高原型河谷城市具有特殊的自然地理与气候特征,生态环境脆弱,城市化引起的生态环境问题日益突出。植被作为其生态系统的载体,响应更加敏感。深入研究高原型河谷城市的植被变化及其影响因素,对促进西部大开发及城市化健康发展,建立良好的城市人居、生态环境具有重要的现实意义。西宁市作为典型的高原型河谷城市,植被覆盖在城市化与退耕还林(草)政策共同作用下变化明显。基于植被-不透水表面-土壤(V-I-S)模型,以西宁市城市规划区1995年与2009年两期landsat TM影像为数据源,利用线性光谱混合模型进行混合像元分解,获取研究区植被覆盖度的空间分布。通过整体分析、转移矩阵分析,格网分析等技术手段,研究植被时空变化特征并分别探讨川道与丘陵植被变化的影响因素。结果表明:研究期内,西宁城市规划区平均植被覆盖度维持在30%左右。2009年与1995年相比植被覆盖度出现下降,植被覆盖空间差异略有减小。在数量上,基本无覆盖、中覆盖、高覆盖等级呈增加趋势,低覆盖、全覆盖呈下降趋势。研究区西北部及西南部丘陵区植被覆盖整体趋于好转,主要由中低丰度植被等级变化而来,原因在于2000-2005年湟中县累计退耕还林(草)54.91km<sup>2</sup>,累计造林247.98km<sup>2</sup>,使研究区西北部植被覆盖等级提高,表明西宁市退耕还林(草)工程对于改善植被覆盖效果明显。同时丘陵区植被变化与气候影响趋势相同,表明其植被变化可能也受气候变化影响。城市扩展方向及强度对其周边植被覆盖的影响突出。市区快速扩张及农业退化使川道内中高丰度植被覆盖整体退化趋势明显。主要原因在于2000年后西宁进入快速发展期,城市用地规模迅速增大,川道内城市周边大量中高覆盖等级植被转变为基本无覆盖等级,造成植被退化。川道内城市区域植被变化与气候影响趋势相反,表明本文研究结果可能低估了城市化对川道内植被变化的影响幅度,相比气候影响,人为活动的影响更加强烈。研究区内植被覆盖等级的变化趋势为植被覆盖较差的等级(基本无覆盖和低覆盖)向高一级别发展,得益于退耕还林(草)工程;中等级别以上的覆盖等级出现一定程度的退化,尤其是位于川道中受到城市化干扰的区域植被退化问题尤为突出,需对这些区域采取植被保育措施,避免植被覆盖高等级区域受到城市化影响造成不可逆转的退化。;The geographical and climatic characteristics of Plateau Valley-City in the Western China are very typical and fragile. The urban ecosystems and environment have been damaged by urbanization recently years. Vegetation coverage plays a key role in the ecosystem of Plateau Valley-City, and it is very sensitive to urbanization. As a typical plateau Valley-City, Xining experienced rapid urbanization and implemented the policy of Converting Cultivated Land into Forest/Grass at the same time. Study on pattern dynamics of vegetation coverage of Xining urban plan zone and its causative factors is of great significance for promoting Western Development and urbanization, and contributes to creating a pleasant urban eco-environment in the Western China. Landsat TM data are the most economical and temporal continuous remote sensing images, however, vegetation pattern analysis on urban scale needs high resolution images. To resolve the problems and provide a more accurate result, Landsat TM data of 1995 and 2009 were unmixed by Linear Spectral Mixing Model (LSMM) in Xining plan zone to calculate proportion of vegetation cover, based on Vegetation-Impervious surface-Soil (V-I-S) model. Linear Spectral Mixture Model comprises five main processes: Minimum Noise Fraction (MNF), Pixel Purity Index (PPI), end-members collection by n-D visualizer, linear spectral unmixing and accuracy test. End-members include vegetation, high albedo surface, low albedo surface and soil. By using statistical analysis, transfer matrix and grid analysis, we evaluated the pattern dynamic of vegetation coverage, and discussed the effect of urbanization, Converting Cultivated Land into Forest/Grass and climate change to vegetation changes in each Valley terrace area and hills area. The results indicate that: at overall level, the average vegetation coverage kept about 30% and showed a downtrend in the study period, meanwhile the regional differences had a little decrease. The areas of no vegetation coverage, moderate and high abundance vegetation coverage showed increasing trend, while the areas of low and full abundance vegetation coverage showed decreasing trend. The vegetation coverage in northwestern and southwestern area emerged in an increasing trend and it was mainly from low and moderate vegetation coverage. Huangzhong County is the main area affected by the policy, and finished Converting Cultivated Land into Forest/Grass and afforestation of 54.91km<sup>2</sup> and 247.98km<sup>2</sup> from 2000 to 2005. It suggested that the policy was very effective. Vegetation dynamics of hills areas may be affected by climate change because that they had same trend to the vegetation dynamics effects of climate change in Tibetan Plateau. Vegetation coverage around urban built-up area changed obviously along with urbanization direction and intensity. Areas of moderate and high abundance vegetation coverage presented a degenerate trend, especially in the urban built-up area influenced by urbanization. The main reason is that urban construction land rapid increased after 2000 and occupied moderate and high vegetation coverage. Effects of human activities on vegetation coverage may be more intense comparing to climate change because that their effect trends on vegetation are opposite. The changing trend of vegetation coverage grade can be summarized that low grades developed to high grades because of converting cultivated land into forest/grass and medium grades presented a degenerate trend because of urbanization. It is necessary to take measures to protect the vegetation around built-up area, and avoid irreversible degradation.