The availability and quality of relevant hydrometeorological data remains a decisive factor for estimating, forecasting and compensating weather-related errors in the satellite interferometric monitoring of slow ground deformations with typical rates of several mm/a. Modern airports with paved runways are supplied with thorough meteorological information and ground control, and therefore represent very appropriate test grounds for studying local weather impacts on synthetic aperture radar interferometry (InSAR). In the present research, basic quantitative relationships between the Sentinel-1 interferometric displacement time series and the timely data recorded at airport meteorological stations were determined using statistical methods of bivariate analysis. In the test airports of Vienna, Bratislava and Graz the phenomena of land subsidence or uplift is rather improbable or is well constrained, and the interferometric displacement signal observed on run- and taxiways represents the cumulative effect of different meteorological factors on the phase delay. The arctic airdromes of Longyearbyen and Svea, both in Svalbard, are known for local deformations of runways. Available reference elevation models with 2.5 m and 25 m postings cover all our test airfields and represent the runway state in the 2010s.

190 Sentinel-1 SAR IW single-look scenes obtained over study aerodromes either from descending or ascending orbits in the period of 2014-2017 were collected and combined chronologically in three regional time series so that the longest sequence included 79 SAR images. Separate scenes in the SAR data time series were ranked and rearranged in accordance with the amount of daily precipitation recorded in the corresponding airfield. For the days with zero precipitation the values of wind speed or air’s humidity were applied as ranking parameters. The newly arranged time series were divided in two equally long parts, each including the same master scene with zero precipitation. The resultant sub-series were processed separately using the standard procedure of persistent scatterer interferometric analysis, and two equally long displacement time series representing the weather influence on “dry” and “wet” persistent scatterers (PSs) were generated for each test airfield. These displacement time series were compared statistically and graphically. Standard and limit values of weather-related errors in persistent scatterer interferometry were calculated. Some of our results were laid over Google Earth imagery for the sake of PSs identification and further analysis.

Typical values of the interferometric coherence observed on airways are somewhat lower than those obtained in built-up areas. Therefore the threshold interval for coherence values was settled between 0.8 and 0.9 thus excluding aside PSs and increasing the number of PSs detected on run- and taxiways. Yet, this led to the appearance of sidelobes at strong scatterers and we had to observe only pixels with local amplitude maxima in order to avoid sidelobe observations. The asymmetric spectrum of most targets brought about additional difficulties in the identification of conjugate wet and dry scatterers. “There would be no luck, but misfortune helped” and in the course of our experiment we discovered the systematic relative range shift between conjugate dry and wet PSs indicating the presence of weather impacts. In steady and flat terrain, typical range shifts varied between 0.1 and 0.5 pixel. We precisely measured the range sub-pixel position of such PCs and mapped the relative range shift at 1:10,000 scale. The spatial resolution of the resultant map depends on the number of conjugate PCs detected in airfields which in turn depends on the orientation of grooved runways with respect to the radar track. Anyway, the resultant plots legibly represent stratified and turbulent artefacts at meso- and local scales that makes them suitable for studying tropospheric turbulences at high horizontal resolution. Interestingly, some artefacts coincided with optically discernible deposits of rubber from aircraft tires that builds up in the start-off and touch-down zones of airport runways. Besides, in our differential PS-InSAR products we recognized new sites with real land deformation which were identified in high-resolution optical imagery as areas with cracks on the surface of concrete slabs.

Further, we designed an original correction of weather impacts based on the iterative remove-restore approach to SAR data analysis and successfully tested it on seven airport runways. In contrast to other known tropospheric correction techniques the proposed technique referred to as “better half” correction accounts, apart from the tropospheric phase delay, for changes in the state of the target surface. The technique is tolerant in terms of SAR scenes connectivity. The removal of noisy SAR scenes taken at the days with the precipitation amount exceeding 2 mm/day and/or the wind speed exceeding 30 km/hour reduced the mean influence of weather impacts to approximately 2 mm/year. Also the variation of residual errors around zero value was essentially reduced. Short-term forecasts of InSAR performance in the test airfields were made by extending the numerical model and using adverse weather data. It was noted that the transition from dry to wet, puddled, flooded, slushed or icy state changes the surface roughness and dielectric properties which, depending on the polarisation and grazing angle of the incident microwave, influences the backscattering coefficient, the phase and the coherence of reflected signals. Thermal effects were neglected in this study because of expansion joints built in runway which absorb thermal expansion or compaction of separate concrete slabs. Also orbital inaccuracies and long-scale ionospheric errors were not treated since we used precise orbits and worked at local scales.

