基于序列高斯协同模拟的多尺度区域森林碳密度空间分布估计
Multi-Scale Regional Forest Carbon Density Estimation Based on Sequential Gaussian Co-Simulation
-
摘要: 以浙江省仙居县为例,基于2008年全县森林资源清查样地数据和2007年2月获取的Landsat TM影像数据,采用序列高斯协同模拟方法,分别在30m×30m和270m×270m空间分辨率水平上模拟森林地上部分碳密度及其分布,并对模拟结果进行对比分析。结果表明:仙居县森林地上碳密度分布具有空间连续变异性,四周为高碳密度有林地集中区,中间大部分为低碳密度无林地集中区,抽样估计研究区域地上森林碳储量为528378963Mg。基于30m×30m分辨率的序列高斯协同模拟结果为569287569Mg,模型确定系数为06203;对比270m×270m像元大小基础上估计得到的森林碳储量503087179 Mg ,模型确定系数02383,小尺度上估计的碳储量总量更多,碳密度分布范围更广,模型精度更高。序列高斯协同模拟考虑了森林碳密度空间分布的差异性,模拟结果接近地面样地估计值,碳密度分布范围合理,能够很好地反映碳分布空间的连续变异性。Abstract: Based on Forest Inventory (plot) data in Xianju County, Zhejiang in 2008 and the Landsat TM image data collected in the same region in 2007,the aboveground forest carbon density and its distributions at 30m × 30m and 270m×270m resolution was estimated and the results analyzed comparatively by applying sequential gaussian cosimulation The results showed that the aboveground forest carbon density of Xianju County was continuously distributed, which was surrounded by high carbon density of forest land and the intermediate region was filled with the majority of low carbon density of nonforest land. The total carbon is 5283789.63Mg based on the estimation by rardomly sampling method. With the sequential gaussian cosimulation, the sum of the carbon is 5692875.69 Mg and the square R of model is 06203 in 30m×30m resolution. Comparing with the result in 270m × 270m resolution, the former total carbon is larger, the range of distribution is wider and the model′s precision is higher. The result showed that sequential gaussian cosimulation which considers the spatial distribution of carbon density is closer to the estimation from the plot data, the carbon density distribution is more reasonable and the ability to represent the continuous changes of carbon distribution is better.