4th GEOSS Science and Technology Stakeholder Workshop
March 24-26, 2015, Norfolk, VA, USA

Scope | Deadlines | Venue | Logistics | Committees | Participation | Output | Abstracts | Registration | Accommodation | Sessions | Schedule | Program

Workshop infos

Co-Located Events

Relevant Links and Documents


Automatization of information extraction to build up a crowd-sourced reference database for vegetation changes

Jonas Eberle, Christian Hüttich, Christiane Schmullius; Friedrich-Schiller-University Jena, Institute for Geography, Department for Earth Observation, Jena, Germany

Scientists can benefit from the wide range of data quantity if their algorithms are made available to the public in an easy-to-use manner. Automated data access in combination with the follow-up execution of algorithms can help to test algorithms in different regions around the world and lead to new information based on the knowledge of local users. In the example of vegetation change analysis based on Earth Observation time-series data, we can provide lots of data for the validation of changes detected by scientific algorithms, e.g., true/false color images, fire data, weather data. Based on these input users can validate the algorithm in their study areas. Furthermore, they can build up a database with change areas that can be used as reference databases on other analysis tools (e.g., change classifications).

Based on the bfast (Breaks For Additive Season and Trend) algorithm we can detect vegetation changes in time-series data. For the validation of a detected “break” we will search automatically for other datasets at the detected date of break and provide these data in an easy-to-use web portal. So users can execute the algorithm for change detection and validate the detected changes. A crowd-sourced reference database can be build up on areas where change occurred and this change was validated by users.

Such a crowd-sourced initiative can help scientists to better understand algorithms for information extraction. The authors of an algorithm can benefit from the input of users that are testing the algorithm. The Web 2.0 leads us to a new way of how algorithms can be tested and how we can build up reference databases with areas around the world. Thus, Earth Observation time-series data are better useable and lead to new knowledge to further improve algorithms and validated reference information.

In this presentation the author will describe the developments made for automated data access in combination with automated data analysis based on Earth Observation vegetation time-series data with no need to process any data by the users.