Advanced Differential Interferometric Synthetic-Aperture Radar (A-DInSAR) technique has proven to be a significant tool to detect and monitor ground deformation over wide urbanized area for studying the displacement time series of buildings, roads, railways, dams and tunnels. London is a megacity of the United Kingdom experiencing a growing density of housing and infrastructures such as the ones related to the extension of the underground Northern Line and the High Speed 2 (HS2) railway construction. In this work, Sentinel-1 SAR data are exploited to measure the London displacements during the 2015-2017 period. The geological interpretation of dense time series provided by Sentinel-1 thanks to its reduced revisit time (i.e. 6-12 days), gives new insights on potential applications guaranteed by the most recent spacecraft sensors for optimizing the surveillance of movements during construction activities and the understanding of the effects on the ground surface generated after the termination of the engineering works such as the uplift phenomena following the cessation of dewatering processes needed for the Crossrail tunnelling in the central part of London.

Area A - Board 119

Rail transport is currently one of main means for conveyance of passengers and goods. The structural health of railway infrastructure such as rails and embankments is an operational requirement of high importance, considering continuously increasing traffic loads and ageing. Early detection of irregular settlement and timely maintenance can guarantee the safety of railway operation and reduce the maintenance cost [1]. As satellite radar missions can regularly deliver almost world-wide ground measurements with millimeter-level precision on a weekly basis, this study will demonstrate a cluster-wise hypothesis testing method for InSAR deformation time series modelling, which can systematically detect structural instability of railway infrastructure in a probabilistic manner.

Essentially, any deformation anomalies in the InSAR deformation time series of a certain railway segment reflects its irregular changes at the past or present phase and can be used to predict its subsequent behavior. To identify the spatio-temporal deformation anomalies, we need first parameterize deformation time series per InSAR measurement point, then estimate and evaluate the parameters. Due to InSAR measurements are always opportunistic, for which the measurement location cannot be predefined, thereby the InSAR mathematical model is always built upon the harsh assumption(s). [2] presents a probabilistic approach to point-wisely determine the optimal functional description of a time series amongst all potential physically realistic deformation models, using multiple hypothesis testing and the B-method of testing. Yet, with the lack of sufficient a-priori information, the stochastic model in [2] initially excludes the spatial correlation between adjacent points. [3] made an improvement in this respect. It shows that using semi-variogram analysis, the spatio-temporally correlated adjacent points can be defined as a cluster, and the associated covariance value can be initialized, and consequently the most probable deformation model per cluster considering both spatial and temporal correlation are determined. However, the signal heterogeneity within a cluster is not shown yet in [3].

To include spatial and temporal correlation into the stochastic model of InSAR deformation time series, and detect signal heterogeneity within a cluster, this work further develops the approach to InSAR deformation time series modelling. We use either externally cartographic information (if available) or internally semi-variogram analysis to categorize spatio-temporally correlated adjacent points as a cluster. The potential spatial deformation patterns, e.g. differential settlement, will be parameterized and introduced in the functional model as pseudo observations and tested by cluster-wise hypothesis testing. Per cluster, the null hypothesis is defined as a unique linear function of time of all points. The alternative hypotheses are designed according to the library of canonical deformation models. Within a cluster, all points are not necessarily following the same deformation model, they can either follow the same deformation model but different parameter values, or distract to different deformation models according to the cluster-wise hypothesis testing. This will allow us more precisely recognizing the deformation pattern and spatial-temporal anomalies. The performance of this approach will be tested for the Dutch railways and evaluated with in-situ survey-train measurements.



[1] H. Wang, L. Chang, an V. Markine (2018). Structural health monitoring of railway transition zones using satellite radar data. Sensors, 18(2), [413].
[2] L. Chang and R.F. Hanssen (2015). A probabilistic approach to InSAR time series post-processing. IEEE Transactions on Geoscience and Remote Sensing, 54(1):421-430.
[3] L. Chang, K. Ou and R.F. Hanssen (2018). Identification of deformation pattern changes caused by enhanced oil recovery (EOR) using InSAR. International Journal of Remote Sensing, PP1-11.

