As an entire ecosystem, Angkor World Heritage site is well known for hundreds of globally renowned monuments, such as Angkor Wat Temple, scattered across a heritage landscape that includes the Kulen Mountain in the north and Tonle Sap Lake in the South. The sustainability of this site has been of increasing concern due to rapid socioeconomic development of the surrounding region. Several questions have been raised by APSARA and its partner organizations, including: What is the cause of collapses of ancient monuments? Is there an impending crisis that could be triggered by excessive groundwater extraction linked to the tourism industry? If yes, how does it pose a threat to the site sustainability over the next decade?
In this study, we used the smart eye of satellite radar (InSAR) for monitoring surface movement trends for the period 2007-2017 within the heritage landscape. A two-tier network PSInSAR approach was applied for the surface displacement inversion (millimeter-level accuracy) considering its capability in deriving an adequate spatial density of measuring artificial targets on ancient temple architectures and Siem Reap urban areas. Standard interferometric data procedures were firstly implemented (including SLC dataset co-registration, interferogram generation, coherence image generation, flattening and topographic phase estimation, differential interferogram generation, and interferogram filtering). Then, criteria of amplitude dispersion and averaged coherence were jointly applied for the extraction of persistent scatterer (PS) candidates. Followed by the definition of a linear deformation model for the simultaneously estimation of unknown parameters, in the first-tier Delaunay triangulation network, an M-estimator and a ridge estimator were sequentially applied for the estimation of linear motion rates and residual heights of reliable PSs. In the second-tier network, remaining PSs were detected by constructing local star networks referring to neighborhood PSs in the first-tier network. Unknown parameters were directly calculated using the M-estimator. After that, un-modeled motions and residual atmospheric artifacts were further isolated by filters taking advantage of their distinguishable characteristics in the spatiotemporal domain (e.g. low frequency in spatial in conjunction with high frequency in time for atmospheric component phases). Finally, the motion time series of PSs with reference to the first acquisition were derived by combining linear and nonlinear displacements.
High resolution TerraSAR-X InSAR data in the observation period of 2011-2013 shown that pumping of groundwater either in public or in private wells does not cause an immediate region-wide sinking hole; instead, the interaction of material decay, thermodynamics, and seasonal surface motion caused a progressive mild to severe structural instability over the long term that threatens the sustainability of monuments. This phase study firstly provided us a new solution to resolve the current controversy of structural collapse of monuments. Then, evolution of InSAR-derived motion trends along with hydraulic heads indicated that the surface in the archaeological zone was stable up to 2011, followed by a mild subsidence during 2011-2017. We investigated interactions among urbanization, tourism, surface subsidence, groundwater table and precipitation. Interdisciplinary analysis coupled with a prediction model provided us quantitative data that pointed to the potential of a crisis due to the explosive tourism modulated by climate change impacts. We recommend specific countermeasures; e.g. capping maximum number of visitors at 2.5 million per year, for the smart management of this site. Our study clearly demonstrates that the future conservation of the Angkor archaeological Park and UNESCO World Heritage site is dependent on the sustainable development of the surrounding ecosystem, including Siem Reap city in the vicinity.

The preventive conservation of historical constructions and heritage sites against natural hazards, such as extreme climate events, floodings and earthquakes, is a major societal and scientific challenge of European countries that is highly stimulating intersectorial and multidisciplinary research efforts. This work is aimed at presenting some results of ongoing research activities concerning the fusion of deformation measurements from satellite synthetic aperture radar (SAR) interferometry (InSAR) data and in situ monitoring technologies, for structural assessment of monumental buildings.

Activities presented in the work have been carried out within HERACLES European research project that has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No. 700395. The project covers three test beds affected by various forms of material and structural degradation, mostly caused by climate-change related hazards. These are: the city of Gubbio in Italy (in particular the ancient city walls and the Palazzo dei Consoli), the Knossos Palace and the Venetian Fortress of Heraklion in Crete.

The sites of Gubbio and Crete have been monitored since 2011 with series of COSMO-SkyMed high resolution and very high resolution SAR acquisitions processed by persistent scatterer pair (PSP) InSAR technology. The satellite technology makes possible to obtain dense (tens of thousand per sqkm) and precise (millimetric) deformation measurements over very large areas at relatively low cost. For these characteristics, this technology can be a powerful tool, complementary to in-situ measurements and structural analysis technologies, for monitoring of historical sites, often large and/or sparse in the territory.

