灌溉耕地空间分布制图研究进展

RECENT PROGRESS IN MAPPING IRRIGATED CROPLAND BY REMOTE SENSING

关键词:  灌溉耕地 制图 特征 方法 进展

Key words:  irrigated cropland  cartography  feature  method  progress

摘要:

［目的］灌溉耕地空间分布是管理水资源、指导农业生产和监控环境变化的基础数据，遥感是获取灌溉耕地空间分布的重要手段。科学总结有关区域灌溉耕地空间分布制图的研究进展，可为遥感在该领域的应用研究提供参考。［方法］文章全面收集近年来国内外利用遥感提取灌溉耕地的文献资料，系统分析了灌溉耕地制图的特征选择和主要技术方法，并展望了未来的发展方向。［结果］从分类特征的角度看，气候、生产方式和设施等农业相关地域条件是灌溉耕地制图的必要参考； 灌溉引起的水分或植被生长状况差异是主要依据； 引入参量的时序变化特征或新的特征参量是对单一影像信息不完善的有效补充。从分类方法的角度看，不同分类方法多局限于局地或区域尺度，规则普适性不高，真值数据收集困难，难以适应大尺度下多样的地域条件，导致无法高效及时地生产更新相应规模且合适分辨率的产品，大区域灌溉制图的主要方式仍依赖空间分配模型。［结论］未来发展而言，自动分类技术和遥感影像资源的丰富已为高效生产灌溉分布地图提供了基础条件，整合数据资源、挖掘特征参量和优化分类方法等3个方面应是未来的主要发展方向。

Abstract:

Precisely geospatial information of irrigation is essential for improving agricultural production and climate change researches，while remote sensing has been a effective way for thisThe objective of this paper was summarizing relevant studies for the guidance of future work on mapping irrigation by satelite imagesThis paper grouped the studies in recent 20 years，and concluded them from the aspects of coverage，data resources，products，and especially features and algorithms for classification.Firstly，climate，production mode and relevant installation were indispensable references for irrigated cropland extractionDifferences of water content and growth conditions which irrigation leaded were the dominant basisIntroducing of time series data or more features enriched information for classificationSecondly，various algorithms have been used in irrigation mapping researches，but most of them were in local scalesThe diversity of regional conditions and data collection problem resulted in the difficulty on generalizing the obtained rules，so as to few regional irrigation maps with moderate resolution and updated messages，as while as the dependence of global products on spatial allocation modelsBriefly，abundant satellite data and automatic classification algorithm have been prepared，for the future sake of regional or global application on dynamic monitoring and management of irrigation，more studies were needed on integrating existing remote sensing and spatial data，extending features and optimizing algorithms.