基于随机森林协同克里金法的区域森林地上生物量制图——以粤北森林为例

doi:10.12302/j.issn.1000-2006.202202015

国家林草科技领军期刊
中国精品科技期刊
中国高校百佳科技期刊
江苏省新闻出版政府奖期刊奖
RCCSE林学权威期刊（A+）
CSCD核心期刊
Scopus数据库收录期刊
中文核心期刊
SCD核心期刊

作者加群：102861116

微信公众号：南京林业大学学报

高级检索

南京林业大学学报（自然科学版） ›› 2024, Vol. 48 ›› Issue (1): 169-178.doi: 10.12302/j.issn.1000-2006.202202015

基于随机森林协同克里金法的区域森林地上生物量制图——以粤北森林为例

周友锋¹(), 谢秉楼²^,^*(), 李明诗¹

1.南京林业大学林草学院、水土保持学院,南方现代林业协同创新中心,江苏南京 210037
2.浙江省森林资源监测中心,浙江杭州 310020

收稿日期:2022-02-18 修回日期:2022-04-07 出版日期:2024-01-30 发布日期:2024-01-24
通讯作者: 谢秉楼
基金资助:
国家自然科学基金项目(31670552)

Mapping regional forest aboveground biomass from random forest Co-Kriging approach: a case study from north Guangdong

ZHOU Youfeng¹(), XIE Binglou²^,^*(), LI Mingshi¹

1. College of Forestry and Grassland, College of Soil and Water Conservation, Co-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing 210037, China
2. Forest Resources Monitoring Center of Zhejiang Prorince, Hangzhou 310020, China

Received:2022-02-18 Revised:2022-04-07 Online:2024-01-30 Published:2024-01-24
Contact: XIE Binglou

摘要/Abstract

摘要：

【目的】森林地上生物量(aboveground biomass,AGB)是森林生态系统健康状态和碳汇潜力评估的重要指标,本研究提出了一种快速、准确地获取区域尺度 AGB 信息的制图方法。【方法】基于 Landsat 5 TM、ALOS-1 PALSAR-1、STRM DEM和国家森林资源清查数据,提取光谱特征、纹理指数、后向散射系数和地形因子等特征因子,采用随机森林进行AGB建模。考虑到AGB是典型的具有空间自相关特性的生物物理参量,针对模型残差,使用以高程为协变量的克里金空间插值法分离残差项中的结构化成分,并将其叠加到随机森林模型预测值上形成最终的AGB预测值,从而提高区域尺度的AGB制图精度。【结果】协同克里金法将高程数据作为协变量,在预测AGB残差结构化成分方面的性能优于普通克里金法,协同克里金法与随机森林协同的AGB预测性能明显提升。经独立AGB数据的模型验证表明,随机森林模型预测的AGB与实际观测的AGB间的决定系数 R²为0.46,随机森林结合普通克里金的验证 R²提高到了0.51,而随机森林结合协同克里金模型的验证R²为0.57。相应的均方根误差(RMSE)分别从32.48 t/hm²降低到31.58、29.80 t/hm²,平均绝对误差(MAE)从27.28 t/hm² 降低到26.63、25.12 t/hm²,相对改进指数为0.03和0.08。【结论】总体而言,本研究提出的随机森林协同克里金模型提供了一种更准确、可靠地进行复杂地形区域的亚热带森林AGB制图新方法,所生成的AGB专题图有助于发展针对固碳效能的森林经营方法,为全球气候变化背景下的森林碳增汇和森林可持续经营提供参考。

关键词: 森林地上生物量, 随机森林, 协同克里金, ALOS-1 PALSAR-1, Landsat 5 TM, 国家森林资源连续清查, 粤北地区

Abstract:

