气候变化问题作为人类社会可持续发展面临的重大挑战,受到国际社会越来越强烈的关注。全球气候变化深刻影响着草地生态系统,定量评估区域和不同类型草地生态系统的生产力,研究其对气候变化的敏感性可以为草地生态系统适应未来气候变化提供基础数据和理论依据。草原综合顺序分类系统（CSCS）将天然草原分为42类（其中中国包含41类）,并将其聚合为10个类组。研究利用改进Carnegie-Ames-Stanford Approach （CASA）模型模拟分析中国天然草地2004—2008年的净初级生产力（NPP）并进行系统聚类,分析了草地NPP与影响因子的相关性和敏感程度。结果表明：2004—2008年中国10个草地类组年均NPP均呈现增长趋势,其中亚热带森林草地增长最快,增长率达38.2%。温带湿润草地增长最慢,增长率为14.3%。通过聚类分析将中国41类草原的年均NPP分为3类：第1类NPP值较小,其湿润度级较低,而热量级较高;第2类NPP值较大,其热量级和湿润度级均较高,水热比适宜植被的生长;其余为第3类,其NPP值介于上述两者之间。可见,不同草地类型NPP分布规律与CSCS划分草地类型的湿润度级和热量级密切相关。草地年均NPP与>0 ℃年积温（Σθ）、降水量、湿润度（K）和NDVI的相关性强,与太阳辐射的相关性弱。草地类组NPP平均值对NDVI最敏感,其次为Σθ、K和降水量,敏感性最弱的为太阳辐射。草地NPP对CSCS量化分类的标准Σθ和K较为敏感,改进CASA模型在一定程度上实现了CSCS与草地生产力的耦合。

As a major challenge to the sustainable development of human society, climate change has attracted increasing attention from the international community. Global climate change significantly affects grassland ecosystems. Quantitative assessment of regional and grassland ecosystem productivity and its sensitivity to climate change could provide basic data on and a theoretical basis for grassland ecosystems adapting to future climate change. The grassland integrated comprehensive sequential classification system (CSCS) divides the natural grassland into 42 classes and aggregates them into 10 super-classes. Using the improved Carnegie-Ames-Stanford Approach (CASA) model, we simulated the net primary productivity (NPP) of natural grasslands in China from 2004 to 2008. A system clustering analysis was also conducted based on the annual average NPP of grassland classes to analyze NPP correlation and sensitivity and its influencing factors. The results show that the NPP showed an increase from 2004 to 2008 for all grassland super-classes. The subtropical forest rapidly increased by 38.2%. The temperate humid grassland slowly increased by 14.3%. The 41 grassland classes of China were clustered into three groups. The NPP of group 1 was low for its low wetness level and high heat level. Group 2 had high NPP for its high wetness and heat level and a suitable moisture and temperature ratio. Others were in group 3. The results indicate that the grassland NPP distribution pattern is related to the wetness and heat level in CSCS. There exists a significant correlation between the annual NPP and annual accumulated temperature greater than zero (Σθ), annual precipitation, wetness (K), and normalized difference vegetation index (NDVI). The correlation between annual NPP and solar radiation was the weakest. NDVI was the most sensitive variable to grassland NPP in China, simulated by the improved CASA model in this study, followed by Σθ, K-values, precipitation, and solar radiation. K-values and Σθ were used for quantitative classification in CSCS, and these two parameters were sensitive to NPP. This study completed the coupling of CSCS and grassland NPP to a certain extent.