Area A - Board 120

Currently under construction, the new international airport of Mexico City is set in the centre of a dried-up lakebed, now a marshy wetland, which has previously not been urbanized due to the setting of the terrain. In addition to the problem posed by the soft sediment, the area of Mexico City is sinking at a high rate due to groundwater extraction from aquifers, and affected by seasonal storms, which increases the risk of flooding. The chosen area for the construction therefore needs to be stabilized on the long term, before it can safely be used by commercial aircraft. The chosen solution for this problem has been the layering of basaltic rock on the construction site, squeezing out the water underneath. The hope is that once final differential settling rates have been reached across the site, the next step of the runway paving can take place. However, monitoring such a vast area using ground based methods is challenging.
Here we present the results of our case study for monitoring this large scale construction project using Sentinel-1 data. Our analysis supports local and off-site project managers and decision makers to constrain the timing of the completion of different construction steps concerning the preparation of the building site. Furthermore, using InSAR time series, we have estimated the ground deformation rates across the airport site. We identify extremely high differential deformation rates and spatial deformation gradients, approaching the limitations as determined by the radar wavelength, the revisit time and the wrapped nature of phase.
The information we provided is highly relevant for assessing the need of follow-up measures for ground stabilization, as well as long-term temporal planning of the airport construction as a whole. Furthermore, quantification of the deformation was only possible through highly customised and target-specific optimization of the processing chain, highlighting the need for case-specific remote sensing solutions.

Area A - Board 121

SAR Interferometry (InSAR) is a technique that processes a stack of SAR images, detects targets having a stable radiometric backscattering (Persistent Scatterers, PS), and measures their displacement (along the line-of-sight of the satellite) and 3D position, with millimetric and metric precision, respectively. The result of InSAR processing is a collection of tens of thousands of PS per square kilometer (and for each PS tens of displacement measurements corresponding to the satellite acquisition times), which describe the displacement of the area with a dense but sparse collection of measurements. This wealth of information is extremely meaningful, but at the same time requires expert knowledge to be converted in a concise description of the underlying phenomena. Therefore, it is very useful to develop techniques that can support end user decisions through automatic information extraction from the PS measurements.

The first PS manipulation we propose, in order to help with the interpretation of the results, is the projection of the PS onto a 3D model of the target structure. PS positioning is achieved with metric precision from the data processing itself, but the availability of a 3D model of the target structures can improve it, by projecting the PS onto the closest model surface within its resolution cell: the correspondence between the structure and the displacement measured by the PS is thus highlighted.

Our second important post-processing procedure consists in a spatial analysis of the PS displacement measurements. A model (e.g. polynomial) of the deformation is reconstructed onto the surfaces of the observed buildings and structures, in order to sum up the information obtained from the PS measurements with a set of statistical indices that can concisely describe the phenomena in act. The spatial analysis can use the PSs found in two distinct InSAR processings, with different line-of-sight, combine the information contained in those two sets, and extract the vertical, the East-West, and (with less sensitivity) the North-South components of the displacement. The resulting models are able to catch the underlying differential displacement phenomena affecting the building and extend their representation to those parts of the building that are not covered by PSs. Statistical indices are then extracted from the models, since they are useful when processing wide areas rather than single buildings, inasmuch they can concisely describe building stability in a wide area (see Figure).

The third idea is to perform a temporal analysis of the PS displacement evolution, and use it to identify time periods in which the displacement changes its underlying model, thus highlighting a possible change in the structure stability. The results of this analysis are then condensed in statistical indices that concisely describe the displacement evolution of each building within the wider investigated area.

The final step is to combine the spatial and temporal analyses into a single spatio-temporal analysis, by means of either model-based or artificial intelligence (AI) algorithms, in order to obtain a warning system for all the buildings and structures in the area of interest.

The proposed system makes it possible to automatically analyze wide areas with a large number of buildings and structures and a huge number of PS measurements. The results can be used by end-users to take decisions regarding the stability of buildings and structures, without the need of a deep knowledge of InSAR.


Figure caption: Mathematical model of the displacement affecting a building, computed from the combination of the PS displacements measured with the ascending and descending geometries; the component visualized is the vertical one.

