利用空间分布模式与样地设计提升森林资源抽样调查精度

胡樱馨; 梅安琪; 徐晴; 侯正阳

doi:10.12171/j.1000-1522.20230061

利用空间分布模式与样地设计提升森林资源抽样调查精度

胡樱馨^{1, 2,},
梅安琪^{1, 2},
徐晴³,
侯正阳^{1, 2, ,}

1.
北京林业大学森林培育与保护教育部重点实验室，北京 100083
2.
国家林业和草原局黑龙江三江平原沼泽草甸生态系统定位观测研究站，黑龙江双鸭山 518000
3.
国际竹藤中心竹藤资源与环境研究所，国家林业和草原局/北京市共建竹藤科学与技术重点实验室，北京 100102

基金项目: 雄安新区科技创新专项（2022XACX1000），国家社会科学基金项目（22BTJ005），国家自然科学基金项目（32001252）。

详细信息

作者简介:
胡樱馨。主要研究方向：森林资源抽样调查。Email：3276888425@qq.com　地址：100083，北京市海淀区清华东路35号

责任作者:
侯正阳，博士，副教授。主要研究方向：森林资源统计监测。 Email：houzhengyang@bjfu.edu.cn　地址：同上。

中图分类号: S757.2
计量
- 文章访问数: 428
- HTML全文浏览量: 65
- PDF下载量: 71
出版历程
- 收稿日期: 2023-03-16
- 修回日期: 2024-01-04
- 网络出版日期: 2024-01-10
- 刊出日期: 2024-01-31

Using spatial distribution patterns and sample plot design to improve the accuracy of forest resource sampling survey

Hu Yingxin^{1, 2,},
Mei Anqi^{1, 2},
Xu Qing³,
Hou Zhengyang^{1, 2, ,}

1.
Key Laboratory for Silviculture and Conservation of Ministry of Education, Beijing Forestry University, Beijing 100083, China
2.
Ecological Observation and Research Station of Heilongjiang Sanjiang Plain Wetlands, National Forestry and Grassland Administration, Shuangyashan 518000, Heilongjiang, China
3.
International Center for Bamboo and Rattan, Key Laboratory of National Forestry and Grassland Administration/ Beijing for Bamboo & Rattan Science and Technology, Beijing 100102, China

摘要

摘要:
目的
森林资源调查中，研究森林属性空间分布模式下的抽样设计，以突破地域限制，为抽样调查提供可推广的经验法则。
方法
利用北京市鹫峰国家森林公园样地调查的实测数据，构建人工总体。以树木死亡率作为为森林属性的代理属性,表达空间自相关。采用系统抽样设计，并通过蒙特卡洛模拟法，评估森林空间自相关、样地大小以及系统抽样设计对抽样精度的影响。
结果
（1）4种不同空间分布模式的总体变异系数，从小到大依次为：死亡率为0%的总体、死亡率为20%的随机模式总体、死亡率为10%的聚集模式总体、死亡率为20%的聚集模式总体。当死亡率为20%，抽样强度为2.73%时，随机模式的变异系数比聚集模式的变异系数低了1.3%。（2）3种不同大小的样地总体变异系数，从小到大依次为：20 m × 20 m、30 m × 30 m、40 m × 40 m。其中，40 m × 40 m的变异系数明显高于20 m × 20 m和30 m × 30 m对应的变异系数。（3）随着抽样强度增大，随机模式下8 × 8的主单元数目设计的人工总体的变异系数比4 × 4的约高0.02%，比16 × 16的约高0.15%；聚集模式下，8 × 8的N设计的人工总体的变异系数比4 × 4的约高0.32%，比16 × 16的约低0.54%。
结论
（1）不同强度的空间自相关都会削弱抽样精度，其中聚集模式相比随机模式的影响更为显著；（2）较小的样地有利于提高抽样精度和精度的收敛速度，但合理大小的样地设计才能有效提升抽样效率；（3）系统抽样中不同主单元数目对抽样精度的影响不明显，实际调查中应避免选择样本量为1的系统抽样，否则抽样误差难以度量。
- 抽样调查 /
- 空间模式 /
- 样地大小 /
- 系统抽样
Abstract:
Objective
In forest resource survey, sampling design under the spatial distribution pattern of forest attributes was studied to break through regional limitations and provide generalizable empirical rules for sampling survey.
Method
Artificial forest populations were constructed with the data field survey at the Beijing Jiufeng National Forest Park. Tree mortality rate was used as a proxy for expressing forest spatial autocorrelation. Systematic sampling design was adopted, and Monte Carlo simulations were implemented to evaluate the effects of spatial autocorrelation, sample plot size and systematic sampling on sampling precision.
Result
(1) The coefficients of variation for the four different spatial distribution patterns increased in the following order: 0% mortality, 20% mortality in random pattern, 10% mortality in aggregated pattern, and 20% mortality in aggregated pattern. When mortality rate was 20% and sampling intensity was 2.73%, the coefficient of variation for random pattern was 1.3% lower than that for aggregated pattern. (2) The coefficients of variation for three different sample plot sizes increased in the following order: 20 m × 20 m, 30 m × 30 m, and 40 m × 40 m. The coefficient of variation of 40 m × 40 m was significantly higher than that of 20 m × 20 m and 30 m × 30 m. (3) With increasing sampling intensity, under the random pattern, the coefficient of variation for the artificial population designed with 8 × 8 main units was about 0.02% higher than that for 4 × 4, and about 0.15% higher than that for 16 × 16. Under the clustered pattern, the coefficient of variation for the artificial population designed with 8 × 8 main units was about 0.32% higher than that for 4 × 4, and about 0.54% lower than that for 16 × 16.
Conclusion
(1) Different degrees of spatial autocorrelation reduce sampling accuracy, among which aggregated pattern has a more significant impact than random pattern. (2) Smaller sample plots help improve sampling accuracy and convergence rate, but reasonable sample plot design can effectively enhance sampling efficiency. (3) The number of main units in systematic sampling has little impact on sampling accuracy. In practice, systematic sampling designs with a sample size of 1 should be avoided because sampling errors would be hard to quantify.
- sampling survey /
- spatial model /
- sample plot size /
- systematic sampling

