Data Fusion of AIS and LPMS data on the Great Lakes used to assess Maritime Transport Efficiency

This data was collected and processed as part of ongoing research to characterize waterway infrastructure performance in the Great Lakes. These dataset enable researchers to evaluate both travel time and vessel carrying capacity in the waterway.;I assembled AIS data from the MarineCadastre website for UTM Zones 15-18 for the years 2015-2017 available in csv format. I combined files for Navigation Seasons, defined as March to January and clipped data for a set of predefined features using a python code (AIS Data Processor.ipynb). The code writes the appended and clipped files to csv for a single Navigation Year. The written files are submitted here:
Trimmed_NY2015_new.csv (n=13,228,824);
Trimmed_NY2016_new.csv (n=18,782,779);
Trimmed_NY2017_new.csv (n=16,816,603);Data fusion of AIS and LPMS used the following algorithm for a subset of 30 vessels on the waterway. Let A be the original AIS data and let B be the subset of records for vessel i within geographic feature j. The script for this analysis is attached (Maritime Data Fusion.ipynb);For Connecting Channels and select segments of the Great Lakes: 1. Subset A for vessel i. Let B_i⊆A | 2. Subset B_i in geographic feature, Gj. Let B_ij⊆B_i | 3. Select tmin for each unique date or any consecutive dates, record as vessel i arrival to feature j, b_ijt | 4. IF feature j is a harbor or lock, select tmax for each unique date or any consecutive dates, record as departure from feature j, b_ijt | 5. Calculate time elapsed between features for each vessel;For vessel passage through the Soo Locks:
1. Subset A for vessel i. Let B_i⊆A | 2. Subset B_i in geographic boundaries (46.5<Lat<46.6, -84.4<Lon<-84.3). Let C_(i,lock)⊆B_i | 3. Select tmin for each unique date or any consecutive dates, record as arrival to Soo Locks | 4. Select tmax for each unique date or any consecutive dates, record as departure to Soo Locks | 5. Calculate time delta between arrival and departure times;The merged dataset is included here along with the raw LPMS data:
Merged_Data_new.csv (n=42,021),
LPMS obscured.csv (n=55,342).
VesselNames have been obscured in these datasets to protect proprietary information for shipping companies.

本数据集为五大湖航道基础设施性能表征持续研究的采集与处理成果。本数据集可供研究人员评估航道通行时长与船舶载货量。
我从MarineCadastre网站获取了2015至2017年UTM（Universal Transverse Mercator）投影分区15-18区的自动识别系统（Automatic Identification System, AIS）数据，格式为逗号分隔值（Comma-Separated Values, CSV）。我将航行季（定义为3月至次年1月）的文件进行合并，并通过Python代码（AIS Data Processor.ipynb）对预定义的地理要素进行数据裁剪。该代码会将合并并裁剪后的文件按单个航行年导出为CSV格式，本次提交的导出文件如下：
"Trimmed_NY2015_new.csv"（样本量n=13,228,824）；
"Trimmed_NY2016_new.csv"（样本量n=18,782,779）；
"Trimmed_NY2017_new.csv"（样本量n=16,816,603）。

针对航道内30艘船舶的子集，本研究采用如下算法完成AIS数据与LPMS的数据融合：设A为原始AIS数据集，B为船舶i在地理要素j内的记录子集。本分析所用脚本已随附（Maritime Data Fusion.ipynb）。

针对连接航道与五大湖部分区段的处理流程如下：
1. 针对船舶i从数据集A中提取子集，记为B_i⊆A；
2. 从B_i中筛选出位于地理要素G_j内的记录，记为B_ij⊆B_i；
3. 为每个唯一日期或连续日期选取最早时间（t_min），记录为船舶i抵达地理要素j的时刻b_ijt；
4. 若地理要素j为港口或船闸，则为每个唯一日期或连续日期选取最晚时间（t_max），记录为船舶i离开地理要素j的时刻b_ijt；
5. 计算每艘船舶在不同地理要素间的通行时长。

针对通过苏圣玛丽船闸（Soo Locks）的船舶通行场景，处理流程如下：
1. 针对船舶i从数据集A中提取子集，记为B_i⊆A；
2. 从B_i中筛选出位于地理边界（46.5°<纬度<46.6°，-84.4°<经度<-84.3°）内的记录，记为C_(i,lock)⊆B_i；
3. 为每个唯一日期或连续日期选取最早时间（t_min），记录为船舶抵达苏圣玛丽船闸的时刻；
4. 为每个唯一日期或连续日期选取最晚时间（t_max），记录为船舶驶离苏圣玛丽船闸的时刻；
5. 计算船舶抵达与驶离时刻的时间差。

本研究已将合并后的数据集与原始LPMS数据一并提交：
"Merged_Data_new.csv"（样本量n=42,021）；
"LPMS obscured.csv"（样本量n=55,342）。
为保护航运公司的专有信息，本数据集已对船舶名称进行脱敏处理。