The Arctic ocean is becoming increasingly accessible due to melting sea ice caused by global warming. This opens for increased exploitation of energy resources and fish stocks, whereas new potential shipping routes can reduce transportation times and fuel costs. Kara Sea region is important both for economic and environmental reasons due to the extensive oil and gas resources as well as cargo transport along the North Sea Route (NSR). The Kara Sea is covered by ice from November to June, and it is characterized by land-fast ice along major parts of the coasts. Freezing period begins in September in the northeastern part and in early October in the central part. The southwestern sector normally remains open until November. The ice cover is in constant motion due to ocean currents and wind. Throughout the fall and the winter, the only important openings in the ice cover occur at the boundary between the fast ice and the pack ice. The ice motion produces leads and polynyas in the divergent areas, and ridges and fields of broken ice in the convergent areas. During the freezing period (September to May), new ice is continually produced in the leads. In the areas of convergence, broken ice freezes into much thicker and stronger masses. The presence of leads in the winter season is very useful for navigation.
Vessels navigation can be tracked by means of Automatic Identification System (AIS) that uses electronic exchange of navigation data between vessels with onboard transceiver. The original purpose of AIS was collision avoidance; however, many applications have developed and proposed in the scientific literature mainly in open ocean applications. On ice-covered waters, AIS can act as an important source of information for ships’ trend and behavior according to different ice types. Information like ships’ speed can be used as sources for sea ice classification and machine learning to improve navigation in ice-covered waters. This is a piolet study to investigate the sea ice regimes in the Kara Sea and their influences on the shipping navigation. The study focuses on the route selection and ship speed in the area by combining different data sources; that is, AIS data with sea ice product data (ice chart) and Synthetic Aperture Radar (SAR) images form Sentinel-1. Data from eight ships; 2 icebreakers, 5 cargo ships and, one oil product tanker, were analyzed. The data for each ship is extracted for the whole year 2015, then it is organized by routes (trips) from start port to the end destination. Each route spatially analyzed with ice chart and Sentinel-1 SAR images. The ice chart and SAR images were used to study the variation in speed in different sea ice types. The results showed that there is a direct connection between the ice cover and the ships’ speeds from the AIS data. Another finding is the routes selection between the Kara gate or Cape Zhelaniya at the tips of Novaya Zemlya for entering Kara Sea is highly influenced by the sea ice condition. The methodology we suggest here can be further developed by combining other information and navigation models to benefit ships navigation and operation in the polar regions.

Persistent scatterer SAR interferometry (PSI) has emerged over the last decade as a standard technology capable of precise
measurements of ground deformation from space. The capability of PSI has increased substantially in the previous years due to the availability of multiple SAR missions and the development of several data processing methodologies. In particular, with the exploitation of high-resolution SAR data, the application domain of PSI moves from wide area deformation/subsidence mapping to monitoring small-scale deformation mechanisms.

Deformation monitoring on infrastructure in built environments has recently gained more attention, especially using high-resolution sensors, such as TerraSAR-X (TSX) or COSMO-SkyMed. Despite the successful application of high-resolution data in deformation monitoring, still, its applicability is limited due to the high price of the X-band data and their limited archive. In contrast to high-resolution SAR imagery, medium, and low-resolution SAR data, e.g., C-band data from Copernicus missions Sentinel1 A/B), have some attractive properties such as large volumes of archived imagery, larger coverage/swath, and important from application point of view, free and open access data policy. Despite these properties, due to the lower resolution, they yield a relatively poor point density in urban environments. This makes the applicability of standard InSAR methodologies more challenging for C-band data in urban environments. In particular there are two related but different challenges: 1) there would be a limited number of PS points on urban infrastructures and 2) the lower PS density cause difficulty in phase unwrapping of deformation patterns with small spatial wavelength (e.g., spontaneous/isolated deformation of a bridge relative to the surroundings). That is for PS points on infrastructures with isolated deformation pattern in time/space, phase unwrapping algorithms (which all assume some degree of spatio-temporal smoothness) are more prone to errors.

The main objective of this study is to improve PSI algorithms by reducing the chance of occurrence of unwrapping errors on line infrastructures. The main focus is on designing an optimal network of 1st order PS points around infrastructures. The first-order persistent scatterers form a reference notwork for phase unwrapping of other persistent scatterers. In original PSI algorithms, the first-order PS candidates are homogeneously distributed over a SAR scene. This is accomplished by defining a grid which directs the distribution of PS candidates. The size of the grids and consequently the arc length (i.e., the distance between nearby 1st order PSs in a network) has a main role in the success/failure of the unwrapping step. Longer arcs are more affected by noise components (e.g., atmosphere, nonlinear deformation) and so more prone to unwrapping errors. The main idea of this paper is to create a network with an adaptive density or adaptive grid size around predefined infrastructures.

