Tel Aviv (TLV) Traffic Drop — 31 h 52 min Without an Airline Movement Across Yom Kippur, September 2026
- A full stop, then a normal restart. No scheduled-airline movement at TLV from 11:38 UTC on 20 September to 19:30 UTC on 21 September (14:38 local Sunday to 22:30 local Monday). The evening restart carried 72 movements, in line with 66–86 in the same hours on the previous four Mondays, and the week after it was ordinary: 607 movements on Tuesday 22 September, 588 on Wednesday the 23rd, 619 on Thursday the 24th, 503 on Friday the 25th, 354 on Saturday the 26th and 580 on Sunday the 27th — Fridays and Saturdays read lower at TLV every week. Monday the 28th has filled in at 555 — 5% under our Monday median, and an ordinary Monday on the independent feed.
- Not our receivers, not the weather. Worldwide position volume was normal on both days and an independent ADS-B feed shows the same collapse; the last weather reports before the stop and the first after it were VFR — 30 °C, 9–13 kt, no thunderstorm. Larnaca and Amman, next door, ran ordinary days.
- What we cannot say from data alone: the notice. The window matches Yom Kippur, the one day a year on which Ben Gurion stands down, and the signature — clean stop, quiet FIR, silent weather station, clean restart — fits a planned stand-down rather than a disruption. Our NOTAM feed carries no closure notice for these dates, so that attribution is labeled as inference. 29 September carries a two-hour hole in our own feed and is plotted as an undercount; 30 September settled at 538. For a specific flight, check your airline; our counts are a floor, not a total.
- Closed 15 October — with one more thing in the record. The stop and the recovery are settled. From Monday 5 October TLV has run 16–27% under its September weekdays on both feeds we hold — 486, 456, 431, 513 and 352 on 5–9 October, 189 on Saturday the 10th (−47%), 488 and 437 on the 11th and 12th — while Larnaca, Amman and Haifa held their ranges. That is a slide, not a stop; we report it without a cause and follow it on the TLV airport brief.
Advisory timeline — how active is this episode
This chart is the page in one picture. It counts only the movements our receiver network observed at TLV between 12:00 and 23:59 UTC — 15:00 to 02:59 in Tel Aviv, the afternoon and evening bank. Across the 24 days from 27 August to 19 September that window never carried fewer than 149 movements (Friday 4 September, a Shabbat eve that also had a notified works closure) or more than 341; ordinary Sundays ran 325–339. On Sunday 20 September it carried 1: a non-airline round trip that had departed at 11:15 UTC and returned at 13:27 UTC. The last scheduled-airline arrival of the day was at 10:30 UTC and the last airline departure at 11:38 UTC (14:38 local). Nothing else moved until an arrival at 19:30 UTC on Monday 21 September (22:30 local) — 31 hours 52 minutes later. The Monday window then carried 72, every one of them after 19:30 UTC, against 66, 76, 72 and 86 in those same late-evening hours on the previous four Mondays. On Tuesday 22 September the window was back to 318, inside the 281–341 of the three previous Tuesdays; on Wednesday 23 September it carried 330, above the 287–303 of the three previous Wednesdays; on Thursday 24 September 333, inside the 310–334 of the three previous Thursdays; on Friday 25 September 208, inside the 149–213 of the four previous Fridays — Fridays run low in this window because the Shabbat-eve wind-down falls inside it; and on Saturday 26 September 201, inside the 158–218 of the four previous Saturdays. Both days we left off the chart at the last update have now arrived: Sunday 27 September carried 296, inside the 282–339 of the four most recent ordinary Sundays — we skip the two Sundays in between, Rosh Hashanah and Yom Kippur — and Monday 28 September 295 — we published it at 285 at the last update, ten late records have since landed in this window, and it now sits inside the 289–340 of the four most recent ordinary Mondays. 29 September is plotted at 227 and starred: our flight record for that day has a two-hour hole in it, worldwide, from 13:00 to 15:00 UTC — inside this window — and two weeks on it has not backfilled, so the figure is an undercount by construction, not a quiet Tuesday. The chart now runs to 12 October. 30 September to 4 October sat inside or next to their September bands — 291, 306, 208, 206 and 306 — and then the bank thinned: 256, 245, 232, 276, 137, 112 and 253 from Monday 5 to Sunday 11 October, each under the band its weekday carried in September (Friday the 9th under the 149 floor, Saturday the 10th under 158), then 222 on Monday 12 October, a provisional day, also starred. That is a different shape from the one this page was written about — a thinner bank every day, not an empty one — and the closing entry below sets it out. 13 and 14 October are not yet in our feed and are left off.
