On the 11 August 2012 a major earthquake hit the region of Ahar in NW Iran with a magnitude of Mw 6.4. Only 11 minutes later a second main shock with Mw 6.3 appeared in the same region, which has not experienced large earthquakes in the decades before. The short interval between these two events made it impossible to distinguish the first and the second earthquake in the available static displacement data. For this reason, to describe the two earthquakes, the cumulative effect in InSAR and GNSS data of them both has to be analysed. In contrast, in regional waveform data, the second main shock signal is well separated. Here we see an opportunity for a combined-data study to resolve the individual sources of the earthquake doublet. Better constrained locations and mechanisms will help to understand the fault system and the current regional stress in this seismically re-activated region.
Earthquake source analysis is an important field for understanding active fault structures. But, depending on the available data, source studies could hold considerable uncertainties. Combined-data studies often benefit from better data constraints that lower the uncertainty of a solution, compared to one-data type studies. Moreover, the combined-data approach mitigates systematic errors in the solution. Particularly in this study we need good source constraints to separate the doublet’s two sources.
The goal of this study is to reduce the uncertainties for the two source mechanisms in a finite source optimisation joining static and dynamic data. We use InSAR data from RADARSAT-2 acquisitions and published co-seismic displacement vectors based on GNSS data. Unfortunately, only measurements of an ascending orbit are available. The seismological data are regional and teleseismic waveform recordings. We do a simultaneous optimisation of two sources of the earthquake doublet. For the modelling we use Green’s functions of a layered regional velocity model and rectangular, constant slip rupture models.
We expect at least for the first main shock an improvement of the source mechanism, compared to published solutions, by additionally using static displacement data. Potentially, we also achieve to better constrain the second event.
For the optimisation the python software package pyrocko, a toolbox available under GNU General Public License, is used. The optimisation code is based on a Bayesian bootstrap optimisation approach.
The RADARSAT-data used in study have been provided through the RADARSAT-2 SOAR-EU loan agreement #16736. This research is further supported by the German Research Foundation DFG through an Emmy-Noether-Grant.

Area A - Board 105

Interferometric Synthetic Aperture Radar (InSAR) has become an interesting remote sensing technique for geodetic applications. The independent satellite based approach allows to detect natural or man-made changes of the surface of the earth with a high temporal and spacial resolution and is thus intended to validate terrestrial measurements. Successful referencing of SAR scenes is contemplated by joining the InSAR time series with those from geodetic benchmarks, such as GNSS reference stations. However, the coincidence between different geodetic techniques does only provide a millimeter precision at best and therefore, a millimeter calibration of SAR data remains a challenging task. The few reliable and usable links between InSAR and GNSS must be improved and expanded. This includes hardware updates of reference stations with active or passive radar backscatterers, which must be validated by terrestrial methods.

The German Federal Agency for Cartography and Geodesy (BKG) provides reprocessings of the time series of its sustainable German geodetic reference network GREF and improves the station design and distribution to establish links to InSAR. The BKG has reprocessed its GNSS time series of its national reference network, resulting in a much higher accuracy and representativity of the data. While the GREF stations with frequent leveling to the German national leveling network (DHHN) and equipped with absolute gravity measurement time series and permanent GNSS measurements are already providing three dimensional coordinates of high stability and quality, more and more of them are converted to non-building bound, ground stations with deep pillar foundations and upgraded with C- and X-band capable corner reflectors and active C-band transponders to enable sub millimeter-precision terrestrial analysis and a geodetic datum connection to InSAR.

The results are shared with the German national terrain motion service of the German Federal Institute for Geosciences and Natural Resources (BGR) and other interested partners. The BKG uses ESA Sentinel-1 InSAR data and high resolution X-band data from the DLR to cross validate the surroundings of its GREF network independent from other institutes or authorities. Data from the computation center for nationwide and European leveling is compared to the results. The poster presents the GREF network and current work for GNSS reprocessing, SAR datum connection approaches and InSAR data processing of the BKG in Germany.

