基于序贯高斯条件模拟的龙竹叶面积指数估测

Estimation of leaf area index for Dendrocalamus giganteus based on sequential Gaussian conditional simulation

叶面积指数(LAI)是评估森林生态系统健康状况的重要参数,为探究遥感数据在估测区域尺度森林LAI的能力,以搭载了先进地形激光测高仪系统(ATLAS)的冰、云和陆地高程卫星(ICESat-2)为主要信息源,协同陆地卫星(Landsat 9)及地形因子,结合51块实测样地数据,采用序贯高斯条件模拟(SGCS)方法与随机森林(RF)、梯度提升回归树(GBRT)、支持向量机(SVM)与K近邻(KNN)模型进行区域尺度的龙竹(Dendrocalamus giganteus)LAI遥感估测。结果表明:(1)在提取的46个ICESat-2/ATLAS光斑参数中,经Pearson相关性分析后,最佳拟合分段地形高度、插值地形表面高度、绝对平均冠层高度、太阳高度角4个特征参数与实测LAI具有显著相关性。(2)在RF、 GBRT、 SVM、 KNN模型中,多源遥感数据结合RF模型估测的LAI结果最佳,决定系数(R^(2))、均方根误差(E_(rms))、平均绝对误差(E_(ma))分别为0.901、0.352、 0.289。(3)使用SGCS方法及RF模型估测的研究区LAI为2.089~2.493,平均值为2.291。基于ICESat-2/ATLAS高密集光斑进行区域尺度LAI估测具有一定优势,光学影像数据及地形因子等辅助数据的加入,可有效提升模型估测精度,为高效率、低成本区域尺度LAI估测提供了案例,还为将ICESat-2/ATLAS数据与其他遥感影像相结合以估测其他森林结构参数提供了新的思路。

Leaf area index(LAI)is a crucial parameter in forest ecosystem studies that serves as a key indicator for assessing ecosystem development.This study utilized data from the ice,cloud,and land elevation satellite(ICESat⁃2)equipped with an advanced topographic laser altimeter system(ATLAS)in conjunction with Landsat 9 satellite imagery and terrain characteristics data.Data from 51 measured plots were incorporated,and the sequential Gaussian conditional simulation(SGCS)method was applied in tandem with machine learning models,including random forest(RF),gradient boosting regression tree(GBRT),support vector machine(SVM),and K⁃nearest neighbor(KNN),to estimate the LAI of Dendrocalamus giganteus on a regional scale.We found the following:(1)Pearson correlation analysis of the 46 extracted ICESat⁃2/ATLAS spot parameters revealed significant correlations of four specific parameters—best⁃fit segmented terrain height,interpolated terrain surface height,absolute mean canopy height,and solar altitude angle—with the measured LAI.(2)LAI estimated using the RF model,which integrated multi⁃source remote sensing data,demonstrated superior performance among the RF,GBRT,SVM,and KNN models,with a coefficient of determination(R^(2))of 0.901,a root mean square error(E_(rms))of 0.352,and a mean absolute error(E_(ma))of 0.289.(3)LAI within the study area derived from combined SGCS and RF models ranged from 2.089 to 2.493,with a mean value of 2.291.Regional⁃scale LAI estimation using ICESat⁃2/ATLAS high⁃density light spots offers distinct advantages.Incorporating auxiliary data,such as optical imagery and terrain factors,can significantly enhance model accuracy,and represents an efficient and cost⁃effective approach for regional⁃scale LAI estimations.We also introduce novel concepts for integrating ICESat⁃2/ATLAS data with other remote sensing imagery to estimate additional forest structural parameters.