【Objective】 Forest aboveground biomass (AGB) is an important indicator for evaluating forest ecosystem health status and carbon sink potential. Accurate and quick mapping regional forest AGB has become intensively researched in forest ecosystem status assessment and global climate change studies in recent years. The major objective of this study was to develop a framework for improving the mapping accuracy of AGB in a subtropical forested area with complex terrain. 【Method】 Spectral features, textural indices, backscattering coefficients, and topographical variables were derived from Landsat 5 TM, ALOS-1 PALSAR-1 data and STRM DEM. Next, in tandem with national forest inventory plot measurements, a random forest/Co-Kriging framework that combines the advantages of random forest (RF) and a geostatistical approach was proposed to map AGB in northern Guangdong Province. 【Result】 The experimental results showed that the ordinary Kriging (OK) and Co-Kriging (CK) were able to predict the distribution of the RF-predicted AGB residuals. The predicted structured components of the residuals adding onto the RF predictions could improve the mapping accuracy of AGB to some extent. After the validation of the independent 20% dataset, the determination coefficient between the predictions and the observations increased from 0.46 (RF) to 0.51 (RFOK) and to 0.57 (RFCK). The root mean square error decreased from 32.48 to 31.58 and to 29.80 t/hm² accordingly. The mean absolute error decreased from 27.28 to 26.63 and to 25.12 t/hm². Overall, co-Kriging, which considers elevation as a co-variable, was better than ordinary Kriging in predicting AGB residuals. 【Conclusion】 The RFCK framework provides an accurate and reliable method to map subtropical AGB with complex topography. The resulting AGB maps contribute to targeted forest resource management and promote forest carbon sequestration and sustainable forest management under global warming scenarios.

Key words: forest aboveground biomass, random forest, Co-Kriging, ALOS-1 PALSAR-1, Landsat 5 TM, national forest inventory, north Guangdong Province

中图分类号:

S771.8

周友锋,谢秉楼,李明诗. 基于随机森林协同克里金法的区域森林地上生物量制图——以粤北森林为例[J]. 南京林业大学学报（自然科学版）, 2024, 48(1): 169-178.

ZHOU Youfeng, XIE Binglou, LI Mingshi. Mapping regional forest aboveground biomass from random forest Co-Kriging approach: a case study from north Guangdong[J].Journal of Nanjing Forestry University (Natural Science Edition), 2024, 48(1): 169-178.DOI: 10.12302/j.issn.1000-2006.202202015.

