Railway track health monitoring and maintenance are crucial stages in railway asset management, aiming to enhance the train operation quality and service life. For this aim, various inspection means (using diverse non-destructive testing techniques) have been applied, however, these means are mostly not able to monitor whole railway track network or track underlying layers (e.g., ballast and subgrade). The use of remote sensing techniques, such as Interferometric Synthetic Aperture Radar (InSAR), can expedite the defect diagnosis process for railway tracks, elevating the scope of health monitoring to a network-wide level. The Ground Penetrating Radar (GPR) has emerged as a particularly reliable method, especially for detecting structural deficiencies in underlying layers. As a result, combining the two distinct non-destructive testing techniques – GPR and InSAR – presents a promising strategy for efficient railway asset management. Recognizing the significance of embracing newer and more advanced monitoring strategies, this paper reviews the fusion of GPR and InSAR methodologies, and explores the potential integration of machine learning models to develop a predictive health monitoring and condition-based maintenance approach for railway tracks.

@en

Synthetic aperture radar (SAR) missions with short repeat times enable opportunities for near real-time deformation monitoring. Traditional multitemporal interferometric SAR (MT-InSAR) is able to monitor long-term and periodic deformation with high precision by time-series analysis. However, as time series lengthen, it is time-consuming to update the current results by reprocessing the whole dataset. Additionally, the number of coherent scatterers varies over time due to disappearing and emerging scatterers due to inevitable changes in surface scattering, and potential deformation anomalies require changes in the prevailing deformation model. Here, we propose a novel method to analyze InSAR time series recursively and detect both significant changes in scattering as well as deformation anomalies based on the new acquisitions. Sequential change detection is developed to identify temporary coherent scatterers (TCSs) using amplitude time series. Based on the predicted phase residuals, scatterers with abnormal deformation displacements are identified by a generalized ratio test, while the parameters of stable scatterers are updated using Kalman filtering. The quality of the anomaly detection is assessed based on the detectability power and the minimum detectable deformation. This facilitates (near) real-time data processing and decreases the false alarm likelihood. Experimental results show that the technique can be used for the real-time evaluation of deformation risks.

@en

Satellite radar interferometry (InSAR) is an emerging technique to monitor the stability and health of line-infrastructure assets, such as railways, dams, and pipelines. However, InSAR is an opportunistic approach as the location and occurrence of its measurements (coherent scatterers) cannot be guaranteed, and the quality of the InSAR products is not uniform. This is a problem for operational asset managers, who are used to surveying techniques that provide results with uniform quality at predefined locations. Therefore, advanced integrated products and generic performance assessment metrics are necessary. Here, we propose several new monitoring products and quality metrics for a-priori and a-posteriori performance assessment using multisensor InSAR. These products and metrics are demonstrated on a 125 km railway line-infrastructure asset in the Netherlands.

@en

Transition zones in railway tracks are locationswith considerable changes in the rail-supporting structure. Typically, they are located near engineering structures, such as bridges, culverts and tunnels. In such locations, severe differential settlements often occur due to the different material properties and structure behavior. Without timely maintenance, the differential settlement may lead to the damage of track components and loss of passenger’s comfort. To ensure the safety of railway operations and reduce the maintenance costs, it is necessary to consecutively monitor the structural health condition of the transition zones in an economical manner and detect the changes at an early stage. However, using the current in situ monitoring of transition zones is hard to achieve this goal, because most in situ techniques (e.g., track-measuring coaches) are labor-consuming and usually not frequently performed (approximately twice a year in the Netherlands). To tackle the limitations of the in situ techniques, a Satellite Synthetic Aperture Radar (InSAR) system is presented in this paper, which provides a potential solution for a consecutive structural health monitoring of transition zones with bi-/tri-weekly data update and mm-level precision. To demonstrate the feasibility of the InSAR system for monitoring transition zones, a transition zone is tested. The results show that the differential settlement in the transition zone and the settlement rate can be observed and detected by the InSAR measurements. Moreover, the InSAR results are cross-validated against measurements obtained using a measuring coach and a Digital Image Correlation (DIC) device. The results of the three measuring techniques show a good correlation, which proves the applicability of InSAR for the structural health monitoring of transition zones in railway track.

@en

Satellite radar interferometry (InSAR) has been used to monitor the structural health of line-infrastructure (e.g. railways, bridges, dams and dikes) in recent years. This enables the retrieval of millimeter-level changes in the line-infrastructure geometry on a bi-weekly basis. However, InSAR is an opportunistic method for which the location of the measurements (coherent scatterers) cannot be guaranteed, and the quality of the InSAR products vary from one case to another. Particularly, this is due to the orientation of the line-infrastructure relative to the satellite position, and its expected deformation magnitude and direction. Hence, the InSAR applicability and performance quality is not uniform. In operational situations, this tends to make asset managers skeptical about the potential of InSAR application on these assets. In this work, following [1] we develop new standard InSAR products for line-infrastructure monitoring, provide tools for predicting optimal multi-sensor SAR data combinations, and propose generic a priori performance assessment metrics for line-infrastructure. These products and metrics are tested on the Dutch railway line-infrastructure asset.@en

