Extensive experience gained from the provision of Maritime Situational Awareness and maritime monitoring activities concludes that the most complete picture of the maritime domain is through the combination of all information sources available. Heterogeneous, unstructured, multi-sensor, free and commercial data always represents the best overall solution when used in combination with one another.
However, the continuous increasing amount of data (e.g. the evolution of existing services and new satellite constellations), characterised by improved resolutions in time and space, requires the continual development of new methodologies for analysis. In this context, the AI and Big Data approach represents the right approach to process, aggregate and fuse all the input data and present a simplified view of analytics to the users allowing to improve the decision making process.
Within the context of monitoring maritime activities the analysis of huge volume of data from tracking systems (e.g. AIS, VMS, LRIT) and satellite Earth Observation (EO) using the current Big Data Analysis approachs, enables to better contextualize the maritime scenarios to improve and optimize the monitoring activities.
Using the right combination of free and open information (e.g. satellite data from the Copernicus program [1]), together with the commercial data, EO analytics can focus on the analysis of maritime traffic, environmental protection and pollution identification over areas where the events of interest can be automatically detected (e.g. anomalous behaviors or oil spills) which in turn could allow for the reduction of costs for aerial and naval patrols by targeting only events of interest that have been triggered and resulted from the analysis instead of general patrol patterns.
For operational purposes, as proved in many initiatives (e.g. ESA LexEO; monitoring of oil spills in the Adriatic Sea and the systematic monitoring of the Mediterranean Sea), the open and free data can in fact be used for the systematic monitoring of oceans providing sustainable maritime monitoring services for knowledge discovery and early warning notification related to the maritime activities.





Figure 1 – Systematic monitoring of the sea, from data to insights
The huge amount of available data can also be combined for the creation of aggregated maps to highlight where specific phenomenon is concentrated, allowing the generation of sets of reference information which can be used in the anomaly detection activities, as well as for the planning of patrolling/exploring activities. Clusterisation of vessel positions can be used to generate density maps and extract Maritime Patterns of Life. The same approach can be applied to detected oil spills and provide indicators of vulnerable areas (strongly impacted by pollution activities) or potential seepage.

Figure 2 –Maritime traffic classification by ship type using the transmitted AIS messages
For the future, and following current trends (e.g. new constellations, RPAS/UAV, HAPS, IoT), there is a continuous growth of new data sources with increasing resolutions (in time and space). The increasing amount of data available will require a complete exploitation of Big Data Analytics for the operational and systematic monitoring of maritime activities. All end users, such as national law enforcement agencies and environmental protection agencies, must be ready and open to this new opportunity, allowing to upgrade their operational procedures and trusting the new methodologies as new way to optimize the allocation of available resources.

Area B - Board 180

Operational maritime surveillance services using satellite SAR imagery started in the early 2000’s based on the European ASAR ENVISAT ASAR and the Canadian Radarsat-1 SAR missions. These services aimed at highlighting the maritime situation in various applications (illegal fishing activities, sea ice and icebergs monitoring, etc). Since then, maritime surveillance has known a remarkable development in Europe notably with the implementation in 2007 of the CleanSeaNet service by the European Maritime Safety Agency (EMSA). The launch of series of commercial SAR EO missions (Radarsat-2, TerraSAR/PAZ and Cosmo SKYMED), the advent of satellite AIS constellations and of course the European Copernicus program were undoubtedly important milestones associated to this development too.

As a service provider, CLS played its part in the definition and provision of maritime services since 2003. Over more than 15 years, CLS has been continuously providing operational services for all applications related the maritime surveillance and the observation of the marine environment. For example, CLS launched in 2003 the pioneer detection service of illegal fishing activities in the southern Indian Ocean. From 2008 until today, detection and tracking services of icebergs in the southern seas around Antarctic have been provided on a yearly basis to support safety for navigation. CLS also supported the growth and the diversification of maritime surveillance services in Europe such as implemented by EMSA for member states and other European agencies. All these services have experienced the loss of historical SAR mission and the venue of new ones. Despite these changes, most of maritime surveillance services were able to evolve in a positive way for the end users, benefiting in the ends from an increased revisit, an improved operational capacity, a better data quality, more acquisitions modes not mentioning a best value for money.

