In this paper, a reflectance-based method is proposed to accurately quantify percent grass cover from TM data for a semiarid grassland in western China. In situ measured percent grass cover was sampled over I in 2 plots at 68 sites. Their ground coordinates were logged with a global positioning system (GPS) receiver and their spectral reflectance measured with a spectrometer. Normalized difference vegetation index (NDVI) was derived from both in situ measured spectral reflectance and radiometrically calibrated Landsat Thematic Mapper (TM) bands 3 and 4. It was found that the NDVI derived from in situ measured spectral reflectance was closely correlated with percent grass cover (R-2 = 0.74), but not with its counterpart derived from the satellite image. After standardization of the latter with the former, the TM-derived NDVI bore a close regression relationship with the in situ measured samples (R-2 = 0.74). This relationship enabled the successful quantification of grass cover from the satellite image at an overall accuracy of 89%. This reflectance-based method can be used to reliably quantify grass cover from TM imagery. (C) 2003 Elsevier Inc. All rights reserved.
[1]
R. H. Haas,et al.
Evaluating Landsat Thematic Mapper derived vegetation indices for estimating above-ground biomass on semiarid rangelands
,
1993
.
[2]
John R. Dymond,et al.
Percentage vegetation cover of a degrading rangeland from SPOT
,
1992
.
[3]
José M. Paruelo,et al.
Range assessment using remote sensing in Northwest Patagonia (Argentina).
,
1994
.
[4]
J. Michaelsen,et al.
Estimating grassland biomass and leaf area index using ground and satellite data
,
1994
.
[5]
Xavier Pons,et al.
A simple radiometric correction model to improve automatic mapping of vegetation from multispectral satellite data
,
1994
.