Can anyone interpret this data for me?

FlightAware ADS-B Flight Tracking
1

After moving my antenna to the roof a week or so ago, my feeder’s performance has improved dramatically. In fact, I noticed today on the FA Statistics page, Nearby ADS-B sites, I now “lead” the local group in POSITIONS, and that is with one day practically off line.

As you can see, I have more than twice as many positions as the feeder with the most flights, but I am about 150 flights less. Just looking at the right hand two columns, I am really cranking out the positions, but not so much, speaking relatively, flights when compared to the sites on the page.

One interpretation would be that all planes slow down when in range of my feeder, thus I get more positions, since they spend more time in range. I do not think this is the cause.

My site is on the east coast, and there are three flight paths, generally speaking, that I track. One that goes almost directly over my site, one parallel, but about 50 miles to the west, and a third, again parallel, about 130 miles to the west. Does the geography, lots of flights, closing directly on my feeder, drive up the number of positions but not the number of flights?

What I don’t want to be the case is that I am not processing or feeding data correctly, so a high percentage of positions do not correlate to specific flights. I dunno.

Any ideas appreciated. As always, thanks in advance

2

Report counts are not comparable between PiAware and Planeplotter because they use different rules to decide when to make the report.
Even with identical input, you’d get substantially different report counts.

3

Thank you. I got it.

However, there are a lot of PiAware feeders with the stats that make mine seem way out of whack.

If there is no obvious issue or problem that is fine with me.

4

I think there are some good questions here in general however. How about when comparing sites within a few miles of each other - all using the latest (or same) basic setup (SD card build, or package), one site may have ~10% more positions as seen by FA, yet their plane count is ~10% lower, or the other way around? I think this is what the OP was going for here and I’ve often wondered the same in general terms. I guess what is unclear is how the algorithm works on what passes and what doesn’t as a message; assuming here that all unique/valid ICAO’s are passed as plane count, perhaps the assumption is incorrect. The position message selection doesn’t seem as clear-cut.

5

See github.com/flightaware/dump1090 … io.c#L1790

6

Thanks mate. At first I was wondering why you supplied a link only and then after review of the current vs. history it became quite apparent that there simply is no better way to summarize or answer without writing a novel. :stuck_out_tongue:

The only somewhat simple summary I can come up with regarding the general rules is that that altitude, speed, heading, etc. needs to have shown enough change (depending on the category and/or mixture of several in some instances) to be considered a packet of interest after passing several validity tests within the tighter time restriction as apposed to older versions to be under the wire - lacking a ton of detail here obviously. So you are walking that fine line of gathering just enough data to be usable, yet efforts taken to keep the ever subtle changes out and to keep the buffer as slim as possible for lack of better terms. Props for your continuing to take this beast on since only a mountain of data and time can pave the way for efficiency. I hope that’s a decent cliff note version for those of us following with the same questions. For those of us trying to compare site to site, there are simply too may variables involved to give a clear cut answer…could be anything from target signal angle, to blocked line of site, nearby noise, etc. Am I warm here Oliver?

7

Yeah that’s exactly why I linked to the code :slight_smile: It’s a large can of worms that is hard to make generalizations about without knowing a lot about the traffic that a particular site sees. For example having a view of a high-altitude flight path will produce quite different results to having a view of a lower-altitude area used for stacking or an approach path or even an airport surface. And most sites will be a combination of all of the above in different proportions… And then the code changes from version to version.

So you can compare between sites that are nearby and have a similar field of view and are running the same version, but anything else gets tricky.

8

Thanks for the efforts to explain this. I do notice that a site farther down the coast, but situated about the same as me relative to these major flight lines.

Again, my only concern is that I was either collecting or preprocessing incorrectly. I am satisfied I am OK. Thank you again.