A specific attention has been therefore devoted (considering in particular the Palazzo dei Consoli in Gubbio test site) to the interpretation of millimetric deformation measurements obtained from PSP InSAR technology through a cross-correlation against data provided by an in-situ structural health monitoring system and through engineering interpretation based on damage survey and structural analysis via Finite Element simulation. A first analysis has been carried out in order to identify anomalies that can possibly indicate the activation of local failure mechanisms consisting of overturning of lateral façades.

Results of such a preliminary analysis have been analyzed in light of the outcomes of a detailed architectural and damage survey, as well as in light of damage scenarios predicted by finite element simulations. This analysis, complemented by visual inspections, has guided the definition and installation of a mixed static-dynamic continuous structural health monitoring (SHM) system deployed over the building, that has been in continuous operation since July 2017.

The availability of some months of continuous monitoring data has represented a notable opportunity to cross-compare remote sensing with terrestrial monitoring data, thus contributing to understanding the potential of InSAR technology to enable large-scale remote structural monitoring of monumental buildings.

The obtained results are very promising, and can be applicable to the structural assessment and monitoring of buildings and infrastructure in general, even though the study has focused in particular on historical sites.

Northern Greece is considered as the birthplace of a significant number of Cultural and Natural Heritage (CNH) sites. However, due to mountainous topography, intense seismic activity, and flash rainfall events, the regional landscape faces severe impacts of damages and losses due to frequent natural disasters, such us landslides. This way, in the area of our interest there are natural Wildlife Refuges and historical monuments of the Byzantine period and the Bronze Age that are at risk. Thus, the estimation of large-scale landslide susceptibility in the region, coupled with the assessment of the propagation of the runout distance of an active landslide is of a paramount importance. Such critical information provides to national and regional authorities the landslide hotspot areas, assisting decision makers in evidence-based disaster preparedness activities and planning of mitigation measures.

Therefore, the main scope of our work is to investigate the contribution of Earth Observation (EO) data and GIS to the estimation of landslide susceptibility and runout distance assessment, in order to provide useful tools for enhancing the resilience of CNH sites.

Firstly, we use Persistent Scatterer Interferometry (PSI) methods to estimate the ground deformation pattern and trends over the last 20 years (1998-2018) in order to construct a new landslide inventory map for the entire area. This dataset is used as a validation tool for the subsequent landslide susceptibility analysis. We then compare two semi-quantitative methods; the Norwegian Geotechnical Institute (NGI) approach and the statistical method Weights of Evidence (WoE). We consider as predisposing and triggering factors the ground conditions (geological map of 1:300.000 scale), the local inclination resulted from a medium resolution DEM (SRTM), the vegetation cover produced from Sentinel-2 data, and the estimation of extreme rainfall distribution of 100-years, based on a CHIRPS monthly rainfall time-series analysis (1980-2017).

The calculation of the slope failure initiation areas (starting points) is compiled through the combination of the initial landslide inventories and the estimated starting zones as determined by the high susceptible zones refined with the concave surface areas and the Topographic Wetness Index (TWI). The runout distance modelling is performed based on a rainfall-triggered scenario, with the combination of the multiple flow direction algorithm of Random Walk, enriched with a Monte-Carlo simulation procedure (Gamma, 2000), and the empirical-based approach of Angle of Reach (Hunter & Fell, 2003). For modelling the future prone affected areas, we use a hydrologically sound DEM, the calculated starting points, and the estimated friction angle values per point. The estimation of the friction angle is based on the fact that it tends toward lower angles with the increase of the catchment area (Wichmann, 2017). The validation of the aforementioned approach is compared with the calculated friction angles of the landslide inventory areas, using the Scheidegger (1973) theory.

Our analysis results indicate that the largest percentage of the recorded landslide events are classified by the WoE method more efficiently, especially in medium-to-very risk classes. The estimated friction angle has less than one degree RMSE, which is highly accurate. Additionally, the runout distance paths and the risk susceptibility index are in a good agreement, as the spatial distribution of the runout is mainly located in very highly landslide susceptible areas.

Our proposed method proves that the contribution of the geomorphological and climatological factors, enriched with GIS and EO approaches is a valuable tool for landslide susceptibility monitoring of CNH sites. The final products can be directly used to assist decision makers and stakeholders for anticipating risk mitigation measures.

