研究目的
To analyze the accuracy enhancement in land cover classification of two forested watersheds when using datafusion of annual time series of Sentinel-2 images complemented with low density LiDAR.
研究成果
The inclusion of LiDAR data in the classification improved the overall accuracy from 73% to 79%. This enhancement is significant for land cover classification in forested watersheds, impacting runoff estimation, forest monitoring, and flood management prediction.
研究不足
The study utilized LiDAR data with a low point density of 0.5 points/m2, which may limit the detail of vegetation structure analysis. The threshold between shrubs and trees was fixed to three meters above the ground, which might not capture all variations in vegetation height.
1：Experimental Design and Method Selection:
The study utilized a pixel-based supervised classification applying the Random Forest classifier (RF) for land cover mapping. The analysis was carried out with the Sentinel Application Platform (SNAP) software.
2：Sample Selection and Data Sources:
The study area comprised two neighboring watersheds in Badajoz province (Spain). Sentinel-2 images from April 2017 to April 2018 and LiDAR data from the Spanish National Plan of Aerial Orthophotography (PNOA) were used.
3：List of Experimental Equipment and Materials:
Sentinel-2 multispectral imagery, LiDAR data, FUSION software for LiDAR data analysis, and SNAP software for classification.
4：Experimental Procedures and Operational Workflow:
Processing of cloud-free Sentinel-2 images, calculation of vegetation and soil indices, analysis of LiDAR data to create canopy height model (CHM) and tree canopy cover factor (TCCF), and supervised classification.
5：Data Analysis Methods:
Comparison of classification accuracies derived from error matrices, including overall accuracy, user's, and producer's accuracies.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