Far from being a threat for commercial SAR missions, the Sentinel satellite missions must be seen as a real opportunity for the consolidation and development of historical services but also for the advent of new services. The objective of this presentation is to illustrate the smooth integration of Sentinel and commercial SAR on several applications of maritime surveillance. The illustrated services will cover multiple applications of maritime surveillance, including inter alia oil spill monitoring, monitoring of Illegal Unreported Unregulated (IUU) Fishing, monitoring of maritime traffic, border surveillance, sea-ice & icebergs mapping as well as the monitoring of oceanographic and meteorological phenomena. Those illustrations will highlight the complementarities of both open access and commercial EO missions in terms of technical and operational point of views. Some examples of emerging services made possible by Sentinel-1 will also be presented.

Area B - Board 181

The European Copernicus Sentinel open data together with the power of cloud infrastructures provide players in the EO sector with the unprecedented opportunity to design operational Earth monitoring services, shifting from the provision of data to the provision of continuous monitoring services. This shifting (i.e., continuous access to information) is the key point upon which addressing real users’ needs in the new Era of Big Data. Moreover, shifting from monitoring services on request to long time information services available under subscription is the real disruptive innovation in the field of EO: end-users pay for the information not for the processing.
At the forefront of this new innovative model there is Rheticus®, a cloud-based hub that automatically processes satellite imagery and geospatial data, and delivers geo-information services ready-to-use by end-users. Designed and developed by Planetek Italia, Rheticus®, from the name of the unique pupil of Nicolaus Copernicus, moves beyond mapping visualisation, thanks to a broad range of advanced geo-analytics. It provides application services, based on open data, such as satellite images and geospatial, environmental and socio-cultural data available online. The main services already available on the platform are based on Sentinel-1, Sentinel-2 and Sentinel-3 satellite data. Thanks to these data, Rheticus® is capable of delivering continuous monitoring services of Earth's surface transformation phenomena, as the urban evolution, landslides, risk management, fires, oil and gas or the quality of marine waters.
Actionable information are provided by means of thematic maps, geo-analytics, pre-set reports and alerts. Contents are dynamically displayed through an intuitive and user-friendly web dashboard, available 24/7 on any device or in Machine-to-Machine (M2M) mode directly within users’ systems.
In this way, engineering and utilities companies, public and private organizations can easily integrate geospatial free and open information in their business processes, without having to worry about technical data analysis and having skills to process data.
Rheticus® allows end-users to gain insight into patterns not easily identified through traditional approaches to better understand the whole story that lives within data related to their assets (e.g. roads, railways, buildings, dams, mines, water supply networks and utilities), combining historical and daily/weekly satellite imagery acquisitions.
An easy to use mobile and web interface provides dynamic indicators and reports derived from the massive data processing of big satellite data. Thanks to Rheticus® applications, as an example, the information derived by the Sentinel-1 RADAR satellite data is quickly converted into business intelligence insights, thus demystifying the complexity of SAR interferometry.
By Rheticus® Platform, Planetek Italia succeeded in creating several monitoring services that provide timely solutions to address users’ needs in various industries and vertical markets. Currently Rheticus services cover six different application fields: Displacement, Marine, Network Alert, Aquaculture, Wildfires and Urban Dynamics. These vertical services transform data into actionable knowledge thanks to our business intelligence tools, overcoming the concept of static maps.
Planetek Italia is continuously working in the creation of new monitoring services through collaborations with academic and research centers. New applications may benefit from multi-source and multi-sensor analysis, as well as from merging data from heterogeneous platforms and exploiting of the most recent Machine Learning paradigms.
Rheticus® platform is accessible by subscription from users.
Rheticus® guarantees always the best quality-price ratio available on the market, thanks to the use of open data, automatic processing procedures and its cloud based architecture.