图/表 8

图1

表1

表2

图2

图3

表3

图4

图5

参考文献 36

[1]	PAN Y D, BIRDSEY R A, FANG J Y, et al. A large and persistent carbon sink in the world’s forests[J]. Science, 2011, 333(6045):988-993.DOI:10.1126/science.1201609.
[2]	张志, 田昕, 陈尔学, 等. 森林地上生物量估测方法研究综述[J]. 北京林业大学学报, 2011, 33(5):144-150.
	ZHANG Z, TIAN X, CHEN E X, et al. Review of methods on estimating forest above ground biomass[J]. J Beijing For Univ, 2011, 33(5):144-150.DOI:10.13332/j.1000-1522.2011.05.026.
[3]	张少伟, 惠刚盈, 韩宗涛, 等. 基于光学多光谱与SAR遥感特征快速优化的大区域森林地上生物量估测[J]. 遥感技术与应用, 2019, 34(5):925-938.
	ZHANG S W, HUI G Y, HAN Z T, et al. Estimation of large-scale forest above-ground biomass based on fast optimizing remotely sensed features from optical multi-spectral and SAR data[J]. Remote Sens Technol Appl, 2019, 34(5):925-938.DOI:10.11873/j.issn.1004-0323.2019.5.0925.
[4]	汤旭光, 刘殿伟, 王宗明, 等. 森林地上生物量遥感估算研究进展[J]. 生态学杂志, 2012, 31(5):1311-1318.
	TANG X G, LIU D W, WANG Z M, et al. Estimation of forest aboveground biomass based on remote sensing data:a review[J]. Chin J Ecol, 2012, 31(5):1311-1318.DOI:10.13292/j.1000-4890.2012.0182.
[5]	韩爱惠. 森林生物量及碳储量遥感监测方法研究[D]. 北京: 北京林业大学, 2009.
	HAN A H. Study on monitoring method of forest biomass and carton storage based on remote sensing[D]. Beijing: Beijing Forestry University, 2009.
[6]	孙雪莲, 舒清态, 欧光龙, 等. 基于随机森林回归模型的思茅松人工林生物量遥感估测[J]. 林业资源管理, 2015(1):71-76.
	SUN X L, SHU Q T, OU G L, et al. Remote sensing estimation of the biomass of artificial Simao pine forest based on random forest regression[J]. For Resour Manag, 2015(1):71-76.DOI:10.13466/j.cnki.lyzygl.2015.01.013.
[7]	欧光龙, 胥辉. 森林生物量模型研究综述[J]. 西南林业大学学报(自然科学), 2020, 40(1):1-11.
	OU G L, XU H. A review on forest biomass models[J]. J Southwest For Univ(Nat Sci), 2020, 40(1):1-11.DOI:10.11929/j.swfu.201907029.
[8]	张雷, 王琳琳, 张旭东. 随机森林算法基本思想及其在生态学中的应用——以云南松分布模拟为例[J]. 生态学报, 2014, 34(3):650-659.
	ZHANG L, WANG L L, ZHANG X D, et al. The basic principle of random forest and its applications in ecology: a case study of Pinus yunnanensis[J]. Acta Ecol Sin, 2014, 34(3):650-659. DOI:10.5846/stxb201306031292.
[9]	BREIMAN L. Random forests[J]. Mach Learn, 2001, 45(1):5-32.DOI:10.1023/A:1010933404324.
[10]	ZHAO Q X, YU S C, ZHAO F, et al. Comparison of machine learning algorithms for forest parameter estimations and application for forest quality assessments[J]. For Ecol Manag, 2019, 434:224-234.DOI:10.1016/j.foreco.2018.12.019.
[11]	HLATSHWAYO S T, MUTANGA O, LOTTERING R T, et al. Mapping forest aboveground biomass in the reforested Buffelsdraai landfill site using texture combinations computed from SPOT-6 pan-sharpened imagery[J]. Int J Appl Earth Obs Geoinformation, 2019, 74:65-77. DOI:10.1016/j.jag.2018.09.005.
[12]	许振宇, 李盈昌, 李明阳, 等. 基于Sentinel-1A和Landsat 8数据的区域森林生物量反演[J]. 中南林业科技大学学报, 2020, 40(11):147-155.
	XU Z Y, LI Y C, LI M Y, et al. Forest biomass retrieval based on Sentinel-1A and Landsat 8 image[J]. J Central South Univ For Technol, 2020, 40(11):147-155.DOI:10.14067/j.cnki.1673-923x.2020.11.018.
[13]	GUO P T, LI M F, LUO W, et al. Digital mapping of soil organic matter for rubber plantation at regional scale: an application of random forest plus residuals Kriging approach[J]. Geoderma, 2015, 237:49-59. DOI:10.1016/j.geoderma.2014.08.009.
[14]	SEKULIC A, KILIBARDA M, HEUVELINK G B M, et al. Random forest spatial interpolation[J]. Remote Sens, 2020, 12(10):1687.DOI:10.3390/rs12101687.
[15]	FAYAD I, BAGHDADI N, BAILLY J S, et al. Regional scale rain-forest height mapping using regression-kriging of spaceborne and airborne LiDAR data:application on French Guiana[J]. Remote Sens, 2016, 8(3):240.DOI:10.3390/rs8030240.
[16]	LI J, HEAP A D, POTTER A, et al. Application of machine learning methods to spatial interpolation of environmental variables[J]. Environ Model Softw, 2011, 26(12):1647-1659.DOI:10.1016/j.envsoft.2011.07.004.
[17]	DOS REIS A A, CARVALHO M C, DE MELLO J M, et al. Spatial prediction of basal area and volume in Eucalyptus stands using Landsat TM data:an assessment of prediction methods[J]. N Z J For Sci, 2018, 48:1.DOI:10.1186/s40490-017-0108-0.
[18]	CHEN L, WANG Y Q, REN C Y, et al. Assessment of multi-wavelength SAR and multispectral instrument data for forest aboveground biomass mapping using random forest kriging[J]. For Ecol Manag, 2019, 447:12-25.DOI:10.1016/j.foreco.2019.05.057.
