基于OpenStreetMap数据的农田地理信息提取

Extraction of farmland geographic information using OpenStreetMap data

关键词:  自发地理信息  OpenStreetMap  农田地理信息  信息提取

Key words:  Volunteered Geographic Information  OpenStreetMap  Farmland Geographic Information  information extraction

摘要:

目的 农田地理信息（位置、面积、空间分布等）能够直观反映一个地区的农业现状，对政府部门制定农业政策和经济计划具有重要的参考价值。该研究利用自发地理信息（Volunteered Geographic Information，VGI）数据提取农田地理信息，以降低农业信息获取成本。方法 VGI通过用户在线协作的方式，以普通手持GPS终端、开放获取的高分辨率遥感影像以及个人空间认知的地理知识为基础，实现地理信息的创建、编辑、管理和维护。文章以VGI中生成和上传方式较为简单、数据体量较大的历史OpenStreetMap（OSM）数据为基础。针对街区面积、道路线密度等进行阈值设定，从而提取农田地理信息；其次，将基于OSM数据提取的农田耕地面积与联合无人机和RapidEye影像数据提取的研究区耕地面积进行比对，确定最佳阈值；最后，基于最佳阈值提取的农田地理信息与OSM自身包含的农田地理信息叠加整合，得到基于OSM数据提取的农田地理信息。结果 针对街区面积，道路密度等各种影响因子进行阈值设定，可提高农田面积的提取精度，以无人机结合遥感影像提取的耕地面积信息为参考，利用OSM数据提取的农田信息精度较高（误差低于15%）。结论 该研究能够实现各种区域尺度的农田地理信息快速提取，可为农田面积动态变化监测和农业生产管理提供数据支撑。

Abstract:

Purpose Farmland geographic information（e.g. location，area，spatial distribution，etc.）directly reflects the status of agriculture in a region，and has important reference value for policy-making. This study uses Volunteered Geographic Information（VGI）data to extract Farmland Geographic information to reduce the cost of agricultural information acquisition.Methods VGI realizes the creation，editing，management and maintenance of geographic information based on the common hand-held GPS terminals，open-access high-resolution remote sensing images and personal spatial knowledge through online collaboration. This paper is based on the historical OpenStreetMap（OSM）data which is easy to generate and upload in VGI and has a large volume of data. In order to extract the Farmland Geographic information，the thresholds of block area and road linear density are set. Secondly，the farmland area extracted from OSM data is compared with the farmland area extracted from UAV and RapidEye images to determine the optimal thresholds. Finally，the farmland extracted from the optimal thresholds is determined. The farmland geographic information extracted from OSM data is obtained by superimposing and integrating the physical information and the farmland geographic information contained in OSM.Results Setting thresholds for block area，road density and other influencing factors can improve the precision of farmland area extraction. The precision of farmland information extracted by UAV combined with remote sensing image is higher（the error is less than 15%）.Conclusion This study can rapidly extract farmland geographic information at various scales and provide data support for monitoring farmland area dynamic change and agricultural production management.