Area B - Board 183

In 2018, Europe experienced a devastating drought and heat wave with scarce precipitation and very high temperatures resulting in severe crop losses, threatening the financial sustainability of millions of farmers in many EU Countries.

Mitigation measures for the farmers have been activated by the EU Commission such as advanced payments (on direct aid and rural development aid) and derogation from specific greening requirements. Additionally to the EU measures several EU member states provided further state aid measures for crisis assistance including substantial ad-hoc disaster payments like Germany (€ 340 million), Sweden and Poland (approx. € 116 million each) to offset the sustained production losses.

While this second major drought since 2003 caused again significant damage to agricultural production in many parts of Europe (and this threat will presumably further intensify with climate change), early and reliable information on the occurrence, regional extent and severity of droughts has become increasingly important.

The Federal Government of Germany for instance assumes damages in the amount of € 680 million, with around 10.000 farms being affected, whereas the German farmer’s association (DBV) estimates an overall 26% decrease of crop yield from the previous 5 years average with a resulting financial damage of around € 3 billion. This big difference assessing the financial consequences is due to a lack of a consistent approach to assess the financial losses using reliable data and technically sound methods.

The key challenge of any drought monitoring system in future is to derive reliable and precise information from vegetation status - which can be monitored with increasing precision over time and space by satellite technology – to quantified crop losses and in a further step to the financial losses caused by drought and heat weaves. In order to fill this gap international insurance expertise in respect of crop loss assessment will be integrated step by step to further develop the monitoring system.

Initiation of any state-aid measures or insurance system to be put into place requires this kind of information. In response to this information demand, GAF developed a fast, precise and cost-efficient information system for monitoring of drought events for a total of 39 European and neighboring countries, covering more than 5.8 million square kilometers. The analysis inter alia revealed that the crop condition in e.g. July was significantly worse than in any of the 15 preceding years in especially the northern European countries.

The European Drought Monitoring Service initially has been developed using a private cloud platform from GAF for implementation. Currently, the system transfer on the Mundi Web Services cloud platform is ongoing, which is an instance of the Copernicus Data and Information Access Service (DIAS) initiative. Hereby, the Mundi cloud platform offers a scalable, high-performance processing environment with direct access to various earth observation data archives. Using a dedicated user interface implemented on the Mundi cloud portal, users can analyze and interact with several datasets in various spatial and temporal dimensions over entire Europe.

