İstanbul (IST) Traffic 'Drop' — Both Istanbul Airports Lost Observed Coverage on the Same Day, August 2026
- Neither Istanbul airport ever showed a disruption signal in any layer we track — the ‘drop’ was our receiver network losing coverage of the Istanbul area from 17 August, not flights stopping. IST and SAW both collapsed on the same day, with opposite one-sided losses (IST missing arrivals, SAW missing departures) — the fingerprint of coverage geography, not operations. The coverage returned on 4–5 September: on 5 September IST read 428 movements (185 departures, 243 arrivals) and SAW 485 (188 / 297), with both one-sided gaps closed on the same day.
- The evidence that they were flying all along: IST’s weather station kept reporting a normal operating day on its usual half-hourly cadence throughout — 48 observations a day, every day, through 6 September — and a separate ADS-B network still saw IST at 301–765 movements a day: 765 on 28 August, 720 on the 29th, 701 on the 30th, 661 on the 31st, 604 on 1 September, 513 on the 2nd and 546 on the 3rd, on days our primary feed showed 12, 27, 7, 21, 22, 10 and 95. That gap has now closed from our side: on 5 September our count read 428 against their 355.
- While our coverage of an area is down we cannot measure its true traffic — observed counts are a floor, not a total, and even now, with IST back at 428 a day against a ~1,500/day norm, we are not claiming to have measured Istanbul. For any specific flight, check your airline as always.
Advisory timeline — how active is this episode
Through mid-August our network heard İstanbul Airport the way you would expect the world’s busiest single-terminal hub to sound: ~1,500 observed movements a day, every day, from 1 to 16 August (1,459 on the 11th, 1,646 on the 15th, 1,393 on the 16th). Then, on 17 August, the count fell off a cliff — 24 movements, then 19, 12, 12, 32, 71 on the 22nd, then 10, 9, 9, 0 on the 26th, 8 on the 27th, 12 on the 28th, 27 on the 29th, 7 on the 30th, 21 on the 31st, 22 on 1 September and 10 on 2 September — eighteen days with a median of 12 movements. Then the coverage came back. 95 on 3 September (67 departures, 28 arrivals), 556 on the 4th (272 / 284) and 428 on the 5th (185 / 243) — the last clean day, and the first since 16 August on which IST’s observed arrivals and departures are of the same order, which is what a heard airport looks like. It is still about 71% under the pre-17-August norm, so we do not call it a measurement of Istanbul’s traffic; what it measures is that we can hear the field again. The gap calls kept moving as the nightly refetch settled, and we print the corrections rather than quietly restate them — including the ones that move against us. 2 September, published here as 10, then re-settled down to 6, has settled back up to 10 (SAW 211, Antalya 1,061), and its whole-network volume now reads ~1% above median — a clean day after all. 3 September, published here as 79 with 53 departures and 26 arrivals, has settled at 95 with 67 departures and 28 arrivals (SAW 239, Antalya 965). 4 September, carried here as a collection gap at 210, has settled at 556 — but network-wide volume that day is still ~33% under median (Antalya 497 against its own ~995 median), so it stays starred and is read as a floor, not a count. The one thing that day does settle is direction: even a third-short day at IST now reads forty-six times its post-cliff median. 5 September is clean (network volume ~8% under median, Antalya 1,105 above its own median) and is the most recent complete day. 6 September is a near-total collection gap on our side (~93% under median; IST 181, SAW 144, Antalya 148), starred and not read in either direction. A real IST cannot shed 99% of its traffic overnight and leave no closure, no NOTAM and no headline — but a receiver network that loses its reach over the Istanbul area can, and then get it back, which is exactly what the corroborating layers below show. The detector tripped on the 22nd because by then its own trailing average had been dragged down to 223, so the flag itself read only about −69%; measured against the genuine pre-17-August norm, the deepest measured drop was a straight −100% (zero observed movements on 26 August, itself a normal collection day) and the last clean day, 5 September, reads about −71.4%:
Fig. 1 — Observed daily movements at İstanbul Airport (IST), 11 Aug – 6 Sep 2026 (4 Sep: network-wide short collection day; 6 Sep: near-total collection gap — both undercounts) · * partial day · Data as of 7 Sep 2026, 12:48 UTC · Source: FlySafe monitoring (crowdsourced ADS-B receiver network).
