Estimation of vegetation coverage of desert grassland based on images from an unmanned aerial vehicle

Effective and accurate monitoring of grassland vegetation coverage is important for sustainable utilization of grassland resources and for restoration and reconstruction of ecosystems. In this study, a threshold method combining the supervised classification with the statistical histogram of visible vegetation index was used to identify grassland vegetation. The vegetation extraction accuracies of 6 Red Green Blue (RGB) vegetation indices were evaluated. The results indicated that the Normalized Difference Green/Red Index was the most accurate index for vegetation coverage extraction (mean absolute error 2.56%, root mean square error 3.06%). The proposed method accurately estimates the vegetation coverage of desert grassland.