Subsidence/Landslides 1

Satellite interferometric synthetic aperture radar (InSAR) data has proven effective and valuable in the analysis of surface deformation making it possible to get qualitative and quantitative information about a range of phenomena, affecting known and new at-risk areas, including:

slow-moving landslides that could trigger fast-moving slides;

volcanism and related dynamics;

subsidence and compaction;

seismic and aseismic fault movement;

buildings instability

other known and previously unidentified phenomena.

Nowadays, SAR data for operational InSAR projects can be provided by RADARSAT-2, RISAT-1, ALOS-2, TerraSAR-X, KOMPSAT-5, Tandem-X, the COSMO-SkyMed constellation (4 SAR satellites) and Sentinel-1A.

The Sentinel-1A European Space Agency (ESA) SAR satellite, that entered its operational phase in late 2014, is the first (civilian) sensor specifically designed for surface deformation monitoring over large areas. Sentinel-1A and the twin sensor Sentinel 1-B (to be launched in 2016) are two C-band, polar-orbiting, satellite systems for the continuation of SAR operational applications, but with significant enhancements, compared to previous ESA missions, in terms of revisiting time, spatial coverage, timeliness and reliability of service. The Interferometric Wide Swath acquisition mode can image a 250-km swath, more than twice that of the ESA ERS mission, with a repeat cycle of 12 days, compared to the 35 days of ERS. Once both sensors will be operational the effective repeat cycle for InSAR analyses will be 6 days. Another key aspect of this mission is the policy of free and open access to data archive, from the Copernicus program, that will increase the range of applications and stimulate the uptake of information based on Earth Observation.

A good example of the potential related to satellite radar data was a large-scale InSAR project financed by the Italian Ministry of the Environment. In 2007, the Italian Government began supporting the largest ever InSAR analysis to be funded by a national government. The project was designed to map unstable areas over the entire Italian territory, using PS data. In January 2008, the Italian Ministry of the Environment awarded a contract to the industrial consortium, composed of TRE, e-geos and Compulab. The main project objective was to expand and update the National Cartographic Portal (PCN) with displacement data for 1992-2010.

This case history could serve as an example for other nations to exploit the Sentinel radar to create a national database of terrain movements. The ability to detect ground displacements that have occurred over hundreds or thousands of square kilometers and even at a national/european scale can be extremely useful, for Civil Protection authorities, to characterize and address areas prone to risk based on homogeneous and reliable measurements.

The huge amount of data available pose new practical challenges and require the definition of new strategies for an efficient data processing aimed at providing surface deformation maps updated every new satellite acquisition. In the paper some of these issues will be addressed, however, it will be demonstrated that a regular update every 12 days over a large area, such as the northern part of Italy, is already feasible.

With the launch of Sentinel-1B it will be possible to reduce the update down to 6 days, this will open new opportunities to design near-early warning systems at a national scale.

An inventory of landslides with the indication of the state of activity is necessary in order to establish hazard maps and therefore take wise decisions towards increased security, reduced vulnerability, and sustainable development. The first step for the landslide hazard assessment is represented by the compilation of a landslide inventory map at regional scale based on the interpretation of aerial or satellite optical images. The photo-interpretation methodology is able to provide information on the landslide’s type, extension, geometry, involved rock mass or soil material. By combing landslide inventory maps together with movement data it is then possible to have information about the state of activity of the mapped phenomena. SAR interferometric approaches represent one possibility for mapping land surface deformation at fine spatial resolution over large areas. Despite limitations due to vegetation cover, the special SAR viewing geometry, atmospheric artefacts, and snow cover, results based on short-baseline differential interferograms (DInSAR) and Persistent Scatterer Interferometry (PSI) are successfully applied in alpine areas to complement photo-interpretation in the inventory of landslides.

In Switzerland, surface displacement was studied in the past using various satellite SAR data. PSI processing was performed using stacks of ERS-1/2 SAR, ENVISAT ASAR, TerraSAR-X, Cosmo-SkyMed and Radarsat-2 data. Differential interferograms were computed for a large number of SAR image pairs, including data from the JERS-1, PALSAR-1, TerraSAR-X and Cosmo-SkyMed satellites with acquisition time intervals of 44, 16, 11 and 4 days time interval.