Area A - Board 106

Artisanal and small-scale mining (ASM) is a cause of major environmental concern in developing countries. In Ghana, ASM is a mixture of legal and illegal operations where illegal mining is locally referred to as “galamsey”. The concerns associated with ASM, and galamsey in particular, include deforestation, water pollution and mercury exposure of miners, as well as a general lack of safety and wider social issues. Recently, the Ghanaian government have made combating galamsey a priority.

Earth observation techniques can assist local governments in regulating ASM activities by providing specific spatial information on the whereabouts of ASM mines. The tropical climate in Ghana, however, hinders the regular flow of useful optical imagery due to a high percentage of cloud cover for most parts of the year. The recent implementation of the Copernicus Sentinel Program now provides regular and reliable radar satellite data capable of overcoming this limitation. The Sentinel-1 satellite produces Synthetic Aperture Radar (SAR) data every six to twelve days for a given location. This SAR data could therefore provide regular spatial information regarding small scale mining activities, allowing the monitoring of ASM over a wide area.

This study aims to develop an approach to automate the discovery and mapping of ASM mining activities on a continual basis. An area on the Ofin River near the town of Obuasi has been identified as the study area.

The detection limits of medium-resolution Sentinel-1 data will be tested against high-resolution KompSAT-5 and/or TerraSAR-X imagery. The performance of a number of filtering methods, machine learning algorithms and image segmentation approaches will be evaluated for the automation process. Given the temporal density of Sentinel-1 data, particular attention will be paid to the benefit of multitemporal image filtering towards improved detection and mapping of ASM activities. The expected outcome of this study is the ability to transfer the approaches and methods developed to other areas in Ghana, ideally providing an efficient system for the tracking of ASM activities to assist in addressing the environmental impacts thereof.

Area A - Board 107

Earthquakes rupture in complex fault patterns and many ruptures have been reported as being segmented. Not only large earthquakes above Mw7 show segmented rupture, also for earthquakes below this magnitude segmentation has been found. It is further suggested that the length of rupture segments and the width of the seismogenic zone are approximately equal. This would imply that the width of the seismogenic zone may determine the maximum size of rupture segments. Definition of segmentation and complexity of faults is depending on the spatial scales of the observations and their resolution. The degree of model complexity is often set arbitrarily in earthquake source modelling or is based on manual fault tracing in satellite images or in the field The discretization of the rupture into segments however could be biased by our model assumptions. To avoid such a bias, we evaluate the rupture segmentation with a data-driven approach, independent of source models.
Using InSAR data by ESA's SAR satellite missions (ERS, Envisat and Sentinel-1) we construct possible segments from lineaments. To justify discretization into segments, significant differences along the fault need to be apparent, e. g. expressed by differences in geometry like bends and/or spatial separations like gaps and step-overs.
We make use of geodetic near-field data from Sentinel-1 and Envisat satellites of several earthquakes in different tectonic regimes and regions. These earthquakes include the 2016 Muji (Pamir), the 1997 Zirkuh (Iran), the 2009 Qaidam (Tibet) earthquakes, the 2016 Amatrice (Italy) and 2008 Zhongba (Tibet).

Our data analysis can broken down into several steps. First we estimate locations that express potential fault traces by using image processing methods on the displacement data, such as gradients in the observed earthquake's near-field displacements. With these traces and by using probability functions of fault-to-fault rupture we compile a set of plausible segmentations for the rupture. Based on the resulting set of possible segments we evaluate the rupture segment characteristics like the segment lengths. Also this set of possible segmented rupture traces can inform source model setups in finite-fault optimizations.

Each set of probable traces can be used to setup a finite fault modelling and constrain the source geometry. Considering this set of possible model geometries lessens the model bias of the finite fault modelling. Based on the modelling results we also investigate if rupture segmentation is linked to the width of the seismogenic zone. To determine that we compare the lengths of the determined possible rupture segments of the earthquakes with the width of the corresponding seismogenic zone. The information about the depth of the seismogenic zone is retrieved from depths of cataloged earthquakes. It would be an important information if we can confirm with a data driven method the assumption of the link between rupture segment length and depth of the seismogenic zone. If the length of segments appears to scale with the depth of the seismogenic zone, we may consider this scaling relation as a characteristic in our rupture modelling to the benefit of less free parameters and more robust modelling results.

