Dagstuhl Seminar 17091
Computer Science Meets Ecology
( Feb 26 – Mar 03, 2017 )
(Click in the middle of the image to enlarge)
Permalink
Please use the following short url to reference this page:
https://www.dagstuhl.de/17091
Organizers
- Benjamin Adams (University of Auckland, NZ)
- Gustau Camps-Valls (University of Valencia, ES)
- Thomas Hickler (Senckenberg Research Centre, DE)
- Birgitta König-Ries (Universität Jena, DE)
Contact
- Susanne Bach-Bernhard (for administrative matters)
Documents
Schedule
In his pioneering work, Jim Gray identified the 4th scientific paradigm, arguing that modern science needs computer supported research. Recent developments in many scientific disciplines suggest this new paradigm to be powerful: Huge amounts of heterogeneous, unstructured and multisource data of different modalities can now be collected routinely, sometimes in a fully automatic manner and be processed to extract knowledge and make inferences.
A discipline that shows great potential but also the challenges of this 4th scientific paradigm is Ecology. Ecology is the study of the interactions amongst organisms and with their physical environment. For a long time, ecological analyses have been primarily realized locally both with respect to the geographical and phenomenological areas of investigation. Today, scientists are increasingly interested in quantifying ecological relations at larger scales or globally and can consider multiple dimensions of interactions between atmospheric, oceanic, and terrestrial processes. Due to the possibilities to record, store and process data ubiquitously, the increase in data resolution and quality as well as the international efforts to document the global distribution of biodiversity, new opportunities arise. These data will enable us to answer questions that are of fundamental importance for the future of our planet.
The aim of the Dagstuhl Seminar is to establish links between (geo-)ecologists, ecoinformaticians and computer scientists in order to leverage computer science expertise for ecology and to identify avenues of future research in computer science of particular importance to ecology.
Based on three concrete use cases regarding automated long-term monitoring of biodiversity, Global Change and Macroecology, and modelling ecosystem and Earth system processes, which we have identified together with researchers working in the field of Ecology, we want to explore particular challenges with respect to
- obtaining and preserving data (Use case: automated long-term monitoring of biodiversity)
- pattern-recognition in highly dimensional and geo-tagged data sets (Use case: Global Change Ecology), and
- model development and model-data-confrontation (Use case: Modelling ecosystem and Earth system processes)
More detailed information about the use cases and relevant areas of computer science can be found here.
The seminar has two main objectives:
- Joint authoring of a book on the state of the art and challenges in the intersection of computer science and ecology. This book shall be based on the results of the working groups. Based on the example scenarios it will introduce three important classes of problems in Ecology. For these, it will provide an introduction to available tools, and will outline challenges for future research. Such a book can serve as a handbook for ecologists wanting to leverage computer science in their research but also as a roadmap for future research activities.
- Define project ideas for cooperation between Computer Scientists and Ecologists and identify suitable funding schemes.
Together, these two objectives will serve to intensify cooperation between the disciplines involved.
Creative Commons BY 3.0 DE
Benjamin Adams, Gustau Camps-Valls, Thomas Hickler, and Birgitta König-Ries
Ecology is a discipline that shows clearly the potential but also the challenges of computer supported research described as the 4th scientific paradigm by Jim Gray. It is increasingly data driven, yet suffers from hurdles in data collection, quality assurance, provenance, integration, and analysis.
We believe that ecology could profit from modern computer science methods to overcome these hurdles. However, usually, scientists in ecology are not completely aware of current trends and new techniques in computer science that can support their daily work. Such support could consist in the management, integration, and (semi-)automatic analysis of resources, like experimental data, images, measurements, in the generation of useful metadata, cloud computing, distributed processing, etc. Ecoinformatics is regarded as an important supporting discipline by many ecologists. However, up to now, very few computer scientists are involved in this discipline; mostly ecoinformatics (or biodiversity informatics) is done by people with a strong background in e.g. ecology and a long (mostly self-taught) experience in data management. It lacks a strong connection to cutting-edge computer science research in order to profit from the results of this area. On the other hand, computer scientists know too little about the domain to be able to offer solutions to relevant problems and to identify potential research avenues.
Motivated by our belief that a stronger bond between the disciplines that goes beyond viewing computer science as a “service provider” is of vital importance, we proposed this Dagstuhl seminar. The aim of the Dagstuhl seminar was to establish such links between (geo-)ecologists, ecoinformaticians and computer scientists.
