Virtual Research Environment (continued)

ESA Swarm data is available for anyone to use via the virtual research platform “VirES for Swarm” (http://vires.services). A highly interactive data manipulation and retrieval interface is provided for the magnetic products of the Swarm constellation mission, continuously synchronized with the official product archive. 

Graphical widgets (globe view, analytics view, time-slider, dashboard, download manager) make it possible to select and plot the individual magnetic elements (scalars, vectors) contained in the Swarm Level 1b products. It is also possible to visualize selected parameters constituting geomagnetic models and to display residuals between Swarm product data points and such models. The geomagnetic models are either pre-loaded (IGRF, SIFM, CHAOS-5) or can be uploaded by the user as custom harmonic coefficients. Also ancillary data characterizing the space weather context can be shown in various displays. A multi-parameter filtering is based on product data and auxilliary data values shown along histogram axes. The saving to local disk of the Swarm Level 1b products corresponding to the specified criteria is possible as as well, whereby different download formats are supported. 

Leading geomagnetists at DTU Space, GFZ Potsdam and mission experts at ESA/ESRIN and ESA/ESTEC have been members of the project team involved in the detailed specification of VirES and its accurate validation at various stages of evolution. The servce is now being routinely used by scientists for a deeper exploitation of Swarm measurements and results.

The service - required by the Earth Observation Directorate of ESA - has been designed and is operated under an ESA service contract since May 2016 by the company EOX. The technical framework entirely consists of Free and Open Source Software (MIT style licence). The client framework includes Cesium JavaScript code and convinces by its simplicity and intuitiveness of use; the server side is implemented using EOxServer technology which has been tuned for very very fast magnetic model evaluation. Users are invited to get involved into branching out software code and to contribute own enhancements to it. The code is maintained and related issues are publicly tracked on GitHub. The operational site is hosted on the servers of EOX. It features continuous testing and deployment set-up; multiple-server operations, failover, and load balancing configuration; as well as OAuth authentication with major social providers.

Arrangements have been granted by ESA and EOX to maintain VirES for Swarm in operations for extended mission periods. This includes a future release plan of new and additional service features and their introduction by incremental software evolutions and service deployments.

In 2015, Meteorological and Environmental Earth Observation (MEEO) established the largest Data Cube implemented within the EarthServer project (http://earthserver.eu). Based on strong heritage of the first phase of the project (September 2011 - August 2014) - a volume of 100+ TB of Earth Observation (EO) products - the EO Data Service (http://eodataservice.org) is now entered in the H2020 phase to facilitate Big Data exploitation, powered by the NASA Web World Wind (http://webworldwind.org).

EO Data Service supports the full data lifecycle: it provides a set of tools to collect EO products from heterogeneous data sources, to store data and build the data cube, and analyse gridded / no-gridded data for decades of EO and related data from multiple space-borne sensors and other acquisition systems, via OGC access processing services and via NASA Web World Wind client.

Among others, the EO Data Service enables advanced data access and retrieval capabilities to the currently-available European coverages of Landsat-8 (full pan-European coverage) and the Australian Landsat archive maintained by Geoscience Australia as a real virtual research environment.

The current study is co-financed by the European Union’s Horizon 2020 research and innovation programme under the EarthServer-2 project (Grant Agreement No. 654367) and the European Space Agency under the MaaS project (ESRIN Contract No. 4000114186/15/I-LG).

Recent decisions by major space agencies have been making unprecedented amounts of imagery available for research and operations. This brings about a unique opportunity to measure the global and local changes in our environment and assess the human impacts on land and the oceans. However, current scientific methods for extracting information for Earth observation data lag far behind our capacity to build sophisticated satellites. These satellites produce massive amounts of data, but only a fraction of that data is effectively used for scientific research and operational applications.  Our work addresses a key scientific problem: How can we substantially improve the extraction and validation of land use and land cover change information from big Earth Observation data sets in an open and reproducible way?

In response to this challenge, we are building and deploying a new type of knowledge platform for organization, access, processing and analysis of big Earth observation data. The platform uses a scientific database based on the open source SciDB innovative array database management system, capable of managing large remote sensing data sets. We are developing an innovative set of spatiotemporal image analysis methods, mostly based in analysis of satellite image time series. These analytical methods use the open source R statistical language. 

The results we have already obtained and will be presented at the conference include:

(a)   A Web Time Series Service (WTSS) for serving time series extracted from remote sensing imagery on the internet.

(b)  A Spatial Data Infrastructure (SDI) architecture for big EO data sets.

(c) A time-weighted version of the dynamic time warping method for land use and land cover classification using remote sensing image time series. This method has achieved 90% accuracy for land use change information in large areas in the Brazilian Amazonia.

(d) A new method for near-rela time change detection in tropical forests, based on Kalman filtering.

(e) A set of collaborative validation tools for land use classification using remote sensing time series.

Using array databases such as SciDB and analytical tools in the R statistical language, the resulting infrastructure is a powerful and open infrastructure for big Earth Observation data analysis.