Area A - Board 122

Despite many applications and improvements offered by new satellites such as Sentinel-1, standardised procedures allowing for the effective use of InSAR data for mapping wide urban areas are lacking. Displacement mapping methodologies are still frequently based on the visual inspection of average line-of-sight velocities, as for example in the case of the ground stability areas detected in the framework of e.g. PanGeo, Terrafirma ESA GMES project or planned European Ground Motion Service (EU-GMS).

However, the spatial distribution of velocities as visible from a standard point cloud map, is often scattered and affected by noises and offers no reliable information on the probability of movement for intervening areas. Moreover, the information extraction from Multi-temporal InSAR (MTI) techniques is sometimes difficult due to a possible large number of persistent scatterers (PS), thereby entailing long interpretation times.

Final deformation maps are usually fragmented once temporal coherence of resulting PS points is not handled properly. While producing final deformation maps, usually thresholds are applied in order to sample only high coherence ([0.7) radar targets from investigated area. Thresholding on coherence might, however, cause misinterpretation of the results over areas undergoing more complex deformation scenarios. If the discrepancies in the areas of moderate coherence share similar behaviour, it is important to take into account their spatial correlation for correct inference.

Our recently proposed data mining approach allows for minimization of outliers in final results while preserving spatial and statistical dependency among low-coherent observations. Such localities might represent areas with fast deformation processes, seasonal changes or other effects. These would remain undetected, especially over wide areas monitored by Sentinel-1, as their coherence is often weakened and by the rules of standard thresholding procedure they would be discarded. The information and full range of coherence [0 1] over low-coherent areas might, however, be exploited in a similar way the coherence is used in thematic mapping applications such as change detection.

Several case studies from monitoring urban environments are presented in this work in order to demonstrate the ability of our data mining approach to detect and disentangle the contributions of different deformation processes. The use of a full range of a coherence value enabled the detection of potential risk zones and areas affected by significant movements or hardly-to-observe deformation processes.

For a better disentangling between technological constraints of InSAR and deformation processes, it is necessary to develop effective identification method, continuous spatio-temporal deformation monitoring and mechanism recognition analysis, in order to adopt the appropriate risk mitigation strategies.

Area A - Board 123

This work describes a procedure developed in the framework of the EU project “Multi-scale observation and monitoring of railway infrastructure threats” (MOMIT, 2018) to semi-automatically identify clusters of persistent scatterers and preliminarily associate them to different potential types of geological-geotechnical hazards affecting infrastructures. This procedure consists of three main modules: a) ADAfinder, aimed at the detection of Active Deformation Areas (ADA) using Persistent Scatterer Interferometry (PSI) data; b) LOS2HV, focused on the decomposition of line of sight (LOS) displacements from ascending and descending PSI datasets into vertical and east-west components; c) ADAclassifier, that semi-automatically categorises each ADA into potential geological-geotechnical hazards (i.e. landslides, subsidence, sinkholes, swelling soils, consolidation processes and thermal effects) using the outputs derived from a) and b), as well as complementary external information (i.e. inventory maps of different geological hazards, a Digital Elevation Model - DEM - , a geological map and a vector map of the infrastructures). The described procedure allows the infrastructures management authorities to identify, classify, monitor and categorize the most critical deformations measured by PSI techniques in order to provide the capacity to implement prevention and mitigation actions over wide areas against geological-geotechnical threats. The municipality of Abbadia San Salvatore (Tuscany, Italy) is used as a case study for illustrating the developed methodology. A dataset of 152 ascending and 149 descending PSI data derived from Sentinel-1A satellite and covering the timespan 12 October 2014 and 29 August 2018 has been used, jointly with the geological map of Italy (scale 1:50.000), the Italian Landslide Inventory map - IFFI project- (scale 1:25.000), a 25 m grid DEM and a cadastral vector map (scale 1:5.000). The application of the above described workflow to this case study highlights a high capability to process wide areas in very short times and a high compatibility with GIS environments for data visualization and representation. The derived products are of high interest to infrastructures and land management as well as decision-making.

References
MOMIT (2018). MOMIT consortium, 2016. Multi-scale observation and monitoring of railway infrastructure threats. Home | MOMIT project consortium. http://www.momitproject.eu/. Accessed: 02 November 2018.