Earth Observation (EO) applications to archaeological research in marginal or remote areas has been often limited due to poor satellite coverage and limited temporal and spatial resolution. These limitations are changing today thanks to a) the availability of global, high-resolution satellite imagery from new EO missions, such as the European Union Copernicus Programme; and b) the implementation of multi-petabyte image catalogues and geospatial datasets in cloud computing environments, allowing for planetary-scale analysis of the Earth's surface.

This paper will present ongoing EO-based research carried out by the Computational and Digital Archaeology Laboratory, McDonald Institute for Archaeological Research at the University of Cambridge and the Landscape Archaeology Research Group at the Catalan Institute of Classical Archaeology, with a special focus on the cultural landscapes of South Asia. The study areas encompass core areas to the development of the Indus Civilisation (ca. 3300-2500 BC) in India and Pakistan, ranging from the semi-arid alluvial plains of Punjab and Haryana to the hyper-arid Cholistan and Thar deserts.

In particular, we will discuss multi-temporal and multi-sensor data integration in Google Earth Engine ranging from optical (Sentinel 2 and Landsat missions), radar (Sentinel-1) and DEMs (ALOS, TanDEM-X). Data mining and analysis have been carried out to 1) accurately detect and map landscape features of archaeological, historical and environmental interest; and 2) identify the historical and modern landscape alterations of anthropic origin that have changed settlement patterns and traditional land and water management strategies, with a specific focus on taphonomic biases affecting archaeological visibility and preservation. In addition, a set of machine learning algorithms has been tested and implemented in GE Engine and R (i.e. RF, CART, SVM) using several single and multi-sensor approaches to test the accuracy of the algorithms and the quality of our training and test data. Preliminary results provide a first large-scale automatic classification of anthrosols corresponding to archaeological mounds and landforms of potential archaeological significance. We argue than GE Engine is a powerful platform to test and perform efficient EO-based analyses at a very large scale of analyses.

COSMO-SkyMed, the flagship Earth observation constellation of the Italian Space Agency (ASI), is well known among the current Synthetic Aperture Radar (SAR) space missions for providing: high spatial resolution up to 1 m in SpotLight mode (and even better) for civilian purposes; and very short repeat cycles up to 1 day thanks to the wide spectrum of revisiting combinations that are achievable according to the constellation deployment strategy and configuration (ASI 2016). These properties already proved very advantageous for cultural heritage applications, particularly for identification of surface features, monitoring and condition assessment (Tapete & Cigna 2016). More recently, we have demonstrated the benefit of COSMO-SkyMed long time series acquired in StripMap mode at regular intervals (Cigna & Tapete 2018) to substantiate known events of damage and document unknown situations, thus allowing regular updates of the digital records about the condition of heritage assets. On the other side, the interferometric granularity offered by the constellation is being increasingly exploited to assess the impacts of shallow and slow-moving geohazards affecting heritage sites, such as land subsidence and ground instability, with an increasing in publication of studies focused on UNESCO World Heritage Sites (WHS) in Europe since 2014 (Tapete & Cigna 2017). Building upon this evidence, we aim to discuss the value of COSMO-SkyMed for monitoring cultural heritage in ordinary times and to assess the condition during crisis, in integration to assessments made using data acquired from the SAR and multispectral constellations Sentinel-1 and Sentinel-2 of the European Commission Earth observation programme Copernicus. We prove the concept through the following case studies in the Middle East and North Africa region, with particular focus on specific processes and hazards that are threatening the integrity of local heritage and cultural landscapes: (i) urban sprawl around Cyrene WHS in Libya; (ii) illegal excavations (namely, looting) and anthropogenic disturbance in the Hellenistic town of Apamea and the ancient Mesopotamian city of Mari. In Cyrene, we exploited the full archive of cloud-free Sentinel-2 data acquired since mid-2015 to generate a multi-temporal map of the spatial distribution of new building blocks and roads clearly detected at 10-m resolution in the visible bands as distinctive features of the unregulated urban development spreading across the landscape (Tapete & Cigna 2018). Although the high temporal revisit up to 5 days allowed us to achieve a high temporal granularity of spatial mapping, we achieved a further improvement in filling the temporal gaps by processing COSMO-SkyMed images that we selected from the long time series acquired in StripMap mode at 3-m resolution over the site since the end of 2011. In particular, we extracted the urban development features through the amplitude change detection approach defined in Tapete et al. (2016) and found that a seamless multi-sensor change detection analysis of urban sprawl can be undertaken. We used the same processing workflow to extract features due to looting activities and human disturbance in Apamea and Mari. Bespoke COSMO-SkyMed Spotlight acquisitions were acquired at an enhanced spatial resolution with respect to the coeval time series provided by Sentinel-1 and Sentinel-2 as per their predefined observation scenarios. We could therefore test the performance of a multi-scale approach of condition assessment, by which a site-scale assessment could be initially done using the Sentinels and patterns of change could be identified and mapped, to be then investigated at higher spatial detail with COSMO-SkyMed. We use the results of these experiments to discuss the mutual complementarity of the different sensors in light of future implementation in operational scenarios.