The seminar: perspective and self-evaluation
Before the seminar. It turned out that it was not an easy task to motivate non-computer scientists to attend the seminar. For many, travel costs were a hurdle ultimately preventing attendance. This resulted in an unusually large number of declined invitations (often accompanied by “I would love to attend, but…” emails.
Despite these initial problems, we believe that the aim to start building links among the communities was reached at the seminar: We had fruitful discussions in numerous working groups resulting in some very concrete plans for future work.
Organization of the seminar. A total of 27 attendees gathered at the seminar. The wide variety of expertise and backgrounds constituted an initial challenge for the organization. The agenda considered a first round of presentations of the individuals and their research groups with a clear outline and items to treat (personal background, Research Areas/Interests, prospective links to „Computer Science meets Ecology“ seminar). After this, the main topics of interest for a wide audience were designed: essentially, three breakout groups were set up in the very first day of the meeting. Over the course of the seminar, these groups were adjusted, split up, or merged, several times. This resulted in quite a number of topics being touched upon with concrete results ranging from a working example for the application of a new method to a modeling problem to concrete plans for publications, a proposal and follow-up activities. Reports on these groups were given in the plenary session, and can be found in this report.
Broad results of the seminar. Results from the seminar can be categorized in three types: (i) collaborative and networking, as new joint works on specific topics came out of the meeting; (ii) knowledge transfer between fields, as computer scientists learned about the main problems in ecology involving data, while ecologists became aware of what kind of problems data scientists can solve nowadays; and (iii) educational, as several young PhD students and postdocs attended and participated in high level discussions.
Conclusions. The seminar brought together top scientists in the fields of ecology and computer science. The group of individuals was largely interdisciplinary, with a wide range of interests and expertises in each community too: from botany and animal science, to machine learning and computer vision. The seminar was organized in two main types of sessions: plenary and working group sessions to better focus on particular topics. Interesting developments and discussions took place in both, and a high level of cross-fertilization and future collaborations was initiated. On top of this, there was a broad consensus among the participants that the seminar should be the start of a series of yearly or bi-yearly meetings. We hope that the success of this first seminar will encourage broader participation in follow-up activities.
Gustau Camps-Valls, Joachim Denzler, Thomas Hickler, Birgitta König-Ries, and Markus Reichstein
- Martin Bücker (Universität Jena, DE) [dblp]
- Tilo Burghardt (University of Bristol, GB) [dblp]
- Gustau Camps-Valls (University of Valencia, ES) [dblp]
- Yun-Heh Jessica Chen-Burger (Heriot-Watt University - Edinburgh, GB) [dblp]
- Joachim Denzler (Universität Jena, DE) [dblp]
- Matthew Evans (University of Hong Kong, HK) [dblp]
- Florian Hartig (Universität Regensburg, DE) [dblp]
- Thomas Hickler (Senckenberg Research Centre, DE) [dblp]
- Donald Hobern (GBIF - Copenhagen, DK) [dblp]
- Forrest Hoffman (Oak Ridge National Laboratory, US) [dblp]
- Kazuhito Ichii (JAMSTEC - Yokohama, JP) [dblp]
- Martin Jung (MPI für Biogeochemistry - Jena, DE) [dblp]
- Birgitta König-Ries (Universität Jena, DE) [dblp]
- Ivaylo Kostadinov (Jacobs University Bremen, DE) [dblp]
- Bertram Ludäscher (University of Illinois at Urbana-Champaign, US) [dblp]
- Miguel Mahecha (MPI für Biogeochemistry - Jena, DE) [dblp]
- Laetitia Navarro (iDiv - Leipzig, DE)
- Shawn Newsam (University of California - Merced, US) [dblp]
- Frank Pennekamp (Universität Zürich, CH)
- Natalia Petrovskaya (University of Birmingham, GB) [dblp]
- Markus Reichstein (MPI für Biogeochemistry - Jena, DE) [dblp]
- Andrew Richardson (Harvard University - Cambridge, US) [dblp]
- Ribana Roscher (Universität Bonn, DE) [dblp]
- Brody Sandel (Santa Clara University, US) [dblp]
- Bernhard Seeger (Universität Marburg, DE) [dblp]
- Johann Wolfgang Wägele (ZFMK - Bonn, DE) [dblp]
- Jakob Zscheischler (ETH Zürich, CH) [dblp]
Classification
- computer graphics / computer vision
- data bases / information retrieval
- modelling / simulation
Keywords
- Biodiversity
- Earth System
- Earch Observation
- Remote Sensing
- Ecology
- Big Data
- Modeling
- Data Integration
- Semantics
- Society
- Citizen Science
- Mobile Computing
