Half a weekend: how to read a week in which we looked less often

The week in numbers

Between Monday 12 October and Sunday 18 October our receivers logged 13,331 different aircraft near the 955 airports where we saw traffic. Counted per airport and per day, that is 20,138 sightings, against 24,265 the week before: 17 percent fewer. Before anyone reads that as a quiet week in the sky, it was not. We took 68,024 snapshots of the airspace, against 81,121 the week before, 16.1 percent fewer. Per snapshot the count barely moved: 29.6 sightings for every hundred snapshots, against 29.9, a difference of 1.0 percent. This week’s story is therefore not about traffic but about us: what our data sees, what it misses, and how to read a week in which we looked less often.

Monday to Friday look like any other autumn week: between 3,024 aircraft a day (Wednesday) and 3,425 (Tuesday 13 October, the busiest day). Then the line falls off a cliff: 1,766 on Saturday 17 October and 1,343 on Sunday 18 October, the two lowest days in the four weeks we hold in this edition. That cliff is in our snapshots, not in the sky.

The weekend we looked half as often

How we count matters here. Several times a day our collection takes a snapshot of the airspace around each airport we follow, and every aircraft in that picture counts once for that airport on that day. Fewer snapshots means fewer chances to catch an aircraft. In week 42 the number of snapshots per day eased from Wednesday onward, from 11,580 on Monday to 10,950 on Friday, and then fell sharply: 7,235 on Saturday and 4,380 on Sunday. Across the weekend we took 11,615 snapshots, against 23,076 the weekend before, 49.7 percent fewer. From Monday to Friday the drop was only 2.8 percent.

Most of the missing snapshots were never attempted. The collection planned 69,583 snapshots for the week, 14.7 percent fewer than the 81,575 of the week before, and 97.8 percent of the planned ones succeeded, against 99.4 percent a week earlier. So the gap lies mostly in how many snapshots were scheduled, and only a small part of it is snapshots that failed. The data does not tell us why fewer were scheduled, and we will not guess; it is a question for our collection, and we are looking into it. It touched the large airports alike: every one of the 60 airports with the most sightings was polled fewer times than the week before.

Why we cannot simply scale the weekend up. Per hundred snapshots, the weekend of week 42 came out higher than the weekend before, 26.8 aircraft against 25.5, and Sunday 18 October even reached 30.7, more than any weekday. That is not a busy Sunday. An aircraft counts once per day however many snapshots catch it, so when there are fewer snapshots, each one finds a larger share of aircraft that no earlier snapshot had caught. Correcting per snapshot overcorrects when snapshots fall this far. The weekly figure of 29.6 per hundred stays close to last week because only two of the seven days were hit hard: a fair summary of the week, not a precise one.

What still works is comparing shares. If every airport lost snapshots in roughly the same way, then the part of the network that a country or an airline accounts for should hold steady even while the totals fall. That is what we see, and it is how we read the rest of this week: the United States accounted for 518 of every 1,000 sightings, against 527 the week before; the United Kingdom 51 against 49; the Netherlands and Japan almost exactly the same as a week earlier. The largest moves among the bigger countries are Australia, up from 53 to 59 per 1,000, and Canada and Germany, down from 45 to 42 and from 36 to 33. Those are moves in share in a week with a gap in the measurement, so we note them and wait for a normal week before we call any of them a shift.

What we promised to watch: Saturday, Madrid and American Airlines

Last week Saturday had slipped four weeks in a row, to 77 percent of the weekday average, and we promised to check whether it would slip again. This week we cannot answer that honestly. Saturday 17 October had 1,766 aircraft from 7,235 snapshots. Per hundred snapshots that is 24.4, a little above the 23.4 of the Saturday before, but as the box above shows, that measure flatters a day with few snapshots. The Saturday question moves to next week.

We also wrote that our count around Madrid-Barajas had fallen from 201 to 38 sightings, and that we treated that first as a question about our measurement. This week Madrid came in at 51 sightings from 19 snapshots: a little more, with fewer snapshots, but still only a quarter of where it stood in early October. The question stays open, and the Spanish total should still be read with care.

The third question was whether American Airlines would hold its 13 percent rise while the other big names moved by a few percent. In share it did: 53 of every 1,000 sightings were American Airlines, against 51 the week before, and its lead over Southwest grew from 118 to 154 sightings. Delta lost the most share among the four largest American carriers, from 45 to 43 per 1,000; United gained, from 35 to 38. FedEx, at 208 sightings, rose from 9 to 10 per 1,000.

Shares also settle the fleet question we have followed since the first article. Counting the five main variants of each family, we saw 2,872 aircraft of the Airbus A320 family against 2,588 of the Boeing 737 family. That is 111 Airbus aircraft for every 100 Boeings, against 108 a week earlier, so the Airbus lead held and widened a little. The Boeing 737-800 remains the single most-seen type with 1,174 aircraft. The Cessna 172 slipped to sixth place with 460, behind the Airbus A321 (551); with half a weekend missing from our data, we read nothing into that.

The United States and the United Kingdom this week

At the 296 American airports in our list we counted 10,425 sightings from 6,712 distinct aircraft, against 12,787 the week before. We polled those airports 4,845 times instead of 5,864, so the share of the network is the better measure: the United States accounted for 518 of every 1,000 sightings, against 527. Seen from inside the country, American Airlines led with 1,036 sightings, ahead of Southwest (920), Delta (816) and United (714). JFK (572), Los Angeles (541) and Las Vegas (492) top the airport list. The Cessna 172 still holds its place among the airliners: 345 of the 460 Skyhawks we saw this week were near American airports, behind the Boeing 737-800 (486), the 737 MAX 8 (406), the Embraer 175 (398) and the Airbus A321 (352).

Across the Atlantic, the 28 British airports in our list produced 1,031 sightings from 764 distinct aircraft, against 1,184 the week before, with 489 snapshots instead of 585. In share the United Kingdom actually gained, from 49 to 51 of every 1,000 sightings. Heathrow leads with 371, Gatwick follows with 202 and Manchester with 88; Stansted (58), Edinburgh (49) and East Midlands (34) complete the top six. easyJet (142) stays ahead of British Airways (130), with Ryanair (82) and Jet2 (39) behind them. Last week’s exact tie in the British fleet is broken: we saw 118 Airbus A320s against 80 Boeing 737-800s, and the A320neo (79) is now right behind the 737-800.

What to watch next week

First our own collection: the thinner snapshots carried on into the first days of week 43, and the next article will show in full how quickly the count returned to normal. Then the two questions this week could not answer. Saturday: in a week with full snapshots, is it still below 77 percent of the weekday average? And Madrid: does our count there get back to where it was in early October? Third, Germany and Canada: whether their lower share holds once the measurement is whole again. The next article covers week 43, 19 to 25 October, the week that ends with the start of the airlines’ winter timetable.