 2001, Tectonic contraction across Los Angeles after removal of groundwater pumping effects. Nature, 412, pp CASU, F., MANZO, M. and LANARI, R., 2006, A quantitative assessment of the SBAS algorithm performance for surface deformation retrieval from DInSAR data. Remote Sensing of Environment, 102, pp COSTANTINI, M. and ROSEN, P., 1999, A generalized phase unwrapping approach for sparse data. In Proceedings of the International Geoscience and Remote Sensing Symposium 99, June 2 July 1999, Hamburg ( Institute of Electrical and Electronics Engineers), pp. Q CROSETTO, M., BIESCAS, E., DURO, J., CLOSA, J. and ARNAUD, A., 2008, Generation of advanced ERS and envisat interferometric SAR products using stable point network technique. Photogrammetric Engineering & Remote Sensing, 74, pp FERNANDEZ, J., ROMERO, R., CARRASCO, D., TIAMPO, K.F., RODRIGUEZ-VELASCO, G., APARICIO, A., ARANA, V. and GONZALEZ-MATESANZ, F.J., 2005, Detection of displacements on Tenerife Island, Canaries, using radar interferometry. Geophysical Journal International, 160, pp FERRETTI, A., PRATI, C. and ROCCA, F., 2001, Permanent scatterers in SAR interferometry. 515 IEEE Transactions on Geoscience and Remote Sensing, 39, pp FIALKO, Y., SANDWELL, D., SIMONS, M. and ROSEN, P., 2005, Three-dimensional deformation caused by the Bam, Iran, earthquake and the origin of shallow slip deficit. Nature, 435, pp GLASER, R., HABERZETTL, P. and WALSH, R.P.D., 1991, Land reclamation in Singapore, Hong 520 Kong and Macao. GeoJournal, 24, pp HERRERA, G., TOMAS, R., LOPEZ-SANCHEZ, J.M., DELGADO, J., MALLORQUI, J.J., DUQUE, S. and MULAS, J., 2007, Advanced DInSAR analysis on mining areas: La Union case study (Murcia, SE Spain). Engineering Geology, 90, pp HOFFMANN, J., 2005, The future of satellite remote sensing in hydrogeology. Hydrogeology 525 Journal, 13, pp HOOPER, A., ZEBKER, H., SEGALL, P. and KAMPES, B., 2004, A new method for measuring deformation on volcanoes and other natural terrains using InSAR persistent scatterers. Geophysical Research Letters, 31, pp JIANG, L.M. and LIN, H., 2010, Integrated analysis of SAR interferometric and geological data 530 for investigating long-term reclamation settlement of Chek Lap Kok Airport, Hong Kong. Engineering Geology, 110, pp JUNG, H.C., KIM, S.W., JUNG, H.S., MIN, K.D. and WON, J.S., 2007, Satellite observation of coal mining subsidence by persistent scatterer analysis. Engineering Geology, 92, pp KAMPES, B.M., 2006, Radar Interferometry: persistent scatterer technique, (Dordrecht: Springer-Verlag). KETELAAR, V.B.H. and HANSSEN, R.F., 2003, Separation of different deformation regimes using PS-InSAR data. In Proceedings of the FRINGE 2003 Workshop (ESA SP-550), 1 5

23 Monitoring reclamation settlement with InSAR 23 December 2003, Frascati, Italy, H. Lacoste (Ed.), p.19.1 ( ESA/ESRIN), published on 540 Q16 CDROM. KIM, S.W., LEE, C.W., SONG, K.Y., MIN, K.D. and WON, J.S., 2005, Application of L-band differential SAR interferometry to subsidence rate estimation in reclaimed coastal land. International Journal of Remote Sensing, 26, pp KIM, S.W. and WON, J.S., 2003, Measurements of soil compaction rate by using JERS-1 SAR 545 and a prediction model. IEEE Transactions on Geoscience and Remote Sensing, 41, pp LANARI, R., MORA, O., MANUNTA, M., MALLORQUI, J.J., BERARDINO, P. and SANSOSTI, E., 2004, A small-baseline approach for investigating deformations on full-resolution differential SAR interferograms. IEEE Transactions on Geoscience and Remote Sensing, 42, pp LIANG, Z.J. and GAO, L., 2000, Geological condition of coastal engineering in Macao District and prospects for sea reclamation. Port Engineering Technology, 2, pp (in Chinese). LIN, H., JIANG, L.M. and PANG, Y.C., 2008, Deformation monitoring by using InASR in Macao, 555 DSCC technical report. (Macao, China: Macao Cartography and Cadastre Bureau). LU, Z., FATLAND, R., WYSS, M., LI, S., EICHELBERER, J., DEAN, K. and FREYMUELLER, J., 1997, Deformation of New Trident volcano measured by ERS-1 SAR interferometry, Katmai National Park, Alaska. Geophysical Research Letters, 24, pp MEISINA, C., ZUCCA, F., FOSSATI, D., CERIANI, M. and ALLIEVI, J., 2006, Ground deformation 560 monitoring by using the Permanent Scatterers Technique: the example of the Oltrepo Pavese (Lombardia, Italy). Engineering Geology, 88, pp METTERNICHT, G., HURNI, L. and GOGU, R., 2005, Remote sensing of landslides: an analysis of the potential contribution to geo-spatial systems for hazard assessment in mountainous environments. Remote Sensing of Environment, 98, pp MORA, O., MALLORQUI, J.J. and BROQUETAS, A., 2003, Linear and nonlinear terrain deformation maps from a reduced set of interferometric SAR images. IEEE Transactions on Geoscience and Remote Sensing, 41, pp PICKLES, A.R. and TOSEN, R., 1998, Settlement of reclaimed land for the new Hong Kong International Airport. ICE Proceedings: Geotechnical Engineering, 131, pp RAUCOULES, D., BOURGINE, B., DE MICHELE, M., LE COZANNET, G., CLOSSET, L., BREMMER, C., VELDKAMP, H., TRAGHEIM, D., BATESON, L., CROSETTO, M., AGUDO, M. andengdahl, M., Validation and intercomparison of Persistent Scatterers Interferometry: PSIC4 project results. Journal of Applied Geophysics, in press. Q21 STRAMONDO, S., BOZZANO, F., MARRA, F., WEGMÜLLER, U., CINTI, F.R., MORO, M. and SAROLI, 575 M., 2008, Subsidence induced by urbanisation in the city of Rome detected by advanced InSAR technique and geotechnical investigations. Remote Sensing of Environment, 112, pp STUYFZAND, P.J., 1995, The impact of land reclamation on groundwater quality and future drinking water supply in the Netherlands. Water Science and Technology, 31, pp TEATINI, P., STROZZI, T., TOSI, L., WEGMÜLLER, U., WERNER, C. and CARBOGNIN, L., 2007, Assessing short- and long-time displacements in the Venice coastland by synthetic aperture radar interferometric point target analysis. Journal of Geophysical Research, 112, F01012, doi: /2006jf TEATINI, P., TOSI, L., STROZZI, T., CARBOGNIN, L., WEGMÜLLER, U. and RIZZETTO, F., 2005, 585 Mapping regional land displacements in the Venice coastland by an integrated monitoring system. Remote Sensing of Environment, 98, pp TOSEN, R., PICKLES, A.R. and JAROS, M.B.J., 1998, Assessment of differential settlement at Chek Lap Kok Airport reclamation site. In Proceedings of HKIE Seminar on ACP Projects. 590 Q17 WEGMÜLLER, U., WERNER, C., STROZZI, T. and WIESMANN, A., 2004, Multi-temporal interferometric point target analysis. In Analysis of multi-temporal remote sensing images, Series