基于Planet遥感数据的沿海滩涂提取和细分类型识别研究  

Research on Extraction and Subdivision Type Identification of Coastal Mudflat Based on Planet Remote Sensing Data

作　　者：刘连胜 温进杰 李慧珊 岳文 LIU Liansheng;WEN Jinjie;LI Huishan;YUE Wen(Guangdong Provincial Institute of Land Surveying&Planning,Guangzhou 510075,China;Key Laboratory of the Land Surface System and Human-land Relationship of the Ministry of Natural Resources,Guangzhou 510075,China;College of Natural Resources and Environment,South China Agricultural University,Guangzhou 510642,China)

机构地区：[1]广东省土地调查规划院,广东广州510075 [2]自然资源部陆表系统与人地关系重点实验室,广东广州510075 [3]华南农业大学资源环境学院,广东广州510642

出　　处：《测绘与空间地理信息》2024年第12期78-81,85,共5页Geomatics & Spatial Information Technology

基　　金：国家自然科学基金(42471340);2022年产学合作协同育人项目(220802313195721)资助。

摘　　要：根据高时间分辨率Planet遥感卫星每天过境时间筛选出5个研究区近3年最高潮位和最低潮位对应日期的遥感数据,采用U-Net深度学习算法识别高潮位和低潮位水陆分界线,提取每个研究区的滩涂范围,并通过目视解译识别沿海滩涂细分类型。5个研究区的结果表明,高时间分辨率的Planet遥感数据可以快速通过最高潮位和最低潮位的水陆分界线提取沿海滩涂范围,并识别沿海滩涂自然资源细分类型,解决了目前沿海滩涂自然资源登记细分类型单元划分的难题,为自然资源资产清查中沿海滩涂实物量量测和沿海滩涂自然资源登记提供新的技术支持。Based on the daily overpass times of high temporal resolution Planet remote sensing satellite,this paper selects imagery corresponding to the highest and lowest tides over the past three years for five study areas.The U-Net deep learning algorithm is employed to identify the land-water boundary at high and low tide and extract the mudflat extent in each study area.Coastal mudflat subdivision types are identified through visual interpretation.The results from the five study areas demonstrate that high temporal resolution Planet remote sensing data can be used to rapidly extract coastal mudflat extents by the land-water boundaries at the highest and lowest tides and identify subdivision types of coastal mudflat natural resources.This approach addresses current challenges in subdividing coastal mudflat natural resource units for registration and provides new technical support for physical measurement of coastal mudflats and natural resource registration in the inventory of natural resource assets.

关 键 词：沿海滩涂 细分类型 U-Net模型 Planet影像 高时间分辨率 

分 类 号：P237[天文地球—摄影测量与遥感]