基于森林资源清查、卫星影像数据与随机协同模拟尺度转换方法的森林碳制图

Mapping of forest carbon by combining forest inventory data and satellite images with co-simulation based up-scaling method

当前,全球变暖对地球生态系统的影响正引起世界的广泛关注。为减缓其影响进程,让决策者获得准确的碳源/碳汇空间分布信息与动态至关重要。目前面临的重大挑战是如何准确估计森林碳的空间分布和分析估计结果的不确定性。本研究基于森林资源连续清查样地数据和遥感影像数据发展了一个森林碳制图的一般方法。基于序列高斯协同模拟算法,结合样地数据与卫星影像数据进行模拟,将森林碳汇分布图的尺度从30m×30m转换到900m×900m(区域、国家和全球森林碳制图单位大小)。以临安市为例,利用全市2004年森林资源清查样地数据和同年度LandsatTM影像数据,进行研究区森林碳(地上部分)模拟和尺度转换。结果显示,方法准确重现了森林碳空间分布和变异规律,在分布上模拟结果与地面样地属性具有较好的一致性,在数量上模拟结果的总体平均值较地面样地的总体平均值低约24.9%;模拟还提供了其估计结果的不确定性,包括估计值的方差和估计值大于一定阈值的概率,这些可用于不确定性传播模型的模拟分析,进而实现对森林碳估计结果的评价。

Global warming is currently a major concern to the sustainability of earth′s ecosystems. To mitigate this effect, it is essential to provide policy makers with accurate information on the distribution and dynamics of carbon sources and sinks. However, one important challenge in the estimation of forest carbon is how to quantify its spatial distributions and corresponding uncertainties. This study developed a general methodology for mapping forest carbon sinks by combining existing National Forest Inventory （NFI） plot data and satellite images. This method was based on sequential Gaussian co-simulation to spatially combine and up-scale plot data and satellite images from small （30 m×30 m） to large map units 900 km×900 km. Those large units are usually required for mapping forest carbon at regional, national and global scales. The proposed method was applied to mapping forest carbon using the 2004 NFI plot data and Landsat Thematic Mapper images for Lin′an County, Zhejiang, China. Results showed that the proposed method accurately reproduced the spatial distribution of the NFI plot data. However, the simulated average carbon was 24.9% lower than that of the NFI plot data. This study also provided quantitative information on variability in forest carbon and uncertainty of its estimates, including variances and probability that estimates would be larger than a threshold value. This information will in turn be useful in further uncertainty propagation modeling and analysis, and management of forest carbon markets. In conclusion, this study provided the solution to overcome some of the current significant gaps in the generation and assessment of forest carbon products and its uncertainty.