In this study, we used a hierarchical approach to construct the 1st order PS network around and on line structures. In this regard, two buffers around the line are defined with an adaptive grid size. The closest buffer to the line has the smallest grid size, and the areas far from the line have larger grid size. One challenge in this algorithm is how to define the maximum allowable grid size. To answer this question, we should define what is the maximum allowable arc-length in the whole network. To define the maximum arc-length, we need a metric to quantify the effect of distance-dependent error sources on the success/failure of the phase unwrapping. Here, we use the concept of success-rate. By definition, the success rate is the probability of correct ambiguity estimation (or correct phase unwrapping). To reduce the chance of unwrapping errors, we require the success-rate close to 1. By choosing high success-rate (in this study 99%), we can define the maximum arclength, and based on that, the maximum grid size in the network is computed. The result provides the size of the grid cells that are far from the line infrastructure. Then, for the buffers closer to the line, a smaller grid-size will be used.

The proposed method has been successfully applied to monitor the deformation induced by metro construction in Tehran (Iran), based on sentinel1 data from 2014 to 2018. Figure 1 shows, as an example, the results obtained for one of the metro lines of Tehran. The adaptive network and also the final velocity map of the area has been visualized. For validation, the results have been compared to the original PSI algorithm. The results show significant improvement in both density of detected PS points and also in reducing the unwrapping errors around the metro line.

Monitoring the deformation of railways is a key element in assessing a railway’s structural health. As InSAR (Interferometric Synthetic Aperture Radar) time series enable the estimation of deformation time series for high coherence scatterers with millimeter-level precision, it is becoming a feasible tool to monitor displacements of the large man-made structures in near real-time. Apart from steady state deformation rates and spatial gradients, temporal anomalies in the deformation time series are important since large deformations within a short time interval may do great harm to infrastructure. Additionally, both surface changes, as well as large deformations, may lead to phase ambiguity anomalies which we cannot distinguish using the wrapped interferometric phase, which increases the probability of false alarms.
Here, we propose a recursive multi-temporal InSAR approach by subsequently analyzing amplitude and phase time updates. The main process contains of three parts, (i) establishing the initial deformation model, (ii) change detection based on amplitude analysis and (iii) detection of deformation anomalies. First, a stack of images is processed using the temporary coherent scatterer InSAR method and the deformation model for each coherent scatterer is obtained. The effects of temperature on interferometric phase cannot be neglected due to the metal structure of the railway, so both a linear deformation model and a temperature model are taken into account in this step. The established model is used to detect deformation anomalies in the new image. Second, the relative calibration factor of the new image is determined and an F-test based on the Rayleigh distribution is used to conduct a single pixel change detection based on a significance level. Finally, the predicted and actual displacement of each coherent scatterer is estimated and a t-test is utilized to see whether there is a significant anomaly. The main advantage of our method is to separate occurrence of surface change from deformation anomalies, thus decreasing the probability of false alarms significantly. The approach is demonstrated along a railway in the Netherlands using a stack of RadarSAT-2 XF images.

This study shows the benefits of integrating InSAR from Sentinel-1 mission into risk management systems for the pipeline industry.
Pipeline corridors are subject to a series of hydrotechnical, geotechnical or seismic hazards typically referred as geohazards. Example of geohazards are landslides, subsidence, seismicity, erosion or floods. Geohazards are possible in any geographic locations and environmental conditions, however mountain regions and watercourse crossings are often associated with an increased geohazard frequency and magnitude. These areas require detailed inspections assessing hazard levels, hazard controls and risk mitigation measures.
The pipeline industry and regulators have recognised the importance of adopting a systematic geohazard and risk management plan to reduce the number of pipeline incidents. BGC has pioneered Cambio™, a platform used to evaluate and manage geohazards along pipeline corridors. Cambio™ ingests different sources of data collected from in-situ instrumentation and field inspections and investigations carried out by expert engineers and geoscientists. However, pipelines are also vulnerable to centimetre scale deformation which can be difficult to detect without precise instrumentation, such as slope inclinometers. Furthermore, in-situ instrumentation are costly and typically are only installed at known areas at risk. Thanks to the ability to cover large areas, satellite InSAR data was used to identify unknown geohazards, such as landslides or subsidence and to prioritize or include areas which require field investigation and installation of in-situ instrumentation. InSAR baselines can also support the routing stage by minimizing the crossing between known geohazards and pipeline routes.
For existing pipelines and known geohazards, InSAR complements the traditional monitoring and surveying systems used to control risks along the pipeline alignment.
Deformation rates maps derived from Sentinel-1, were integrated into Cambio™ to support geohazard ranking and prioritization for field inspections. Frequency of updates would be dependent on the risk level, the potential landslide kinematics and the deformation rates measured.
Sentinel-1 data were tested over different geographic locations and climate conditions. The quality of the results is dependent, among other factors, on the vegetation and snow cover. Therefore, the operation uses of Sentinel-1 InSAR data is limited in countries characterized by dense vegetation and constant snow cover through winter months. In arid and semi-arid regions, Sentinel-1 data provided a cost-effective option to be integrated in geohazard management systems such as Cambio™.