HTML全文

图 1 鹫峰森林人工总体构建框架

图中颜色由深到浅变化反映了树木个体蓄积由大到小的梯度变化。The color gradient in the image reflects a gradient change of tree biomass from large to small.

Figure 1. Construction framework of artificial forest in Jiufeng

下载: 全尺寸图片幻灯片

图 2 鹫峰地区树种径阶分布

Figure 2. Distribution of tree species diameter classes in Jiufeng area

下载: 全尺寸图片幻灯片

图 3 森林空间自相关表达

图中坐标为UTM Zone 50N投影坐标。红色表示活树，灰色表示死树；死树是任何依赖于树木生死状态的森林属性或生态指标的载体；死亡模式反映了任何替代属性或指标的空间分布模式。The coordinates in the image are UTM Zone 50N projected coordinates. Living trees in red, dead trees in gray; dead trees are surrogate for any forest attributes or ecological indicators relying on living status; mortality patterns represent the spatial distribution of any surrogated attributes or indicators.

Figure 3. Forest spatial autocorrelation expression

下载: 全尺寸图片幻灯片

图 4 系统抽样布设方法

Figure 4. System sampling layout method

下载: 全尺寸图片幻灯片

图 5 森林空间自相关对抽样精度的影响

Figure 5. Effects of forest spatial autocorrelation on sampling accuracy

下载: 全尺寸图片幻灯片

图 6 样地大小设置对抽样精度的影响

Figure 6. Effects of sample size setting on sampling accuracy

下载: 全尺寸图片幻灯片

图 7 系统抽样设计对抽样精度的影响

R.随机模式；C.聚集模式；两种空间模式下死亡率均为20%，样地大小均为20 m × 20 m。R, Random; C, Cluster; The mortality rate is 20% under both spatial patterns, with plot sizes of 20 m × 20 m in each case.

