Hyperspectral discrimination of Eichhornia crassipes covers, in the red edge and near infrared in a Mediterranean river - Institut Agro Rennes-Angers Accéder directement au contenu
Article Dans Une Revue Biological Invasions Année : 2023

Hyperspectral discrimination of Eichhornia crassipes covers, in the red edge and near infrared in a Mediterranean river

Résumé

Hyperspectral data are important for determining and discriminating the spectral signature of invasive aquatic species. Water hyacinth (Eichhornia crassipes) is a dangerous invasive species that has occupied the Al Kabir River in Lebanon since 2006. It forms a dense carpet of different physiological and morphological states and coexists with other vegetation species. This study aimed to use hyperspectral data acquired by a field spectroradiometer to (1) determine the spectral signature of monospecific water hyacinth cover in different morphological and physiological states to discriminate it from those of (2) mixed water hyacinth covers and (3) other plant targets. To this end, six hyperspectral bands (740, 783, 842, 865, 945, and 705 nm) were selected, the first five of which are present as Sentinel-2 bands (B6-B9, respectively) in the red-edge and near-infrared (NIR) spectral zones. Water hyacinth covers and those of other species in these bands were discriminated based on three main criteria: value, shape, and slope. Ten slopes (S-i) and 12 vegetation indices (10 Normalized Vegetation Indices (NVIs), Cl red-edge, and NDRE2) were calculated. Principal component analysis and hierarchical cluster analysis were also applied to classify the targets. The spectral signature of water hyacinth covers differed from those of other plant targets. It had high reflectance, a specific shape of the spectral curve, and the lowest slope and NVIs, especially using combined red-edge and NIR bands (783-865 and 783-954 nm, respectively). In addition, slopes were considered more useful than NVIs for discriminating water hyacinth covers. This hyperspectral spectroradiometer study can be used and combined with Sentinel-2 data in the these red-edge and NIR bands to detect and monitor invasive species at different spatial scales (20 and 60 m resolutions).
Fichier non déposé

Dates et versions

hal-04188091 , version 1 (25-08-2023)

Licence

Copyright (Tous droits réservés)

Identifiants

Citer

Youssra Ghoussein, Ghaleb Faour, Ali Fadel, Jacques J. Haury, Hussein Abou-Hamdan, et al.. Hyperspectral discrimination of Eichhornia crassipes covers, in the red edge and near infrared in a Mediterranean river. Biological Invasions, 2023, 25 (11), pp.3619-3635. ⟨10.1007/s10530-023-03131-2⟩. ⟨hal-04188091⟩
42 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More