Area B - Board 184

In November 2018 the Institute of Geodesy and Cartography completed the demonstration project entitled: “Advanced Sustainable Agricultural Production – ASAP” (contract no.: 4000110860/14/NL/CBi).
The Project lasted 36 months and aimed at development of the one-stop-shop platform for automatic, real-time satellite data processing and analysis, with the possibility to synergize it with ancillary data and further to develop products adjusted to various User needs and requirements. The modular system architecture was designed for possible extensions with additional processing chains, adjusted to newly identified clients’ requirements as well as to add new sources of data.
Within the project execution, various Users and Customers representing the agricultural market were identified and their needs and requirements towards the EO based solutions supporting their work, were thoroughly analyzed.
The Insurance sector was identified as prospective customer of the System developed within the project. In order to demonstrate the system capabilities and validate EO based products accuracy and applicability, two pilot demonstrations were conducted. The pilot activities comprise of application of EO data and ancillary data for overwintering of crops assessment and drought impact estimation.
Pilot demonstration, focused on assessment of crops damage due to negative winter conditions, was designed according to the requirements of the one of the biggest insurance company in Poland. In order to thoroughly assess the crops damage during the winter conditions, the EO based Prototype of the System was developed. The System is based of four components. Three of them are applied in order to estimate the influence of main factors impacting the crops conditions during the winter: surface water (Sentinel 1), surface temperature and snow cover (MODIS AVHRR) as well as air temperature and wind speed (meteorological data from WRF model). Additionally the crops reduction was assessed on the basis of NDVI index decrease between Autumn and Spring conditions (before and after winter conditions). The System will be subsequently enhanced with the models for calculation of number of crops before and after winter conditions which is being used as a basis estimation of crops damage in the claim based insurance (opposite to index-based one). The models will be based on the in-situ measurements performed currently in the winter rape and winter cereals fields in Poland. The in-situ results will be used as a training samples for the aerial imagery collected for this areas. Further the aerial imagery will be applied into neural networks for development of the model of crops number estimation on the basis of Sentinel 2 satellite data. Finally the developed models and anticipated to be used as an exclusive source of data on the claim resolution by the insurance companies (instead of costly “on-spots“ controls).
Subsequently, due to severe drought conditions in Poland, the Executive Agency of Ministry of Agriculture (KOWR) approached Institute of Geodesy and Cartography for development of the Prototype of the System for drought assessment on the level of individual agricultural plot. The products of the ASAP system: crops classification in the selected NUT3 area and NDVI index estimated for the entire growing season 2018, was integrated with the NOAA AVHRR based drought assessment system (developed in the Institute of Geodesy and Cartography in 1997). The delivered System demonstrates the assessment of drought impact on the particular crop type in the individual agricultural plot as well as delivers the NDVI based map of crops condition decrease after the period of drought influence. Further the System was upgraded with the meteorological data comprising of: daily sum of precipitation, average temperature (of the air and land surface) and soil moisture, obtained from the WRF model. The Delivery of the presented System expanded with the model for yield forecasting and yield reduction calculation is expected to be the Milestone for the insurance sector in increasing the drought insurance claims sale. The weight transfer of drought compensation from Government to insurance sector is highly anticipated.
Both the presented Prototypes were assessed very positively by their receivers. Since though further validation of the models are required, the Contract Change Notice toward the ASAP Contract was submitted to Polish National Delegation to ESA and European Space Agency. The project extension would enable the Institute of Geodesy and Cartography and commercial partner – OPEGIEKA sp. z o.o. to provide further pilot demonstrations which are anticipated to deliver fully operational and comprehensive system and service enabling to apply it for claim based insurance for crops (yield) reduction due to negative overwintering of crops and drought.

Area B - Board 185

Data sets of the Chinese Zi Yuan 3a satellite (ZY-3) were investigated to derive forest structural information. These information should support decisions on the strategical/tactical level of a forest enterprise at an envisaged scale of 1:10.000. ZY-3 is combining four bands VIS/NIR imaging at 5.8 m spatial resolution with three line panchromatic imaging at 2.5m nadir and 3.5 m off nadir stereo imaging capability. The analysis focused on complementarities and synergies by combining height and angular information from stereo bands and spectral information from the multispectral bands of ZY-3. The stereo bands derived digital surface model provide canopy heights with an accuracy of appr. +/- 3 m. Angular signatures, giving answer on the anisotropy behavior of the surfaces, were derived via the difference in intensity of the forward and backward stereo band reflectance (anisotropy ratio). Forest type classifications were performed separately based on angular and multispectral signature and compared with the approach combining these signatures. Main forest types “coniferous”, “broadleaved” and “mixed” were classified with all approaches. By introducing the height information the three forest type classes could be expanded to nine overall classes by distinguishing the forest management relevant height classes <12 m (establishment), 12-24 m (stabilization, qualification) and >m (dimensioning). Verification of the classifications was performed on behalf of digital CIR photographs and LIDAR height products.
The best classification result into forest types was received by the combined method with the highest overall accuracy of 81 % and with the KHAT statistic of 63.8 %. In the next step the classification output is used to place inventory points into area objects of the respective strata. If the object is large enough to completely cover the 500 m² circle of an inventory point then the condition for becoming a candidate for an inventory point assumed fulfilled.
A statistically calculated number of 77 inventory points are randomly selected from the set of objects fulfilling the above condition. Tree species composition and volume estimation of the ZY-3a data based inventory is compared against the result of the traditional regular grid inventory (348 inventory points) of the forest enterprise. Tree species above a diameter at breast height of 0,7 m found in both inventory approaches were the same. In the RS data steered inventory coniferous were underrepresented (53,9 % compared to 61 %). With regard to the volume estimation, the stratified remote sensing based inventory showed a relative standard error of 4.2 %, at a mean volume per hectare of 332,7 m³/ha when using the ZY-3a derived heights and a standard error of 4,6 % with an estimated mean volume of 356,8 m³/ha when using the measured heights at the ZY-3a selected inventory points. In comparison, the relative standard error of the regular grid inventory was 2.8 %, with a mean volume per hectare of 336.2 (m³/ha). These differences are in the range of the typical variations of volume estimations on base of inventory point or angle count methods. As conclusion it can be stated that the stratified remote sensing-based approach leads to significant cost savings at comparable outcomes in species composition and timber volume. On the one hand, due to the reduced sample size (77 instead of 348) and, on the other hand, due to the shorter time required to search for the sampling point in the terrain. The method seems appropriated for inventories without established regular sample grid like it is common in private and most of the communal owned forests.