[19]	SILVEIRA E M O, SILVA S H G, ACERBI-JUNIOR F W, et al. Object-based random forest modelling of aboveground forest biomass outperforms a pixel-based approach in a heterogeneous and mountain tropical environment[J]. Int J Appl Earth Obs Geoinformation, 2019, 78:175-188.DOI:10.1016/j.jag.2019.02.004.
[20]	QIN Q Q, WANG H Y, LEI X D, et al. Spatial variability in the amount of forest litter at the local scale in northeastern China: Kriging and co-Kriging approaches to interpolation[J]. Ecol Evol, 2020, 10(2):778-790.DOI:10.1002/ece3.5934.
[21]	CHATTERJEE S, SANTRA P, MAJUMDAR K, et al. Geostatistical approach for management of soil nutrients with special emphasis on different forms of potassium considering their spatial variation in intensive cropping system of West Bengal, India[J]. Environ Monit Assess, 2015, 187(4):183.DOI:10.1007/s10661-015-4414-9.
[22]	卢月明, 王亮, 仇阿根, 等. 一种基于主成分分析的协同克里金插值方法[J]. 测绘通报, 2017(11):51-57,63.
	LU Y M, WANG L, QIU A G, et al. A co-Kriging interpolation method based on principal component analysis[J]. Bull Surv Mapp, 2017(11):51-57, 63.DOI:10.13474/j.cnki.11-2246.2017.0347.
[23]	SHEN W J, LI M S, HUANG C Q, et al. Quantifying live aboveground biomass and forest disturbance of mountainous natural and plantation forests in northern Guangdong,China,based on multi-temporal Landsat,PALSAR and field plot data[J]. Remote Sens, 2016, 8(7):595.DOI:10.3390/rs8070595.
[24]	LU D S, CHEN Q, WANG G X, et al. A survey of remote sensing-based aboveground biomass estimation methods in forest ecosystems[J]. Int J Digit Earth, 2016, 9(1):63-105.DOI:10.1080/17538947.2014.990526.
[25]	AVTAR R, SAWADA H, TAKEUCHI W, et al. Characterization of forests and deforestation in Cambodia using ALOS/PALSAR observation[J]. Geocarto Int, 2012, 27(2):119-137.DOI:10.1080/10106049.2011.626081.
[26]	宋茜, 范文义. 大兴安岭植被生物量的ALOS PALSAR估算[J]. 应用生态学报, 2011, 22(2):303-308.
	SONG Q, FAN W Y. ALOS PALSAR estimation of vegetation biomass in Daxing’anling Region[J]. Chin J Appl Ecol, 2011, 22(2):303-308.DOI:10.13287/j.1001-9332.2011.0074.
[27]	王璟睿, 沈文娟, 李卫正, 等. 基于RapidEye的人工林生物量遥感反演模型性能对比[J]. 西北林学院学报, 2015, 30(6): 196-202.
	WANG J R, SHEN W J, LI W Z, et al. Performances comparison of multiple non-linear models for estimating plantations’ biomass based on RapidEye imagery[J]. J Northwest For Univ, 2015, 30(6):196-202.DOI: 10.3969/j.issn.1001-7461.2015.06.36.
[28]	杜虎, 曾馥平, 王克林, 等. 中国南方3种主要人工林生物量和生产力的动态变化[J]. 生态学报, 2014, 34(10):2712-2724.
	DU H, ZENG F P, WANG K L, et al. Dynamics of biomass and productivity of three major plantation types in southern China[J]. Acta Ecol Sin, 2014, 34(10):2712-2724.DOI:10.5846/stxb201212121788.
[29]	李明诗, 谭莹, 潘洁, 等. 结合光谱、纹理及地形特征的森林生物量建模研究[J]. 遥感信息, 2006(6):6-9,66.
	LI M S, TAN Y, PAN J, et al. Modeling forest aboveground biomass by combining the spectrum, textures with topographic features[J]. Remote Sens Inf, 2006(6):6-9, 66.DOI:10.3969/j.issn.1000-3177.2006.06.003.
[30]	徐婷, 曹林, 佘光辉. 基于Landsat 8 OLI的特征变量优化提取及森林生物量反演[J]. 遥感技术与应用, 2015, 30(2):226-234.
	XU T, CAO L, SHE G H. Feature extraction and forest biomass estimation based on Landsat 8 OLI[J]. Remote Sens Technol Appl, 2015, 30(2):226-234. DOI:10.11873/j.issn.1004-0323.2015.2.0226.
[31]	ASLAN A, RAHMAN A F, WARREN M W, et al. Mapping spatial distribution and biomass of coastal wetland vegetation in Indonesian Papua by combining active and passive remotely sensed data[J]. Remote Sens Environ, 2016, 183:65-81.DOI:10.1016/j.rse.2016.04.026.183:65-81.
[32]	ISAAKS E H, SRIVASTAVA R M. An introduction to applied geostatistics[M]. Oxford: Oxford University Press,1989.
[33]	SAKATA S, ASHIDA F, ZAKO M. An efficient algorithm for Kriging approximation and optimization with large-scale sampling data[J]. Comput Methods Appl Mech Eng, 2004, 193(3/5):385-404.DOI:10.1016/j.cma.2003.10.006.
[34]	ZIMMERMAN D L, ZIMMERMAN M B. A comparison of spatial semivariogram estimators and corresponding ordinary Kriging predictors[J]. Technometrics, 1991, 33(1):77-91.DOI:10.1080/00401706.1991.10484771.
[35]	李云, 张王菲, 崔鋆波, 等. 参数优选支持的光学与SAR数据森林地上生物量反演研究[J]. 北京林业大学学报, 2020, 42(10):11-19.
	LI Y, ZHANG W F, CUI J B, et al. Inversion exploration on forest aboveground biomass of optical and SAR data supported by parameter optimization method[J]. J Beijing For Univ, 2020, 42(10):11-19.
[36]	ZHAO P P, LU D S, WANG G X, et al. Forest aboveground biomass estimation in Zhejiang Province using the integration of Landsat TM and ALOS PALSAR data[J]. Int J Appl Earth Obs Geoinformation, 2016, 53:1-15.DOI:10.1016/j.jag.2016.08.007.

