Eelgrass (Z. marina) extent at sites along the Central Coast, British Columbia

This polygon dataset contains a time series for the spatial extent of eelgrass (Zostera marina) meadows delineated from aerial imagery at several long-term monitoring sites located on the Central Coast of British Columbia. The areal extent of eelgrass (Zostera marina) meadows is an important metric for understanding the broader dynamics of eelgrass habitats. Since 2015, the Hakai Geospatial Program has been using remotely piloted aerial systems (RPAS, aka drone) imagery to record change in eelgrass meadow extent at monitoring sites on the Central Coast of BC. Prior to 2015, eelgrass extent at monitoring sites has been conducted using aerial fixed wing aircraft and helicopter imagery and towed underwater video (see 2014 metadata record for eelgrass sites delineated using towed underwater video). Existing methods for mapping eelgrass (e.g. low-tide walking surveys, towed video, or boat-based SCUBA surveys) can be difficult to apply repeatedly and efficiently across large geographic extents. RPASs are portable and efficient tools for mapping nearshore marine environments, such as eelgrass habitat. Despite some limitations of RPAS mapping (e.g. weather, flight time, survey size), RPASs are able to collect high-resolution data (<10 cm) during short windows of opportunity (e.g. low tide). Core monitoring sites at the Hakai Institute include three eelgrass beds (Pruth Bay, Choked Pass and Koeye) that are mapped annually and three others (Triquet Bay, Goose SW, and McMullin North) which are mapped bi-annually (Fig. 1). Other sites have been mapped advantageously or as part of shorter-term projects and are included in this data package. For each site, eelgrass extent was delineated using a combination of object-based image analysis (eCognition Developer 9) and manual delineation within a GIS environment (ArcGIS, v 10.8). In brief, image objects area created via a multiresolution segmentation algorithm within eCognition and the site orthomosaic. These objects are exported to a GIS environment where an analyst classifies objects as eelgrass. The minimum mapping unit for a patch of eelgrass or patch within a bed is 4 m2.

本多边形数据集包含不列颠哥伦比亚省中部海岸多个长期监测点位的鳗草（Zostera marina）草甸空间范围的时间序列数据，该范围系通过航空影像解译勾绘得到。
鳗草草甸的面积范围是理解海草生境整体动态的重要指标。自2015年起，哈凯地理空间项目（Hakai Geospatial Program）采用遥控飞行系统（remotely piloted aerial systems, RPAS，又名无人机）获取的影像，记录BC省中部海岸监测点位的鳗草草甸范围变化。2015年之前，监测点位的鳗草范围则通过固定翼航空飞机、直升机影像以及拖曳式水下视频开展监测（详见2014年采用拖曳式水下视频勾绘鳗草点位的元数据记录（metadata record））。
传统的鳗草制图方法（例如低潮徒步调查、拖曳式视频或船载水肺潜水（SCUBA）调查等），难以在大面积地理范围内重复且高效地开展。遥控飞行系统是绘制近岸海洋环境（如鳗草生境）的便携高效工具。尽管RPAS制图存在一定局限（例如天气条件、飞行时长、调查规模），但它能够在短暂的适宜作业窗口（例如低潮时段）采集分辨率低于10厘米的高分辨率数据。
哈凯研究所的核心监测点位包括3个每年开展年度制图的鳗草床：普鲁斯湾（Pruth Bay）、乔克德水道（Choked Pass）和科伊湾（Koeye），另有3个点位：特里奎特湾（Triquet Bay）、西南鹅湾（Goose SW）和麦克马林北点（McMullin North），每半年开展一次制图（图1）。其他点位则根据项目需求或作为短期项目的一部分完成制图，并被纳入本数据包。
对于每个监测点位，鳗草范围通过面向对象影像分析（object-based image analysis, eCognition Developer 9）结合ArcGIS v10.8地理信息系统（GIS）环境中的人工勾绘相结合的方式确定。简言之，先通过eCognition内的多分辨率分割算法（multiresolution segmentation algorithm）生成影像对象，并结合站点正射镶嵌影像（orthomosaic）。将这些对象导出至GIS环境，由分析人员将对象分类为鳗草。鳗草斑块或草甸内斑块的最小制图单元为4平方米。