Multi-pass annotation about nps-activespace

[dzucker7]

This stack overflow mentions a method for finding the max time gap which you could use to see if there is a time gap greater than 10 mins for example.

https://stackoverflow.com/questions/56115086/how-to-find-maximum-gap-of-observations-for-the-whole-duration-of-the-time-serie

[kirbyh]

From what I remember, the main problem is not determining whether or not an airplane trail sufficiently left a study area, it is in the continuity of data between querying the database (flightsdb). Splitting the flights (see e.g. geopandas clip) is okay for a first pass through, with the exception of saving the annotations because now one flight ID corresponds to two separate flights.

The approach (and one possible solution) here would be to decide while reading a flight ID from the database if it needs to be clipped. If the answer is yes, then one would need to assign a (new) appropriate flight ID, perhaps 5022 becomes 5022-0 and 5022-1. Then clip the track, separating into distinct linestrings based off of the times (one could determine the "split times" based off of the points that were thrown away i.e. not in the study area which would divide the time range into zones). Then each individual linestring would be interpreted as a new flight with the same process of densify -> spline interpolate -> annotate -> save.

The most immediate problem is in querying from flightsdb again, presumably to re-annotate or possibly just load and view the flights. Maybe the study area is slightly different this time so the flight gets clipped differently. Is the approach to "look up" the flight (e.g. 5022) in the annotations file and see where it was clipped before, and mirror the same transformation?

A harder problem is what happens when the study area is too big. Then a plane could conceivably perform a route that is audible, flies out of audibility, and then does a second audible pass, without leaving the study area. This use-case raises the need for an option to manually clip the tracks into two, and annotate each separately. This could certainly be done, but then the need to "look up" past annotations when querying flightsdb at a later date is even more crucial (so the clips are all made at the same place, or worse, that one flight e.g. 5022 is not saved as 5022 AND as a split 5022-0 and 5022-1).