Evaluation of Landsat TM vegetation indices for estimating vegetation cover on semi-arid rangelands: a case study from Australia

The accurate monitoring of vegetation cover of globally extensive arid and semi-arid environments is important and challenging. This study examined the capacity of five Landsat Thematic Mapper (TM) spectral bands and 17 vegetation indices (VIs) in estimating saltbush and total vegetation cover in semi-arid rangeland environments. It investigated the relationships between ground-surveyed vegetation cover and VIs derived from Landsat TM images and coincident ground reflectance measurements at Lake Mungo and Fowlers Gap, New South Wales, Australia, both vegetated by perennial chenopod shrublands, principally saltbush. In comparison with spectral measures that rely only on visible and near-infrared wavebands, mid-infrared wavelengths and VIs derived from them were better at characterizing vegetation cover. TM5, TM7, SRVI-1, SRVI-2, and SRVI-3 were identified as generally suitable for large- scale surveys in semi-arid rangelands, where a cost-effective evaluation of vegetation cover is required.