Fig. 1 — Observed arrivals + departures at TLV between 12:00 and 23:59 UTC (15:00–02:59 local), per UTC day, 27 Aug – 12 Oct 2026 · * partial day · Data as of 15 Oct 2026, 01:06 UTC · closed · TLV series settled through 10 Oct; 11 Oct a clipped day plotted as a recount; 12 Oct provisional; 13–14 Oct not yet collected; 29 Sep plotted as an undercount (two-hour feed hole); NOTAM set refreshed to 4 Oct · Source: FlySafe monitoring (crowdsourced ADS-B receiver network).
Airports in the exposure zone
| Airport | Size | Observed signal | FIR |
|---|---|---|---|
| Tel Aviv Ben Gurion International (TLV) | large | — · last airline departure 11:38 UTC 20 Sep · first airline arrival 19:30 UTC 21 Sep · 21 Sep 72 movements, all after 19:30 UTC (−88% vs Mon baseline 584) · 22 Sep back to 607, 23 Sep 588, 24 Sep 619, 25 Sep 503, 26 Sep 354, 27 Sep 580 · 28 Sep 555 · 29 Sep 478 (two-hour hole in our feed, undercount) · 30 Sep – 4 Oct 538, 574, 471, 356, 570 · 5–11 Oct 486, 456, 431, 513, 352, 189, 488 — 16–27% under the September weekday medians, Sat 10 Oct −47% · 12 Oct 437 (provisional) | LLTA |
| Haifa (HFA) | medium | — · 21 movements 20 Sep, last at 10:52 UTC · none recorded 21 Sep · 63 on 22 Sep, 53 on 23 Sep, 69 on 24 Sep, 45 on 25 Sep — back inside or above its 40–57 band of 14–19 Sep · 31 on Saturday 26 Sep, an ordinary Saturday there · 49 on 27 Sep, 53 on 28 Sep · 34–56 a day from 29 Sep to 12 Oct, inside or next to its 40–57 band — no October slide here | LLTA |
Distance alone does not determine impact — wind direction and the flight levels the ash reaches decide which airports and routes are affected on a given day. FIR = Flight Information Region.
What our monitoring shows: a full stop with clean edges, not a slide
FlySafe tracks observed departures and arrivals at 493 airports from crowdsourced ADS-B receiver data. Observed counts are a floor, not a total — receiver coverage varies — but an airport measured against its own baseline is a reliable shape. Tel Aviv Ben Gurion International's last 30 days:
Fig. 2 — Observed daily movements at TLV, 25 Aug – 12 Oct (latest complete UTC day). Data as of 15 Oct 2026, 01:06 UTC · closed · TLV series settled through 10 Oct; 11 Oct a clipped day plotted as a recount; 12 Oct provisional; 13–14 Oct not yet collected; 29 Sep plotted as an undercount (two-hour feed hole); NOTAM set refreshed to 4 Oct · Source: FlySafe monitoring (ADS-B receiver network).
- Baseline: 584 observed movements — the median of the four previous Mondays (24 Aug 573, 31 Aug 584, 7 Sep 570, 14 Sep 623). TLV has a strong weekly rhythm — Fridays run 405–507 and Saturdays 291–360 — so the honest comparison is same-weekday, not the rolling average.
- 21 Sep: 72 observed movements (16 departures, 56 arrivals) — −88% against the Monday baseline, −83% against the settled 7-day average of 425. The day has firmed up since we first published it at 65, as we said it might: seven late records have landed and the figure is now settled. The morning bank that carries 261–295 movements by noon UTC on a Monday carried none. All 72 fall after 19:30 UTC.
- 20 Sep: the day the stop began. 228 movements (154 departures, 74 arrivals), −62% against a Sunday median of 600 — and 227 of the 228 fall before 12:00 UTC. This day has now settled: the TLV airport brief read 177 when we first published, because a rolling refresh had clipped the first hours of the day, and it caught up to the flight-level recount of 228 on the next nightly pass, as we said it would.
- The restart ran at the normal late-evening rate. From 19:30 to 23:59 UTC on 21 September we observed 72 movements — the whole day's traffic — against 66, 76, 72 and 86 in the same hours on the previous four Mondays; 7 September carried exactly the same 72. Arrivals first (the first departure came at 20:46 UTC), then a normal night bank. No catch-up surge.
- 22 Sep: an ordinary Tuesday, not a catch-up. 607 movements (342 departures, 265 arrivals), inside the 553–609 of the four previous Tuesdays and 3% above their 590 median. The day has stopped moving, and the airport brief stores the same 607. The backlog question we left open four updates ago is answered: the network absorbed the stand-down without one.
- 23 Sep: an ordinary Wednesday too. 588 movements (333 departures, 255 arrivals), inside the 503–589 of the four previous Wednesdays and level with their 587 median. The refresh clip we flagged on this day at the last update has cleared: the airport brief now stores 588 for it too, the same figure we published from the flight-level recount.
- 24 Sep: a third ordinary day, and the busiest of the five. 619 movements (366 departures, 253 arrivals), inside the 579–621 of the four previous Thursdays and 1% above their 612 median. The refresh clip we flagged on this day at the last update has cleared: the airport brief now stores the same 619 we published from the flight-level recount, in place of the 46 it held then.
- 25 Sep: a fourth ordinary day, on a Friday's own terms. 503 movements (295 departures, 208 arrivals), inside the 405–507 of the four previous Fridays and 7% above their 470 median. TLV's Fridays sit well below its mid-week days, so this is a same-weekday comparison, not a shortfall. We published this day at 499 at the last update and four late departures have since landed. The clip we flagged on this day at the last update has cleared: the airport brief now stores the same 503, in place of the 14 it held then.
- 26 Sep: a fifth ordinary day, on a Saturday's own terms. 354 movements (191 departures, 163 arrivals), inside the 291–360 of the four previous Saturdays and 1% under their 357 median. We published this day at 347 at the last update and seven late records have since landed. Saturdays are TLV's quietest day of the week — the Shabbat wind-down — so a number near half of a mid-week day is the norm here, not a residue of the stand-down five days earlier. The airport brief stores the same 354.
- 27 Sep: a sixth ordinary day — a Sunday measured against ordinary Sundays. 580 movements (333 departures, 247 arrivals), inside the 567–656 of the four most recent ordinary Sundays and 3% under their 600 median. We compare against 16, 23 and 30 August and 6 September rather than the two Sundays immediately before, because those were Rosh Hashanah (426) and Yom Kippur itself (228) — and we say so rather than quietly picking a friendlier window. The refresh clip we flagged on this day at the last update has cleared: the airport brief now stores the same 580, in place of the 279 it held then.
- 28 Sep: a Monday a little under its band on our count, ordinary on the cross-check. 555 movements (307 departures, 248 arrivals) — 5% under the 584 median of the four most recent ordinary Mondays, and 15 movements below the bottom of their 570–623 band. We published this day at 542 at the last update and 13 late records have since landed, as we said they might; the day has now had the three daily passes our counts take to fill, so we do not expect it to move much further. The second ADS-B aggregate reads 580 for the day, inside its own 576–598 Mondays. We report the small gap on our count without reading a cause into it. The refresh clip that sat on this day at the last update (the airport brief stored 175) cleared on the next nightly pass: it now stores the same 555.
- A second ADS-B aggregate agrees, on the stop and on the recovery. The independent feed we cross-check reads TLV at 226 on 20 September and 96 on 21 September, against 582–626 on its ordinary Sundays and 576–598 on its Mondays — then 620 on 22 September, 628 on the 23rd, 639 on the 24th, 528 on Friday the 25th against 421–526 on its own recent Fridays, and 382 on Saturday the 26th against 289–368 on its own recent Saturdays — then 582 on Sunday the 27th, at the bottom of the 582–626 band of its ordinary Sundays, and 580 on Monday the 28th, inside its 576–598 Mondays. It also reads 597 for Tuesday 29 September, inside its 576–636 Tuesdays, and 572 for Wednesday 30 September, inside its 568–628 Wednesdays — which is how we knew the 464 movements we then held for the 29th, with a two-hour hole in them, and the 46 we held for the 30th were our own ingest, not a quiet airport; the 29th now holds 478, still holed, and the 30th settled at 538. This feed runs consistently a little above ours on the same day, because it aggregates a different receiver population; what matters is that the shape agrees. Two pipelines, one collapse, one clean return.
- 29 September – 12 October: one more ordinary week, then a lower October. The week to Sunday 4 October was ordinary on both feeds: 538 on Wednesday 30 September (−8% against the 587 Wednesday median), 574 on Thursday 1 October (−6% against 612, five under the 579–621 band), 471 on Friday the 2nd (level with the 470 Friday median), 356 on Saturday the 3rd (level with 357) and 570 on Sunday the 4th (−5% against 600, inside 567–656); the independent feed read 597, 572, 605, 486, 368 and 592 from Tuesday to Sunday, each inside its own band. From Monday 5 October the counts dropped and stayed down: 486, 456, 431, 513, 352, 189 and 488 from Monday to Sunday — −17%, −23%, −27%, −16%, −25%, −47% and −19% against their September same-weekday medians — then 437 on Monday 12 October (−25%, provisional). The independent feed reads 541, 502, 489, 549, 386, 216, 519, 486 and 438 for 5–13 October, every day below its September band on that feed too. Larnaca, Amman and Haifa held their ranges over the same days, and our worldwide rollup ran at or above its median, so this is Tel Aviv, not our receivers. It is a slide, not a stop — the opposite shape to the event on this page — and we report it without reading a cause into it; the closing entry below says what we hold and what we do not.
- Coverage and gap-day bookkeeping at close-out — the clip now sits on 11 October, and 29 September is plotted as the undercount it is. Our gap-day rule marks any day whose stored global rollup falls under 60% of the 14-day median (today: median 242,490, floor 145,494). The 28 September clip we named at the last update cleared on its next nightly pass: the day stores 555 for TLV and 248,730 worldwide, the recount figures, and every day from 20 September to 10 October now stores the same figure as the flight-level recount. The rolling refresh has moved onto 11 October, re-counted from a 21:25 UTC cutoff: the airport brief stores 31 movements for TLV against 488 on the recount (288 departures, 200 arrivals), and we can account for the clip exactly, because all 31 fall after that minute. The neighbours carry the same fingerprint — Larnaca 10 against 234, Amman 20 against 239, Haifa 2 against 40 — the day stores 30,043 worldwide, 12% of the median, and the airport brief marks it as a gap day. The chart plots the recount. 12 October stores 437, equal to its recount, on a worldwide rollup of 230,161 (95% of the median); it is provisional until its nightly pass on 16 October. 13 and 14 October are not measurable: our flight record was last ingested at 02:34 UTC on 14 October and holds 11,999 departures worldwide for the 13th against 104,694–115,441 on the three days before, and nothing for the 14th; TLV reads 32 for the 13th, all departures, where the independent feed reads 438. Both days are off both charts. 29 September did not backfill. Two weeks on, the 13:00 and 14:00 UTC hours still hold 795 and 727 departures worldwide against 7,827 and 7,991 a day earlier, and 8 movements at TLV against 78. The day stores 478 (277 departures, 201 arrivals) — up from the 464 we held, late records on — and we now plot that figure on both charts, marked as an undercount, rather than leave a hole in a series that is otherwise complete; the independent feed reads 597, an ordinary Tuesday. We said we would say so if the hole did not fill. It did not.
Observed counts are a floor, not a total. Live view: TLV airport brief.
What it means for flights
What fired, and why the edges matter more than the percentage. Our detection layer flagged TLV when 21 September first landed at 65 observed movements against a rolling 7-day average of 417 — a −84% traffic anomaly at Israel's main airport. On settled counts the day reads 72 movements against an average of 425, −83%, and −88% against the Monday baseline of 584; the shape is the same and the conclusion does not move. A daily percentage on its own can mean a slow erosion, a receiver hole, or a stop; the clock answers that. On Sunday 20 September the airport ran a normal night and morning — 227 movements by 12:00 UTC, inside the usual Sunday morning band — then the last scheduled-airline arrival came at 10:30 UTC and the last airline departure at 11:38 UTC. From there to 19:30 UTC on 21 September our record holds exactly one movement: a non-airline round trip that departed at 11:15 and returned at 13:27 UTC. That is 31 hours 52 minutes with no airline movement, a stop as clean as the restart that followed it. Arrivals wound down before departures, and the restart came arrivals-first — the first departure after the pause left at 20:46 UTC. Both are what a scheduled stand-down looks like: the airline network empties the field ahead of time and refills it from outside afterwards.