Area A - Board 124

Monitoring of dams is a crucial and essential practice considering the risk associated with erosion, water load, hydraulic gradients or even external factors. Integration of several techniques could help on verifying the structural health of dams and guaranteeing the safety of these infrastructures and surrounding areas.
This work is aimed on testing the potential of space-borne Synthetic Aperture Radar (SAR) data and multi-temporal interferometric techniques on supporting the traditional dam monitoring systems, helping on identifying the dam structural health with an adequate spatial and temporal resolution.
The relevance of this process emerges when environmental or logistic conditions do not allow monitoring dams through traditional geodetic and numerical techniques. In such cases, results obtained from SAR data combined with a modelling stage constitute a reliable diagnostic tool of dam structural health to avoid any extraordinary failure that may lead to loss of lives.
Here, we focused in the Turkish Ataturk dam, located on the mainstream of the Euphrates River.
Ataturk dam was constructed between the 1983 and the 1990 and consists of a rockfill dam with a clay core, having a length of 1.8 km and a height of 169 m and a total storage capacity of 48.7 million m3. Since its completion, the dam was instrumented and systematically monitored through different sensor types, verifying the structure internal states but also the surface deformation through periodic geodetic measurements. It is estimated that the maximum settlement on the central part of the crest reached 7 m. Interferometric multi-temporal techniques have been applied to ERS, Envisat-ASAR and Sentinel-1 SAR datasets spanning the periods 1992 – 2011, 2003 – 2010 and 2014 – 2018 respectively, all acquired with descending orbit. The availability of archive data allowed reconstructing the temporal evolution of the settlements affecting the structure soon after its completion, enabling a consistent comparison with the existing geodetic measurements along the dam crest. Results showed a maximum rate of Line-of-sight deformation associated to ERS dataset and a reduction in time of the compaction rate, which is conform to trend variation of soil consolidation curves. Moreover, thanks to the availability of Sentinel-1 datasets acquired with both ascending and descending orbits, it was possible projecting the displacement and distinguish between vertical and east-west deformation components, identifying non- linearity affecting the time-series of deformations due to the variation of the water level on the dam basin. Different behavior of surface deformation has been analyzed on the dam and on the surrounding area too, highlighting some dramatic variation on the deformation evolution strongly connected to fault systems and tectonic discontinuities.
The obtained results are part of a multidisciplinary project, STEADY (Satellite Synthetic Aperture Radar interferometry to model dam stability), finalized on assessing possible damages affecting a dam through remote sensing techniques and civil engineering investigations.

Area A - Board 125

Structural health assessment is an important practice to guarantee the safety of infrastructure in general. In case of dam monitoring, it is necessary to control the structure itself and the adjacent water reservoir, to pledge efficient operation and safety of surrounding areas. Ensuring the longevity of the structure requires the timely detection of any behavior that could deteriorate the dam and potentially result in its shutdown or failure. Traditional structural dam monitoring requires the identification of soil movements, tilt, displacements, stress and strain behavior. maintenance works to guarantee the efficiency and stability of these structures.
The detection and monitoring of surface displacements is increasingly performed through the analysis of satellite Synthetic Aperture Radar (SAR) data, thanks to the non-invasiveness of their acquisition, the possibility to cover large areas in a short time and the new space missions equipped with high special resolution sensors, having a short revisiting time. Nevertheless, the availability of SAR satellite acquisitions from the early 1990s enables to reconstruct the historical evolution of dam behavior, defining its key parameters, possibly from its construction to the present. Furthermore, the progress on SAR Interferometry (InSAR) techniques through the development of Differential InSAR (DInSAR) and Advanced stacking techniques (A-DInSAR) allows velocity maps and displacement time-series to be obtained.
The importance of these techniques emerges when environmental or logistic conditions do not allow to monitor dams applying the traditional geodetic techniques. In such cases, A-DInSAR constitutes a reliable diagnostic tool of dam structural health to avoid any extraordinary failure that may lead to loss of lives.
In the framework of STEADY project, numerous dam deformation analysis have been performed, exploiting SAR data acquired by different SAR missions, mainly ERS, Envisat-ASAR, COSMO-SkyMed, TerraSAR-X and Sentinel-1, characterized by different bands, frequencies, resolution, revisiting time. Main advantage of each sensor, possible limitations and attempted integrations of the results are presented. Reconstruction of the evolution in time of deformation was focused on some emblematic cases, which stories are strongly connected with critical political strategies and issuing urban developing planning: among them, the Iraqi Mosul dam, the Turkish Ataturk dam, the Indian Mullaperiyar dam.