Figure 7. Effects of systematic sampling design on sampling accuracy

下载: 全尺寸图片幻灯片

图 8 样地大小与空间模式的联合作用

Figure 8. Joint action of sample plot size and spatial pattern

下载: 全尺寸图片幻灯片

表 1 鹫峰乔木树种调查基本信息统计

Table 1 Basic information statistics of Jiufeng arbor species investigation

树种 Tree species	胸径 DBH/cm				树高 Tree height (H)/m
树种 Tree species	最小值 Min. value	最大值 Max. value	平均值 Mean	标准差 SD	最小值 Min. value	最大值 Max. value	平均值 Mean	标准差 SD
栓皮栎 Quercus variabilis	3.0	59.2	15.9	9.2	1.4	31.0	10.0	4.9
侧柏 Platycladus orientalis	1.7	23.5	10.3	3.5	2.2	14.9	8.0	2.3
五角枫 Acer mono	4.1	24.2	10.6	4.8	3.0	23.0	10.8	4.9
千金榆 Carpinus cordata	4.2	30.6	9.3	4.4	2.3	16.9	7.5	3.2
槲栎 Quercus aliena	5.0	36.4	13.5	5.1	2.3	23.6	9.5	3.4
朴树 Celtis sinensis	5.0	24.5	10.9	5.5	2.4	20.2	10.1	4.7
白蜡 Fraxinus chinensis	5.0	21.6	7.8	3.2	3.1	14.2	6.5	2.0
刺槐 Robinia pseudoacacia	5.2	32.6	12.4	9.0	2.1	14.0	6.3	3.1
油松 Pinus tabuliformis	6.3	19.1	13.3	3.5	3.7	11.5	8.1	2.5
华北落叶松 Larix principis-rupprechtii	8.4	17.1	12.4	2.6	4.7	9.4	6.2	1.3

下载: 导出CSV

表 2 基于鹫峰数据的人工总体构建汇总

Table 2 Summary of artificial population construction based on Jiufeng data

序号 No.	总体 Population	空间模式 Spatial pattern	死亡率 Mortality/%	样地尺寸 Sample size	总蓄积 Total volume/m³	单位蓄积量/(m³·hm⁻²) Unit volume/（m³·ha⁻¹）	聚集辐射距离 Cluster radius distance/m
1	M0-P20		0	20 m × 20 m	35 642.97	49.58
2	M0-P30		0	30 m × 30 m	35 642.97	49.58
3	M0-P40		0	40 m × 40 m	35 642.97	49.58
4	R-M20-P20	随机 Random	20	20 m × 20 m	28 584.23	49.70
5	R-M20-P30	随机 Random	20	30 m × 30 m	28 584.23	49.70
6	R-M20-P40	随机 Random	20	40 m × 40 m	28 584.23	49.70
7	C-M10-P20	聚集 Cluster	10	20 m × 20 m	31 829.55	49.69	50
8	C-M10-P30	聚集 Cluster	10	30 m × 30 m	31 829.55	49.69	50
9	C-M10-P40	聚集 Cluster	10	40 m × 40 m	31 829.55	49.69	50
10	C-M20-P20	聚集 Cluster	20	20 m × 20 m	28 125.76	49.72	70
11	C-M20-P30	聚集 Cluster	20	30 m × 30 m	28 125.76	49.72	70
12	C-M20-P40	聚集 Cluster	20	40 m × 40 m	28 125.76	49.72	70

下载: 导出CSV

表 3 各树种异速生长模型参数统计

Table 3 Parameter statistics of allometric growth model of various tree species

树种 Tree species	材积参数 Timber volume parameter			树高曲线参数 Tree height curve parameter
树种 Tree species	a	b	c	c₀	c₁	c₂
栓皮栎 Quercus variabilis	0.000 057 469	1.915 56	0.926 60	10.010 30	0.155 53	1.788 38
侧柏 Platycladus orientalis	0.000 091 972	1.863 98	0.831 57	4.298 95	0.458 07	4.320 85
五角枫 Acer mono	0.000 057 469	1.915 56	0.926 60	10.017 40	0.292 15	4.296 35
千金榆 Carpinus cordata	0.000 062 324	1.825 58	0.977 49	12.584 10	0.120 40	1.670 84
槲栎 Quercus aliena	0.000 057 469	1.915 56	0.926 60	10.010 30	0.155 53	1.788 38
朴树 Celtis sinensis	0.000 057 469	1.915 56	0.926 60	19.487 60	0.040 81	0.801 35
白蜡 Fraxinus chinensis	0.000 057 469	1.915 56	0.926 60	19.487 60	0.040 81	0.801 35
刺槐 Robinia pseudoacacia	0.000 071 182	1.941 49	0.814 87	13.725 10	0.130 98	1.818 21
油松 Pinus tabuliformis	0.000 066 492	1.865 56	0.937 69	9.285 56	0.219 29	3.400 51
华北落叶松 Larix principis-rupprechtii	0.000 053 403	1.808 07	1.072 42	19.614 70	0.014 09	0.432 46

下载: 导出CSV

参考文献(30)

