在国际异质景观中的土地覆盖类型的自动绘制,使用Sentinel-2图像与辅助地理空间数据
Kristofer Lasko1, Francis D O'Neill1, Elena Sava1
1Geospatial Research Laboratory, Engineer Research and Development Center, 7701 Telegraph Road, Bldg 2592, Alexandria, VA 22315, USA.
Sensors (Basel, Switzerland)
|March 13, 2024
概括
这项研究引入了一种使用有限的卫星图像和浅层机器学习进行自动化土地覆盖分类的新方法. 该方法实现了高精度,可与深度学习相提并论,而不需要广泛的时间序列数据.
科学领域:
- 地球观测 地球观测
- 机器学习 机器学习
- 遥感 遥感 遥感 遥感
背景情况:
- 现有的土地覆盖分类方法通常需要广泛的时间序列卫星图像和复杂的模型.
- 在使用稀疏的时间数据进行土地覆盖地图的自动化培训数据生成方面存在差距.
研究的目的:
- 开发和评估使用浅层机器学习和低密度时间序列图像进行自动化土地覆盖分类的框架.
- 从有限的Sentinel-2数据生成准确的训练数据,用于土地覆盖面的绘制.
主要方法:
- 在七个国际地点利用了两次Sentinel-2图像日期 (冬季和非冬季).
- 使用光谱,纹理和距离决策函数与辅助数据相结合,创建用于训练数据生成的二进制面具.
- 应用随机森林分类器,逐步调整值,并评估全球和区域适应值,包括使用NDVI复合材料进行时间校正.
主要成果:
- 与全球值相比,区域适应值提高了九类 (73.1%),六类 (82.8%) 和五类 (85.1%) 计划的土地覆盖分类准确度.
- 暂时纠正的模型进一步提高了准确性,特别是对于五类方案 (85.1%).
- 开发的五级和六级模型的准确性与手动标记的深度学习模型 (Esri) 相当.
结论:
- 使用浅层机器学习与低密度时间序列图像实现自动化培训数据生成和土地覆盖分类的近全球框架是可行的.
- 该方法提供了准确的土地覆盖地图,而不需要完整的年度时间序列,提供了计算效率高的替代方案.
- 结果突出了浅层机器学习方法在不同地区强大的土地覆盖地图绘制方面的潜力.
相关概念视频
Selected Data About Geographic Locations
27
Geographic Information Systems (GIS) rely on two core types of data: spatial data and attribute data.Spatial DataSpatial data defines the physical location of features within a coordinate system, typically expressed in terms of latitude and longitude. It provides precise positioning for elements like roads, rivers, or buildings.Attribute DataAttribute data complements spatial data by adding descriptive information about these features. For example, a road's spatial data includes its start and...
27
Manipulation and Analysis
23
GIS manipulation and analysis functions are vital for decision-making and planning. These activities range from data retrieval tasks, such as selecting information based on specific criteria, to advanced analytical techniques that address complex spatial problems.One critical GIS analysis method is overlaying, which combines multiple data layers to examine impacts. For example, overlaying a river-dammed lake boundary with road networks can identify affected infrastructure. Another common...
23
Thematic Layering in GIS
36
In the past, planning projects such as schools or public facilities required extensive manual effort to gather and compile data. Information such as property boundaries, soil characteristics, road networks, zoning regulations, and flood zones had to be sourced individually from courthouses, utility providers, and registry offices. Assembling these datasets into a coherent format often took several months, delaying project timelines.The introduction of Geographic Information Systems (GIS)...
36
Introduction to GIS
65
Geographic Information Systems (GIS) are tools for storing, analyzing, and displaying spatial data alongside related attributes. Unlike traditional information systems that address general queries, GIS incorporates spatial components, enabling users to answer "where" and "how far." For example, GIS can process housing data linked to geographic locations like zip codes, allowing insights into population density or housing distribution through thematic maps.GIS integrates technologies such as...
65
GIS Software, Hardware, and Sources of GIS Data
59
A Geographic Information System (GIS) combines specialized software and hardware to effectively manage, analyze, and present spatial and related data. GIS software includes critical functionalities such as a user interface for easy navigation, database management tools for handling spatial and attribute data, and data retrieval features for efficient access. Analytical tools transform raw data into insights, while display functions produce maps and reports in various formats for effective...
59
Applications of GIS: Disaster Management and Emergency Response
78
Geographic Information System (GIS) technology is essential for risk identification, action prioritization, and resource optimization in critical situations like flooding and earthquakes. By integrating spatial and demographic data, GIS provides a comprehensive framework for emergency response.GIS integrates data layers, like rainfall intensity, topography, elevation profiles, and river levels, to model high-risk flood zones. These layers assess areas susceptible to flooding based on their...
78


