DIDSON video collection of Coastal Lake Erie Wetland, Lucas Co, Ohio in 2011

The data set includes quantitative fish abundance counts estimated from video data files collected by a Dual-frequency IDentification SONar (DIDSON) placed at the junction of a water control structure located between Pool 2B and Crane Creek (41.62133N, -83.20769W) within Ottawa National Wildlife Refuge (Lucas County, Ohio). One-hour-long video data files were used to estimate fish activity in a restored coastal wetland within Ottawa National Wildlife Refuge in the Western Basin of Lake Erie. Over 800 video hours of fish movement were recorded at the fully open fish passage structure; 136 video data files were chosen at random to represent the full video data file set and are available upon request. Fish counts from the DIDSON video data files using four different methods are provided, comparing two semi-automated and two manual counting methods. Fish tracking software Echoview paired with an "Automod" script from Milne Technologies (Keene, Ontario, Canada) as well as DIDSON viewer software were used to semi-automate the fish counting process and estimate the total number of fish passing through the structure within a video hour. Manual counts were performed by individuals counting fish from video playback given a specific set of criteria to follow for either the full hour of playback or a set of three two-minute intervals. In addition to the count data, calculated values of percent differences are included offering a comparison of the semi-automated and manual count values. Environmental data (turbidity) associated with the time periods captured in the video data files are also included.

本数据集包含基于双频识别声呐（Dual-frequency IDentification SONar, DIDSON）采集的视频数据文件估算得到的鱼类丰度定量计数结果。该声呐布设于俄亥俄州渥太华国家野生动物保护区（卢卡斯县，俄亥俄州）内2B池与克兰溪之间的水利枢纽交汇处（41.62133N, -83.20769W）。本研究采用时长为1小时的视频数据文件，对伊利湖西盆地渥太华国家野生动物保护区内一处修复型滨海湿地的鱼类活动进行评估。该开放式鱼类通行枢纽累计记录了超800小时的鱼类洄游视频，研究人员随机选取136个视频数据文件作为完整视频数据集的代表样本，该数据集可应申请获取。本数据集提供了基于DIDSON视频数据文件、采用四种不同方法得到的鱼类计数结果，涵盖两种半自动计数法与两种人工计数法的对比分析。鱼类追踪软件Echoview搭配来自加拿大安大略省基恩市米尔恩科技（Milne Technologies）的“Automod”脚本，以及DIDSON查看器软件，被用于实现鱼类计数流程的半自动化，进而估算单小时视频内通过该枢纽的鱼类总数量。人工计数环节由研究人员依据特定标准，对完整1小时的视频回放或三段2分钟的采样片段开展鱼类计数。除计数数据外，数据集还包含半自动计数与人工计数结果对比得到的百分比差异计算值。此外还收录了与视频数据采集时段对应的环境数据（浊度）。