Continuous hydrocarbon production and steam/water injection cause compaction and expansion of the reservoir rock, leading to irregular downward and upward ground movements. Detecting such anthropogenic ground movements is of importance, as they may significantly influence the safety and sustainability of hydrocarbon production activities, in particular, enhanced oil recovery (EOR) and even lead to local hazards, e.g. earthquakes and sinkholes. As InSAR (Interferometric Synthetic Aperture Radar) can routinely deliver global ground deformation observations on a weekly basis, with millimetre-level precision, it can be a cost-effective, and less labour intensive tool to monitor surface deformation changes due to hydrocarbon production activities. Aimed at identifying the associated deformation pattern changes, this study focuses on InSAR deformation model optimization, in order to automatically detect irregularities, both spatially and temporally. We apply multiple hypothesis testing to determine the best model based on a library of physically realistic canonical deformation models. We develop a cluster-wise constrained least-squares estimation method for parameter estimation, in order to directly introduce contextual information, such as spatio-temporal correlation, into the mathematical model. Here a cluster represents a group of spatially correlated InSAR measurement points. Our approach is demonstrated over an enhanced oil recovery site using a stack of TerraSAR-X images.

@en

The geolocation of coherent radar scatterers, used for InSAR deformation analysis, is often not accurate enough to associate them to physical geo-objects. The imaging geometry of satellite InSAR results in (i) biases in the entire point field, and (ii) quite elongated and skewed confidence ellipsoids in the range, azimuth and cross-range direction. The metric defined by the covariance matrix of the InSAR results defines the optimal way to associate scatterers with geo-objects. Laser scanning point clouds, stemming from aerial or terrestrial laser surveys, yield very dense geometry of geo-objects and topography. Here we combine InSAR and laser point clouds, taking the covariance metrics of the InSAR data into account. This enables us to correct the positions of InSAR data, to provide a geometric match with geo-objects. We demonstrate how this allows for adding contextual information as attributes to individual scatterers, which improves the interpretation of the InSAR results. @en

PS-InSAR time series yield large volumes of data points, observed during many epochs. While traditional processing algorithms use a single parameterization for the behavior of all points, in reality this behavior will differ significantly between points and over time. It is a challenge to find the optimal parameterization for this behavior, and to assess the quality of the measurements per point and per epoch. Here we propose a post-processing method to improve the model estimation of PS-InSAR phase time series. The method combines machine learning (ML) algorithms and hypothesis testing (HT) into the ML/HT method efficiently leading to significant improvements in data interpretation, parameterization, as well as the quality of the estimated parameters. Moreover we show that we can find structure in the data regardless of spatial location and temporal complexity. In contrast to conventional assumptions that nearby points behave in the same way, with unchanged characteristics over time, a method is developed that takes individual behavior into account. Demonstrating that we can move from spatial and temporal analysis tools to semantic-based analysis.@en

In persistent scatterer (PS) interferometry, the relatively poor 3D geolocalization precision of the measurement points (the scatterers) is still a major concern. It makes it difficult to attribute the deformation measurements unambiguously to (elements of) physical objects. Ground control points (GCP's), such as corner reflectors or transponders, can be used to improve geolocalization, but only in the range-azimuth domain. Here, we present a method which uses only one GCP, visible in only one single radar acquisition, in combination with a digital surface model (DSM) data to improve the geolocation precision, and to achieve an object snap by projecting the scatterer position to the intersection with the DSM model, in the metric defined by the covariance matrix (i.e. error ellipsoid) of every scatterer.@en

InSAR time series analysis involves the processing of extremely large datasets to estimate the relative movements of points on Earth. The estimated movements may reveal geophysical processes, or strain in anthropogenic structures. In parametric estimation methods, it is important to chose the optimal mathematical functional model relating the satellite observations to the kinematic parameters of interest. A standard approach is to parameterize the kinematic behavior, in first order, as a linear function of time, but it is unlikely that all objects behave in this purely linear way. Ideally, the kinematic parameterization should be optimized for each individual measurement point in the area of interest. In this work, following [1] we introduce a method to select the optimal functional model, with a minimum but sufficient number of free parameters using a probabilistic method based on multiple hypotheses testing.

@en

Satellite synthetic aperture radar interferometry (InSAR) has the capability to monitor railway tracks and embankments with millimeter-level precision over wide areas. The potential of detecting differential deformation along the tracks makes it one of the most powerful and economical means for monitoring the safety and stability of the infrastructure on a weekly basis. Yet, the mere capability to detect such small deformations is not sufficient for an operational application of the technique. Handling huge data volumes, homogenizing independent datasets, and the connection with expert knowledge to identify risk areas are challenges to overcome. Here, we use a probabilistic method for InSAR time series postprocessing to efficiently scrutinize the data and detect railway instability. Moreover, to detect high-strain segments of the railway, we propose a short-arc-based method to focus on localized differential deformation between nearby InSAR measurement points. Our approach is demonstrated over the entire railway network of the Netherlands, more than 3000 km long, using hundreds of Radarsat-2 acquisitions between 2010 and 2015, leading to the first satellite-based nationwide railway monitoring system.@en