Area A - Board 108

Geodetic techniques for deformation monitoring such as InSAR, levelling, and GNSS are complementary in spatial density and coverage, temporal density and coverage, 3D (GNSS) or
3D‐to‐1D projection (levelling and InSAR), precision and datum. However, this complementarity also poses challenges in the integration of the various datasets. When the integration is not performed properly, data analysis, model selection, and predictions, will be sub-optimal.

We developed the CUPiDO tool (Connecting Undifferenced Points in Deformation Observations) to cope with this challenge. The objective of CUPiDO is to provide a standardized way to convert the original geodetic observations to a generic set of double‐difference deformations. Combined with a description of its stochastic properties, the format can be used as a standard in the analysis or data inversion. The approach is based on a two-step procedure: 1) generation of standardized datasets, and 2) application of a conversion tool to generate the optimal set of double-difference observations, together with their corresponding covariance matrix.

In the first step, the original multi-format geodetic datasets are converted to a common NetCDF format. Depending on the nature of the original observations, the data is stored in single difference (levelling, GNSS) or double difference (InSAR) form. Apart from the observations, also the associated covariance matrix describing the measurement noise is inserted.

Second, a conversion tool is used to generate the optimal set of double-difference observations for the user-defined area of interest and period of interest. The advantage of using double-differences is that the effect of different reference points and datums is eliminated, whereas for deformation analysis, the relative (spatial and temporal) motion is of importance anyway.

A first version of the CUPiDO tool is already made publicly available (see https://github.com/TUDelftGeodesy/CUPiDO). In our contribution, we will present the standardized data format, and the functionality of the tool. The approach is applied to an illustrative case in The Netherlands, using levelling, GNSS, as well as ENVISAT, RadarSAT-2 and Sentinel-1 based InSAR data.

Area A - Board 109

The peninsula of Methana in Greece is located northeastern of Peloponnese and is part of the northwestern edge of the active Aegean volcanic arc. The area is characterized by active volcanic activity and extensive hydrothermal alterations (Cloutis et al., 2017). The aim of this study is to detect potential alterations in the dacite lavas of the geothermal system of Methana using multispectral Sentinel-2 data.

The volcanic activity in Methana probably started at the Pliocene–Pleistocene boundary, although the oldest dated rocks are dated around 3.5 ± 0.9 Ma. However, there are very few bibliographic references on the subject and the volcanic history of the peninsula has mainly been reconstructed from geomorphological data. Dietrich and Gaitanakis (1995) attributed the old undated volcanic series to the late Pliocene followed by seven groups of volcanic rocks of the Quaternary period. Radiometric measurements date the most recent of these volcanic groups between 0.39–0.25 Ma, while the older series between 0.77–0.90 Ma (Pe-Piper and Piper, 2002). The most recent volcanic activity (around 230BC) was described by Strabo. The result of this activity was a flank eruption of andesitic composition.

Silica-rich volcanic rocks, predominantly andesite and dacite lava domes and flows, extend radially from the center of the peninsula above the limestone basement (D' Allesandro et al., 2008). They consist predominantly of plagioclase feldspar, clinopyroxene, quartz, hornblende, biotite, and titanomagnetite.

In this study, a multispectral Sentinel-2B image is used, acquired in 29/8/2017. The pre-processing phase consists of image sub-setting to the Methana area and the application of atmospheric corrections. The ground reflectance image consisting of 10 spectral bands with 10m spatial resolution is retained for further processing.

Nine regions of interest (ROIs) are selected, each one consisting of dacite lavas based on the existing 1:50.000 geologic map (Suesskoch et.al, 1984). The nine dacitic lavas belong to three different volcanic periods: a) the old series and the volcaniclastic apron (3.5 ± 0.9 Ma) (D1, D2 and D3, shown in Fig. 1), b) volcanic rocks of intermediate age (0.75 ± 0.2; 0.90 ± 0.25 Ma) (D4 and D5, shown in Fig. 1) and c) younger domes and flows (0.32 ± 0.07 Ma) (D6, D7, D8 and D9, shown in Fig. 1). Three dacite lavas (D1, D2 and D3, shown in Fig. 1), located to the central and eastern side of the peninsula, are known to present alterations (Pe-Piper and Piper, 2013).