[1]	白降丽, 彭道黎, 庾晓红. 我国森林资源调查技术发展研究[J]. 山西林业科技, 2005(1): 4−7. Bai J L, Peng D L, Yu X H. Study on the development of forest resource inventory technology in China[J]. Shanxi Forestry Science and Technology, 2005(1): 4−7.
[2]	张会儒, 雷相东, 李凤日. 中国森林经理学研究进展与展望[J]. 林业科学, 2020, 26(9): 130−142. Zhang H R, Lei X D, Li F R. Research progress and prospects of forest management science in China[J]. Scientia Silvae Sinicae, 2020, 26(9): 130−142.
[3]	陈盼盼. 地面摄影测量在森林资源调查中的关键技术研究[D]. 北京: 北京林业大学, 2020. Chen P P. Study on key technologies of ground photogrammetry in forest inventory[D]. Beijing: Beijing Forestry University, 2020.
[4]	冯仲科. “互联网+”推进森林资源调查走进精准林业新时代[J]. 国土绿化, 2019(2): 16−19. Feng Z K. “Internet+” promotes the forest resource survey to enter the era of precision forestry[J]. Land Greening, 2019(2): 16−19.
[5]	罗仙仙, 亢新刚, 杨华. 我国森林资源综合监测抽样理论研究综述[J]. 西北林学院学报, 2008, 23(6): 187−193. Luo X X, Kang X G, Yang H. A review on the sampling theory of forest resources comprehensive monitoring[J]. Journal of Northwest Forestry University, 2008, 23(6): 187−193.
[6]	曾伟生, 黄国胜, 党永峰, 等. 全国森林资源宏观监测的抽样设计与估计方法探索[J]. 林业资源管理, 2016(3): 1−6. Zeng W S, Huang G S, Dang Y F, et al. Discussion on sampling design and estimation methods of national forest resources macro-monitoring[J]. Forest Resources Management, 2016(3): 1−6.
[7]	高宏. 森林资源抽样调查技术方法[J]. 林业勘查设计, 2018, 188(4): 87−88. doi: 10.3969/j.issn.1673-4505.2018.04.040 Gao H. Technical methods of sampling survey of forest resources[J]. Forest Investigation Design, 2018, 188(4): 87−88. doi: 10.3969/j.issn.1673-4505.2018.04.040
[8]	李春干, 代华兵. 中国森林资源调查: 历史、现状与趋势[J]. 世界林业研究, 2021, 34(6): 72−80. Li C G, Dai H B. Forest management inventory in China: history, current status and trend[J]. World Forestry Research, 2021, 34(6): 72−80.
[9]	Tomppo E. Forest inventory: methodology and applications[M]. Dordrecht: Springer, 2006: 179−194.
[10]	李明阳, 姜文倩, 徐婷, 等. 基于总体表面属性特征的森林资源抽样调查方法比较[J]. 东北林业大学学报, 2011, 39(9): 49−51, 64. doi: 10.3969/j.issn.1000-5382.2011.09.016 Li M Y, Jiang W Q, Xu T, et al. Comparison of sampling survey methods of forest resources for populations with different surface attributes[J]. Journal of Northeast Forestry University, 2011, 39(9): 49−51, 64. doi: 10.3969/j.issn.1000-5382.2011.09.016
[11]	Annika K M. Forest inventory: methodology and applications [M]. New York: Springer-Verlag, 2006.
[12]	Cochran W G. Sampling techniques[M]. 3rd ed. Hoboken: John Wiley & Sons, 1977: 227.
[13]	Mcroberts R E, Westfall J A. Propagating uncertainty through individual tree volume model predictions to large-area volume estimates[J]. Annals of Forest Science, 2016, 73(3): 625−633. doi: 10.1007/s13595-015-0473-x
[14]	Särndal C E T, Hoem L, Lindley J M, et al. Design-based and model-based inference in survey sampling[J]. Scandinavian Journal of Statistics, 1978(5): 27−52.
[15]	Schreuder H T, Gregoire T G, Wood G B. Sampling methods for multiresource forest inventory [M]. New York: John Wiley & Sons, 1993: 446.
[16]	金勇进, 郝一炜. 非概率样本的模型推断[J]. 数学的实践与认识, 2019, 49(5): 246−255. Jin Y J, Hao Y W. Model-based inference for non-probability sample[J]. Mathematics in Practice and Theory, 2019, 49(5): 246−255.
