This study addresses the scale effect problem in remote sensing estimation of forest biomass, focusing on the soft broadleaf forest in Qifeng Forest Farm, Heilongjiang Province. We innovatively constructed a scale conversion theoretical framework integrating land cover heterogeneity information. Using Sentinel-2 (10 m) and GF-2 (4 m) data, we applied multiple linear regression and random forest models for aboveground biomass (AGB) estimation and scale effect analysis. We proposed an ECv weighting method based on the regional true mean invariance principle for land cover information for developing a scale correction algorithm.The results show that: (1) GF-2 (4 m) data achieved the best estimation performance, with the random forest model performing best (R² is 0.610 1, RMSE is 26.821 4 t/hm2); (2) The land cover entropy-based scale correction method significantly improved estimation accuracy, reducing the RMSE of Sentinel-2 data from 30.798 2 t/hm2 to 23.125 5 t/hm2.This study establishes a theoretical-methodological framework for cross-scale integration of forest biomass estimation, providing important applications for regional carbon neutrality monitoring.
当前尺度校正研究存在3个关键技术瓶颈。第一,在理论层面,缺乏对异质地表辐射传输过程的精确建模,现有线性混合模型难以表征植被参数的尺度非线性响应特征[10,17];第二,在方法层面,多数研究依赖统计回归而忽视物理机制,特别是未充分考虑地类空间配置对尺度转换的影响[18];第三,在验证环节,缺乏可靠的跨尺度真值参照体系,制约了校正方法的普适性评价[19]。针对这些问题,Chen[10]通过引入土地利用类型精细分类,将叶面积指数(leaf area index,LAI)尺度误差降低了15%;王培娟等[17]则基于结构分析法,使初级生产力(net primary productivity,NPP)尺度校正精度提升至88%。这些进展为生物量尺度校正提供重要启示,但针对森林AGB的专用校正框架仍有待建立。
模型构建后采用十折交叉验证评估模型的泛化性能,即将数据集分成10份,轮流用其中9份训练模型,剩下的1份测试模型,重复10次后取平均性能指标。选取决定系数R2、均方根误差(root mean square error,RMSE,式中记为RMSE)、相对均方根误差(relative root mean square error,rRMSE,式中记为rRMSE)和平均绝对误差(mean absolute error,MAE,MAE)4种评判指标进行精度评价,公式如下。
本研究采用基于地表空间异质性特征的权重确定方法,通过引入信息熵进行权重分配,建立地类信息熵变异数权重法(entropy coefficient of variation,EVC),在此基础上,进一步构建尺度校正法,以有效降低粗分辨率数据的总体误差。该方法的核心原理在于利用各要素包含的信息确定权重,其本质是反映事物间的离散程度或者差异程度[32]。具体公式如下。
SHAOW R, DUW B, WEIC D,et al.Research on forestry carbon sequestration construction and CCER development methodology for shrub forests in northwest desert regions under carbon neutrality background[J]. Environmental Ecology,2021,3(11):100-102.
[3]
HOUGHTONR A, HALLF, GOETZS.Importance of biomass in the global carbon cycle[J].Journal of Geophysical Research Biogeosciences,2009,114(3):G00E03.
DENGM W.Development opportunities of forestry carbon sinks and corporate action choices under carbon peak and neutrality goals[J].Sustainable Development Economic Guide,2021(Z1):26-29.
ZHANGS W, HUIG Y, HANZ T,et al.Large-area forest aboveground biomass estimation based on rapid optimization of optical multispectral and SAR remote sensing features[J].Remote Sensing Technology and Application,2019,34(5):925-938.
[8]
ZHUY, LIUK, MYINTS W,et al.Integration of GF2 optical, GF3 SAR, and UAV data for estimating aboveground biomass of China's largest artificially planted mangroves[J].Remote Sensing,2020,12(12):2039.
[9]
HAYASHIM, SAIGUSAN, YAMAGATAY,et al.Regional forest biomass estimation using ICESat/GLAS spaceborne LiDAR over Borneo[J].Carbon Management,2015,6(1-2):19-33.
[10]
ZHANGX, NI-MEISTERW.Biophysical applications of satellite remote sensing[J].Remote Sensing of Forest Biomass,2014:63-98.
[11]
NIANV.The carbon neutrality of electricity generation from woody biomass and coal, a critical comparative evaluation[J].Applied Energy,2016,179:1069-1080.
ZHUX H, FENGX M, ZHAOY S,et al.Scale effects and error analysis of crop LAI in remote sensing[J].Journal of Remote Sensing,2010,14(3):579-592.
