Old-growth forests (OGFs) are extremely valuable relict ecosystems for studying natural disturbance dynamics. Small-scale disturbances caused by tree crown mortality of one or few individuals, i.e. gap dynamics, are the most frequent events occurring in OGFs. Understanding these processes requires information on the spatial arrangement of forest patches dominated by different tree species and forest canopy gaps at a fine spatial scale. Here, we aimed at mapping different land-cover classes including conifers, broad-leaved trees, and forest canopy gaps using two very-high-resolution satellite images, i.e. Pléiades images, in the mixed fir-spruce-beech OGF reserve of Biogradska Gora (Montenegro). Specifically, we coupled an Object-Based Image Analysis (OBIA) approach and a Random Forest classifier, trained with samples partly derived from field data. The adopted approach showed high accuracy for the main land-cover classes (conifers, broadleaved trees, grasslands, bare ground, and water), e.g. producer’s and user’s accuracy higher than 92% and 95%, respectively. Conversely, forest canopy gaps were classified with lower accuracy, e.g. minimum producer’s and user’s accuracies of 75% and 54%, respectively. Despite the exploitation of textural metrics during both image segmentation and classification, the lack of remote sensing data providing information on the vertical structure of the forest stand prevented us from accurately map forest canopy gaps.

Land-Cover Mapping in the Biogradska Gora National Park with Very-High-Resolution Pléiades Images

Cagliero E.;Marchi N.;Lingua E.
2022

Abstract

Old-growth forests (OGFs) are extremely valuable relict ecosystems for studying natural disturbance dynamics. Small-scale disturbances caused by tree crown mortality of one or few individuals, i.e. gap dynamics, are the most frequent events occurring in OGFs. Understanding these processes requires information on the spatial arrangement of forest patches dominated by different tree species and forest canopy gaps at a fine spatial scale. Here, we aimed at mapping different land-cover classes including conifers, broad-leaved trees, and forest canopy gaps using two very-high-resolution satellite images, i.e. Pléiades images, in the mixed fir-spruce-beech OGF reserve of Biogradska Gora (Montenegro). Specifically, we coupled an Object-Based Image Analysis (OBIA) approach and a Random Forest classifier, trained with samples partly derived from field data. The adopted approach showed high accuracy for the main land-cover classes (conifers, broadleaved trees, grasslands, bare ground, and water), e.g. producer’s and user’s accuracy higher than 92% and 95%, respectively. Conversely, forest canopy gaps were classified with lower accuracy, e.g. minimum producer’s and user’s accuracies of 75% and 54%, respectively. Despite the exploitation of textural metrics during both image segmentation and classification, the lack of remote sensing data providing information on the vertical structure of the forest stand prevented us from accurately map forest canopy gaps.
Communications in Computer and Information Science
978-3-030-94425-4
978-3-030-94426-1
File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate

Caricamento pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11577/3418625
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact