文章摘要
赵伟丽,任 畑,陈盼芳,蒙 颖,罗德洪,罗志富.基于GF-2 影像数据的马尾松地上生物量反演研究[J].林业调查规划,2025,50(5):10-16
基于GF-2 影像数据的马尾松地上生物量反演研究
Aboveground Biomass Inversion of Pinus massoniana Based on GF-2 Image Data
  
DOI:
中文关键词: 马尾松  生物量  随机森林模型  GF-2 影像  遥感反演  植被因子
英文关键词: Pinus massoniana  biomass  random forest model  GF-2 image  remote sensing inversion  vegetation factors
基金项目:贵州省科技计划项目(黔科合支撑[2023]一般176).
作者单位
赵伟丽 贵州省第一测绘院,贵州 贵阳 550025 
任 畑 贵州省第一测绘院,贵州 贵阳 550026 
陈盼芳 贵州省第一测绘院,贵州 贵阳 550027 
蒙 颖 贵州省第一测绘院,贵州 贵阳 550027 
罗德洪 贵州省第一测绘院,贵州 贵阳 550027 
罗志富 贵州师范大学 地理与环境科学学院,贵州 贵阳 550025 
摘要点击次数: 873
全文下载次数: 511
中文摘要:
      基于GF-2影像对贵州省长顺国营林场马尾松林进行遥感生物量反演,提取多光谱波段、纹 理因子和植被指数因子,构建马尾松生物量一元回归模型、多元线性回归模型和随机森林模型。结果表明,可见光抗大气植被指数(VARI)、红绿植被指数(RGVI)、比值植被指数(RVI)、大气阻抗植被指数(ARVI)是影响生物量的主要因子;经过对比发现,随机森林模型整体上具有最好的表现,回归模型拟合度较高,R2为0.71,绝对误差为-0.45 hm2,相对误差为-0.40%;基于随机森林反演马尾松生物量平均值为112.86 t/hm2,空间上分布破碎,与实际情况相符。本研究构建的反演方法能够有效处理大面积特征植被的生物量遥感反演,提升整体精度,为合理调节林业资源分布和提升生态环境生产力提供了重要的参考方法。
英文摘要:
      Based on GF-2 images, remote sensing biomass inversion was conducted for Pinus massoniana forests in the Changshun State-Owned Forest Farm of Guizhou Province. Multispectral bands, texture factors, and vegetation index factors were extracted to construct simple linear regression models, multiple linear regression models, and random forest model for Pinus massoniana biomass. The results showed that the visible atmospherically resistant index(VARI), red-green vegetation index (RGVI), ratio vegetation index (RVI), and atmospherically resistant vegetation index(ARVI) were the main factors affecting biomass. Comparative analysis revealed that the random forest model overall demonstrated the best performance, and the regression model had high regression model fit, with R2 of 0.71, an absolute error of -0.45 t/hm2, and a relative error of -0.40%. The average aboveground biomass of Pinus massoniana inverted based on the random forest model was 112.86 t/hm2, showing a fragmented spatial distribution, which was consistent with the actual situation. The inversion method constructed in this study can effectively deal with remote sensing biomass inversion of large-area characteristic vegetation, improve overall accuracy, and provide an important reference method for rationally regulating the distribution of forestry resources and enhancing ecological environmental productivity.
查看全文   查看/发表评论  下载PDF阅读器
关闭