Satellite radar interferometry (InSAR) has the capability of monitoring rails and embankments over wide areas. It has been demonstrated that the sub-centimeter-scale deformation of millions of InSAR measurements over railway infrastructure can be measured routinely using InSAR. Yet, to handle such huge data volumes and to recognize anomalies (such as localized differential deformation) in an efficient way, still limits the potential for operational use in wide areas, e.g. for monitoring a nation-wide railway network. In this work, we develop and demonstrate a systematic InSAR methodology that can scrutinize data and automatically detect anomalies. The method is mainly based on statistical testing theory. Particularly, we use a ‘short arc’ method to focus on detecting localized differential deformation between two nearby InSAR measurement points over the railway. Our approach is applied to the entire railway network of the Netherlands, with a total route length of more than 3000 km. We used 210 Radarsat-2 descending data from three different tracks which were acquired between 2010 and 2015. A differential deformation and anomaly map are produced. This method will be further investigated for all railways in China. This research is supported by the Young Research Scientists Support Program, in the framework of the Dragon cooperation 2013 – 2016 (Dragon 3).@en

Multitemporal interferometric synthetic aperture radar (InSAR) is increasingly being used for Earth observations. Inaccurate estimation of the covariance matrix is considered to be the most important source of error in such applications. Previous studies, namely, DeSpecKS and its variants, have demonstrated their advantages in improving the estimation accuracy for distributed targets by means of statistically homogeneous pixels (SHPs). However, these methods may be unreliable for small sample sizes and sensitive to data stacks showing large time spacing due to the variability of the temporal sample. Moreover, these methods are computationally intensive. In this paper, a new algorithm named fast SHP selection (FaSHPS) is proposed to solve both problems. FaSHPS explores the confidence interval for each pixel by invoking the central limit theorem and then selects SHPs using this interval. Based on identified SHPs, two estimators with respect to the despeckling and the bias mitigation of the sample coherence are proposed to refine the elements of the InSAR covariance matrix. A series of qualitative and quantitative evaluations are presented to demonstrate the effectiveness of our method.

@en

Irregular settlement of railways, either due to the loading of the trains or local ground deformation, impacts its structural stability and the safety of passengers on board. Conventional methods for structural monitoring of railway use in-situ measurements, from GPS, leveling or special survey trains. These methods are expensive and can only be applied on a limited scale, either in space or time. Moreover, they are usually only used at locations where structural deformation is suspected, requiring a-priori knowledge which may not be available everywhere. Using satellite InSAR, we are able to complement these conventional methods and monitor the kinematic behavior (deformation) of railways with millimetric precision, to detect irregular settlement. Here we use a probabilistic method for InSAR time series post-processing for the automatic detection of anomalies (e.g. railway irregular settlements). It is based on statistical hypothesis testing and the B-method of testing. In this method, we first (1) build a library of canonical kinematic functions, based on physically realistic behavior, such as linear, seasonal, temperature-related, step-wise discontinuities and exponential behavior. Then, (2) we find the best model per InSAR measurement point using multiple hypotheses testing. Particularly to detect irregular settlement of railways, the localized differential deformation between two nearby points (i.e., over ‘short arcs’) is more important for railway stability than the large deformation of certain point with respect to a far-away reference point. Therefore, we apply the testing methodology on short arcs. Finally, (3) we evaluate the quality of the estimated parameters, and classify the InSAR measurement points along the railway in terms of their temporal behavior. We conclude that irregular settlement of railways can be recognized. Since there are more than 100,000 InSAR measurement points for testing, we use the B-method of testing to increase the computational efficiency and define the optimal testing settings such as the level of significance and the power of the test. Our method is applied to all railways in the Netherlands. The kinematic time series of InSAR measurement points are derived from 73 Radarsat-2 acquisitions between June 2010 and August 2015.@en

Many areas in the world are susceptible to sinkholes and the associated risk of a sudden collapse of the surface. About 13% of the world’s land surface is covered by carbonate rocks which are sensitive to erosion by running water, leading to cavities and potentially sinkholes. In urban areas, sinkhole risks are a direct threat to human lives. Detecting sinkholes is notoriously difficult, as techniques such as ground-penetrating radar (GPR), electrical resistivity tomography (ERT), seismic methods, and microgravity, usually have a very localized range, and are difficult to deploy between buildings. Yet, recently it has been demonstrated that the detection of small depressions using radar interferometry can be indicative for imminent sinkhole collapse site identification. In an earlier study, we have demonstrated that a sinkhole occurring in the south of the Netherlands appeared to be observable as gradual deformation years before the actual collapse. Building on this experience, we designed a warning system to detect locations where the spatio-temporal behavior of the surface, or objects on the surface, has the characteristic fingerprint of a subsurface cavity, or an imminent sinkhole. We apply a robust method of hypothesis testing based on time series of TerraSAR-X and Radarsat-2 SAR data. We report on the characteristics of the method, the ways to deal with false alarms, and the potential for operational deployment.@en