欢迎访问林业科学,今天是

林业科学 ›› 2014, Vol. 50 ›› Issue (11): 13-22.doi: 10.11707/j.1001-7488.20141102

• 论文与研究报告 • 上一篇    下一篇

基于空间仿真的仙居县森林碳分布估算

张茂震1,2, 王广兴3, 葛宏立1,2, 徐丽华1,2   

  1. 1. 浙江农林大学环境与资源学院 临安 311300;
    2. 浙江农林大学 浙江省森林生态系统碳循环与固碳减排重点实验室 临安 311300;
    3. Department of Geography and Environmental Resources, Southern Illinois University at Carbondale Carbondale 62901
  • 收稿日期:2013-08-25 修回日期:2014-10-01 出版日期:2014-11-25 发布日期:2014-12-04
  • 基金资助:

    国家自然科学基金项目(30972360;41201563).

Estimation of Forest Carbon Distribution for Xianju County Based on Spatial Simulation

Zhang Maozhen1,2, Wang Guangxing3, Ge Hongli1,2, Xu Lihua1,2   

  1. 1. School of Environmental & Resource Sciences, Zhejiang A&F University Lin'an 311300;
    2. Zhejiang Provincial Key Laboratory of Forest Ecosystem Carbon Cycling and Sequestration, and Emission Reduction Zhejiang A & F University Lin'an 311300;
    3. Department of Geography and Environmental Resources, Southern Illinois University Carbondale Carbondale 62901
  • Received:2013-08-25 Revised:2014-10-01 Online:2014-11-25 Published:2014-12-04

摘要:

以浙江省仙居县为研究区,基于2008年森林资源二类调查样地(清查样地)数据和Landsat TM影像,用序列高斯协同仿真方法模拟全县森林碳储量及其分布.在此基础上,用总体估计值一致性(OEC)、仿真变动系数均值(ACV)和相对均方根误差(RRMSE)指标分析仿真精度; 用设置于清查样地周围的临时样地(验证样地)数据与LandsatTM数据进行森林碳序列高斯块协同仿真,分析清查样地的空间代表性和森林碳分布空间仿真的尺度上推方法.结果表明: 仙居县2008年森林总碳储量仿真估计值为2 667 878 Mg,大部分分布在南部和北部山区,中部东西向条带状低海拔区域分布较少; 区域碳密度仿真估计值为0~65.66 Mg·hm-2,无论是全部样地还是减少一半样地,仿真结果总体均值均在抽样估计置信区间以内; 基于清查样地与基于加密的验证样地森林碳仿真结果表明30 m × 30 m水平样地位置碳密度相关系数达0.95,以清查样地为中心1 km × 1 km块的碳密度相关系数为0.85,说明1 km × 1 km样地仍具有较好的代表性,块仿真效果满意; 以(1-RRMSE)/n定义成本效益,则使用一半样地得到的成本效益优于使用全部样地的结果,用此指标有望找到满足给定精度的最经济的样地数量.

关键词: 森林碳制图, 空间协同仿真, Landsat TM, 精度评估, 尺度上推

Abstract:

In this study, forest carbon stock and its spatial distribution of Xianju County, Zhejiang, were simulated by using a sequential Gaussian co-simulation algorithm based on data, collected in a forest resource inventory from sample plots in 2008, and the Landsat Thematic MapperTM image. The obtained forest carbon map was assessed using three accuracy measures of overall estimate consistency (OEC), average coefficient of variation (ACV), and relative root mean square error (RRMSE). Moreover, temporary sample plots were selected surrounding the permanent sample plots and the data were collected. The permanent and temporary plot data sets were respectively combined with the TM image to scale up forest carbon stock from a spatial resolution of 30 m × 30 m to 1 km × 1 km using a sequential Gaussian block co-simulation algorithm. The up-scaled map from the temporary plot data were used to assess the accuracy of the corresponding map from the permanent plot data and to analyze their spatial representatives. The results showed that the 2008's forest carbon stock estimated for the county was 2 667 878 Mg. Most of the stock was found in the southern and northern mountainous areas and less amount of the stock in the central west to the east narrow areas that had lower elevation. The forest carbon density varied from 0 to 65.7 Mg·hm-2. Whether all the permanent plots or half of them were used, the population estimates of forest carbon fell into the confidence intervals of the plot data at a significant level of 5%. When the data from the permanent sample plots were compared with those from the temporary plots, the coefficient of correlation for carbon density was 0.95 and 0.85 at the spatial resolutions of 30 m × 30 m and 1 km × 1 km, respectively. This result implied that when forest carbon stock was scaled up from a spatial resolution of 30 m × 30 m to 1 km × 1 km, the obtained map using the permanent plots re-produced the spatial distribution of the forest carbon density very well. In addition, if a cost efficiency was defined as (1-RRMSE)/number of sample plots, using half of the sample plots showed a higher cost efficiency than using all the sample plots, implying this cost efficiency indicator can be used to determine an appropriate number of field plots in sampling design for generating spatial distribution of forest carbon stock.

Key words: forest carbon mapping, spatial co-simulation, Landsat TM, accuracy assessment, scaling up

中图分类号: