【目的】基于Landsat8遥感影像数据，反演西藏日喀则地区高寒草原植被盖度。【方法】采用比较常用的3种植被盖度反演模型，即像元二分模型、基于归一化植被指数（NDVI）与土壤调整植被指数（SAVI）的回归模型、改进的三波段梯度差模型，对稀疏高寒草原植被盖度进行反演，并采用照相法实测植被盖度进行反演精度分析。【结果】3种模型对高寒草原植被盖度的反演精度以像元二分模型最高，反演精度为82.02%，其他两种模型均小于80%。【结论】像元二分模型相对于回归模型和改进的三波段梯度差模型更适用于稀疏高寒草原植被盖度的反演。

【Objective】Based on Landsat8 Multispectral images. Vegetation coverage of sparse Alpine Grassland in Shigatse city was estimated.【Method】The dimidiate pixel model、regression model based on NDVI (Normalized Difference Vegetation Index) and SAVI (Soil Adjusted Vegetation Index) 、modified three-band gradient difference model were used to inversion vegetation coverage of sparse Alpine Grassland. The measured vegetation coverage calculated by photographic method, which was used to validate the estimation results. 【Results】The dimidiate pixel model had a higher precision than the other two models, its precision is 82.02%. The estimation accuracy of the other two models were less than 80%. 【Conclusion】The dimidiate pixel model is more suitable for the inversion of sparse alpine grassland vegetation coverage.