[17]	Thompson S K. Sampling[M]. 3rd ed. Hoboken: John Wiley & Sons, 2012.
[18]	Legendre P. Spatial autocorrelation: trouble or new paradigm[J]. Ecology, 1993, 74(6): 1659−1673. doi: 10.2307/1939924
[19]	Reed D, Burkhart H. Spatial autocorrelation of individual tree characteristics in loblolly pine stands[J]. Forest Science, 1985, 31(3): 575−587.
[20]	Penttinen A , Henttonen H M. Marked point processes in forest statistics[J]. Forest Science, 1992, 38(4): 806−824.
[21]	Tomppo E. Models and methods for analyzing spatial pattern of trees[J]. Communicationes Instituti Forestalis Fenniae, 1986(138): 65.
[22]	Chou Y H. Spatial patterns and spatial autocorrelation[J]. Lecture Notes in Computer Science, 1995(988): 365−376.
[23]	Zeng W S, Tomppo E, Healey S P, et al. The national forest inventory in China: history-results-international context[J]. Forest Ecosystems, 2015(2): 23.
[24]	宋新民, 李金良. 抽样调查技术[M]. 北京: 中国林业出版社, 2007. Song X M, Li J L. Sampling survey technology[M]. Beijing: China Forestry Publishing House, 2007.
[25]	Hou Z, Xu Q, Hartikainen S, et al. Impact of plot size and spatial pattern of forest attributes on sampling efficacy[J]. Forest Science, 2015, 61(5): 847−860. doi: 10.5849/forsci.14-197
[26]	庞丽峰, 雷渊才, 陆元昌, 等. 森林经营单位级系统群团抽样设计比较[J]. 西北农林科技大学学报(自然科学版), 2015, 43(6): 141−152. Pang L F, Lei Y C, Lu Y C, et al. Comparison of cluster systematic sapling at management unit level[J]. Journal of Northwest A & F University (Natural Science Edition), 2015, 43(6): 141−152.
[27]	嘎拉泰, 朱双双, 徐道春, 等. 鹫峰森林多层土壤温度变化规律的研究[J]. 林业工程学报, 2018, 3(3): 136−141. Ga L T, Zhu S S, Xu D C, et al. Investigation of temperature changing patterns of multi-layer soil in Jiufeng forest[J]. Journal of Forestry Engineering, 2018, 3(3): 136−141.
[28]	孟芮萱. 城郊型森林公园游憩功能评价及结构优化方法研究: 以鹫峰国家森林公园为例[D]. 北京: 北京林业大学, 2020. Meng R X. The study on recreation function evaluation and structure optimization methods in the suburban forest park: take Jiufeng National Forest Park as a case[D]. Beijing: Beijing Forestry University, 2020.
[29]	蒲莹, 曾伟生, 阳帆. 北京市树高胸径回归模型研建及一元立木材积表检验[J]. 林业资源管理, 2021(3): 62−66. Pu Y, Zeng W S, Yang F. Tree height-diameter regression models development and one-variable tree volume tables examination in Beijing[J]. Forest Resources Management, 2021(3): 62−66.
[30]	Berrill J P, O’hara K L. Influence of tree spatial pattern and sample plot type and size on inventory estimates for leaf area index, stocking, and tree size parameters[C]//Proceedings of coast redwood forests in a changing California: A symposium for scientists and managers. Albany, CA: Pacific Southwest Research Station, 2012: 485– 497.

施引文献(6)

期刊类型引用(4)

1.	张苗苗，罗于洋，王树森，张丽娜，马成功，于胜利，王景圆. 内蒙古旺业甸华北落叶松人工林空间结构分析及其优化. 西北林学院学报. 2024(01): 81-87+107 . 百度学术
2.	荆媛，魏爽，史文辉，马梓贺，王德宇，戎可. 天然次生林中小斑啄木鸟的取食偏好. 野生动物学报. 2024(01): 84-94 . 百度学术
3.	孙宇，刘盛，田佳歆，程福山，赵士博，王诗俊. 基于空间结构优化的长白落叶松人工林分间伐模型构建. 中南林业科技大学学报. 2023(01): 72-83 . 百度学术
4.	刘鑫，黄浪，卿东升，李建军. 基于Voronoi空间单元的林分空间结构智能优化研究. 林业资源管理. 2023(04): 27-35 . 百度学术