De la conception d'un système d'observation à large échelle au déploiement et à l'exploitation de son système d'information : application à l'observation des habitats coralligènes et à la colonisation de récifs artificiels (ARMS)

Abstract : In the field of marine environment, observation protocols developed in many settings produce a large volume of heterogeneous data, which are difficult to aggregate since they focus on the use that is often specific to a profession. However, access and sharing of data on a large scale is essential to better understand the challenges behind protecting biodiversity and marine resources, and to anticipate their irreversible deterioration. To address these challenges, there is a need to strengthen the effectiveness of knowledge acquisition systems, and to organise data access for all potential users. Improving the consistency between observational systems and informational systems is the main objective of this work. Effective implementation of large-scale observational systems requires an inventory of knowledge, available skills and locks to be lifted. Monitoring networks for environmental protection become multi-user and must produce sufficient reliable descriptors to develop an effective reporting. The present work proposes to develop i) methods, protocols and recommendations to build and / or support the establishment of networks for monitoring biodiversity (from gene to species to habitats), operational and sustainable at all scales, (ii) innovative usage of data, their conservation and the organisation of their permanent access. Two case studies were chosen: coralligenous habitats at the Mediterranean scale and the colonisation of artificial reefs (ARMS) in various regional seas, focusing on: the construction of sustainable monitoring and observational networks; efficient sharing and inter-operability of long-term knowledge along with data analysis methods exploiting advances in the field of Big Data, and the analysis of heterogeneous data in the form of graphs. The testing of the proposed protocols developed during this thesis shows that a large-scale experiment must be described very explicitly in standardised terms beyond the disciplinary field of marine ecology and to be based on the simplest possible measurement methods. These tests also demonstrated the importance of operator’s training and iterative intercalibration. The work on the architecture of informational systems highlights the inevitability of a modular management system, business-oriented and decentralised. A mechanism for coupling data of different origins accessible in the form of parameterisable and open flows is proposed (field observations and data describing the contexts). It is based on the requalification of heterogeneous descriptive factors built on a collaborative arbitration between specialists. Thanks to the development of a prototype, a method of analysis and data mining based on graph theory has been experimented. Diverse types of possible demonstrations based on the data were constructed through the organisation of multidisciplinary curation and graphical data visualszation workshops. In conclusion, we recommend that the information produced is iteratively controlled in real time and that the curation processes of the data is implemented at the same time as the design of the observational processes. Standard definitions and access to context data require collaborative, interdisciplinary, iterative, long-term, international work. They must be considered for all their possible uses. In perspective, the use of the computation grid to carry out graph mining in a parallel way was prepared during the workshops, with the challenge of scaling up distributed and very heterogeneous data and forming graphs with more than one billion nodes and several hundred billions links.
Complete list of metadatas

Cited literature [41 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/tel-01839376
Contributor : Romain David <>
Submitted on : Tuesday, August 21, 2018 - 4:08:43 PM
Last modification on : Friday, March 22, 2019 - 11:47:59 AM
Long-term archiving on : Thursday, November 22, 2018 - 12:18:45 PM

Identifiers

  • HAL Id : tel-01839376, version 1

Citation

Romain David. De la conception d'un système d'observation à large échelle au déploiement et à l'exploitation de son système d'information : application à l'observation des habitats coralligènes et à la colonisation de récifs artificiels (ARMS). Biodiversité et Ecologie. Aix Marseille Université, 2018. Français. ⟨tel-01839376⟩

Share

Metrics

Record views

967

Files downloads

287