Airports in the exposure zone
| Airport | Size | Observed signal | FIR |
|---|---|---|---|
| İstanbul Airport (IST) | large | — · ~1,500/day through 16 Aug → median 12/day for 17 Aug–2 Sep, missing side arrivals; coverage back 4–5 Sep (5 Sep: 428 = 185 dep / 243 arr, still ~71% under norm) | LTBB |
| Istanbul Sabiha Gökçen (SAW) | large | — · ~1,000/day through 16 Aug → median 244/day for 17 Aug–3 Sep, missing side departures; coverage back 5 Sep (485 = 188 dep / 297 arr, still ~51% under norm) | LTBB |
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 showed: coverage lost, not traffic — two airports at once, and both back
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. İstanbul Airport's last 30 days:
Fig. 2 — Observed daily movements at IST, 25 Jul – 5 Sept (latest complete UTC day). Data as of 7 Sep 2026, 12:48 UTC · Source: FlySafe monitoring (ADS-B receiver network).
- Pre-collapse baseline: ~1,500 observed movements/day at IST through 1–16 August (average 1,499), with no day below 1,385. This is the honest baseline; the detector’s own trailing figure had been eroded to 223 by the collapsed days, which is why the 22 August flag reads −69% rather than the true ~−95%.
- From 17 August the count collapsed: 24, 19, 12, 12, 32, then 71 on the 22nd, then 10, 9, 9, 0 on the 26th, 8 on the 27th, 12 on the 28th, 27 on the 29th, 7 on the 30th, 21 on the 31st, 22 on 1 September and 10 on 2 September — eighteen consecutive measured days with a median of 12 movements, a ~95–100% loss against the real norm on every one of them.
- The decisive tell was that it happened at two airports at once. Sabiha Gökçen (SAW), 64 km across the city in the same Istanbul FIR (LTBB), collapsed on the exact same day — from ~1,000 movements/day (955 on 16 Aug, ~993/day average for 1–16 August) to 181, 148, 248, 253 and 238 through the 21st, then 265, 248, 232 and 226 through the 25th, and 105, 248, 296, 302, 265, 229, 253, 211 and 239 on 26 August through 3 September (median 244). Two of the world’s busiest airports do not lose 70–100% of their traffic on the same morning by coincidence of operations; they do when the receiver coverage they share degrades.
- Closing signal: the coverage came back on 4–5 September, at both fields. On 5 September — the most recent clean collection day — IST read 428 movements (185 departures, 243 arrivals) and SAW read 485 (188 departures, 297 arrivals): 36× and 2× their respective post-cliff medians, on a day when our en-route count for the whole FIR sat at 272 against a trailing average of ~252. IST is still ~71% under its pre-17-August norm and SAW ~51% under, so this is a statement about hearing the airports again, not a measurement of what they flew.
- The asymmetries were opposite — and both have closed. Through the window the missing side at IST was arrivals (2 Sep: 10 departures against 0 arrivals; 31 Aug: 18 against 3; 29 Aug: 27 against 0; 20 Aug: 11 against 1; 17 Aug: 20 against 4), while at SAW it was departures (3 Sep: 8 departures against 231 arrivals; 31 Aug: 2 against 227; 29 Aug: 17 against 285; 20 Aug: 6 against 247; 17 Aug: 10 against 171). No single operational cause can suppress arrivals at one airport and departures at another 64 km away — but two receiver footprints losing different close-in geometry can. On 5 September both sides are back in proportion at the same time: IST 185/243 and SAW 188/297, against SAW departures that had not cleared 22 in a day for eighteen days.
- A real schedule is a conserved system: over a full day an airport’s arrivals and departures balance to within a handful of positioning flights. IST landing 1 and launching 11, while SAW lands 247 and launches 6 (both on 20 August), is not a schedule — it is what half-heard airspace looks like on each side. The balance returning at both fields on the same two days, after nineteen days of one-sided counts, is the clearest evidence in this dataset that what changed was the listening, not the flying.
- Known gaps and corrections, including the ones that move against us: the labels kept moving as the nightly refetch settled, in both directions, and this is the final set. 2 September — published here first at IST 10, then re-settled down to IST 6 — has settled back at IST 10 (SAW 211, Antalya 1,061) on a day whose whole-network volume now reads ~1% above median, so it is a clean day, and the “~17% under median, read it as a floor” caveat we published for it is withdrawn. 3 September, published here as IST 79 with 53 departures and 26 arrivals, has settled at IST 95 with 67 departures and 28 arrivals (SAW 239, Antalya 965; network volume at median). 4 September, carried here as a collection gap at IST 210, has settled at IST 556 (SAW 190, Antalya 497) — but network-wide volume that day is still ~33% under median and Antalya is at half its own median, so it stays starred and is read as a floor. 5 September is clean (~8% under median network-wide; Antalya 1,105, above its ~995 median) and is the most recent complete day. 6 September is a near-total collection gap on our side (~93% under median; IST 181, SAW 144, Antalya 148) and is starred and unread — the recovery in this brief rests on 5 September, not on it. 26 August remains a normal collection day, so IST’s 0 that day is a genuine observed zero rather than under-collection.