Area A - Board 126

Title: On challenges and opportunities for a continental scale ground motion service - from a continent to a single building

A5.04 Design and monitoring of transport, energy and utility infrastructure with Copernicus data

Authors: Carles Debart (Kongsberg Satellite Services AS (KSAT)), Charlotte Bishop (Kongsberg Satellite Services AS (KSAT)), Yngvar Larsen (Norut), Tom-Rune Lauknes (Norut), Petar Marinkovic (PPO.labs)

Abstract text:

The timely availability of quantitatively and qualitatively consistent data is an intrinsic characteristic of the Copernicus program. It creates a whole range of possibilities to develop a new paradigm for EO services and products. The valorisation of such services and products often hinges on the capabilities of the service providers to address the gap between market needs and technology. While the developer driven process of “If you build it, they will come” seems appealing due to full control over the innovation process, this approach often leads to a mismatch between the product and the actual needs of the mainstream end user. This mismatch can be resolved by explicitly adding a step in the development process that focuses on the translation of the bare results into interpretable results for the end users.

Specifically for the case of Ground Motion Services (GMS), using Copernicus Sentinel-1, the data available allows for generation of large scale deformation maps (on regional/national or even continental scale) providing consistent baseline deformation information. While it can be assumed that production of such baseline maps should be possible with current time series satellite radar interferometry (InSAR), the challenge is to continuously maintain and consistently update (both temporally and spatially) this map on a wide spatial scale. Moreover, specific user needs require a product flexible enough to be tailored, scaled and/or filtered; as well as potentially complemented by additional radar data, eg., ESA historical data ERS1/2 and ENVISAT, and high resolution commercial alternatives like TerraSAR-X, COSMO-SkyMed, RADARSAT-2.

This contribution will focus on the service deployed within the KSAT-GMS partnership (http://gms.ksat.no) to address this three-level (time, space, user) challenge. We will outline our technology roadmap, and primarily focus on presenting a number of relevant application case studies from the end user perspective. Specifically, we will demonstrate the flexibility of our service to reliably scale over the entire continental Europe.

Area A - Board 127

Interferometric Synthetic Aperture Radar (InSAR) is a powerful tool for detecting large-scale displacement caused by groundwater extraction, mining, oil & gas pumping, urban development, regional tectonics or other such processes. The advent of SAR satellite constellations with large image footprints and small temporal gaps, including Sentinel-1, facilitates monitoring of these large areas on a regular basis and in near real time. However, existing InSAR processing technology lacks the speed and scalability for wide area operational monitoring. Another significant issue for correctly estimating subtle large-scale displacement is that atmospheric signals which contaminate the interferograms can completely dominate wide area displacement signal. For instance, spatial filters designed to remove atmospheric signal, typically attenuate displacement signal also. The aim of this CARIC (Consortium for Aerospace Research and Innovation Canada) funded project is to eliminate these shortcomings and develop Wide Area Monitoring System capable of processing large areas (~1 million square kilometers) and recovering displacement on scales of 10’s to 100’s of kilometers. This will enable the identification of potential risk areas and mitigate future hazards.

The primary test site used for this project is the entire state of California (400 000 km2) due to the presence of widespread ground displacement from natural and man-made sources. The goal of this study is to find optimal ways to remove atmospheric contamination from InSAR measurements whilst preserving time dependent ground displacement at all length scales.

The performance of these algorithms are compared to the “Generic Atmosphere Correction Model for Interferometric Synthetic Aperture Radar Observations”, Chen. et al., [2018]. With this project, 3vG seeks to find an atmospheric correction algorithm which does not rely on external weather data, as delays in data availability of these models provides a critical limiting factor for rapid analysis of InSAR data. These weather models, however, are used to tune denoising filters to better detect signals which are characteristic of the water vapor contaminants on large spatial extents. Once the solutions are optimized using the external weather models, they can be used independently to decontaminate interferograms from atmospheric noise for near real-time monitoring. Future work could incorporate weather data into the Wide Area Monitoring System to refine the initial displacement signal estimations.

Area A - Board 128