GlobeLand30湿地细化分类研究

发布时间：2018-04-20 01:42

本文选题：湿地细化 + 分层分类　；参考：《测绘通报》2017年10期

【摘要】：基于30 m地表覆盖数据产品完成湿地精细化分类,能够更好地满足当前较高分辨率及较详尽全球湿地数据的应用需求。本文在深入分析湿地分类体系与细化方法的基础上,提出以湿地细化类别的定义、多元知识的分层分类、亚类数据精细化提取为主线的总体研究思路,制定了基于先验知识的对象系统筛选、基于森林数据的同位像元提取、基于最佳阈值的极大似然掩膜的主体分类方法,并应用于数据生产实践获得8个亚类信息。该方法克服了常规手段普遍存在的周期长、效率低等弊端,实现了全球较高分辨率湿地亚类数据的快速精确制图,总体分类精度达82.6%,对地理世情及其他地表覆盖研究具有借鉴意义。
[Abstract]:Based on the 30 m surface cover data product, wetland classification can be completed, which can better meet the needs of higher resolution and more detailed global wetland data application. Based on the in-depth analysis of wetland classification system and refinement methods, this paper puts forward a general research idea which takes the definition of wetland classification, the hierarchical classification of multiple knowledge, and the refinement and extraction of subclass data as the main line. An object system selection method based on prior knowledge, a method of extracting the same pixel based on forest data and a maximum likelihood masking method based on optimal threshold are developed. The method is applied to data production to obtain eight subclasses of information. This method overcomes the disadvantages of long cycle and low efficiency, and realizes the rapid and accurate mapping of global high-resolution wetland subclass data. The overall classification accuracy is 82.6, which is useful for the study of geography and other land cover.
【作者单位】：中南大学地球科学与信息物理学院;国家基础地理信息中心;湖南科技大学地理空间信息技术国家地方联合工程实验室;
【基金】：国家高技术研究发展计划(863计划)(2013AA122802)
【分类号】：P237

【相似文献】