[23]
CHAMBERSJ Q, NEGRON-JUAREZR I, MARRAD M,et al. The steady-state mosaic of disturbance and succession across an old-growth Central Amazon forest landscape[J].Proceedings of the National Academy of Sciences, 2013,110(10):3949-3954.
WANGP J, XIED H, ZHANGJ H,et al.Spatial scaling methods for forest net primary productivity in Changbai Mountain Nature Reserve based on process models[J].Acta Ecologica Sinica,2007(8):3215-3223.
[28]
刘美辰.大海林林业局中国雪乡旅游营销策略研究[D].哈尔滨:黑龙江大学,2016.
[29]
LIUM C.Research on tourism marketing strategies of China Snow Town in Dahailin Forestry Bureau[D].Harbin:Heilongjiang University,Harbin: 2016.
[30]
ZHENGY, DAIQ, TUZ,et al.Guided image filtering-based pan-sharpening method: a case study of gaofen-2 imagery[J].International Journal of Geo-Information,2017,6(12):404.
[31]
KOBAYASHIN, TANIH, WANGX,et al.Crop classification using spectral indices derived from Sentinel-2A imagery[J].Journal of Information and Telecommunication,2019,4(1):1-24.
[32]
董利虎.东北林区主要树种及林分类型生物量模型研究[D].哈尔滨:东北林业大学,2015.
[33]
DongL H.Research on biomass models of major tree species and forest types in Northeast China forest region[D].Harbin:Northeast Forestry University,2015.
[34]
HEROLDM, CARTERS, AVITABILEV,et al.The role and need for space-based forest biomass-related measurements in environmental management and policy[J].Surveys in Geophysics: An International Review Journal of Geophysics and Planetary Sciences,2019,40(4):757-778.
[35]
LINX, SHIY, FANGH,et al.Vegetation carbon stocks driven by canopy density and forest age in subtropical forest ecosystems[J].Science of the Total Environment,2018,631-632:619-626.
CHIH, HUANGJ L, QIUJ,et al.Forest biomass estimation using GLAS spaceborne LiDAR and Landsat/ETM+ data[J].Science of Surveying and Mapping,2018,43(4):9-16,23.
[38]
LINC, WANGY Q, RENC Y,et al.Assessment of multi-wavelength SAR and multispectral instrument data for forest aboveground biomass mapping using random forest kriging[J].Forest Ecology and Management,2019,447:12-25.
VARGAS-LARRETAB, LÓPEZ-MARTÍNEZJ O, GON-ZÁLEZE J,et al.Assessing above-ground biomass-functional diversity relationships in temperate forests in northern Mexico-Forest Ecosystems[J].Forest Ecosystems,2021,8(1):14.
[43]
ANTÓNIOF, SASSANS, CLÉMENTM,et al.Airborne lidar estimation of aboveground forest biomass in the absence of field Inventory[J].Remote Sensing,2016,8(8):653.
[44]
JIANGJ, JIX, YAOX,et al.Evaluation of three techniques for correcting the spatial scaling bias of leaf area index[J].Remote Sensing,2018,10:22.
[45]
MASINDAM M, LIF, LIUQ,et al.Prediction model of moisture content of dead fine fuel in forest plantations on Maoer Mountain,Northeast China[J].Journal of Forestry Research,2021,32(5):2023-2035.
[46]
WHELENT, SIQUEIRAP.Coefficient of variation for use in crop area classification across multiple climates[J].International Journal of Applied Earth Observation & Geoinformation,2018,67:114-122.
CAIX Y.Estimation of aboveground biomass in natural secondary forests using multi-source remote sensing data and random forest bias correction[D].Harbin:Northeast Forestry University,2021.
YuX T.Multi-source remote sensing estimation of forest biomass based on SAR polarimetric decomposition and TM data[D].Harbin:Northeast Forestry University,2019.
[51]
SANTOROM, CARTUSO, MERMOZS,et al.The global forest above-ground biomass pool for 2010 estimated from high-resolution satellite observations[J].Earth System Science Data, 2021,13(7):3927-3950.
[52]
ZHAOP, LUD, WANGG,et al.Forest aboveground biomass estimation in Zhejiang Province using the integration of Landsat TM and ALOS PALSAR data[J].International Journal of Applied Earth Observation and Geoinformation,2019,78:345-354.
HUANGT B, OUG L, WUY,et al.Multi-source Remote Sensing Estimation of Forest Biomass Based on Machine Learning Algorithms [J]. Journal of Northwest Forestry College, 2024, 39(1): 10-18.
[55]
KAMOSKEA G, DAHLINK M, STARKS C,et al.Leaf area density from airborne LiDAR:Comparing sensors and resolutions in a temperate broadleaf forest ecosystem[J].Forest Ecology and Management,2019,433:364-375.