基于多源遥感数据的森林优势树种提取  

Extraction of Dominant Forest Tree Species Based on Multi-Source Remote Sensing Data

作　　者：苏显辉 杨军 SU Xianhui;YANG Jun(Heilongjiang Forestry and Grassland Survey and Planning Institute,Harbin 150008)

机构地区：[1]黑龙江省林业和草原调查规划设计院,哈尔滨150008

出　　处：《林业勘查设计》2025年第2期78-81,共4页Forest Investigation Design

基　　金：中央财政林业科技推广示范项目(黑〔2023〕TG10)。

摘　　要：卫星遥感技术以其高空间分辨率、广泛覆盖范围和高效时效性等特点而著称,在林业领域,利用遥感影像对特定区域内的森林优势树种进行分类确定已成为研究热点。鉴于传统多光谱影像在区分光谱特征相近的树种时面临挑战,高光谱影像技术逐渐受到重视并成为研究的新趋势。研究选取哨兵2号、Landsat8和高分5号3种卫星遥感数据,采用随机森林分类方法对其森林优势树种进行分类。通过对比3种不同数据源影像分类准确性,旨在验证高光谱影像在识别森林优势树种方面的优越性,从而为森林资源调查在进行优势树种确定时提供科学依据。Satellite remote sensing technology is renowned for its high spatial resolution,extensive coverage,and efficient temporal monitoring capabilities.In the forestry sector,the classification and identification of dominant forest tree species within specific regions using remote sensing imagery have become a key research focus.Given the limitations of traditional multispectral imagery in distinguishing tree species with similar spectral characteristics,hyperspectral imaging technology has gained increasing attention and has emerged as a new research trend.This study utilizes three types of satellite remote sensing data—Sentinel-2,Landsat 8,and Gaofen-5—and applies the random forest classification method to classify dominant forest tree species.By comparing the classification accuracy of these three data sources,the study aims to validate the superiority of hyperspectral imagery in identifying dominant tree species,providing a scientific reference for forest resource survey in determining dominant species.

关 键 词：多光谱遥感 高光谱遥感 优势树种提取 随机森林 

分 类 号：S771.8[农业科学—森林工程]