It was not our receivers — with the one caveat an empty sky always carries. Every traffic page we publish has to rule out its own measurement first. Here the honest starting point is that a receiver hearing nothing and a sky with nothing in it look identical in a position feed: between 12:00 UTC on 20 September and 19:00 UTC on 21 September our network heard at most one aircraft per hour over Israel and its immediate surroundings, against 15–54 distinct aircraft per hour on the comparison days of 6–7 and 13–14 September — and for most of those hours it heard none. What separates the two readings is everything around that silence. Worldwide coverage was normal: 857,471 position reports on 20 September and 878,095 on 21 September, against 826,844–925,916 on the four comparison days. A second, independent ADS-B aggregate saw the same collapse: 226 movements at TLV on 20 September and 96 on 21 September, where its ordinary Sundays run 582–626 and its Mondays 576–598. The edges are sharp and they line up with the calendar, and the first thing the network heard when the silence ended, in the 19:00 UTC hour on 21 September, was ordinary airline traffic — nine aircraft, three of them on the ground at the airport. A receiver outage does not choose a holiday to start and a holiday's end to finish, and does not report back to a second pipeline. We read this as a real stop, and we say why.
The whole FIR went quiet — FIR-level, labeled as such — and so did the weather station. This was not one airport's schedule. Our FIR-level detector for the Tel Aviv FIR — which counts aircraft across the airspace, not movements at a runway — logged a shortfall at every half-hour check from 12:15 UTC on 20 September to 18:45 UTC on 21 September: 0–8 aircraft against a same-slot expectation of 12–30, bottoming at 2 against 23 at 14:45 UTC on the 20th (−91%) and at zero against 14 at 03:15 UTC on the 21st. A handful of aircraft did remain — the FIR was quiet, not empty — and a FIR-level count is a coarser instrument than an airport count, so we label it as what it is. Haifa (HFA), the other civil airport in our feed, logged its last movement at 10:52 UTC on 20 September and nothing at all on the 21st, against 40–57 movements on the weekdays of 14–19 September — then 63 on the 22nd, 53 on the 23rd, 69 on the 24th — its busiest day of the fortnight — 45 on the 25th, 31 on Saturday the 26th, inside the 28–49 of the three Saturdays before it for which we hold usable coverage, 49 on Sunday the 27th and 53 on Monday the 28th; we hold no usable coverage for Ramon (ETM) and publish nothing for it. And the airport's own weather station fell silent: the last METAR before the stop was the 12:00 UTC report on 20 September and the next came at 11:00 UTC on 21 September — 23 hours, long enough that our station-silence monitor raised a severe alert at 14:00 UTC on the 20th and cleared it at 12:15 UTC on the 21st. That is a useful discriminator, because we have a works closure at TLV in the same record to compare against: on 4 September a notified closure (15:00–19:55 UTC) produced the same hole in movements — 3 in the 15:00 hour, none from 16:00 to 18:59, 3 in the 19:00 hour — while the station kept reporting, 11 METARs through the closure. A works closure stops the runway; this stopped the aerodrome.
It was not the weather, and neighbouring airports had an ordinary day. The last reports before the stop were CAVOK: 30 °C, wind 11–13 kt from the north-west, no significant change forecast. The first reports after the station came back on 21 September were VFR — 9,999 m visibility, a few clouds at 4,000 ft, 9–11 kt — and stayed VFR or marginal-VFR through the restart. Across the last 30 days, none of the 1,379 observations at TLV were IFR. Next door, Larnaca (LCA) ran 243 and 235 movements on 20–21 September against 237–285 on the six days before, and Amman (AMM) 282 and 238 against 249–258 — Amman's Monday sits 4% under that band, nowhere near a stand-down. Whatever stopped Tel Aviv did not touch the region. The airport brief carries no airport-level GNSS-interference ratio for TLV, so we make no claim about that layer here either way.
What the calendar and the signature suggest — labeled as inference. Yom Kippur ran from sundown on Sunday 20 September to nightfall on Monday 21 September 2026 (local time is UTC+3). Ben Gurion's stand-down for it is an annual, publicly announced practice, and the shape we measured fits it exactly: the airline network emptied the field by early afternoon on the 20th — roughly four hours before sundown — the FIR went quiet, the weather station stopped reporting, and traffic returned about three hours after nightfall, arrivals first, at the airport's normal late-evening rate. The same record shows the contrast with Rosh Hashanah a week earlier, when TLV ran reduced but never stopped: 405, 291 and 426 movements on 11–13 September against 507, 360 and 600 on the comparable Friday, Saturday and Sunday. What we have not seen is the notice itself. Our NOTAM feed carries aerodrome-closure notices for TLV — it holds the 4 September works closure and several later ones — but none for 20–21 September, so we cannot cite the operational instrument that stopped the airport. Our movement record also begins on 28 April 2026, so this is the first Yom Kippur we have measured and we have no earlier year to compare against. We therefore publish the measured stop and the fitting explanation side by side, and mark the explanation as inference from calendar and signature.
