Assessing the Accuracy of Spectral Indices in Vegetation Cover Mapping at Vegetation Type and Across Vegetation Type Scales, Using TM Sensor Data in Southern Zagros Regions

In this study، 14 spectral (vegetation) indices have been calculated from different groups، using spectral bands. The correlation of these vegetation indices with canopy cover has then been determined for three rangeland types and one forest type. The percentage of canopy cover in all vegetation types is measured using step-point method in radial direction (6000 points per vegetation type). The correlation of these vegetation indices with canopy cover is then assessed at the scale of vegetation type and across the vegetation types using SPSS and Erdas Imagine softwares. Different groups of vegetation indices، including slope-based، distance-based، orthogonal transformations and plant–water sensitive ones are extracted from Landsat TM satellite reflectance data and evaluated against field cover data. The results indicated that vegetation indices had higher relationships within vegetation types up to 81% (p<0. 001) but the correlations across vegetation types was decreased due to diversity of vegetation spectra and the effects of background features. Overall، the results indicated that each spectral index was suitable for mapping a specific vegetation type because of structural and biological differences in vegetation cover as well as background soil and these characteristics must be considered for selecting the best indices for mapping and stratification of vegetation types.