FA Stats - is there a second systematic error?


#1

already pointed out that the different flight rotation plans in the us and europe combined with the new ads-b counting method sets european sites back extremely. this is well documented by european ads-b decrease of 50-60% while us decrease was only 25-30%.

now i was thinking about a second thing:

-> a day in fa-stats for all sites worldwide is a day from 0h utc to 24h utc
-> for europe this means the next day begins one hour after midnight and is very similar to the ‘real day’ over here
-> in the us the next day starts middle of the day - varies between east- and west-coast

but what does this mean for ‘aircrafts per day’?

=> in europe when over night there is nearly zero traffic -> nearly zero aircrafts are split in two
=> in the us hundrets of aircrafts count into both days

i’m not 100% shure about this thing at the moment - maybe i overlooked an aspect …


FlightAware ADS-B Statistics - Ranking
30 Day statistics/rankings
Weighted ranking?
#2

I see this in NYC. It would be the same for any location more than a few hours from UTC/GMT(inc Sydney, Singapore, HK and Japan).

A few minutes after midnight UTC and normally I have 50-100 aircraft in my totals.

It showed up even more with our big snow storm today.
I hit 800 aircraft on a day when I normally see 2000. The majority of these aircraft 80+% were in the 5 hours before midnight local time as they tried to moved people and aircraft before the storm.
The position reports slowly dropped from 4000/hour 8PM to 500/hour at midnight.

It the first hour after UTC today(I see 20 aircraft). The last two Saturdays it was over 200.

FYI, most of the aircraft I saw today were above FL300. Most were to/from Europe, via the North Atlantic, heading to parts of the U.S. not affected by the storm. One was heading to Miami,FL.

Not sure it will be fixed as it only affects rankings and individual site stats.


#3

hi jon - thanks for your reply!

in my opinion the base of good stats is reliability and comparability. if this is not the case no stats are better than wrong stats - as they lead to wrong insights and actions. that’s why fa should be interested in doing statistics the right way while spending lots of money and effort in this ‘stats thing’.

as you said - this part of the statistics affects rankings and individual site stats. but this is their main instrument to attract feeders and motivate them to deliver highest/best possible input. in addition they already (or plan to) sell data they aggregate - therefore it’s not the best recommendation to collect/hold/show screw-up data.

regarding the ‚site ranking‘ - it should simply reflect traffic/reach/uptime. as these are the main aspects every professional ads-b network would install their sites first to collect maximum traffic with minimum money. in a second step they would go for sites that deliver ground traffic on highly frequented airports. the third step then is to fill all lower frequented areas with sites with maximum reach.

having this in mind there should be three rankings with proper and comparable data:

  • best overall site (max. traffic/reach/uptime)
  • best ground view (max. ground traffic/uptime)
  • best reach (max. field of reach/uptime)
    every single site should be one location only as this is what the word ‘site’ means

anyway - who cares …

edit: wow - didn’t realize what massive snow storm is/was over there at your hometown - hope that all gets well soon!


#4

I will admit it has been “fun” to be in the top 10 overall results, but also want to make sure that it is warranted. (hence the thread I started http://discussions.flightaware.com/ads-b-flight-tracking-f21/30-day-statistics-rankings-t36560.html) It would benefit all of us to have a fair and consistent way to compare our results around the globe. I pay more attention to my local results since these feeders have a similar view of the sky.

Cheers!
LitterBug

PS That snowstorm to my east dumped 1 snowflake on my yard, while 50 miles away got 1 foot or more. It has severely dropped my stats over the weekend but did seem to increase my stats leading up to the event as planes were moved around to prepare for the storm.

#5

hi litterbug,
thanks for reply :slight_smile: i read your mentioned thread - and yes i fully agree what you wrote there and above! i think it’s simply funny to measure results worldwide - it’s a bit like a world-championship - but moreover it’s interesting to see what event brings different results in worldwide air traffic.

snow is really a weird thing - at your place nearly nothing and some miles away a huge load. over here in bavaria over the last years i had often 2 feet or more on my roof - and i thought my site would have nearly a blackout as the antenna sits under the roof. but this winter it was just one foot - and did not decrease my receiving too much :slight_smile:


#6

So, if the World stats are from 00:00UTC to 24:00UTC and here in the UK (as is with TomMuc) at 24:00UTC, we see hardly any flights during these hours. What happens around the rest of the World at 24:00UTC and the clock resets. Do the aircraft’s get counted again, as they were seen on both sides of 24:00UTC. So, in Europe. probably about 5 aircraft at this time of night, whilst in USA (-5hrs East Coast to -8hrs West Coast), RUSH HOUR and probably twice the a/c count when the 24:00UTC is reset :open_mouth:


#7

yep - that’s what i’m talking about. us sites for example with high traffic get in their rush-our 200-300 aircrafts double counted. combined with the other systematic error that’s like standing in wimbledon final and the european player has to use a 40% smaller racket. some programmers here say this is high end math - some would disagree …


#8

This does need looking at, as my (count on one hand) aircraft’s on the reset of 24:00UTC are recounted at 00:01UTC, as I’ve watched it. Never occurred to me until you mentioned it Tom. As a phrase from a horror movie goes; “ding, ding, ding” :laughing:


#9

:))) karl, the only real horror movie is what happens here in germany. but i’ve already a new ring-tone - and hopefully can make use of it soon ‘ding dong the witch is dead …’


#10

No, don’t say that. It’ll be sorted, hopefully :wink:

Updated 23:47UTC. In my receiving range of 250+ nm I am tracking 4 ADS-B A/C and 0 mlat. It’s that time of night in Europe :unamused:


#11

… though it would be simple to work out when the local midnight is for each statiom, since the longitude is known.


#12

If only timezones were that simple!


#13

http://www.nationsonline.org/oneworld/map/US-timezone_map.htm


#14

yep - all rocket-science - and sadly mr. von braun already passed away 1977 …


#15

The map was to point out some intricacies of the US Time Zone system. For the country observes Daylight Savings Time, unless your in Arizona, which doesn’t observe DST, but if your on Tribal land in Arizona, then you do observe DST. And time zones don’t fall on latitudes they fall on state boundaries (mostly, some states are split), so you can go north and south to differing time zones. I’d like to see some of the code that could map GPS co-ords to timezones…


#16

developers.google.com/maps/docu … zone/intro

but there are 1000 ways to do it better than now without need for this …


#17

That’s a good resource, but

The standard Google Maps Time Zone API supports up to 100,000 requests daily.


#18

don’t think they have more than 100,000 additional sites per day!


#19

It’s quite simple to be honest. Just write a sub-routine into the next PiAware that when it resets the PiAware.log file at midnight local, that it also sends a command to the FA Server to reset the stats. No need for this Universal UTC stuff :slight_smile:


#20

Configuration param for site similar to location…

Cheers!
Litterbug.