2000～2013年中国植被叶片聚集度系数时空变化特征

Spatial-temporal characteristics of foliage clumping index in China during 2000–2013

聚集度系数（clumping index,CI）表征了植被叶片在空间分布的随机程度,作为一个重要的植被结构参数,它强烈影响冠层辐射传输、光合作用及陆气交互过程.本文利用MODIS BRDF模型参数产品提取了中国区域2000~2013年每8天的500 m分辨率CI,并分析其时空变化特征.研究结果表明CI在空间分布上具有明显的地域差异,并且与地表覆盖类型密切相关.平均而言,落叶针叶林和常绿针叶林具有最小CI,代表叶片聚集程度最大,常绿阔叶林、落叶阔叶林和常绿灌木林次之,耕地和草地的CI值最大.各种植被类型CI具有明显的季相变化,但是不同植被类型CI季相表现差别很大.CI与年均温度和年均降水量相关性不显著.基于MODIS数据提取的CI能够捕捉到不同植被类型叶片空间分布特征的时空变化规律,这对于提高我国生态、水文、气候及其他地表过程模型的模拟精度具有重要价值.

The distribution of natural vegetation foliage is often clumped at different scales of three-dimensional structures such as shoots, branches, whorls, tree crowns, and tree groups. Clumping index（CI） quantifies the level of foliage grouping relative to a random distribution in space. As an additional vegetation structural parameter of comparable importance to leaf area index（LAI）, CI strongly influences canopy radiation regimes, photosynthesis and land-atmosphere interactions. It is a correction factor to convert the effective LAI which assumes a random distribution of leaves in space to the true LAI, and is also needed to separate the canopy into sunlit and shaded leaves for accurate simulations of process-based ecological and land surface models. However, due to the lack of clumping index available, clumping index values are often assumed to vary only with land cover types in most operational LAI retrieval algorithms and ecological models, which might induce large uncertainties in these model outputs. Therefore, it is highly desirable to map the spatial and temporal variations of the clumping for different land cover types. The bidirectional reflectance distribution function（BRDF） provides the anisotropic nature of surface scattering including both the spectral and angular signatures, which can be used to retrieve structural information of vegetated surfaces. In this paper, based on the moderate resolution imaging spectroradiometer（MODIS） BRDF parameters product, a modified Ross Thick-Li Sparse Reciprocal model was used to simulate the reflectance at hotspot（the solar zenith angle was set as 45°, and the relative azimuth angle between the sun and sensor was 0°） and darkspot（the solar zenith angle was set as 45°, while the relative azimuth angle was 180°） in the near infrared band, which was then used to calculate the angular index named normalized difference between hotspot and darkspot（NDHD）. With the relationship between CI and NDHD simulated by the 4-Scale geometrical model, the clumping index over China at 500 m resolution was retrieved every 8 days during the period from 2000 to 2013. The effect of topography on the retrieved CI was corrected using a topographic compensation function calculated from the digital elevation model at 90 m resolution. A locally adjusted cubic-spline capping method was used to remove the noises in the seasonal trajectory of CI. Then, the spatial and temporal variations of clumping index in China and its relationships with temperature and precipitation were analyzed. Results show that there is obvious difference in the dimensional distribution of CI which closely relates to the land cover types. Among all cover types, the needle-leaved deciduous forests and the needle-leaved evergreen forests have the smallest mean CI（CI are equal to 0.63 and 0.65, respectively） during the growing seasons（from April to September） from 2000 to 2013 in China, indicating the most clumped canopy structure of coniferous forests. The broad-leaved evergreen forests have the second smallest mean CI（CI=0.68）, followed by the broad-leaved deciduous forests（CI=0.75）, and the shrub evergreen forests（CI=0.76）. On average, grasslands and croplands have the highest CI（least clumped）. All cover types have smaller CI values in mid-summer than in other seasons due to the fact that the canopy becomes denser after the spring growth, but the CI seasonal trajectories driven by site/species phenology are quite distinct for different land cover types. MODIS CI can reflect the influence of extreme weather and climate events on the vegetation structure, but the correlation between CI and annual average temperature or annual average precipitation is not significant. MODIS CI can effectively capture the spatial-temporal pattern of the structural difference among land cover types, and provide more reasonably accurate and up-to-date vegetation structure information used in the modeling of terrestrial energy, water, and carbon cycle processes, particularly in heterogeneous vegetated areas.