EO big data analytics for the discovery of new trends of marine species habitats in a changing global climate

Abstract

Climate change has been observed using multiple methods of Earth Observation (EO) including in situ, air-borne and space-borne sensing methods. These use multi-modal observation platforms, with various geospatial coverages, spatio-temporal resolutions and accuracies. The resulting EO Big Data from heterogeneous sources constitute valuable sources for scientists to investigate on the manifested responses of natural species behaviour to climate change. In the EO4wildlife1 research project, we have access to Copernicus and Argos EO Big Data for conducting studies on the changes of habitats for a variety of marine species. The challenge is to discover causality of Metocean environmental observations and their relationship with the changing habitats of species. Nevertheless, there is a need to deploy Big Data technologies for connecting, ingesting, processing of EO data, as well as implementing specialised open data analytics services in this study. The particular services shall be made accessible to the scientific community for setting up modelling scenarios concerning the potential discovery of new trends of marine species habitats due to climate change. Three marine species are being studied in the EO4wildlife project. They include the Bluefin Tuna in the Atlantic-Mediterranean migratory regions, the black-footed albatross seabirds across the sub-tropical Atlantic Ocean and Loggerhead sea turtles along the North West coast of the African continent and Cape Verde. Large data representing geospatial migratory tracks and settlements of these respective marine species have been acquired in the project over period of times together with Metocean EO data from Copernicus and Argos satellites. These are currently analysed and modelled with a set of features obtained by searching in a large space of possible measured and derived Metocean parameters. A two-step search was used involving significance measurement and an iterative breadth first search based wrapper type feature selection algorithm. Furthermore, the analysis is useful for improving the performance of our habitat prediction models across the three marine species in the study. The discovery of new habitats geospatial and temporal trends which may be associated to the changing climate under these analyses will be achieved through the deployment of web-enabled data mining and analytics open services. A dedicated Big Data platform supported by generic data management services in the cloud is therefore deployed for assuring the scalability of the data processing and analytics services.

More information

Identifiers

ORCID for Zoheir Sabeur: orcid.org/0000-0003-4325-4871

ORCID for Gianluca Correndo: orcid.org/0000-0003-3335-5759

Catalogue record

Export record