遇见数据集

Multi-Temporal Sentinel-2 and NEON LiDAR Canopy Height Dataset for Few-Shot Semi-Supervised Domain Adaptation

收藏
Zenodo2026-01-08 更新2026-05-26 收录
官方服务:

资源简介:

This dataset supports the experiments presented in the paper: Learning Canopy Height Locally with Minimal Labels: A Semi-Supervised Domain Adaptation Approach It consists of multi-temporal Sentinel-2 Level-1C image tiles paired with LiDAR-derived canopy height models (CHMs) from the National Ecological Observatory Network (NEON). It is designed to study few-shot and semi-supervised domain adaptation for canopy height estimation under strong geographic and ecological domain shifts. Geographic Coverage The dataset includes four NEON forest sites used as target domains: CUPE – Río Cupeyes, Puerto Rico (2018) GUAN – Guanica Dry Forest Reserve, Puerto Rico (2018) WLOU – West St. Louis Creek, Colorado (2019) HOPB – Lower Hop Brook, Massachusetts (2019) Each site is represented by multiple 1 × 1 km tiles, aligned across all modalities. The accompanying CSV file (df_tiles.csv) indexes all tiles and includes: Tile identifiers and NEON site codes Acquisition year Paths to Sentinel-2 time series Paths to aligned canopy height ground-truth rasters Number of temporal observations per tile WKT geometry for spatial reference Dataset split labels

提供机构:
Zenodo
创建时间:
2026-01-08
二维码
社区交流群
二维码
科研交流群
商业服务