For the six remaining dacites, there is no bibliography on the existence of alteration. Four of these lavas (D6, D7, D8 and D9 shown in Fig. 1) belong to the “young” volcanic period and two to the “intermediate” one (D4 and D5 shown in Fig. 1).

In the present study, the spectral unmixing approach is adopted, in order to reveal patterns associated to alterations, using as reference the three dacite lavas (D1, D2, D3) known to exhibit alteration. Then, based on these patterns, we investigate potential alterations within the six remaining dacite lavas.

More specifically, based on the mineralogy of the Methana dacites, the spectral signatures of the following seven minerals were considered: albite, augite (HCP), hornblende, hypersthene (LCP), magnetite, quartz and biotite. These were derived from the USGS Spectral Library Version 7, convolved to the Sentinel-2 MSI sensor (Kokkaly et al., 2017). In addition, an eighth spectral signature for vegetation was also considered. The above spectral signatures are then used as endmembers in the spectral unmixing process. We note here that the linear mixing model has been adopted, where it is assumed that the spectral signature of a pixel is written as a linear combination of the selected endmembers. Then, the abundances (coefficients of the spectral signatures in the above combination) are estimated via the unmixing process, which in our case utilizes the least squares error criterion (Shimabukuro and Smith, 1991).

We consider each dacite lava separately. More specifically, we apply unmixing to each pixel of each dacite lava. Then, we characterize each lava by the set of maximum abundance values of the endmembers, over its pixels.

Dacite lavas D1, D2 and D3, which are our reference for establishing a pattern associated with alterations, show a similar mineral abundance pattern, which is mainly characterized by relatively low abundances of magnetite and biotite (however, the abundance of magnetite is slightly higher than the biotite corresponding one). Furthermore, albite, hornblende and quartz are the prevailing minerals with the highest abundance values while augite, biotite, hypersthene (Fig. 1) and magnetite show relatively lower ones. According to Pe-Piper and Piper (2013), these three dacitic lavas contain considerable faulting and alteration which is evidence of the active hydrothermal field of Methana.

The ‘’intermediate’’ lavas D4 and D5, located at western and northern part of the peninsula, also present the same pattern with D1, D2 and D3.

The D6, D7, D8 dacites are located to the central and eastern part of the peninsula. They present the same mineral pattern with the previously identified altered dacite lavas. The D6 dacite is located in the central part of Methana. The mineral pattern also follows the pattern of the alterated lavas. However, albite, augite, biotite, hornblende, hypersthene, magnetite present lower abundances than in the other dacites. This may be due to a less advanced alteration. On the other hand, D9 presents higher magnetite abundances combined with very low biotite abundances, probably due to an advanced alteration along the edges of biotite (mica), amphiboles (hornblende) and pyroxenes (hypersthene and albite) of the lava. The proximity to the one of the two major geothermal fields of the peninsula increases the probability of the presence of alteration.

Based on the aforementioned preliminary observations, the six dacites (D4, D5, D6, D7, D8 and D9) which lack information about alteration, show similar patterns with the three alterated dacites (D1, D2 and D3). However, the degree of alteration seems to vary. Four dacites present almost similar alteration patterns with the known alterated dacites, one dacite seems to present advanced alteration (due to relatively higher magnetite and lower biotite abundances) and one dacite less advanced alteration (due to relatively lower magnetite abundance).

Concluding, the preliminary results of this study have shown that, in principle, Sentinel-2 data can be utilized to detect alteration in volcanic formations. This is an ongoing study. Current investigations include identification of specific alteration type and degree on the Methana dacites, study of silica content, differentiation between dacite and andesite alterations in the Methana peninsula as well as the connection of different lava types and observed alterations with the geological-geothermal processes in the area.

