The Ship Detection project provides up to date information on gobal maritime activities by running an object detection algorithm on global Sentinel-1 SAR imagery. Typically, imagery is processed with the detections uploaded to the website within a day of the image being acquired.The main applications of this data include monitoring global shipping activities, commercial fishing operations, identifying offshore oil and gas operations and monitoring the size, extents and construction progress of offshore wind farms.
The data is visualized by colouring each detection according to the age of the image which observed it from red to blue. Each colour represents a 12 day period which matches the revisit period for the Sentinel-1 satellite. 10 colours are used for visualization with up to 120 day old detections being displayed.
Each colour should usually represent one image acquired over a given location. However, this isn't always the case due to ascending and descending orbits covering overlapping areas, and once Sentinel-1C joins Sentinel-1A in orbit, the revisit time over most of the europe will be reduced to 6 days.
The Ship Detections project uses Sentinel-1 satellite imagery provided by the Alaska Satellite Facility. The Sentinel-1 constellation is a fleet of radar imaging satellites operated by the European Space Agency (ESA) for the Copernicus Earth observation program.
The program downloads the Sentinel-1 imagery from the ASF using the ASF's API to search and download the most recent data in each of the subdivided processing AOIs.
Once the Sentinel-1 imagery is downloaded, the program uses the ESA's Sentinel Application Platform (SNAP) and its Graph Processing Tool (GPT) module to extract the ship locations from the radar images using it's provided object detection algorithm.
Once the detections are processed, they are extracted and formatted into a webmap compatible format and vector tiles are generated.
The website is hosted using github pages so commits are automatically made as the program works through each of the processing AOIs with the tiles generated once all AOIs have finished processing.
The Ship Detections project was created by Drew Branson. You can learn more about Drew on his website drewbranson.github.io.
Credit is also due to Barry Carter who developed the vector tiling feature which allows this website to function. He can be reached at sheeps@barrycarter.info