Area B - Board 186

For more than twenty years, spaceborne Synthetic Aperture Radar (SAR) has been used to target environmental applications, as such user awareness of the potential of SAR is exponentially rising, partly due to the increased availability of data. As an Earth observation sensor, the application range is the widest possible and the undeniable advantage of being weather-independent, is catching the attention of stakeholders dealing with locations with frequent cloud cover. This talk discusses the use of SAR in the first six years of operations at the Satellite Applications Catapult, a UK-government backed organisation. The Catapult’s main mission is to support academic, governmental and business innovation by driving economic growth through the exploitation of downstream satellite data.

Four main sectors are continuously being explored: 1) sustainable living, including agriculture, extractive industries, and sustainable development 2) infrastructure and transport 3) blue economy and 4) knowledge exchange. The applications developed in these sectors are designed with a human centred approach, with solutions based on an in-depth understanding of user needs. Indeed, the use of SAR often comes as consequence of the user requirement stage, when the concept is finalised after a preliminary technical analysis and a requirement prioritisation. Following the design stage, SAR data procurement is generally carried out. Here, depending on the requirements, one or more SAR mission are chosen. Sentinel-1 is often preferred due to the free of charge policy but can be complemented or substituted with commercial missions such as Radarsat-2, TerraSAR-X and Cosmo Skymed if the application requires higher resolutions, a dedicated data acquisition plan, and/or a particular system configuration (e.g. polarisations and viewing geometry). To be noted, when developing research projects, commercial data can actually be more suitable and obtained free of charge through scientific proposals. The data processing and information extraction is then performed with personal, open-source or commercial software packages. All these aspects are conceived in the application design phase and generally involve several institutions carrying specific portions of the research and/or development.

National and international projects are the funding base for the applications. The project scale is varying depending on budget, and can target a single individual problem, e.g. a farmer who wants to improve the crop yield, or a group of nations, e.g. trying to build resilience to climate change. Considering the four large sectors mentioned above, successful stories based on the use of SAR are briefly described in the following.

Under the sustainable living and blue economy categories, successfully delivered and ongoing examples include ‘SAR Crop Index’, a project that used a Sentinel-1 data to derive a crop growth model over the UK for selected crops; ‘Satellites for Grassland Management’, which used Radarsat-2 data to support the precision management of grassland agriculture for the dairy farmer industry; ‘IPP Common Sensing’, a very large collaborative project to support three Commonwealth nations (Fiji, Solomon Islands, Vanuatu) particularly exposed to climate change and where Sentinel-1 data will be used to monitor coastal erosion, flooding and disaster monitoring through the exploitation of an innovative analysis ready data (ARD) framework for Sentinel-1; ‘Earth and Sea Observation System (EASOS)’, another very large project to support Malaysia in three main areas: flooding, marine pollution and deforestation. Here SAR data, specifically Sentinel-1 and Cosmo Skymed, is playing a key role; ‘Hephaestus’, focused on mining operations in Chile’s Coquimbo region and where TerraSAR-X data has been used to derive deformations over tailings dams; ‘Lithium’, a research project to detect lithium hot-spots in Cornwall, UK, where PALSAR and Sentinel-1 data have been the main input to check the terrain stability of the area for environmental baseline; ‘IPP Peru’, where Sentinel-1 and Cosmo Skymed are the input data to build an early warning system to predict tailings dam collapse and estimate landslide risk; and finally ‘Oceanmind’, recently spun out as an independent company, using advanced machine learning and several SAR data sources to detect and classify fishing vessels, with the objective to help stop Illegal, Unreported and Unregulated (IUU) fishing.