What we watched next — as written on 1 October; each open item is resolved in the closing entry below. The stop has ended, the restart looked normal — 72 movements from 19:30 UTC on 21 September against 66–86 in the same hours on the previous four Mondays — and the six days after it were ordinary: 22 September settled at 607 movements, inside the 553–609 band of the four previous Tuesdays, 23 September at 588, level with the 587 median of the four previous Wednesdays, 24 September at 619, 1% above the 612 median of the four previous Thursdays, 25 September at 503, 7% above the 470 median of the four previous Fridays, 26 September at 354, 1% under the 357 median of the four previous Saturdays, and 27 September at 580, 3% under the 600 median of the four most recent ordinary Sundays. The network absorbed the stand-down with no catch-up backlog. Monday 28 September has now filled in at 555 — up from the 542 we published, 13 late records on — which leaves it 5% under the 584 median of the four most recent ordinary Mondays and 15 movements below their 570–623 band. The independent aggregate we cross-check reads the day at 580, an ordinary Monday on its own scale; we report the small gap on our count without reading a cause into it, and will restate the figure if it moves again. The bookkeeping item we flagged last time has closed: the clip on 27 September has cleared — the day now stores 233,963 worldwide, the figure our flight-level record held all along, and the airport brief stores its 580 movements for TLV. What is open now is the same refresh artefact, one day on, and a hole of a different kind: 28 September was re-counted from a 15:25 UTC cutoff, so the airport brief stores 175 movements for TLV against 555 on the recount and 116,442 worldwide against 248,730 — 47% of the day, under the 60% floor, so this time our gap-day rule flags it. The neighbours show the identical clip (Larnaca 93 stored against 262, Amman 93 against 231), and we publish the recount. 29 September is held off the charts for another reason: our flight record for that day has a two-hour hole, worldwide, from 13:00 to 15:00 UTC, which the gap-day rule does not see because the day as a whole clears the floor. We hold 464 movements for TLV against 597 on the independent feed, an ordinary Tuesday there; we will plot the day if the hole backfills and say so if it does not. 30 September is still arriving on our side — 46 movements, all of them departures, against 572 on the independent feed, an ordinary Wednesday there. On the event itself, we will cite the operational notice if it appears in the public record we track. Automated multi-signal analysis keeps re-checking TLV every three hours alongside the Tel Aviv FIR count, weather and the regional layers; this page and the TLV brief update from the same cycle. Whether a given flight operates is decided by the airline, the airport and Israel's aviation authorities, not by this page.
What the data showed — close-out, 15 October. The number. Monday 21 September settled at 72 observed movements (16 departures, 56 arrivals), every one after 19:30 UTC — −88% against the 584 median of the four previous Mondays and −83% against the 425 seven-day average that tripped the detection layer — at the end of 31 hours 52 minutes without a scheduled-airline movement, from 11:38 UTC on Sunday 20 September. The figure has not moved since its third nightly pass. The recovery. Tuesday 22 September measured 607 observed movements (342 departures, 265 arrivals), 3% above the 590 median of the four previous Tuesdays and inside their 553–609 band — the first full day after the restart, and the number this page closes on. The six days that followed each sat inside their own weekday band: 588, 619, 503, 354, 580 and 555 from Wednesday 23 to Monday 28 September, the last 5% under its Monday median and ordinary on the independent feed (580). No catch-up surge appeared on any of them; the network emptied the field before the stop and refilled it from outside afterwards. The week to Sunday 4 October was ordinary too, on both feeds: 538, 574, 471, 356 and 570 on our count from Wednesday to Sunday (−8%, −6%, level, level and −5% against their September medians), with 29 September an undercount we explain below, and 580, 597, 572, 605, 486, 368 and 592 on the independent feed, every one inside its own September band. Open items from the last update, resolved. The refresh clip on 28 September cleared on the next nightly pass: the airport brief stores the same 555 we published, and the day's worldwide rollup stands at 248,730, the figure our flight-level record held. The two-hour hole on 29 September did not backfill: two weeks on, our worldwide record still holds 795 and 727 departures in the 13:00 and 14:00 UTC hours against 7,827 and 7,991 in the same hours a day earlier, and 8 movements at TLV against 78. We said we would say so if it did not, and we now plot the day at the 478 movements we hold (277 departures, 201 arrivals), marked as an undercount on both charts; the independent feed reads 597 for it, an ordinary Tuesday. 