红树林
地理
大数据
计算机科学
渔业
数据挖掘
生物
作者
Xiao Han Xiao Han,Su Fenzhen Su Fenzhen,Fu Dongjie Fu Dongjie,Yu Hao Yu Hao,Ju Chengyuan Ju Chengyuan,Pan Tingting Pan Tingting,Kang Lu Kang Lu
标识
DOI:10.11922/sciencedb.01019
摘要
Mangroves are intertidal ecosystem with high ecological value. With the increase of human activities in the coastal zone and the intensification of climate change, the distribution of mangrove in the world has declined sharply in the past half century. Accurate and efficient monitoring of global mangroves is not only the foundation of mangrove protection, but also one of the indicators reflecting the ecological environment of the coastal zone. However, the current mangrove distribution data has problems such as short time-series, imprecise description of patches, and uneven spatiotemporal accuracy. This study took the advantages of big data and cloud platforms to propose a multi-method compound mangrove extraction strategy based on big data, which follows the principle of optimal data and optimal algorithm. According to the global mangrove classification sample database based on the field studies in Southeast Asia, together with the multi-source remote sensing data and other spatial data related to mangrove growth conditions, this study integrated multiple band selection methods, pixel-based and object-oriented composite methods, and space-spectrum combined coastal topography-targeted deep learning methods to realize the 10-m global mangrove distribution classification of 2018-2020. There are 14.3484 million hectares of mangroves in the world in 2018-2020. The average overall accuracy of the product is 91.62%. Data uploaded here is mangrove distribution data in vector .shp format of 10-meter and raster .tif format of 100-meter.
科研通智能强力驱动
Strongly Powered by AbleSci AI