Publications des scientifiques de l'IRD

Pottier Auréa, Catry Thibault, Tregarot E., Marechal J. P., Fayad V., David Gilbert, Cheikh M. S., Failler P. (2021). Mapping coastal marine ecosystems of the National Park of Banc d'Arguin (PNBA) in Mauritania using Sentinel-2 imagery. International Journal of Applied Earth Observation and Geoinformation, 102, p. 102419 [16 p.]. ISSN 1569-8432.

Titre du document
Mapping coastal marine ecosystems of the National Park of Banc d'Arguin (PNBA) in Mauritania using Sentinel-2 imagery
Année de publication
2021
Type de document
Article référencé dans le Web of Science WOS:000700848100002
Auteurs
Pottier Auréa, Catry Thibault, Tregarot E., Marechal J. P., Fayad V., David Gilbert, Cheikh M. S., Failler P.
Source
International Journal of Applied Earth Observation and Geoinformation, 2021, 102, p. 102419 [16 p.] ISSN 1569-8432
Coastal marine ecosystems ensure fundamental hydro-ecological functions and support high levels of biodiversity, besides being an important resource for local populations. These biocenosis have been increasingly threatened by human pressures (e.g. pollution, overfishing) along with climate change, which may have a dramatic impact on them. The National Park of Banc d'Arguin (PNBA) in Mauritania, one of the biggest parks in Western Africa, is a RAMSAR zone (classified by UNESCO since 1989) that plays a major role in (i) the maintenance of marine biocenosis, (ii) the protection of the ecosystems and (iii) the sequestration of carbon dioxide. Ecosystem databases and associated maps of the PNBA are out of date and limited to the southern and central parts of the park: updating is thus needed. In this paper, a supervised Support Vector Machine (SVM) was deployed using high-resolution images from Sentinel-2 combined with field data to map marine biocenosis of the PNBA. The results highlight that Sentinel-2 shows good classification accuracy for mapping marine biocenosis (>80% overall accuracy and a kappa index of 0.75), including seagrass beds. Also, the use of high-resolution sensors like SPOT-6 (1.5 m pixels) can overcome the limitations of Sentinel-2 (10 m pixels) when it comes to detecting small ecosystems distributed in patches. The use of freely-downloadable Sentinel-2 data, processed using geoinformatic freeware, make the methodology reproducible, affordable and easily transferable to local actors of biodiversity conservation for long term usage.
Plan de classement
Sciences du milieu [021] ; Ecologie, systèmes aquatiques [036] ; Etudes, transformation, conservation du milieu naturel [082] ; Télédétection [126]
Description Géographique
MAURITANIE ; BANC D'ARGUIN PARC
Localisation
Fonds IRD [F B010083125]
Identifiant IRD
fdi:010083125
Contact