REFERENCES
1. Cloutis, E. A., Jonatanson, V., Bandfield, J. L., Amador, E. S., Rivera-Hernández, F., Mann, P., and Mertzman, S. A., 2017. Hydrothermally-altered dacite terrains in the Methana peninsula Greece: Relevance to Mars. Planetary and Space Science, 138, 55-74.
2. D’Allesandro, W., Brusca, L., Kyriakopoulos, K., Michas, G., Papadakis, G., 2008. Methana, the westernmost active volcanic system of the south Aegean arc (Greece): insight from fluids geochemistry, J. Volcanol. Geotherm. Res., 178, 818–828.
3. Dietrich, V., Gaitanakis, P., 1995. Geological Map of Methana Peninsula (Greece). ETH Zürich, Switzerland.
4. Kokaly, R.F., Clark, R.N., Swayze, G.A., Livo, K.E., Hoefen, T.M., Pearson, N.C., Wise, R.A., Benzel, W.M., Lowers, H.A., Driscoll, R.L., and Klein, A.J., 2017. USGS Spectral Library Version 7: U.S. Geological Survey Data Series 1035, 61 p.
5. Pe-Piper, G. and Piper, D.J., 2013. The effect of changing regional tectonics on an arc volcano: Methana, Greece. J. Volcanol. Geotherm. Res., 260, pp. 146-163.
6. Pe-Piper, G., Piper, D.J.W., 2002. The Igneous Rocks of Greece. Bornträger, Berlin.
7. Shimabukuro, Y.E., and Smith, J.A., 1991. The least-squares mixing models to generate fraction images derived from remote sensing multispectral data. IEEE Transactions on Geoscience and Remote Sensing, 29(1), pp. 16-20.
8. Suesskoch, H., Bannert, D., Kalkreuth, W., Walliner, P., Strauss, M., Fytikas, M., Mazzuoli, R., 1984. Methana sheet, 1:50.000 Geological Map of Greece. IGME, Athens.

Area A - Board 110

ESA’s observation program has a consolidated heritage lasting for more than two decades with C-band Synthetic Aperture Radar Systems (SAR).
In this framework, the current C-band Sentinel-1 observations constitute an essential component of Copernicus program for the monitoring of the Earth characterized by wide-swath systematic global observations of the Earth to build-up a systematic, long-term archive of globally consistent data. The next generation of C-Band Sentinel-1 mission represents the next logical step in ESA’s C-band line, conceived to serve more applications and services through increased capabilities, first among the others, the improvement of the mapping capability with wide swath, high resolutions and full-polarimetry.
The possibility of providing simultaneously 400 km wide swath and high resolution of 5 x 5 m2 single-look resolution represents a factor 5 to 6 improvement with respect to Sentinel-1’s IWS mode. A 280 km swath width, with the same single-look resolution but with full-polarimetry enabled, could be thought as even more attractive and highly beneficial for services and scientific applications.
It is well known, however, that wide unambiguous swath coverage and high azimuth resolution pose contradicting requirements on the design of spaceborne SAR systems. Nevertheless, Digital Beam Forming (DBF) techniques, such as Scan-on-Receive (SCORE), and multiple azimuth channels, allow to overcome these fundamental limitations of conventional SAR systems. A multichannel system based on this technique was proposed for a follow-on of the Sentinel-1 system[1, 2]. However, new operational modes allow to reach the same imaging capabilities, both with planar and with reflector-based antennas [3].
The next generation of C-Band Sentinel-1 mission, would ensure a revisit time better than the current one of 6 days for the constellation of Sentinel-1 flying over a 12 days sun-synchronous orbit at an obit height of 693 km. For the next generation of Sentinel-1 satellites, a mission repeat cycle of 4 days is required, which is achievable when three satellites fly over the same orbit of Sentinel-1 and displaced of 120 degrees. Furthermore, with this configuration and enabling acquisitions with 400m km swath width, daily acquisitions are ensured for higher latitudes (i.e., starting from 45 degrees), as required by several applications, especially those dedicated to the maritime surveillance and sea ice monitoring.
Further solutions in terms of orbits are available, however at the cost of breaking the continuity with Sentinel-1. A first possibility is to deploy two satellites, displaced of 180 degrees, over a 8 days – orbit that with 400 km swath width are able to map the global Earth surface, as in the previous case, in 4 days. In this case however, areas at 45° latitude are observed every 2 days benefiting from swath overlap at higher latitudes. An additional option would be to deploy a constellation of satellites over a higher orbit (in the order of 1000 km - 1300 km), also in this case over 12 days or 8 days sun- synchronous repeat orbits. The great advantage for a higher orbit is the possibility of enabling even a wider swath of 600 km, that allows to fully cover the North Pole, keeping the access range limited, otherwise prohibitive for lower orbits. This means in other words that the instrument is less complex in this case than for lower orbits.
In the final paper a more detailed assessment of the mission performance would be provided.



