1. DInSAR technique in monitoring of active landslides along the coastal line of North-East Bulgaria
Hristo Stoyanov Mila Stoyanova Atanasova-Zlatareva2
1Space Research and Technology Institute, 2National Institute of Geophysics, Geodesy and Geography
Bulgarian Academy of Sciences, Bulgaria
Nov 2014 – Apr 2015
Nov 2014 – Dec 2015
Jan 2015 – Dec 2016
Figure 3 Resulting raster maps for displacements after phase unwrapping and geocoding
Introduction
Landslides are one of the major effects occurring after natural disasters such as short-term and intensive rains or after show melting. According to the national authority responsible for landslides monitoring and mitigation in the last two years their number almost doubled.[3] This is reason why elaboration of fast and accurate method for landslides monitoring is needed. One possible solution to achieve this goal is to use DInSAR derived information regarding Earth’s crust deformations such as subsidence and horizontal movement. In the last years this kind of information proved to provide reliable results for the said tasks and together with the dependable source of operational data from SAR missions such as Sentinel-1 form a solid basis for establishing a procedure for creating maps of the vulnerable regions prone to landslides in Bulgaria. One of those regions is located in the North-East of Bulgaria and is well known for several large active and potential landslides. In this research we used SAR data in order to obtain information about the ongoing Earth’s deformation processes in the abovementioned zone. This study was focused on relatively narrow strip along the coast where most of the slopes are highly susceptible to landslides. The achieved final results are in the form of interferograms witnessing the Earth’s crust motions and could be used for determination of areas needing more detailed field surveys. Thus the method used provides cost effective manner for regular monitoring of the sites investigated and can provide complementary data in updating the current landslide thematic maps.
Fig. 1. Endogenic and exogenic landslide factors of the country
1. Seismicity above VIII degree by MSK-64; 2. Active faults; 3. Regions with recent vertical movements > +2 mm/a; 4. Regions with river and sea erosion slope destruction; 5. Regions with precipitation > 700 mm/a; 6. Levels fluctuation of rivers and reservoir more than 5 m; earthquake epicenters with magnitude M ³ 5 (for period 1940—2000); 8. Regions with low degree of endogenic and exogenic landslide factors.
Motivation
The tectonic structure of the Black Sea basin is a complex zone of collision between the African and Eurasian plates and a movement around the various microplates has created an area that has the potential for occurrence of landslides. The Northern part of the Black Sea coast used in our research falls within the eastern part of Moesian platform. However, the majority of tectonic activity in the area of study dates from before Quaternary and this the reason it is commonly adopted that the majority of the faults present in the region of study are not highly active. In previous researches considering the geology of this region it was concluded that geomorphological conditions are highly favorable for landslides formation. One particularity of the area investigated is the abrasion process which is an additional factor in triggering landslide activities. Other factors for landslides generation in this area are the increased construction activities of houses and roads which introduce additional instability on the slopes.
The river erosion (deep and lateral) also plays a role for the activation and development of landslide processes. This type of erosion is caused both directly by water and by the carried in the stream rock materials, applying their action on the bottom and around the riverbed. (Fig.2). The influence of this type of erosion on the activation of landslides is expressed in carrying away the material from the lower retaining parts of the slopes and disturbing the force equilibrium.
The basic influence of sea erosion provoking sliding is the washing away and transporting of materials from the retaining part of the landslide, which is the reason for reduction of the retaining force magnitude. Sea erosion is witnessed approximately on 70% of the length of the Bulgarian Black Sea coastline (Fig.1). The highest values for sea erosion are observed in the investigated coastal sections near Kranevo, Balchik and Kavarna.
Figure 4 Raster heat map based on displacement values at the points of landslides (left) and excerpt with extracted displacement and coherence in vector layer
Fig. 2 Map of landslides distribution in the region
master
slave
Time
(Days)
Bperp
Modeled coherence 1
26 Nov 2014
07 Apr 2015
132
66.11
0.83 2
27 Dec 2015
391
-4.22
0.63 3
01 Jan 2015
21 Dec 2016
720
21.95
0.34
Table 1 List of candidate scenes for final selection
Data
In this research we used SAR data from Sentinel-1A mission of ESA which become operational after November First consideration taken into account was to select ascending orbits of the satellite since most of the expected ground motions are supposed to have east direction. Next we set the time between two scenes to be six months since no major landslide event was registered during the period. Third requirement we had to fulfill was to have minimum vegetation cover due to well know effect of signal attenuation and coherence decrease caused by it. Other constraint we had to cope with was to find those scenes having not large perpendicular baseline and high modeled coherence. From the images available in the data archive the scenes satisfying the first three said criteria are shown in Table 1 and from them selected for subsequent processing were those meeting the last one. This way we formed three interferometric pairs (InP) use in the next steps for DInSAR processing.
Processing
In this research for SAR data processing the SNAP software was used following widely adopted methodology implemented in it . For interferogram formation adopted was usage of DEM having 1arcsec resolution since it was essential to have as high high spatial resolution as possible. Since we are interested in only part of the SAR image independently processed were only two of the available subswats namely IW1/2 which were merged after producing the differential interferogram. Next step was to extract from the new data set a smaller polygon containing the areas of interest. All subscenes were multilooked for filtering and obtaining square pixels, and unwrapped to transform the phase into ground displacements in the line of sight of the satellite.. In the final product we included the coherence band as well because it was used as quality indicator and the value of 0.2 was set as threshold for it. . Finally the resulting images were geocoded, the sea areas were removed and subsequently exported into GE for overlaying the map
Conclusions
In this research the set task to provide information on the Earth’s surface motion was completely accomplished. Based on the results obtained during this study it can be concluded that landslide monitoring can be successfully implemented by spaceborne radar interferometry as open SAR data are processed by freely distributed software. This way SAR instruments based on satellites can assist in monitoring of and mitigating the damages caused by the landslides creating regularly updated risk and disaster management maps in support of the decision-making processes for civil protection activities. Other possible application of those results is in production of large scale maps for the regions susceptible to landslides that could be used in complementing the documentation before construction activities and in the period of exploitation.
Results and discussion
The results achieved from this study are in form of raster maps presenting the calculated displacements from the unwrapped phase information and in vector format combining values extracted from the displacement and coherence bands (see Figures 3&4). Both kind of maps produced has been transferred into KML format for usage in the GE. Raster maps exhibit good correspondence is observed between areas prone to landslides (see white polygons on Figure 3 – as reported to Pangeo Project) and areas having high displacement values. On the vector maps one can check the values for displacement and coherence at the points designated as landslides by the state authorities as shown on Figure 4.
It is to mention that typical fringe pattern found in interferograms can not be observed because no big ground motion phenomena (i.e. earthquake) has happened during the period investigated. But nevertheless the zones being designated by the authors as susceptible to said processes are clearly visible and coincide with the geological findings of underground layers composed by sands and clays in the area investigated – starting from city of Varna up north to Kaliakra cape.
References
Bruchev, I., Dobrev, N., Frangov, G., Ivanov, Pl., Varbanov, R., Berov, B., Nankin, R. Krastanov, M. The landslides in Bulgaria — factors and distribution, Geologica Balc. 36, 3-4; pp 3-12, 2007.
Ministry of Regional Development and Public Works, Map of landslide - accessed May 2017
Coverage Map of PanGeo towns,
Fringe 2017, the 10th International Workshop on “Advances in the Science and Applications of SAR Interferometry and Sentinel-1 InSAR”, Aalto University in Helsinki, Finland, June