Identification of Tundra Land Cover near Teshekpuk Lake, Alaska Using SPOT Satellite Data

• Main Authors: Carl J. Markon, Dirk V. Derksen
• Other Authors: The Pennsylvania State University CiteSeerX Archives
• Format: Text
• Language: English
• Published: 1994
• Subjects: Arctic; Arctique*; toundra; Tundra; Alaska
• Online Access: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.485.5556; http://pubs.aina.ucalgary.ca/arctic/Arctic47-3-222.pdf

ABSTRACT. Tundra vegetation in the Teshekpuk Lake area of the Alaskan Arctic Coastal Plain was mapped to assess distribution and abundance of waterfowl habitats. Three SPOT satellite scenes were acquired and registered to a 20 m Universal Transverse Mercator grid. Two clustering techniques were used to develop statistical parameters by which the SPOT data were spectrally classified. A maximum likelihood algorithm that correlated spectral classes with land cover types was applied to the SPOT data. Field data were used to assist in spectral class labeling and vegetation descriptions. Twelve cover classes were mapped. The most common type was moist sedge meadow tundra (13.5%); the least common was moss/peat shoreline (0.2%). The moss/peat shoreline type, important to moulting geese and other waterfowl, was spectrally identified using supervised clustering techniques. All other land cover types were identified using unsupervised clustering techniques. Cover classes were described, and a tundra landscape profile produced. Key words: tundra land cover mapping, SPOT satellite data, Teshekpuk Lake, Alaska RÉSUMÉ. On a dressé une carte de la végétation de toundra dans la région du lac Teshekpuk de la plaine côtière arctique de l’Alaska afin d’évaluer la distribution et l’abondance de l’habitat de la sauvagine. On s’est procuré trois scènes prises par le satellite SPOT, qu’on a alignées sur un quadrillage de 20 m de la projection transverse universelle de Mercator. On a utilisé deux techniques d’agrégation pour élaborer des paramètres statistiques grâce auxquels on a classé en fonction du spectre les données provenant du SPOT. On a appliqué aux données du SPOT un algorithme de vraisemblance maximale corrélant les classes spectrales avec les types de couvert végétal. On a utilisé des données de terrain pour aider à nommer les classes spectrales et à décrire la végétation. On a cartographié 12 classes de couvert végétal