Plans of economic growth in Bhutan, a small developing country in the eastern Himalayas, include a large number of hydropower plants, to exploit the potential of water resources and high topographic gradients. The economic importance of such projects is very high, given that hydropower is estimated to provide about 25% of Bhutan’s GDP and almost half of government revenue (expected to soar to 75% in coming years). Emerging environmental problems related to the construction of these dams include increased landslide activity in the impounded reservoir area, widespread disruption to forests, river systems and habitats of endangered species, flooding of agricultural land, obliteration of paddy land and drying of underground springs. Here we address an even more fundamental and safety related problem of the construction of hydropower plants, which results from insufficient geological investigations and geotechnical understanding of the selected dam site location.
Among the new power plants under construction, the 1200 MW Punatsangchhu I Hydroelectric Project (located 80 km east of the capital Thimphu and 20 km south of Punakha) is discussed in this contribution. An autonomous body has been set up for the implementation of this mega project by the two governments of Bhutan and India. Most of the underground excavations of this project have been completed, consisting of 4 intake tunnels each of 6.4 m diameter and total length of 1926.78 m, 4 underground desilting chambers each of 420 m (L) x 17/19 m (W) x 24.74 m (H) and an underground power house located 15 km downstream the diversion dam at Kamechu, Dagar Gewog. Currently under construction with long delays is a concrete gravity diversion dam of 91 m height and 223.8 m length. During excavation works for this gravity dam, which began in November 2008, a 1 million m3 rockslide was triggered at the right abutment in July 2013, causing severe safety and operational issues. The failure caused the construction to come to a stop in favour of stabilisation measures and led to an estimated threefold increase in the expected costs and raised public concern.
We have investigated the right side slope in the area of the diversion dam with an InSAR time series analysis between February 2007 and July 2017 obtained with ALOS-1, ALOS-2 and Sentinel-1 data, and complemented these results with field observations collected in October 2017. Our results indicate that the dam site is not only confronted with excavation induced slope failures, but that the dam site is located within a very large and active landslide body encompassing the entire east facing valley slope. The instability extends for more than 5 km along the valley axis and 4 km upslope, covers an area or about 14 km2 and the relief between the toe of the slope (at 1200 m a.s.l.) and the head scarp (at around 2300 m a.s.l.) is 1200 m. During the observation period (2007-2017) LOS displacements rates have increased on average from 3 cm/year to almost 10 cm/year over different sectors of the slope. The area over which displacements are seen in the observation period covers at least 10 km2, this suggesting potential total volumes between 0.5 and 1 km3, for average depths between 50 and 100 m. The fastest displacements are observed immediately upslope of the slope failure that occurred in 2013, where large scarps are visible. No signs of decelerations are observed neither in the InSAR time series nor in the local displacements at the dam site recorded by a total station installed after the failure (data from the latter kindly provided by construction company). This indicates that the stabilisation measures that were put in place have not been effective and that stabilisation may be impossible, due to the depth of the large instability. This is confirmed by boreholes drilled after the 2013 event, which have not reached the basal shear zone. Although the satellite based time series are unable to resolve the sharp accelerations observed in the local in situ data, the overall trends and cumulative displacements at the dam site match very well. The area currently monitored by the project is only within the secondary slide at the slope toe and covers an area much smaller than the landslide inferred from displacements observed in the InSAR velocity maps, morphological features on ALOS DSM and outcrop data.
The main concerns are related to the feasibility of project completion, to the likely decrease of slope stability during reservoir impoundment, to the long term dam stability and resulting final capacity of the reservoir. A notable case in the Italian Alps, the Beauregard dam site, appears similar to Punatsangchhu-I in terms of size, surface displacement patterns and poor rock conditions. A maximum reservoir capacity at Beauregard was imposed at 6.8 million m3 as opposed to the 70 million m3 for which it had been designed and this may be a likely scenario for Punatsangchhu, for safety to be ensured. Our results show how EO data and methods can play a vital role in providing information for decision making prior and during the construction of large infrastructures, can monitor the efficacy of stabilisation measures, and also shed light on future hazard potential.