分布于我国半干旱区的温带疏林草原生态系统，是森林到草原之间过渡的一种生态系统类型，是在独特的气候和地形条件下发育于沙地的地带性顶级植被群落。疏林草原具有乔木、灌木和草本植被混合生长的特点、空间异质性高。植被遥感监测难度大，至今仍是全球范围内最不精确的土地覆盖类型。如何兼顾精度和范围，实现温带疏林草原区域尺度不同类型植物生长状态监测是当前干旱区植被遥感的热点和难点。本研究基于机器学习算法，通过近地面无人机遥感观测平台获取地表植被类型信息构建训练数据集，结合高分辨率卫星影像，建立疏林草原木本、草本植物覆盖度估算模型，实现了由无人机到卫星的温带疏林草原木本和草本植物覆盖度的同步估算，并比较了两种高分辨率卫星影像对疏林草原木本和草本植被覆盖度估算的差异。研究结果表明：（1）利用无人机近地遥感影像能够准确分类地表覆盖类型，为区域温带疏林草原木本、草本植物覆盖度估算模型提供大量精确的训练样本数据；（2）基于机器学习算法，利用高分六号（GF-6）和哨兵二号（Sentinel-2）两种高分辨率卫星影像建立的疏林草原覆盖度模型均可较好的实现木本、草本植物覆盖度的估算。其中，基于GF-6的疏林草原木本、草本覆盖度估计结果与无人机观测值的决定系数分别为0.72和0.66，均方根误差分别为6.76%和10.69%，估算精度分别为46.31%和77.88%；基于Sentinel-2的疏林草原木本、草本覆盖度估计结果与无人机观测值的决定系数分别为0.72和0.81，均方根误差分别为6.53%和8.20%，估算精度分别为54.30%和83.17%；（3）基于Sentinel-2卫星影像的疏林草原木本和草本植物覆盖度估算精度稍高于GF-6卫星，基于两个卫星影像的草本植物覆盖度的估算精度都要显著高于木本植物。本研究为实现疏林草原木本植物和草本植物覆盖度由景观尺度扩展到区域尺度的估测提供了新的思路，从无人机到卫星跨尺度协同观测的方法能够为区域温带疏林草原不同生活型植物生长状况监测提供有效的方法支撑，未来基于多时相的高分辨率卫星数据可进一步实现区域尺度温带疏林草原木本植物和草本植物的动态监测。

Objective: China temperate sparse forest grassland is a transition ecosystem between forest and grassland, also the top level ecosystem that evolve under the unique climate and topography in northern China. Sparse forest grassland is characterized by the mixed woody and herbaceous vegetation. It is hard to directly identify separately the woody vegetation and herbaceous vegetation on sparse forest grassland by remote sensing even using the high spatial resolution satellite data. Therefore, it’s challenging to map Fractional Woody and Herbaceous Vegetation Cover(FWHVC) on this ecosystem. How to monitor precisely the growth status of woody and herbaceous vegetation on the sparse forest grassland at the regional scale is a hot and difficult topic of vegetation remote sensing on dryland. Method: This study proposed a novel method of FWHVC estimation in temperate sparse forest grassland based on UAV aerial images and high spatial resolution satellite images (GF-6 and Sentinel-2) by machine learning algorithm. Training dataset of FWHVC was derived from the very high-resolution aerial image (2.41 cm/pixel). Meanwhile, this study compared the results of FWHVC estimated from GF-6 and Sentinel-2 images. Result: The results showed that: (1) Unmanned aircraft vehicle platform can capture precisely the land cover type and provide the large amount of reliable training dataset of FWHVC; (2) FWHVC could be estimated well at regional scale based on both GF-6 and Sentinel-2 high-resolution satellite data by machine learning algorithm. The FWHVC derived from GF-6 images and UAV aerial images had the determination coefficient R2 of 0.72 and 0.66, Root Mean Square Error (RMSE) of 6.76% and 10.69%, and Estimated Accuracy (EA) of 46.31% and 77.88%. The FWHVC derived from Sentinel-2 images and UAV aerial images had the determination coefficient R2 of 0.72 and 0.81, RMSE of 6.53% and 8.20%, and EA of 54.30% and 83.17%; (3) The EA of FWHVC estimated from Sentinel-2 is slightly better than GF-6. And the EA of fractional herbaceous vegetation cover estimation were higher than woody vegetation for both satellite images. Conclusion: This paper provides a new way to estimate FWHVC in temperate sparse forest grassland at the regional scale by using multi-source remote sensing data and machine learning algorithm. The multi-scale approach could provide the new methodology support to accurately monitor woody and herbaceous vegetation cover in temperate sparse forest grassland. In the future, FWHVC in sparse forest grassland can be monitored dynamically by the long term and high spatial resolution satellite remote sensing data at regional scale.