Sure, I'll give you a scenario. Let's perform a search for vessels that have been inside some search box / geometry in the past 72 hours. In this case I'll do

Sure, I'll give you a scenario. Let's perform a search for vessels that have been inside some search box / geometry in the past 72 hours. In this case I'll do a search south of Africa. The search area looks like this: https://i.imgur.com/z2LTC7k.png https://i.imgur.com/z2LTC7k.png Which will return the following list of vessels: https://i.imgur.com/cKm1YlO.png https://i.imgur.com/cKm1YlO.png And if you plot the tracks for that area, there won't be much: https://i.imgur.com/Max72J4.png https://i.imgur.com/Max72J4.png Now, let's save the ships (by MMSI number), and search where those ships have been in the past 72 hours? Here's the resulting tracks: https://i.imgur.com/SqS8GH9.png https://i.imgur.com/SqS8GH9.png --- So, what happened here? 1) As you can see from the list with vessels, some of the rows only have a MMSI number - nothing more. A MMSI number is not static/constant, whereas an IMO number is - think of the IMO number as the VIN number of a vessels, it will following the ship from build to scrapping. So a big problem is that we're not just looking at one track, but every AIS message that has that MMSI number associated. And generating those can be trivial. Anyone with some AIS transponder can send out AIS messages with some MMSI number can achieve that 2) Because these AIS messages are being spammed at random, they will still appear in geometry / area searches. Usually just as one point / message, but you still have to filter them out. Now, with that said, junk like that is easier to spot - because they usually do not have much more data than some MMSI, and maybe a flag state.