The Sentinel-1 mission represents a new approach to SAR mission design with acquisitions regularly available over large areas every 12 days. Nevertheless, the spatial resolution of Sentinel-1 is lower than that of other current missions (TerraSAR-X, Cosmo-SkyMed, Radarsat-2) and represent a challenge in the case of surface motion affecting small areas. Starting from the end of the snow season in late spring 2015 we are using Sentinel-1 data for a systematic monitoring of known landslides in Switzerland. Given the limited number of snow-free images available so far, only DInSAR processing with consideration of a high quality Digital Elevation Model (2 m posting) and time-series retrieval based on the singular value decomposition inversion was performed, while PSI processing has not yet been applied. In our presentation will show selected results obtained with Sentinel-1 data on well known landslides (Aletschwald, Breithorn and Zevreila) and permafrost landforms (in particular rockglaciers in the Matter and Saas valleys) and will make a comparison to products obtained with other past and current sensors.

With this contribution we will illustrate the simultaneous use of SAR data acquired by various remote sensing satellites (ALOS, RADARSAT-2, Envisat, ERS, Sentinel-1, TerraSAR-X, TanDEM-X and Cosmo-Skymed) for monitoring various natural and anthropogenic hazards, such as: volcanoes, earthquakes, mass movements, ground swelling/subsidence related to fluid injection/withdrawal, and post-mining deformation.

            Unwrapped interferograms of all compatible SAR pairs are computed using conventional InSAR processing software (e.g. DORIS, Gamma, ROI-PAC or the InSAR suite from the Centre Spatial de Liège). All stacks of unwrapped and geo-referenced interferograms acquired in different geometries, orbits, wavelengths and polarization are then simultaneously inverted for the 2-D time series at each pixel using the advanced Multidimensional Small Baseline Subset (MSBAS) technique proposed by Samsonov and d’Oreye (2012). The MSBAS method produces vertical and horizontal (mainly East-West) time series of displacements of coherent pixels located in the common footprint of all these SAR acquisitions modes. Compared to conventional times series methods such as Small Baseline Subset or Permanent Scatterers Interferometry (Berardino et al. 2002, Ferretti et al. 2001), the MSBAS method is not limited to measuring displacements in the satellite line of sight (LOS), nor restricted to the time steps that correspond to a given satellite revisiting time. The vertical and horizontal time series have a lower noise and a denser temporal resolution than typical for time series resulting from SBAS/PS analysis. The 2-D time series offer additional constraints for deformation source inversions as compared to the 1-D LOS time series and present improved noise characteristics because various sources of noise such as atmospheric, topographic, and orbital are compensated by the MSBAS algorithm.

We demonstrate how the method has been successfully applied in various case studies: volcano monitoring in Democratic Republic of Congo,  Piton de la Fournaise, and Italy; ground swelling/subsidence related to fluid injection/withdrawal in Canada and Mexico; and post-mining deformation at the French-German border. Special attention is dedicated to validation of the MSBAS method by comparison with ground based data (leveling, GPS, seismicity) and error analysis. 

Landslides are a worldwide natural hazard causing thousands of fatalities and severe monetary losses every year. To predict and thus reduce the landslide risk in the future, a profound knowledge about the past and recent landslide activity is of utmost importance. For this purpose, the records about the landslide activity have to be as complete as possible in time and space, in order to derive spatial and temporal probabilities of landslide occurrence as a crucial prerequisite of landslide hazard and risk assessment. However, for most regions of the world such comprehensive landslide records are not available, because the conventional manual mapping of landslides is an extremely time-consuming and labor-intensive task.

This study presents an automated approach for efficient multi-temporal identification of landslides at regional scale based on optical remote sensing time series data. The developed approach allows for retrospective analysis of long-term landslide occurrence and for monitoring recent landslide activity. In case of the long-term analysis, a combined usage of multiple optical sensors is required to achieve best possible temporal data coverage for the longest possible time span. For this study, such a database has been established for a landslide-affected area of 12000 km² in Southern Kyrgyzstan, Central Asia. It consists of about 900 orthorectified multispectral satellite remote sensing datasets acquired by Landsat-(E)TM, SPOT, IRS-1C (LISS3), ASTER and RapidEye during the last 30 years. For monitoring the landslide activity of the last 5 years, high spatial and temporal resolution RapidEye data have been acquired in the frame of the RapidEye Science Archive (RESA) program.

The developed approach comprises automated multi-sensor pre-processing and multi-temporal change detection methods allowing spatiotemporal identification of landslides in an object-based form. The change detection builds on the analysis of temporal NDVI-trajectories, representing footprints of vegetation changes over time. Landslide-specific trajectories are characterized by short-term vegetation cover destruction and longer-term revegetation rates resulting from landslide related disturbance and dislocation of the fertile soil cover. In combination with DEM-derivatives the developed approach enables automated identification of landslides of different sizes, shapes and in different stages of development under varying natural conditions.