树种 species	异速生长方程 allometric growth equations	树种 species	异速生长方程 allometric growth equations
杉木 Cunninghamia lanceolata	W_干=0.340 15 D^{-0.392 39}H^{0.408 90}V	马尾松 Pinus massoniana	W_干=0.292 89 D^{0.146 21}H^{0.008 924}V
	W_枝=0.271 40 D^{1.072 61}H^{0.281 71}V		W_枝=0.125 32 V
	W_叶=0.464 93 D^{-0.328 92}H^{0.046 374}V		W_叶=0.079 612 D^{-0.352 63}H^{0.015 724}V
阔叶树 broad-leaved tree	W_干=0.297 00 D^{0.212 72}H^{0.046 374}V	桉树 Eucalyptus robusta	W_干=0.237 19 D^{0.315 57}H^{-0.022 517}V
	W_枝=0.545 41 D^{0.274 01}H^{-0.165 65}V		W_枝=0.090 123 D^{-0.302 67}H^{0.019 109}V
	W_叶=0.225 26 D^{-0.388 74}H^{-0.219 25}V		W_叶=0.052 063 7 D^{-0.216 66}H^{0.014 372}V
毛竹 Phyllostachys edulis	W_干=0.000 096 7 D^{0.217 5}N	杂竹 miscellaneous bamboo	W_干=0.001 Ne^{3.274 82-9.674 4/}^D
	W_枝=0.000 831 98 D^{1.177 4}N^0.648		W_枝=0.001 N/(0.685+12.893e^-^D)
	W_叶=0.000 509 9 D^{1.177 4}N^0.648		W_叶=0.001 N/(1.056+48.560 9e^D)

