New Era for Open Science (continued)

The educational web-based remote sensing software BLIF (Blickpunkt Fernerkundung, Point of View Remote Sensing) was initially developed at the Research Group for Earth Observation (^rgeo), Department for Geography at Heidelberg University of Education. The software is continuously evaluated and enhanced ever since. Currently, version 1.1 is developed further in the framework of the project “Space4Geography”, funded by the German Aerospace Center (DLR), Space Administration. BLIF 2.0 will comprise proven tools and characteristics of earlier versions but also new features that take into account current developments in Earth Observation as well as up-to-date e-learning and education principles.

BLIF allows students to independently work with original satellite images featuring a didactically prepared tool set of basic image analysis functions. Preprocessing steps like histogram stretching and image enhancement are included in every workflow. Further possible analysis steps comprise color composites, vegetation indices, supervised and unsupervised classification as well as change detection. Considering prior knowledge, the displayed analysis steps and the degree of guidance vary with the chosen individual level ranging from beginners to experts. The software is embedded in the educational concept of problem- and action-oriented learning. While a variety of satellite images already is available via an integrated webserver, user can also upload own datasets for analyses. These datasets then are available for all users. To enable an easy implementation in schools and other educational institutions, BLIF is available free of charge requiring only a user registration.

The new version of BLIF features supports additional satellite sensors. While the original software was limited to Landsat 5-8 data, BLIF 2.0 supports high resolution imagery of the satellites RapidEye and TerraSAR-X. Hence, RADAR remote sensing methods such as speckle filters and threshold classification algorithms are added to the set of available tools. On top of this, new features comprise a list of new band indices, e.g. for detecting water, ice or snow, and easier navigation in the workflow. The entire user interface is updated to meet the requirements of an up-to-date and intuitive web design. Moreover, BLIF 2.0 is now accessible on mobile devices like tablet computers. From a technical point of view, the processing performance has been improved significantly to manage multiple user accesses with a limited bandwidth which meets the situation in schools.

In October 2013 the European Council stressed the key role of technologies such as Big Data and Cloud Computing as important enablers for productivity and better services. The European Commission and key international organizations such as the European Space Agency are currently implementing several initiatives for the development of Big Data and Cloud Computing based solutions.

The key challenge in the Space and Security domain is to improve the capacity to access and analyse the increasing amount of data produced with high velocity by a variety of sources to timely provide decision-makers with clear and useful information. To face this challenge taking advantage from Big Data and Cloud Computing, the European Union Satellite Centre (SatCen) is currently participating in two Horizon 2020 projects: BigDataEurope and EVER-EST.

BigDataEurope (Integrating Big Data, Software & Communities for Addressing Europe’s Societal Challenges) is collecting requirements by stakeholders tackling the seven Horizon 2020 Societal Challenges (Health, Food & Agriculture, Energy, Transport, Climate, Social Sciences and Secure Societies) in order to design, integrate and evaluate a Big Data Aggregator platform. The BigDataEurope platform, comprising key open-source Big Data technologies and taking advantage of the latest European RTD developments, is being validated through the development of an appropriate pilot for each Societal Challenge. SatCen is building a Secure Societies community, eliciting Big Data requirements for Secure Societies and implementing a pilot in the Space and Security domain.

EVER-EST (European Virtual Environment for Research - Earth Science Themes) is creating a state-of-the-art Virtual Research Environment (VRE) in order to enhance collaborative research and knowledge exchange in the Earth Sciences. EVER-EST, whose data processing infrastructure is based on a Cloud Computing approach, is building on existing e-infrastructures and developing additional functionalities (including Research Object technologies) to meet the needs of Earth Science researchers, represented in the project by four Virtual Research Communities (Sea Monitoring, Land Monitoring, Natural Hazards and Supersites). SatCen is collecting the requirements of the Land Monitoring Community (focusing on the Space and Security domain) in order to define and implement a relevant pilot to validate the EVER-EST VRE.