References:
- ASI - Italian Space Agency. 2016 COSMO-SkyMed Mission and Products Description (31 May 2016). DOC n. ASI-CSM-PMG-NT-001, 151 pp. Available online: http://www.e-geos.it/images/documents/COSMO-SkyMed Mission and Products Description.pdf
- CIGNA F., TAPETE D. 2018. Tracking human-induced landscape disturbance at the Nasca Lines UNESCO World Heritage site in Peru with COSMO-SkyMed InSAR. Remote Sensing, 10 (4), 572, 13 pp. doi:10.3390/rs10040572
- TAPETE D., CIGNA F. 2017. Trends and perspectives of space-borne SAR remote sensing for archaeological landscape and cultural heritage applications. Journal of Archaeological Science: Reports, 14C, 716-726. doi:10.1016/j.jasrep.2016.07.017
- TAPETE D., CIGNA F. 2018. Appraisal of Opportunities and Perspectives for the Systematic Condition Assessment of Heritage Sites with Copernicus Sentinel-2 High-Resolution Multispectral Imagery. Remote Sensing, 10, 561. doi:10.3390/rs10040561
- TAPETE, D., CIGNA, F., DONOGHUE, D.N.M. 2016. “Looting marks” in space-borne SAR imagery: Measuring rates of archaeological looting in Apamea (Syria) with TerraSAR-X Staring Spotlight. Remote Sensing of Environment, 178, 42-58. doi:10.1016/j.rse.2016.02.055

Two crowdsourcing projects for archaeological prospection are presented here. Both use human interpretation of optical imagery to detect buried archaeological structures through crop marks in short vegetation. One project aims to identify large features, such as earthworks, in the UK using Sentinel-2 imagery. The other focusses on detecting remains of buried Roman buildings in Italy using Birds Eye imagery on Bing Maps.

Rural areas in many parts of Europe are declining due to urbanisation and industrial activity. The pressure of construction is putting at risk undocumented archaeological structures in countries that have a rich cultural heritage. In many such countries, prior to development, surveys are required by law to ensure that no buried archaeological structures are present in the area to be developed. This is referred to as "rescue archaeology". Remote sensing techniques can improve the efficiency of archaeology survey, which is traditionally carried out through costly and time consuming ground based methods.

Archaeological features, such as buildings or earthworks, may be completely buried beneath the ground. However, their interaction with surface processes may enable them to be seen, particularly in images acquired from above. In dry periods, for example, vegetation growing at the surface may be forced to place deeper roots into the ground to find moisture. If a buried wall hinders root growth, the overlying vegetation will wilt. Conversely, a buried earthwork, such as an ancient moat or canal, may contain soil with increased moisture content. The roots of vegetation overlying the buried earthwork may therefore grow more abundantly. This differential vegetation growth can sometimes be seen in remotely acquired imagery.

The purpose of the crowdsourcing projects are to systematically document the abundant cropmarks over two selected areas in the UK and Italy. Over the UK, advantage is taken of the exceptionally dry summer of 2018, when many cropmarks were revealed. Here users of the crowdsourcing app would view time series of Sentinel-2 imagery over a large part of the south of England to detect cropmarks through their evolution over the summer period. Over Italy, users view single images of very high resolution oblique air photos available in Bing Maps, to detect the abundant crop marks in the rapidly disappearing countryside surrounding the city of Rome, up to a 25 kilometer radius from the city centre.

This work has been carried out by researchers based in the newly created Phi-Lab of the European Space Agency. The two projects have been published in the web-based service “Crowdcrafting”, which uses the PYBOSSA Open Source framework for crowdsourcing.

Documented archaeological sites will help the relevant authorities to update their records of known sites. The documented sites will also serve as training data to apply Machine Learning techniques to automatically detect similar archaeological crop marks in a wider region. The detection and documentation of unkown archaeological sites is essential if they are to be preserved for future generations, especially given the pressure of development in rural landscapes.