30 September, still arriving at the last update with 46 movements, settled at 538. What October shows — a second pattern, and not this page's. From Monday 5 October the counts changed shape. Against the September same-weekday medians this page has used throughout, our count read 486 on Monday the 5th (−17%), 456 on Tuesday the 6th (−23%), 431 on Wednesday the 7th (−27%), 513 on Thursday the 8th (−16%), 352 on Friday the 9th (−25%), 189 on Saturday the 10th (−47% against a 357 Saturday median), 488 on Sunday the 11th (−19%, recounted from raw records because our refresh clipped the stored day to 31) and 437 on Monday the 12th (−25%, provisional until 16 October). The independent feed agrees on the shape: 541, 502, 489, 549, 386, 216, 519, 486 and then 438 on Tuesday 13 October — every day from the 5th to the 13th below its September band on that feed, by 6% on Monday the 5th (541 against a 576 floor) to 25% on Saturday the 10th (216 against 289). Week on week, 5–11 October against 28 September–4 October, the independent feed is down 16% and ours 18%, with the holed Tuesday depressing our earlier week. It is not coverage, and it is not the region: our worldwide rollup ran 242,490–265,788 movements a day on 5–9 October, at or above its 14-day median; Larnaca read 218–268 and Amman 206–287 over the same days, inside their ranges of the week before (227–282 and 202–281); Haifa ran 35–56 a day, next to its 40–57 band. This is a slide spread across days, with a deep Saturday in it — the opposite shape to the clean 31-hour stop this page measured. Our FIR-level detector logged 16 half-hour shortfalls for the Tel Aviv FIR on 10 October, the most on any day since the 36 of 21 September, and 3 on the 11th, the last at 04:15 UTC; a single one followed late on 14 October. We hold no notice for any of it — our NOTAM set was last refreshed on 4 October — and we read no cause into it. We publish the numbers because they are in the record, and because a page that said the airport was back to its September counts would be wrong by omission: the data say it has run lower for ten days. The question this page opened — was the 20–21 September stop real, and did traffic come back — is answered; the October question is a different one, and the TLV airport brief carries it forward on every cycle. The other layers. The weather station that fell silent for 23 hours over the stop has reported without interruption since: from 22 September to this close-out the longest gap between its reports is 60 minutes, and none of its 1,395 observations in the last 30 days were in IFR conditions. No closure notice for the aerodrome covering 20–21 September exists in our NOTAM set, active or archived, re-checked at close-out; the set was last refreshed on 4 October and speaks for the period to that date. It does hold the next notified works closure, 15:00–19:55 UTC on 16 October — a notice of the kind the September stop never had. Our closure-events layer has carried a standing medium-severity, conflict-category entry for the Tel Aviv FIR in overlapping windows since July — 1–7 September, 10–21 September, 24 September–4 October and 7–14 October — built from notices and news signals at low confidence; it was in place before, during and after the stop, and it identifies neither the stop nor the October slide. The GNSS-interference ratio for the FIR ranged from 0.5% to 12% of observed aircraft per day between 20 September and 13 October, with no step that tracks the stop or the restart; it read 11.3% on 10 and 11 October, at the top of that range, and we list it without connecting it to the traffic. The record. Every figure from 20 September to 10 October now stands at the value its last nightly pass left; the days this page published, withdrew or restated as our feed settled are logged above, and the three bookkeeping artefacts we named — the walking refresh clip, the 29 September hole and the ingest lag on the newest day — are each resolved or labelled: 29 September an undercount, 11 October a recount, 12 October provisional, 13 and 14 October not yet collected (our flight record was last ingested at 02:34 UTC on 14 October and holds about a tenth of 13 October). What this adds up to. A full, clean stop at Israel's main airport for 31 hours 52 minutes, matching Yom Kippur to the hour, confirmed by a second feed, the FIR count, the weather station and the neighbours, with the notice itself never seen — followed by an immediate return to ordinary same-weekday counts and two ordinary weeks, then a lower October that we report and do not explain. We publish no verdict on operations at Tel Aviv; the operating picture belongs to the airport, the airlines and Israel's aviation authorities. What would reopen this page. Nothing: a new stand-down, or a day without an airline movement outside a notified closure, would be a new page with its own clock. The October counts are followed on the airport brief, and if they resolve into an event with edges we will write it up on its own terms.