林分类型 stand type	样地数量 sample size	平均胸径/cm mean DBH	平均树高/m mean tree height	森林地上生物量AGB/(t·hm^-2)
林分类型 stand type	样地数量 sample size	平均胸径/cm mean DBH	平均树高/m mean tree height	最小值 min	最大值 max	平均值 mean	标准差 SD
阔叶纯林pure broad-leaved forest	39	10.0	12	2.97	168.70	38.51	33.73
针叶纯林pure coniferous forest	43	10.6	10	6.48	87.78	27.05	23.65
阔叶混交林mixed broad-leaved forest	97	12.0	10	2.83	203.16	79.36	49.99
针叶混交林mixed coniferous forest	9	13.7	10	11.76	131.42	55.43	40.53
针阔混交林mixed broadleaf-conifer forest	17	12.5	10	13.78	91.15	49.92	24.02
竹林bamboo forest	26	6.6	9	2.08	202.50	60.10	60.10
灌木林shrubwood	14	8.9	7	12.37	158.53	67.26	58.22
总计total	245	9.2	11

克里金法 Kriging	变异函数模型 variogram models	块金值 nugget	基台值 sill	块金值/基台值 nugget/sill	变程/km range	R²	残差平方和 RSS
	指数函数exponential	376.32	387.94	0.97	72.32	0.153	8 921.4
OK	球面函数spherical	400.50	407.20	0.98	78.21	0.161	8 784.6
	高斯函数Gaussian	372.30	387.03	0.96	72.70	0.162	8 754.5
	指数函数exponential	390.86	411.43	0.95	61.34	0.188	8 287.4
CK	球面函数spherical	365.24	376.54	0.97	50.46	0.192	8 234.6
	高斯函数Gaussian	372.37	393.68	0.95	48.47	0.194	8 145.2

基于随机森林协同克里金法的区域森林地上生物量制图——以粤北森林为例

Mapping regional forest aboveground biomass from random forest Co-Kriging approach: a case study from north Guangdong

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献 36

相关文章 10

编辑推荐

Metrics

本文评价

作者

读者

审稿

[1]	唐依人, 贾炜玮, 王帆, 孙毓蔓, 张颖. 基于TLS辅助的长白落叶松一级枝条生物量模型构建[J]. 南京林业大学学报（自然科学版）, 2023, 47(2): 130-140.
[2]	孙铭辰, 姜立春. 基于机器学习算法的樟子松立木材积预测[J]. 南京林业大学学报（自然科学版）, 2023, 47(1): 31-37.
[3]	王甜, 王雪峰, 刘嘉政. 基于RFE_RF算法的幼龄沉香叶片含水率预估模型[J]. 南京林业大学学报（自然科学版）, 2022, 46(4): 177-184.
[4]	吴炅, 蒋馥根, 彭邵锋, 马开森, 陈松, 孙华. 结合树冠体积的油茶树高与产量估测研究[J]. 南京林业大学学报（自然科学版）, 2022, 46(2): 53-62.
[5]	巨一琳, 姬永杰, 黄继茂, 张王菲. 联合LiDAR和多光谱数据森林地上生物量反演研究[J]. 南京林业大学学报（自然科学版）, 2022, 46(1): 58-68.
[6]	纪烨琳, 苏喜友, 于治军. 基于随机森林模型的美国白蛾在中国的潜在生境预测[J]. 南京林业大学学报（自然科学版）, 2019, 43(6): 121-128.
[7]	赵颖慧, 张大力, 甄贞. 基于非参数分类算法和多源遥感数据的单木树种分类[J]. 南京林业大学学报（自然科学版）, 2019, 43(5): 103-112.
[8]	杜学惠,孟春,刘美爽. 基于单个特征分类准确率的特征选择方法研究[J]. 南京林业大学学报（自然科学版）, 2019, 43(04): 109-116.
[9]	张密芳,胡曼,李明阳. 基于PALSAR全极化数据的城市森林蓄积量估测[J]. 南京林业大学学报（自然科学版）, 2016, 40(06): 56-62.
[10]	刘海娟,张婷,侍昊,徐雁南,吴文龙,余佩玲. 基于RF模型的高分辨率遥感影像分类评价[J]. 南京林业大学学报（自然科学版）, 2015, 39(01): 99-103.