ALL PUBLICATIONS

Mapping invasive alien Acacia dealbata link using ASTER multispectral imagery: a case study in central-eastern of Portugal

  • Articles in SCI Journals
  • Oct, 2016

Martins, F., Alegria, C., & Gil, A. (2016) Mapping invasive alien Acacia dealbata link using ASTER multispectral imagery: a case study in central-eastern of Portugal. 

Forest Systems, 25(3), e078. DOI:10.5424/fs/2016253-09248 (IF2016 0,893; Q3 Forestry)
Summary:

Aim of the study: Acacia dealbata is an alien invasive species that is widely spread in Portugal. The main goal of this study was to produce an accurate and detailed map for this invasive species using ASTER multispectral imagery. Area of study: The central-eastern zone of Portugal was used as study area. This whole area is represented in an ASTER scene covering about 321.1 x 103 ha. Material and methods: ASTER imagery of two dates (flowering season and dry season) were classified by applying three supervised classifiers (Maximum Likelihood, Support Vector Machine and Artificial Neural Networks) to five different land cover classifications (from most generic to most detailed land cover categories). The spectral separability of the land cover categories was analyzed and the accuracy of the 30 produced maps compared. Main results: The highest classification accuracy for acacia mapping was obtained using the flowering season imagery, the Maximum Likelihood classifier and the most detailed land cover classification (overall accuracy of 86%; Kappa statistics of 85%; acacia class Kappa statistics of 100%). As a result, the area occupied by acacia was estimated to be approximated 24,770 ha (i.e. 8% of the study area). Research highlights: The methodology explored proved to be a cost-effective solution for acacia mapping in central-eastern of Portugal. The obtained map enables a more accurate and detailed identification of this species’ invaded areas due to its spatial resolution (minimum mapping unit of 0.02 ha) providing a substantial improvement comparably to the existent national land cover maps to support monitoring and control activities.