Under the infrastructure and transport category, ‘Asset Monitoring using Satellite and IoT technologies (AMSIT)’ used both TerraSAR-X and CosmoSkyMed data to monitor the condition of several key infrastructure assets, including bridges and buildings in London; ‘S1 for land reclamation assessment’ demonstrated the utility of Sentinel-1 to derive deformation over landfill areas subject to huge deformation patterns in Turkey and ‘Brigital’ is using Radarsat-2 data to estimate the structural health of bridges in Canada.

Under the knowledge exchange category, ‘FDL-Europe’ is a programme set up to accelerate the use of artificial intelligence coupled with Earth observation data for European researchers. In its first year, a team developed a disaster response solution based on the use of Sentinel-1 data to assess earthquake damage; ‘Cosmo-Skymed radar science and innovation (CORSAIR)’ is a programme to distribute Cosmo-Skymed products to researchers to develop applications; ‘SEDAS’ is a developed portal enabling end-users to search and download data from the constellation of Sentinel satellites. Finally, ‘AirSAR’ has been a mission to demonstrate the use of S-band radar data and it is a precursor to ‘NovaSAR’, successfully launched in September 2018.

Area B - Board 187

Since the launch of the European Space Agency (ESA) Sentinel-1 (S1) mission on April 2014, large collections of Synthetic Aperture Radar (SAR) data over the Earth’s surface have been acquired. Moreover, the features of the S1 sensor such as a short revisiting time, a high resolution in range, its wide coverage, its improved radiometric accuracy and the possibility to work with dual polarimetric data have enhanced the feasibility of a wide range of SAR applications. Surface displacement monitoring, land cover mapping, crop monitoring and flood detection are successful examples of these applications, and they are being increasingly employed by both public administrations and private companies to improve their processes (for instance in the mining or oil & gas sector) or to provide a rapid response in front of disaster risk management (earthquakes, floods, etc.).
This paper presents the benefits of using dense multi-temporal SAR Interferometry (InSAR) to measure ground motion from regional to local scale based on DARES TECHNOLOGY experience using S1 sensor. As known, classical InSAR technique allows exploiting phase differences, i.e., differential interferograms, between multi-temporal pairs of single look complex (SLC) images in order to retrieve the displacement [1,2]. Unfortunately, the usability of the interferometric phase is limited by geometrical and temporal decorrelation phenomena [3] and the presence of complex atmospheric artefacts. In order to overcome these limitations, different Persistent Scatterers Interferometry (PSI) techniques have been developed during the last two decades. They exploit large datasets of multi-temporal SAR images to generate multiple phase relationships, which allows the estimation of both the linear and the non-linear component of ground deformation and the atmospheric phase screen for each acquisition. In this context, DARES TECHNOLOGY makes use of the Coherent Pixels Technique (CPT). This PSI technique has been developed by the Remote Sensing Laboratory (RSLab) of the Universitat Politècnica de Catalunya (UPC) since 2002, and it has shown reliable estimations of the linear and non-linear components of ground displacement, along with the topographic error [4].
There are a wide range of examples in the literature showing the benefits of S1 in the characterizing tectonic activity, calibrating geo-mechanical models in the oil and gas sector or detecting risky areas to ensure safety in mining operations. Nonetheless, the monitoring of subsidence and uplift processes in urban scenarios affected by aquifers, tunnelling works, etc., or the characterization of landslides in mountainous areas are typically addressed by using X-band sensors (such as TerraSAR-X or the COSMO-SkyMed constellation) and rarely investigated using C-band sensors (such as S1). In this framework, the monitoring of local landslides and localized subsidence phenomena in urban environments with S1 will be reviewed, as well as its performance in measuring ground deformation when wide area tectonic activity is present. In addition, the different particularities in the PSI processing for each of these scenarios will be thoroughly discussed.

