论文部分内容阅读
在区域和全球尺度上估算植被总初级生产力(GPP)对理解陆地生态系统的碳循环具有重要意义。由于地表异质性的存在,局限在站点尺度上的观测数据无法直接扩展到更大空间尺度的区域上。通过与地面观测数据相结合,遥感成为实现植被GPP空间扩展的主要工具。但是现有模型对气象数据依赖较多,且在不同气象数据集的驱动下,模拟结果间会有差异,进而产生不确定性。建立以遥感数据为主的GPP模型(简称遥感GPP模型),使其易于在区域和全球尺度上应用,是解决上述问题的一个可行方案。该研究使用TG(temperature and greenness model)和VI(vegetation index model)两个遥感GPP模型,结合中国通量观测研究联盟(China FLUX)的台站数据,对中国典型植被类型的GPP进行了模拟、比较与评估,旨在进一步提高遥感GPP模型在中国区域的适用性。结果表明:(1)TG和VI模型选用的遥感参数均与GPP观测值有较高的相关性,都可以得到可信的光合转换系数m和a。基于与夜间地表温度平均值的相关关系,m和a在空间尺度上得到了扩展,这使得TG和VI都可以应用到区域尺度上。(2)TG和VI模型的模拟值与实测值间的相关性大多较高,决定系数(R~2)多在0.67以上。但不同台站间的误差变动较大,TG模型的均方根误差为0.29–6.40 g·m~(–2)·d~(–1),VI模型的均方根误差为0.31–7.09 g·m~(–2)·d~(–1)。(3)总体而言,TG模型的表现优于VI,尤其在海拔或纬度较高、以温度限制为主的台站,TG模型的模拟效果较好。上述结果初步揭示遥感GPP模型具备了在区域尺度上应用的潜力。
Estimating the total primary productivity of vegetation (GPP) at regional and global scales is important for understanding the carbon cycle in terrestrial ecosystems. Because of the heterogeneity of the surface, the observation data limited on the site scale can not be directly extended to the larger spatial scale area. By combining with surface observation data, remote sensing becomes the main tool to realize the spatial expansion of vegetation GPP. However, the existing models are more dependent on the meteorological data, and under the different meteorological data sets, there will be differences between the simulation results and further uncertainty. The establishment of GPP model based on remote sensing data (referred to as remote sensing GPP model) makes it easy to apply at the regional and global scales and is a feasible solution to the above problems. In this study, two remote sensing GPP models, TG (temperature and greenness model) and VI (vegetation index model), were used to simulate the GPP of typical vegetation types in China based on the station data of China FLUX. Comparison and evaluation, aims to further improve the applicability of remote sensing GPP model in China. The results show that: (1) Both the remote sensing parameters used in the TG and VI models are highly correlated with the observed GPP values, and both the reliable photosynthetic transformation coefficients m and a can be obtained. Based on the correlation with the average night surface temperature, m and a are expanded on the spatial scale, which makes both TG and VI applicable to the regional scale. (2) The correlation between simulated and measured values of TG and VI model are mostly high, and the determination coefficient (R ~ 2) is more than 0.67. However, the error between different stations fluctuated greatly. The root mean square error of TG model was 0.29-6.40 g · m -2 · d -1, and the root mean square error of VI model was 0.31-7.09 g · M ~ (-2) · d ~ (-1). (3) In general, the performance of TG model is better than that of VI, especially in the stations with high temperature or altitude, especially in the stations with the temperature limitation. The TG model has better simulation results. The above results preliminarily reveal that the remote sensing GPP model has the potential to be applied at the regional scale.