The multi-sensor long-term analysis of a 2500 km² region resulted in the identification of 1583 landslides ranging in size between 50 m² and 2.8 km². The highest overall landslide rates occurred in 2003 and 2004 exceeding the long-term annual average rate of 57 landslides per year by more than five times. For monitoring the recent landslide activity the approach has been applied to the RapidEye time series acquired between 2009 and 2015 for the whole 12000 km² study area. The combination of high spatial resolution (5 m) and frequent data acquisition (up to several days/weeks) of the RapidEye data has allowed for the systematic assessment of the whole variety of landslide processes also including small slope failures, which often represent precursors for subsequent larger and more hazardous landslides. Thus, the approach can provide valuable information in the context of early warning. Currently, the applicability of the approach is investigated for assessing the aftereffects of the disastrous Nepal earthquakes of April and May 2015 that triggered thousands of landslides. First results have shown the general transferability of the approach to the differing natural environment of Nepal and its general suitability to operate within a rapid response system. Together with the newly available Sentinel-2 data, this approach has the potential to be developed into a globally applicable landslide mapper, which will open up new opportunities to analyze spatiotemporal landslide activity over large areas facilitating further development of probabilistic landslide hazard and risk assessments.

Worldwide, landslides represent one of the major natural hazards. Large and destructive events are often caused by other natural hazards, such as earthquakes, volcanoes, intense rainfalls and floods. In the result, thousands of landslides can be triggered by single events, such as the 2015 Nepal earthquakes.  For improved hazard and risk assessment profound spatiotemporal knowledge about landslide processes is of great importance, especially under changing environmental conditions. Such an improved process knowledge requires the analysis of past landslide activity resulting in the establishment of dynamic and up-to-date landslide inventories. At the same time monitoring of ongoing slope movements is needed in order to evaluate the current activity state of landslide prone slopes For many parts of the world strongly affected by landslides such information only exist in a limited form because of the lack of a systematic assessment and monitoring of landslide processes covering larger areas over longer periods of time.

The potential of the combined use of optical and radar satellite remote sensing for large-area spatiotemporal analysis of landslide activity has been investigated within a 12000 km study area along the eastern rim of the Fergana Basin representing the area of highest landslide activity in Kyrgyzstan, Central Asia.  Although landslide investigations have already been carried out in this area since the 1950s, there has been a lack of regular large-area monitoring especially during the last 25 years.  In order to perform a comprehensive analysis of landslide activity, a multi-temporal satellite remote sensing database has been established for the study area in Southern Kyrgyzstan containing a multitude of optical data acquired during the last 30 years as well as TerraSAR-X and ALOS-PALSAR radar data acquired since 2007.

The optical data have been used for creating a multi-temporal inventory of backdated landslide activity.  For this purpose an automated approach for object-oriented multi-temporal landslide detection has been developed which is based on the analysis of temporal NDVI-trajectories derived from multi-sensor time series data complemented by relief information to separate landslide-related surface changes from other land cover changes.  The results show constantly ongoing landslide activity with varying intensity between the years indicating the need for the establishment of a remote sensing based  monitoring system for high spatial and temporal resolution analysis.

TerraSAR-X and ALOS-PALSAR (L-band) radar data were analyzed using the Small Baseline Subset (SBAS) time-series technique in order to assess the deformation related to the activation of landslide prone slopes.  Analysis of the results in combination with optical data and DEM information has revealed that most of the derived deformations are caused by slow movements in areas of already existing landslides including ancient landslide deposits which are affected by widespread formation of cracks leading to degradation of the arable land.  Comparison of the X- and L-band results have shown that the  L-band data are especially suitable for longer-term deformation monitoring of larger areas, whereas X-band data preserve well the spatial details and short-term deformations. Moreover, the pronounced relief of the mountainous terrain requires the availability of radar data acquired from different look directions in order to achieve a high degree of spatial completeness.

The combined analysis of the results obtained from optical and radar remote sensing  has shown the big potential  of the complementary use of these techniques for spatiotemporal monitoring of landslide activity at different temporal and spatial scales.  In this context, the newly available Sentinel-1 and Sentinel-2 data will facilitate large-area monitoring of landslide hotspots at a global scale and thus enable systematic and comparable assessment of landslide processes under variable and constantly changing environmental conditions.