Passenger questions, answered from the data
Nothing in our data indicates a continuing closure. We observed no scheduled-airline movement at TLV between 11:38 UTC on 20 September and 19:30 UTC on 21 September 2026; from 19:30 UTC traffic resumed and ran 72 movements by midnight, in line with the 66–86 we see in those hours on an ordinary Monday, and the two weeks after were ordinary — 607, 588, 619, 503, 354, 580 and 555 on 22–28 September, then 538, 574, 471, 356 and 570 on 30 September–4 October, each inside or next to its own weekday band. Since 5 October traffic has run lower than in September — 431–513 on weekdays, 189 on Saturday the 10th — on our count and on the independent feed we cross-check, with no closure notice in our set, which was last refreshed on 4 October. A lower count is not a closure, and we read no cause into it. We measure signals, not operations — for a specific flight, check your airline or the airport's own channels.
We have not seen an official notice, so we state the reason as inference. What we measured: the airport stopped from 11:38 UTC on 20 September to 19:30 UTC on 21 September, the Tel Aviv FIR went quiet (FIR-level), Haifa stopped on the same clock, the airport's weather reports paused for 23 hours, receiver coverage was normal, and the weather was VFR. That window matches Yom Kippur — sundown 20 September to nightfall 21 September — when Ben Gurion stands down every year. The signature fits a planned stand-down; the notice itself is not in our NOTAM feed.
Yes, on the count we publish: 72 observed movements on 21 September against a 425 seven-day average (−83%) and a 584 median of recent Mondays (−88%). We first published the day at 65 movements and −84%; seven late records have since landed, and the figure is now settled. A second, independent ADS-B aggregate reads 96 for the day against its own 576–598 Mondays. The day before, 20 September, reads 228 (−62% against a 600 Sunday median), with 227 of those before noon UTC. Observed counts are a floor, not a total.
The evidence says a real stop. Worldwide position volume was normal on both days (857,471 and 878,095 reports against 826,844–925,916 on comparison days), an independent ADS-B feed shows the same collapse, the stop and restart have clean edges that line up with the calendar, and the first thing our receivers heard on the restart was ordinary airline traffic. One caveat we raise ourselves: our rolling refresh clips the stored count of a recent day until its nightly pass — it did so to 27 and 28 September in turn, and each cleared the next night (28 September stores 555 for TLV and 248,730 worldwide, the recount figures); at close-out the clip sits on 11 October, which stores 31 against a 488 recount, and we publish the recount. Separately, our flight record for 29 September has a two-hour hole from 13:00 to 15:00 UTC, worldwide, that never backfilled, so that day is plotted as an undercount and marked. Neither touches the counts this page turns on: the 72 movements of 21 September all fell after 19:30 UTC and were never clipped. We are explicit that an empty sky and a silent receiver look alike in a position feed — which is why we lean on the cross-checks rather than the silence itself.
We publish measured signals, not safety verdicts — go/no-go decisions belong to airlines, airports and Israel's aviation authorities, who operate from complete data. In the layers we track, TLV weather is VFR, neighbouring airports ran ordinary days through the window, and civil traffic at TLV resumed at 19:30 UTC on 21 September at its usual evening rate.
For media — the full kit
Everything on this page — charts, CSV, quotes — is published under CC BY 4.0: free to reuse with attribution to FlySafe.zone. Charts come in dark (social/web) and light (print/light layouts) versions; attribution and license are embedded in every file.
“Tel Aviv's Ben Gurion airport logged no scheduled-airline movement for 31 hours 52 minutes, from 11:38 UTC on 20 September to 19:30 UTC on 21 September, a window in which the Tel Aviv FIR itself went quiet and the airport's weather reports paused for 23 hours; receiver coverage, a second ADS-B feed and neighbouring airports all show the stop was real and Israel-specific, and traffic restarted at the airport's normal late-evening rate, according to flight-monitoring service FlySafe.zone.”
Boilerplate — about the source (for your article footer)
FlySafe.zone monitors airspace risk and observed airport activity worldwide: 334 flight-information regions scored daily, GPS-interference signals, volcanic-ash exposure and daily observed movements at 493 airports from the crowdsourced ADS-B receiver network. Methodology: flysafe.zone/methodology.
Need a different cut of the data, a comment, or numbers for another airport? Leave an email — we answer fast during active events.

