Part of the slick mapping project we are working on for the iCRAG Marine Spoke is to develop a method for locating, classifying and quantifying slicks in Irish Coastal Waters. We use remote sensing to find the slick and measure the spatial extents - but to help with the classification element - i.e. is it a natural or anthropegenic seep, what material is it - we can really benefit from having additional contextual GIS information to combine with the spectral information in the satellite imagery. There is a wealth of GIS data that we can use, giving environmental, meteorological, geological, legal, petroleum industry, fisheries layers etc and this has all been brought into the MAROBS platform that we are developing for spatial analysis. What we do not have is free access to the automatic identification system (AIS) on vessels, so if we find a seep we cannot check that against the mapping and position of shipping at the time the slick was sighted to see if it is possible pollution, blowing tanks, etc. We can pick up ships in the Sentinel 1 SAR imagery, but it is just a snapshot and we do not know the direction of travel - so the AI would be really useful. This is particularly useful when looking at motion of slicks in the spatial tests because if a slick recurs in a single location over time, it is very possibly a seep, and not a spill.
|
Example plot with AIS data included for Dublin port approaches |
So in the course of my search for a similar, free, GIS-friendly dataset I was interested to see on the news recently that Leonardo DiCaprio launched a citizen science platform for tracking fishing vessels and monitoring illegal fishing.
|
Interpolated heatmap of fishing activity for a week in Irish coastal waters |
I signed up, followed the link and it seems that this and are other excellent examples of citizen science being undertaken by the Google Earth Outreach team - one looking at fishing and the other looking at deforestation. The goal is to have citizens help monitor fishing vessels that are fishing where they shouldn't be, or illegal logging in forests around the world. The users can also help contribute in the machine learning part of the project - helping to calibrate the algorithm, as each method has more than a simple yes/no when the feature is spotted.
More information here