REFERENCES

[1] Massonnet, D.; Feigl, K.L. Radar interferometry and its application to changes in the Earth’ssurface. Rev. Geophys. 1998, 36, 441–500.
[2] Bürgmann, R.; Rosen, P.A.; Fielding, E.J. Synthetic Aperture Radar interferometry to measure Earth’s surface topography and its deformation. Ann. Rev. Earth Planet. Sci. 2000, 28, 169–209.
[3] Hanssen, R.F. Radar Interferometry: Data Interpretation and Error Analysis; Kluwer Academic Publishers: Dordrecht, the Netherlands, 2001.
[4] Mora, O.; Mallorqui, J.J.; Duro, J. Generation of deformation maps at low resolution using differential interferometric SAR data. In Proceedings of 2002 IEEE International Geoscience and Remote Sensing Symposium, IGARSS ’02, Toronto, ON, Canada, 24–28 June 2002.

Area B - Board 188

Satellite SAR interferometry is a well assessed technique to detect and monitor slow terrain movements with millimeter accuracy, covering long-term period with frequent updates, and large areas with tens of thousands of measurements per square kilometer. This is particularly suitable for oil & gas reservoir monitoring.
For oil & gas reservoir monitoring, it is crucial to understand the response of the surface, in term of subsidence and uplift, to the injection and extraction activities. Correlating the surface response to such activities is essential to evaluate possible impacts that oil & gas storage/extraction operations can have on the local infrastructures and the surrounding environment. The characterization of these phenomena requires high density of measurements in the time and space domain that are not affordable with standard monitoring techniques as precise optical leveling, GPS, etc.
In this work we present a method to analyze the correlation between fluid extraction and injection activities and ground deformations measured by satellite SAR. In order to perform this correlation analysis, the terrain displacement measurements in correspondence of some oil reservoir and gas storage sites (located in the Italian territory) have been obtained from RADARSAT-1/2, Sentinel-1, COSMO-SkyMed and TerraSAR-X data by means of the Persistent Scatterer Pair (PSP) technique [1]. Thanks to the availability of a huge archive of SAR data, it has been possible to collect a very long series of ground displacement measurements, starting from 2003 with medium resolution SAR data (C-band), and from 2009 with high-resolution SAR data (X-band). Starting from 2014 with Sentinel-1 (C-band) constellation is possible to monitor regularly all the areas of interest with a new acquisition every six days in both observation geometries.
The terrain deformation measurements have been correlated to the temporal evolution curves of the gas amount injected to or extracted from the analyzed storage sites (expressed in m3). The correlation analysis showed a good agreement between SAR measurements and storage activities and it allowed estimating the delay of the response of the surface, the amplification coefficient (which straightforward relates the cubic meters of gas injected or extracted to the millimeters of measured terrain deformation) and above all the footprint of impact of the reservoir activities.
We obtained very interesting results, but there are still questions that must be more deeply analyzed in order to improve the effectiveness of the proposed method. One for all the fact that, typically, this kind of storage activities has a seasonal component; in fact, gas reservoirs increase during the summer periods and, on the contrary, decrease during the winter periods. Since the aim of the presented analysis is to find the correlation between the reservoir activities and the terrain deformations, it is important to remove from the SAR measurements the seasonal signal component due to the thermal dilatation phenomena that can be highly time correlated with these human activities and therefore falsify the correlation estimation. This signal filtering is a quite complex task and requires a specific spatial-time filter that exploits the fact that, generally, these two periodical deformations (the one related to the human activities and the one related to the thermal dilatation phenomena) have different spatial scales.

References

[1] Costantini, M., Falco, S., Malvarosa, F., Minati, F., Trillo, F., Vecchioli, F., “Persistent Scatterer Pair Interferometry: Approach and Application to COSMO-SkyMed SAR Data,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 7, no. 7, pp 2869 – 2879, July 2014.

Area B - Board 189