IEEE Geoscience and Remote Sensing Magazine - September 2016 - 26

tHE LiFE cYcLE oF BiG, LiNKED, aND oPEN
Eo Data aND tHE tELEios/LEo soFtWarE stacK
Developing a methodology and related software tools that
support the complete life cycle of linked open EO data has not
been tackled by any research project in the past, although there
is plenty of such work for linked data, e.g., by project LOD2
and others [4], [5]. Capturing the life cycle of open EO data
and the associated entities, roles, and processes of public bodies that make this data available is the first step in achieving
LEO's main objective of bringing the linked data paradigm to
EO data centers and re-engineering the life cycle of open EO
data based on this paradigm.
The life of EO data starts with its generation in the ground
segment of a satellite mission. The management of this socalled payload data is an important activity of the ground
segments of satellite missions. Figure 2 gives a high-level
view of the life cycle of linked EO data as we envision it in
TELEIOS and LEO. Each phase of the life cycle and its associated software tools is discussed in more detail below.
INgESTION, PROCESSINg, CATALOgINg,
AND ARCHIvINg
Raw data, often from multiple satellite missions, is ingested, processed, cataloged, and archived. The processing results in the creation of various standard products (level 1, 2,
and so forth in EO jargon; raw data is level 0) together with
extensive metadata describing them.
TELEIOS has developed two technologies that are important for the phases of ingestion and processing: 1) the SciQL

data model and query language [6] and 2) the data vaults
technology [7]. SciQL is an SQL-based query language for
scientific applications with arrays as first class citizens [6].
It allows the stating of complex satellite image-processing
functions as declarative SciQL queries, and it substantially
eases the development of processing chains run by EO data
centers today. The data vault is a mechanism that provides
a true symbiosis between a database management system
(DBMS) and existing (remote) file-based repositories such
as the ones used in EO applications [7]. The data vault keeps
the data in its original format and place, while at the same
time enabling transparent data and metadata access and
analysis using the SciQL. SciQL and the data vault mechanism are implemented in the well-known column store
MonetDB (http://www.monetdb.org/).
CONTENT ExTRACTION, kNOwLEDgE DISCOvERy AND
DATA mININg, AND SEmANTIC ANNOTATION
In the DLR knowledge discovery and data mining (KDD)
framework developed in TELEIOS [8], traditional raw data
processing has been augmented with content extraction
methods that deal with the specificities of satellite images
and derive image descriptors (e.g., texture features and spectral characteristics of the image). Knowledge discovery techniques combine image descriptors, image metadata, and
auxiliary data (e.g., GIS data) to determine concepts from
a domain ontology (e.g., forest, lake, fire, or burned area)
that characterize the content of an image. Hierarchies of the
domain concepts are formalized using ontologies encoded

Ingestion

Cataloging

Archiving
Processing
Search/Browse/
Expolre/Visualize

Content Extraction

Interlinking

KDD

Publishing

Semantic
Annotation

Transformation
into RDF

Quality
Assurance
Storage/Querying

FiGUrE 2. The life cycle of linked open EO data.

ieee Geoscience and remote sensinG maGazine

september 2016

http://www.monetdb.org/

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - September 2016