Pilot for Secure societies – protecting freedom and security of Europe and its citizens.
The aim of the Secure Societies pilot is to automatically identify areas with changes on land cover/use and to interpret those changes by integrating information from heterogeneous sources. This is a key issue for the Secure Societies aims, where useful information can be derived not only from satellite data but also from data coming from a number of different sources (e.g. media and other geospatial data). Moreover, the continuous increase in archives’ size from EO and other sources requires new methodologies and tools for information mining and management. An automatic chain reduces the human assistance in the data analysis and can be designed to retrieve information in Near Real Time (NRT).
The pilot was conceived by the European Union Satellite Centre, together with the University of Athens and the NCSR “Demokritos”, and considers two different workflows of data (Fig. 1), where data from remote sensing (satellites images) and from social sensing (news and social media) sources are used. The Change Detection workflow ingests satellite images to detect areas with changes on land cover or land use by using change detection techniques; the identified Area of Interest (AoI) is then associated with social media and news agencies’ items and presented to the user for cross-validation. The Event Detection is triggered by social media and news agencies’ information, where trending topics with geospatial connotation constitute an event that is localised in time and space. Relevant keywords on selected AoIs could activate the Change Detection, where corresponding satellite images are acquired and processed in order to check for changes in land cover or land use.
There are two primary sources of data.
Remote sensing data: the pilot uses data from the Sentinel satellites operated by ESA in the framework of the Copernicus programme funded and managed by the European Commission. The free, full and open data policy adopted for the Copernicus programme foresees access available to all users for the Sentinel data products. The initial satellite data ingested in the pilot comes from Sentinel-1 that has a revisiting time of 6 days above the same areas at mid-latitudes.
Social sensing data comes from two complementary sources of information: social media, represented by Twitter, and news agencies, represented by Reuters. Recently, Twitter has emerged as a major platform for on-time detection of events, both in research and in industry. The pilot uses the free Twitter Public Streams API that provides a random sample of its content. The content provided by news agencies through public RSS feeds is also free of charge. Within the pilot, the content from Twitter is used for the extraction of metadata, while the user is pointed to the original source for more information to abide by intellectual property rights requirements.
The architecture of the pilot was designed to accommodate the Change Detection workflow and the Event Detection workflow. It consists of the three main blocks depicted in Fig. 2: the user interface, which is a modified version of the web-application Sextant; the storage block, which involves the four components represented as cylinders in the middle of the diagram; and the core block, which consists of the remaining six components that support both the Change Detection in land cover / land use from satellite images and the Event Detection from social media and news agencies.
The first block runs on the client-side as a Web application that can be deployed on a variety of platforms. The other two blocks of components run on the server-side (i.e. on the BigDataEurope infrastructure) in order to offer scalability and high efficiency.
In both workflows, the user is able to select an Area Of Interest, drawing a polygon with Sextant, and a time interval for its analysis (selecting the starting and ending dates). In the Change Detection workflow the selected dates correspond to Sentinel 1 images, automatically downloaded and processed in a parallelized way. The output at the end of the process (areas with changes/no changes) will be displayed in the Sextant interface. In the Event Detection workflow the user can add to the AoI and to the time interval some specific keywords that will narrow the search of specific topics. Corresponding social media and news agencies items are clustered into events and mapped on Sextant based on their specific location.
Current efforts are dedicated to improve the scalability of the Change Detection process (following a tile approach, i.e. an approach based only on the specific AoI instead of the whole image to speed up the process) and to optimize the change detection algorithm as well as the results visualization. For the Event Detection, the multi-core approach is currently being adapted to run on Spark. In the future, other data (e.g. further news and HR satellite images) will be ingested, cybersecurity mechanisms (e.g. authentication / authorization) will be adopted and the user interface will be updated with new functionalities according to the user feedback. This will allow to enhance the information coming out from the analysis, to assure the security of the platform and to improve whole user experience within the pilot functionalities.
For more information
For a detailed view of the pilot, see Albani, S., Lazzarini, M., Koubarakis, M., Taniskidou, E. K., Papadakis, G., Karkaletsis, V. & Giannakopoulos, G. (2016). Integrating remote and social sensing data for a scenario on secure societies in a Big Data platform. In Proceedings of Living Planet Symposium 2016, Prague (CZ), 9–13 May, ESA SP-740, 2016.