Identification of Invasive Species Using Remote Sensing and Vegetation Indices, (Case Study: Vazroud Rangelands, Iran)
Authors
Abstract
Biological invasions form a major threat to the provision of ecosystems products
and services and can affect ecosystems across a wide spectrum of bioclimatic conditions.
Therefore, it is important to systematically monitor the spread of species over broad regions. It
has long been recognized that remote sensing and geographical information system could
contribute to this capacity. This paper aims to investigate the efficiency of Landsat TM images
in identifying and classifying invasive species Cirsium arvense and Stachys byzanthina in
Vazroud rangelands of Iran. For optimizing results, the Cos(t) model was used for atmospheric
correction on the image. Then multiple vegetation indices, by extracting the digital mean of
pixels related to training samples of the corrected image, were calculated. A supervised
algorithm using minimum distance of mean was used as a classification technique for
evaluation against ground truth map. The results indicated that NDVI, Ratio, RVI, TVI and
NRVI were the most suitable indices for the discrimination of Cirsium arvense species. The
best indices for the Stachys byzanthina species were DVI, NDVI, PVI 1, PVI 2, RVI and
WDVI. Of all the indices analyzed, DVI and WDVI were able to discriminate both species but
with varying degrees of separation.
