Automate Weather Radar Bird Roost Path Detection

I work with weather radar archives that clearly show the streams of crows leaving and returning to a communal roost. I can see the outbound and inbound streams by eye, but I now need code that does this for me. I use the bioRad package to extract weather radar reflectivity data and turn it into gridded latitude-longitude (plan position indicator, PPI) format. Starting from the PPI data, the script should evaluate different radials from the roost point, identify those radials that do and don't contain streams of birds, extract the geometry of the streams, and summarize the strength of the path, direction, distance, and temporal profile. I expect there will be various parameters I can tune to capture only stronger streams or also weaker ones, etc. See the attached graphic, which shows a few stronger and a few weaker streams. I can provide many example dates/times when the paths are clearly seen through, for instance, the RadarScope app.

Everything can run offline—batch processing is fine; real-time speed is not required. I will run this against archived radar data.

Please build the solution so that I can use it in RStudio and document the workflow so I can rerun it on new dates and another roost location with manual tuning. I am an R neophyte, FYI.