CLM3.0-DGVM中植物叶面积指数与气候因子的时空关系

作为陆面模型里植被的特征量，叶面积值数（LAI）和植被覆盖度在陆地-大气相互作用的相关研究里被广泛应用。LAI的模拟是动态植被模式（DVM）的核心任务之一，需要对模拟的LAI与气候因子间的时空关系进行评估以更好的了解模式性能以及理解植被-大气反馈过程。用1950—1999年的气象数据驱动通用陆面模式的动态植被模式（CLM3.0-DGVM）模拟得到的全球潜在植被的LAI和2001—2003年MODIS观测资料衍生出的LAI数据进行对比，并在此基础上研究当前气候条件下不同植物功能型（PFT）的LAI与不同气候因子在年际尺度上的时空关系，包括运用Moran系数理论分析空间自相关性、运用逐步回归算法构建空间最优一阶线性回归方程、分析模式LAI与气候因子间的滞后相关性。研究表明：1）以MODIS衍生数据作参照，改进后的CLM3.0-DGVM能较好地模拟不同PFTs的LAI年最大值的空间分布型，但是在物候模拟即LAI的季节循环上存在不足；2）植物LAI的分布具有正的空间自相关性。对潜在植物LAI和气候因子进行拟合时不同气候因子对不同PFTs的方差贡献不一样，一般降水最大、风速最小。这反映了陆地生态系统和气候间复杂的相互关系；3）模式模拟的LAI和气候因子有显著的1～2年的滞后相关，其中光照、降水和LAI的滞后相关性波动较大，而温度、比湿的较小，风速的不明显。这些基于CLM3.0-DGVM的结论在自然界的植物–气候相互作用系统中具有普遍意义：不同地区不同植物受不同气候因子的影响不一样；找出不同PFT的主要气候影响因子和理解其中最关键的生物物理和生物化学过程是至关重要的。进一步工作需要用更精确和更高分辨率的气候数据以及局地观测的LAI对DGVM做评估，同时DGVM本身也需要继续改进（例如加入农作物和灌溉过程的模拟）。

Spatiotemporal relationship of leaf area index simulated by CLM3.0-DGVM and climatic factors

Leaf area index (LAI) and fractional vegetation cover are widely used to characterize vegetation in land models for land-atmosphere interaction studies. The prediction of LAI is one of the core tasks of Dynamic Vegetation Models (DVMs), and the spatiotemporal relationship between the climate and simulated LAI or other vegetation variables simulated by DVMs needs to be evaluated and better understood. In this work, the Dynamic Global Vegetation Model in the Community Land Model version 3.0 (CLM3.0-DGVM) is utilized to address this issue by evaluating the simulated LAI using the new plant function type (PFT) LAI parameters derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) data developed by The National Center for Atmospheric Research. The Moran's autocorrelation index is used to determine the degree of clustering of the simulated annual LAI maximum (LAIx) in the absence of ancient agricultural practices and modern industrialization. Then the stepwise regression algorithm is applied to construct an optimal multivariate linear regression equation for each PFT between LAIx (dependent variable) and climatic factors (independent variables). In this way the dominant and secondary factors as well as their statistical significance can be quantified. Furthermore, the temporal relationship between LAIx and five climatic factors for each PFT at interannual time scales under current climatological condition is investigated using time-lagged correlation analysis. The conclusions from these model-data analyses are: (1) the modified CLM3.0-DGVM is able to simulate well the mean LAIx value for each PFT and reproduce the global biogeographic patterns of LAI, but it still has deficiencies in the simulation of PFT phenology (i.e. LAI seasonal cycle); (2) there is a positive spatial autocorrelation of LAIx within each PFT. The spatial distribution pattern of LAIx is strongly influenced by the climatic factors, and this effect differs for different PFTs. Generally, solar radiation and precipitation are the first-order impact factors, followed by specific humidity; (3) the interannual trends of LAIx simulated by CLM3.0-DGVM has a significant 1-year or 2-year lag relationship with some climatic factors, mainly because LAIx is calculated from previous year's net primary production in this model. Among the five climatic factors, solar radiation and precipitation have larger correlations with the LAIx in the subsequent one or two years than temperature and specific humidity, while the wind has a negligible correlation with LAIx. The implications of these characteristics of LAI and LAI-climate relations revealed in this DGVM for the general vegetation-climate interactions in nature are: different climatic factors have different effects on different plants in different regions, the biogeographic pattern of vegetation is a composite of individualistic responses to climatic factors of different plants which implies that terrestrial ecosystem exhibits complex behavior at different spatiotemporal scales; and for each PFT, it is crucial to identify the dominant climatic factor and understand the main biophysical and biogeochemical processes. Work is still needed to further evaluate the DGVM using more accurate and higher-resolution climatic data and in-situ LAI data. In addition, the DGVM needs to be further improved (e.g., by including crops and irrigation).