1982~2020年NDVI数据驱动随机森林模型模拟中国森林凋落物碳密度的研究  

NDVI-driven Model of Random Forest to Simulate Forest Litter Carbon Density over China during 1982~2020

作　　者：王昭生[1] WANG Zhaosheng(National Ecosystem Science Data Center,Key Laboratory of Ecosystem Network Observation and Modeling,Institute of Geographic Sciences and Natural Resources Research,Chinese Academy of Sciences,Beijing 100101,China)

机构地区：[1]中国科学院地理科学与资源研究所,国家生态科学数据中心,中国科学院生态系统网络观测与模拟重点实验室,北京100101

出　　处：《遥感技术与应用》2023年第6期1381-1389,共9页Remote Sensing Technology and Application

基　　金：国家自然科学基金项目(41801082);第二次青藏高原科学考察项目(2019QZKK0305)。

摘　　要：基于大量站点观测的森林凋落物碳密度数据,利用长时期遥感观测NDVI数据,驱动构建的随机森林模型,模拟评估1982~2020年中国森林凋落物碳密度的时空变化特征。凋落物碳密度与NDVI非线性的联系,初步论证了研究方法的可行性。凋落物碳密度模拟结果显著高正相关与观测值(r=0.65,p<0.001,n=4882),其较小的误差百分率(0.96%),表明随机森林模型模拟精度较高。1982~2020年凋落物碳密度的时空变化特征分析,表明中国森林凋落物碳密度整体呈显著的增加趋势(r=0.81,p<0.001),其中常绿针叶林、落叶阔叶林和落叶针叶林的凋落物碳密度增加显著。可见,长期连续的NDVI数据,驱动先进的随机森林模型,能有效模拟监测国家尺度上森林凋落物碳密度的时空动态。这为利用遥感观测数据动态监测大尺度森林凋落物提供了新的技术方法,也拓展了遥感数据的应用场景。Based on a large number of forest litter carbon density data and NDVI data,a random forest model was constructed to simulate the spatial and temporal variation characteristics of forest litterfall carbon density in China from 1982 to 2020.The nonlinear relationship between litterfall carbon density and NDVI proves the fea⁃sibility of the research method.The simulated results showed a significantly higher positive correlation with the observed values(r=0.65,P<0.001,n=4882),and a smaller error percentage(0.96%),indicating that the simulation accuracy of the random forest model was higher.The temporal and spatial variation characteristics of the carbon density of litters in China during 1982~2020 showed a significant increasing trend(r=0.81,P<0.001),and the carbon density of litters in evergreen coniferous forest,deciduous broad-leaved forest and deciduous coniferous forest increased significantly.It can be seen that long-term continuous NDVI data can ef⁃fectively simulate and monitor the temporal and spatial dynamics of forest litter carbon density at the national scale,driving advanced random forest models.This provides a new technique for dynamic monitoring of large scale forest litter using remote sensing observation data,and also expands the application scenarios of remote sensing data.

关 键 词：NDVI 随机森林 森林凋落物 模型模拟 中国 长期 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]