2026-10-12T01:53:26Z https://www.scidb.cn/oai

10.57760/sciencedb.43314 2026-07-16T15:55:04Z

2026-07-16 AIRCAS-BathyPhotonV1.0 doi:10.57760/sciencedb.43314 en AIRCAS-BathyPhotonV1.0 is a benchmark dataset for ICESat-2 shallow-water photon semantic classification. It is constructed from ICESat-2 ATL03 geolocated photon data and contains 1,136 beam-level along-track photon profiles with 59,268,059 photon records from global coastal, island-reef, and shallow-water regions. Each photon is manually annotated using a unified four-class labeling scheme: bathy photons, surface photons, land photons, and other photons, corresponding to labels 40, 41, 1, and 0, respectively.The dataset provides photon-level CSV files, file-level metadata, and fixed train/validation/test split files for reproducible evaluation. The metadata include region, ATL03 granule, beam, scene attributes, photon counts, class composition, spatial extent, and signal characteristics. The dataset covers diverse observation conditions, including coastal land-water transition, clear shallow water, high-background noise, deep-water weak bathymetric signal, and complex seafloor relief. It is designed to support fair comparison and robustness evaluation of ICESat-2 shallow-water photon classification methods. The associated DASNet code is available at https://github.com/zwshi-pku/DASNet. ICESat-2; photon classification; shallow-water bathymetry; bathymetric photons; photon-counting lidar; benchmark dataset; semantic segmentation; coastal remote sensing; island reefs; ATL03 史振伟 PUBLIC https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank