An Improved Dimidiated Pixel Model for Vegetation Fraction in the Yarlung Zangbo River Basin of Qinghai-Tibet Plateau

The vegetation fraction is a key factor used in many research fields, including soil and water conservation, and ecological evaluation. In this paper, based on the traditional dimidiated pixel model, NDVI soil extraction is modified to improve the accuracy of the vegetation fraction obtained, thus also providing an improved method for obtaining vegetation factors in other areas of research. Due to the unique regional location of the Yarlung Zangbo River basin, terrain factors are critical for the stability of vegetation change. The improved model is here applied to inverse the vegetation coverage of the study region; the spatial patterns of change intensity during the past 15 years (1998–2012) are then analyzed. Considering the area’s alpine climate, results show that terrain factors have a significant effect on the distribution of vegetation coverage change intensity. Before specific thresholds are reached, terrain factors such as elevation, topographic relief, and slope exhibit a positive correlation with change in the vegetation fraction. In areas with the same longitude, the higher the latitude the greater the change intensity, while vegetation change intensity also increases with an increasing variety among the upper, middle, and lower reaches of the Yarlung Zangbo River. In the river’s middle-upper and lower reaches, vegetation coverage is prone to increase and decrease, respectively. The results presented here could greatly enhance the inversion precision of vegetation coverage and reveal the spatial and temporal heterogeneity between vegetation coverage, which are of great ecological significance and practical value for the protection of eco-environment in the Yarlung Zangbo River basin.