Scholarlink

<p>Accurate estimation of aboveground biomass (AGB) in tropical forests is essential for carbon accounting and effective forest conservation, particularly in the biodiverse and threatened Atlantic Rainforest. Satellite-derived vegetation indices (VIs) offer potential for large-scale assessment of forest biomass and condition. We evaluated the utility of four Sentinel-2-derived vegetation indices (normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), ratio vegetation index (RVI), and optimized soil-adjusted vegetation index (OSAVI)) for classifying forest successional stages and estimating AGB in eighteen permanent plots spanning Late-, Disturbed-, and Secondary Forest stages. Vegetation index values were extracted from six cloud-free Sentinel-2 images (cloud cover &lt; 10%). Multinomial logistic regression and principal component analysis (PCA), combined with bootstrap resampling, were applied to classify forests by successional stage and biomass class and to predict plot-level AGB using field inventory data as reference. Classification accuracy for successional stages was high (90.3%), while accuracy for biomass classes was moderate (63.5%) and improved in the higher biomass class (~ 68%). Combined classification of successional stage and biomass class was less reliable (54.0%), likely reflecting high structural heterogeneity and overlap due to historical disturbance. The first principal component captured 93% of the variance in VI data but explained only 37% of the variation in AGB. These findings underscore the limitations of current VI approaches for precise AGB estimation in structurally complex tropical forests. Integrating additional remote sensing data and refined modeling techniques will be necessary to support more accurate forest conservation and management in the Atlantic Rainforest.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evaluating Sentinel‑2 vegetation indices for aboveground biomass estimation and forest management in the fragmented Atlantic Rainforest

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月24日

查看全文
  • Felipe Góes de Moraes,
  • Deicy Carolina Lozano-Sivisaca,
  • José Raimundo de Souza Passos,
  • Rafael Barroca Silva,
  • Ludmila Ribeiro Roder,
  • Antonio Ganga,
  • Gian Franco Capra,
  • Iraê Amaral Guerrini

摘要

Accurate estimation of aboveground biomass (AGB) in tropical forests is essential for carbon accounting and effective forest conservation, particularly in the biodiverse and threatened Atlantic Rainforest. Satellite-derived vegetation indices (VIs) offer potential for large-scale assessment of forest biomass and condition. We evaluated the utility of four Sentinel-2-derived vegetation indices (normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), ratio vegetation index (RVI), and optimized soil-adjusted vegetation index (OSAVI)) for classifying forest successional stages and estimating AGB in eighteen permanent plots spanning Late-, Disturbed-, and Secondary Forest stages. Vegetation index values were extracted from six cloud-free Sentinel-2 images (cloud cover < 10%). Multinomial logistic regression and principal component analysis (PCA), combined with bootstrap resampling, were applied to classify forests by successional stage and biomass class and to predict plot-level AGB using field inventory data as reference. Classification accuracy for successional stages was high (90.3%), while accuracy for biomass classes was moderate (63.5%) and improved in the higher biomass class (~ 68%). Combined classification of successional stage and biomass class was less reliable (54.0%), likely reflecting high structural heterogeneity and overlap due to historical disturbance. The first principal component captured 93% of the variance in VI data but explained only 37% of the variation in AGB. These findings underscore the limitations of current VI approaches for precise AGB estimation in structurally complex tropical forests. Integrating additional remote sensing data and refined modeling techniques will be necessary to support more accurate forest conservation and management in the Atlantic Rainforest.