bitsofbits
commented
7 years ago The new SQL looks good to me (and didn't all the extra shenanigans make the resulting diffs much clearer 😁 ). I didn't inspect the resulting MMSI files, I assume that you've looked them over carefully. Feel free to merge.
Updated lists using new query to include vessels with >500 active points in a year and reporting fishing >99% of the time
I recently explored what all the vessels who have very few positions are doing. Here are some of the data explorations:
Our previous algorithm included likely fishing vessels as only those that have at least 1000 positions in a year. I found it was better to count "active positions" (positions with a speed > .1 knots), thus ignoring vessels that stayed at port, and lower the threshold to 500 positions in a year. This way we include roughly 95% of the total fishing effort.
I also found that we should include vessels that broadcast that they are fishing vessels >99 percent of the time, as opposed to the previous 100% of the time. The reason is that due to errors in broadcast, some fishing vessels have a bad message where they don't broadcast that they are fishing vessels. Including these vessels increases the total number of fishing vessels by a few percentage points.
This updated version increases the number of fishing vessels by ~10,000 per year. It is currently in the branch
version 1.0, and needs to be merged to master.https://github.com/GlobalFishingWatch/treniformis/issues/45