The SatCen pilot implemented in the BigDataEurope platform is focusing on the integration of data coming from Remote Sensing (e.g. Sentinel-1 data) and Social Sensing (e.g. Twitter and Reuters), also making available tools for satellite data analysis in order to add value to the current data exploitation practices. As output of the pilot, an alert is sent to the user when a change (to be possibly verified with additional information from Social Sensing) is identified between two or more images on specific areas of interest. The SatCen pilot implemented in the EVER-EST VRE is complementing the work performed in BigDataEurope by adding other satellite sources (e.g. Sentinel-2) as well as tools for analyzing qualitatively the detected changes and investigating in their nature.

Starting from the implementation of the pilots, SatCen is also looking for possible synergies between the BigDataEurope and EVER-EST solutions to improve its capabilities and in general those of the Space and Security community.

During the concept assessment and feasibility studies for the ESA Earth Observation activities, the mission performance up to the final data products needs to be predicted by means of end-to-end (E2E) performance simulators; later on, these simulators become a coherent test bed for Ground Processor Prototypes (GPP) and support not only the verification of space segment performance and associated instrument and data processing chain sensitivity analysis up to level 1 but also the mission scientific validation including level 2.
A mission E2E performance simulator is able to reproduce all significant processes, design and steps that impact the mission performance as well as generating simulated data products in the mission’s native format. The main purpose is to assess the final performances of an observation system before the system operation by checking the consistency of platform and payload specifications with respect to the system and mission requirements. The core of an E2E performance simulator is composed by a set of software modules simulating the space segment, its data output and the subsequent ground retrieval (Level 1 and Level 2). The execution of these software modules needs to be orchestrated including in particular invocation and provision of input data. Commonalities in the structure of these E2E simulators highlighted the need for a common framework, which was specified with a modular approach that could be used for these missions and the ones yet to come. These software modules typically consist (but not only) of scene generators, instrument or platform simulators and processors (Level 1 and Level 2) and analysis and evaluation tools.
To benefit from commonality between missions and simplify the development effort, a set of standard conventions and requirements has been defined, which the modules have to adhere to, allowing the use of a common orchestrating framework. These general conventions have been built based on the experience in E2E development for a number of different Earth Observation missions as well as from the experience gathered in the development, evolution and usage of the current standard ESA E2E orchestrating framework openSF.
These conventions have been formalised into a common and generic interface specification for software modules [1], which in addition to provide harmonisation allows the direct integration of modules into ESA’s E2E performance simulator orchestrating infrastructure openSF. This interface has been designed to be simple and lightweight, addresses invocation, configuration/parameter control and logging and can be easily added to existing software modules.
openSF is a software framework and GUI that supports standardized E2E simulation capabilities allowing the assessment of the science and engineering goals with respect to the mission requirements. Scientific models and product exploitation tools can be plugged in the orchestrating environment with ease using a well-defined integration process. Besides its advanced orchestration environment openSF also enables automatic parameters span (for sensitivity testing), automatic time/scenario span and structured collection/organization of results. It is a multiplatform framework with free maintenance and regularly updated made available by ESA to all parties involved in supporting ESA missions [2].
The openSF project is “descendent” of ESA’s project ECSIM that provided an E2E simulator for the EarthCARE mission. openSF contains the abstract and re-usable simulation functionalities that ECSIM simulator had in common with other E2E simulators. Since the original openSF development, it’s been used by other ESA projects (e.g. GERSI, AIPC, SEPSO, S3-OGPP/OSPS). openSF is now in use by the different actors (Sentinels) to support E2E simulation activities and planned for use in the upcoming Biomass and Flex ESA Earth Observation missions.

REFERENCES

[1] ESA and DME openSF Team, “ESA generic E2E simulator Interface Control Document”, PE-ID-ESA-GS-464, 2016-04-13.
[2] openSF support website

A tile-based search engine for satellite imagery is presented. This engine shall be the core segment of a future Content Based Image Retrieval (CBIR) prototype service, for finding similar patches over large amounts of data. Both optic and radar scenes can be used. Each tile will be characterized based on specific descriptors adapted to EO image particularities. Positive and negative examples will point a specific class and then semantic annotation will be performed by means of machine learning algorithms. Several visualization approaches of the search results will be considered, aiming to highlight the patches or the scenes most similar to a specific query area.

The results presented in this poster derive from the ongoing ESA GSTP project:  Open Source Image Retrieval - Integration of Developed tools (OSIRIDE).