[1] M.Zonno et al. “Performance assessment for the high resolution and wide swath (HRWS) post-Sentinel-1 SAR system”, ESA Living Planet Symposium 2016

[2] P.Lopez-Dekker et al., “Application-level performance and trade-offs for the post-Sentinel HRWS SAR Systems”, proceedings of EUSAR 2016

[3] F. Bordoni et al., “Operational Modes for Sentinel-1 Next Generation C-Band Synthetic Aperture Radar ”, ESA Living Planet Symposium 2019

Area A - Board 111

In April 2014 was launched the first satellite of the Sentinel-1 (S1) mission, and two years later, with the launch of the second one, the constellation was completed. Each S1 satellite is expected to operate for at least seven years, and will be replaced by a second satellite, in order to guarantee a further continuation of the service supported by the mission [R1].
Though the relative youthfulness of the mission, the preparations for the next generation (NG) of the S1 mission (S1-NG) are ongoing, and the European Space Agency (ESA) is currently supporting a phase 0 study with the aim to define a baseline system concept and architecture for the space segment.
Given the user-driven nature of the mission, the design of S1-NG satellite is strictly related to the requirements imposed by the final users of the S1 data. Specifically, main applications of S1 data include: monitoring of the European marine environment, of the Artic and see-ice zones, of forests and crop; monitoring and assessment of land surface motion risks; water, forest fire, and flood management; surveillance of the open ocean.
A remarkable feature of the S1 mission, key of its success within the user community, is the use of a reduced number of operational modes, each dedicated to systematic acquisitions over specific areas. In particular, the interferometric wide swath (IW) mode is the main mode for acquisitions over land and costal areas, and the extra wide swath (EW) mode for see-ice monitoring. The corresponding Synthetic Aperture Radar (SAR) images are acquired with a mission repeat interval of six days, in single and dual polarization. Moreover, they are characterized by a spatial resolution of 5 m x 20 m over swath of about 250 km in IW mode, and by 20 m x 40 m over a 400 km swath in EW mode.
With respect to S1, S1-NG is expected to improve significantly the answer to the user needs, in terms of SAR data quality. In fact, the SAR images will be characterized by a spatial resolution of 5 m x 5 m, over a swath width of 400 km. For see-ice applications, the coverage could even reach 600 km. Moreover, the mission repeat interval will be reduced to four days, and the instrument will be provided by fully polarimetric capability.
The necessity to guarantee the continuity of the service, i.e. to preserve as much as possible compatibility between S1 and S1-NG on one side, and that of improving the quality of the delivered data pose severe constraints on the design of S1-NG operational modes [R2], [R3]. In this work possible options will be discussed; the achievable imaging performance analyzed; the challenges and the opportunities that they offer discussed, both from the point of view of the applications and the SAR system.

[R1] Torres, R., et al.: “GMES Sentinel-1 Mission”, Special Issue of Journal of Remote Sensing of Environment “The Sentinel Missions – New Opportunities for Science”, Vol. 120, pp. 9-24, May 2012.
[R2] F. Bordoni, P. López-Dekker, G. Krieger, “Beam-Switch Wide-Swath Mode for Interferometrically Compatible Single-Pol and Quad-Pol SAR Products“, IEEE Geoscience and Remote Sensing Letters, Vol.15, No. 10, pp. 1565 - 1569, Oct. 2018.
[R3] M. J. Sanjuan-Ferrer, F. López-Dekker, I. Hajnsek, F. Bordoni, G. Adamiuk, R. Hanssen, F. van Leijen, H. Skriver, R. Danielson, T. Nagler, L. Pedersen, O. Lang, A. Gabriele, M. Ludwig, A. Lecuyot, “High resolution wide swath SAR applications study: an overview”, in Proc. ARSI & KEO, Noordwijk, NL, 2014.

Area A - Board 112

Taiwan is among the activest neotectonic place in the World as it is the result of the rapid collision of both Eurasian and Philippine Sea Plates, with an annual average convergence rate of more than 9 cm.y-1 (Yu S.-B. et al., 1997). This active tectonic shortening generates the Taiwan relief which is characterized by two major mountain ranges: (1) the metamorphic Central Range from Eurasian continental origin, and (2) the eastern Coastal Range characterized by a volcanic Philippine Sea Plate affinity. In between both runs the crustal suture so-called the Longitudinal Valley Fault (LVF), 125km long and N020°E trending, which presents both inter-seismic (Champenois et al., 2013) and earthquake seismic crisis (Hsu Y.-J. et al., 2018). Using high resolution (HR) images (aerial photos, satellites images (SPOT, Sentinel 2) as well as UAS orthosphots) and using UAS HR DTMs datasets, we have up-dated the published Central Geological Survey Coastal Range and the Longitudinal Valley geological mapping (last CGS map, 1993).
Moreover, we used PSInSAR datasets (Champenois et al., 2013) in order to reveal the interseismic active tectonic structures (folds, faults, tiltings...). We present herein several key studied areas (e.g.: Tapo, Peinanshan...), following the preliminary Pingting terrasse test-site (see Deffontaines et al., 2016) where various active geological, tectonics as well as geomorphological processes are revealed, described and explained that gives interesting case example to understand any Mountain Building processes.
As Eastern Taiwan is the place of a maximum earthquake epicenters, the detailed structural study of this major active seismic geodynamic plate suture zone is a major concern for any Taiwan citizens and has great influence on the Taiwan Natural Hazards.

Area A - Board 113

European Sentinel-1 satellite imaging radar provides especially valuable data for mapping of remote, hardly accessible and poorly studied Arctic regions. The possibility of data acquisition independently on cloud cover and daylight with high temporal resolution allows to reveal latent properties of landscapes and to monitor presumable changes due to global warming. The new data are now widely used to establish surface displacement of low-land areas caused by intense permafrost thawing in some regions. However, these studies usually cover relatively small areas. A large amount of data obtained during the first years of operation of this system allows an assessment of these data information content for research and mapping of the vast areas of the Arctic region. Our research is based on multitemporal Extra Wide Swath (EW) Ground Range Detected (GRD) Sentinei-1 images, with two case studies: 1- the Kotelny Island that is one the New Siberian Islands archipelago, located above 74 deg. North and 2- the Wrangel Island, which is the most eastern island in Arctic ocean, located above 71 deg. North.
The Kotelny island is one of the largest islands, it consists of three parts and is characterized by well-defined landscapes diversity for these latitudes. The Wrangel Island is three times smaller and its landscapes are similar to some extent to some parts of the Kotelny island.
The first stage of the research included the phenology studies of the Kotelny island for 10 typical landscapes chosen with the help of optical images. A stack of co-registered multi-temporal calibrated (sigma nought) images was generated after a pre-processing of original data in SNAP software. These typical landscapes i.e. sandy desert, lowland, river valley, thermokarst depression, plateau and some other are rendered to have distinctly different curves of the seasonal variation of backscattering (Fig.1). The analysis of the curves character and the radar images interpretation were implemented relying on a set of additional information sources (maps, optical satellite images, meteorological data). Integration of the multi-temporal radar images and the co-registered spatial data with the weather information allowed revealing the main factors influencing the backscatter values and its dynamics. The largest annual backscatter variations observed within the studied sites are associated with the surface dielectric properties changes due to active layer freezing and thawing. Backscatter differences within each image were related to peculiarities of the topography (slopes aspects and steepness, landforms and etc.), soils, the type and density of vegetation, surface wetness. Temporal backscatter signatures of lakes depend on their depth during winter period and on wind conditions during summer. Individual events, such as rainfalls or snowmelt, that occurred at the time of Sentinel-1A images acquisition, were also detected based on backscatter signatures.
At the second stage modified multitemporal radar images were created, which adapted for visual and automated images interpretation; then the classification of multitemporal SAR images was implemented. The resulted temporal curves of classes were associated with typical landscapes. Comparing the classification results for both islands it was revealed that the Kotelny island landscapes has more definitely pronounced phenology than that of the Wrangel island, that may be caused by climatic features. Nevertheless, the proposed method allows to define the boundaries of main landscapes features, basing on seasonal changes, and thus to create rather detailed maps (scale of 1:300 000 to 1:750 000) of natural landscapes which are characterized by different combination of the relief, vegetation and soils (Fig.2).
This work is supported by the Russian Foundation for Basic Research project No. 18-05-60221

Area A - Board 114

This paper introduces the new prospects for the automated geological mapping
using Sentinel-2 Multispectral Instrument data. The Sentinel-2 was launched on June 2015 as a second mission for Europe's Copernicus environment monitoring pro-gramme. The multispectral imager provides no-cost, high-resolution data in 13 spectral bands, including the visible, near infrared and shortwave infrared bands. Based on the early sample products, the authors have analysed the possible uses of the data for mineral prospection. As an example of Sentinel data interpretation, an automatic identification of rocks, mineral groups and influence of dense green vegetation recognition quality for the areas of South part of Europe was performed.
In the performed analysis, the laboratory reference spectrums were obtained for
a number of field samples. Subsequently, these spectra were used to calculate and evaluate mineral prospection indices. It should be added that those indices are suitable for identification of particular groups of minerals. The obtained results were also compared to Landsat 7 ETM+ and Landsat 8 OLI data for the same area. The detailed comparison of all results was presented and discussed in details. The main, observed differences are result of higher spatial resolution of Sentinel imaginary and additional bands located in short infrared wavelengths.

Area A - Board 115

Naturally occurring asbestos minerals are a potential hazard to human health, in particular, the serpentine rocks weathering leads to the dispersion of asbestos fibers in the environment, thus increasing the risk of their inhalation, which can lead to asbestosis, a form of lung cancer. Among these minerals are chrysotile (a member of the serpentine group) and the fibrous forms of several amphiboles, including tremolite and actinolite. In Italy, such minerals are predominantly, but not exclusively, found in serpentinite and utramafic rocks, which are common in the Alps and Apennines Mountains (Liguria, Toscana, Piemonte and Basilicata regions). Correspondingly, it is important to identify and monitor the exact location and extent where these particular types of rocks occur, so that government officials can take appropriate advisory or regulatory action to protect the public from potential exposure to naturally occurring asbestos that might result from physical disturbance of these rocks. Conventional field mapping of the distribution of highly serpentinized areas is time consuming and requires detailed sampling and laboratory analysis.
For this purpose, in situ and laboratory 0.4 to 2.5m reflectance spectra have been acquired, using an ASD field spectrometer, showing correspondingly the Mg¬OH absorption feature caused by absorption in the magnesium silicate minerals (lizardite, chrysotile and antigorite) that compose the serpentinites (Mg6Si4O10(OH)8). Serpentinization is responsible for a decrease in contrast of olivine-pyroxene iron absorption features and an appearance and increase in OH– absorption features near 1.4 m, 2.3 m associated with serpentine minerals.
The wide distribution as well as the variety and extent of asbestos-bearing rocks make the selected areas in Italy a significant case study for the evaluation of the feasibility of Sentinel 2 multispectral satellite observations for serpentinites outcrops monitoring, in those areas where the density of vegetation allows their spectral identification. For this purpose, Sentinel 2 data were acquired between 2017 and 2018 over the Italian Alps and Appennines where the most serpentinites outcrops in Italy occur.
Suitable Sentinel 2 spectral regions were used in a spectral classification procedure to produce accurate serpentinites outcrops and secondary outcrops maps occurring in the selected images.
A reasonable agreement between ground spectral measurements, laboratory analysis and satellite data was found, also verified by field visual checking on some accessible areas.
The method presented here could result in reduction of time and effort in the field, because it leads to an efficient mapping of weathered serpentinites outcrops even by using Sentinel 2 satellite data, but by no means replace the field geologist. It is an aid for mineral outcrops selection as initial survey for geologists and it could be a first step in the successful resolution of the lacked adequate information and knowledge about asbestos risks of the local government and in creating awareness of where hazardous products are located.

Keywords: Multispectral satellite data, serpentinites outcrops mapping, asbestos minerals

Area A - Board 116