Observed counts are a floor, not a total, and they are only meaningful while receiver coverage holds. Over the Istanbul area that assumption broke on 17 August and was restored on 4–5 September; the cross-checks below are how we know it was coverage, not the airports. Live view: IST airport brief.
What it means for flights
What fired, and why we publish it anyway. Our detection layer flagged IST when the 22 August observed count fell far below its trailing average of 223 — the flag read −69%, a traffic anomaly at one of the world's busiest airports. That figure understates it: the trailing average was itself contaminated by the collapsed days, and against IST's genuine pre-17-August norm of ~1,500 movements the drop is closer to 95%. Automated multi-signal analysis then did what it always does — checked the anomaly against every other layer we track: closure signals, weather-reporting continuity, GNSS interference, a second independent receiver network, neighboring airports, and the composition of the count itself. The layers do not corroborate an airport disruption; they corroborate a coverage failure. We publish the evidence for that, as we did for the Monterrey coverage collapse and the Palermo departure anomaly — publishing observed counts honestly means publishing their failure modes too.
Two airports, same morning, opposite gaps. The single most decisive fact is not the size of the drop but that it hit two airports at once. İstanbul Airport and Sabiha Gökçen sit 64 km apart on opposite sides of the city, in the same Istanbul FIR (LTBB), and both fell off on 17 August — IST from ~1,500 movements/day to a handful, SAW from ~1,000 to a few hundred. Two of the busiest airports on earth losing 70–100% of their traffic on the same morning would be one of the largest aviation events of the year; there was no closure, no NOTAM and no news, because it did not happen in the air. And the shapes contradict any shared operational cause: at IST the missing side is arrivals (no arrivals at all against 6 departures on 2 September; 1 against 11 on 20 August), while at SAW the missing side is departures (8 departures, 231 arrivals on 3 September; 6 against 247 on 20 August). One weather system, one strike, one airspace restriction cannot suppress arrivals at one field and departures at another — but two receiver footprints, each losing a different slice of close-in, low-altitude reception, can. A departure is heard first on the runway and in initial climb, the shortest-range signal there is; an arrival descends along a predictable profile still catchable by distant receivers. Which side vanishes depends on which antennas you lost — and the two airports lost different ones. The recovery reads the same way in reverse, and that is what closes this brief. The sides came back in the order the geometry predicts — IST’s arrivals moved first, 28 against 67 departures on 3 September — and then both fields squared up together on 4–5 September: IST 272/284 and 185/243, SAW 59/131 and 188/297, with SAW’s departures clearing 22 in a day for the first time since 16 August. An operational cause that had suppressed opposite sides at two airports would not resolve both sides at both airports over the same forty-eight hours; restored reception does exactly that.
The airports were flying all along — here is how we know, and how the instruments converged. Two independent lines of evidence showed İstanbul operating normally through the very days our count read near-zero. First, its weather station never missed a beat: 48 METAR observations a day, every day, straight through 17 August–6 September, and on the same cadence again on 7 September (26 observations by 12:20 UTC) — a fully staffed, fully operating airport reporting on its normal half-hourly cadence, with IFR conditions just ~1% of the last 30 days and the latest report (7 September, 12:20 UTC) reading 10 km or more visibility with a scattered cloud layer at 3,300 ft, a 20-knot wind and no significant weather — an ordinary VFR afternoon, no storm flags, on the day we close this brief. Second, a separate crowdsourced ADS-B network, sharing none of our receiver hosts, still saw IST at 301–765 movements a day across 17 August–3 September — a softer dip in its own coverage in the middle of the window (a low of 301 on the 23rd), then 765 movements on 28 August, 720 on the 29th, 701 on the 30th, 661 on the 31st, 604 on 1 September, 513 on the 2nd and 546 on the 3rd, inside its own pre-17-August band (362–762 across 10–16 August), on days our feed read 12, 27, 7, 21, 22, 10 and 95. At its worst, 513 against our 10 on 2 September, the two instruments disagreed fifty-fold about the same runways. That disagreement has now closed, and closed past parity: on 4 September our count read 556 against their 504, and on 5 September 428 against their 355 — the first days since the cliff on which our feed heard more of IST than theirs did. (Their 699 on 6 September against our 181 is our own collection gap, not a reopening of the split.) The airport always had one reality; the gap between the two instruments was the instruments’ problem, and it resolved on the instrument that had broken.
The loss follows receiver geography, not Turkish airspace. If this were an airspace event it would show across the region; it does not. On our own primary feed, Antalya (AYT) on the Mediterranean coast kept running near its norm through the window — roughly 766–1,225 movements a day on clean collection days (842 on 17 Aug, 1,225 on 22 Aug, 965 on 3 Sep and 1,105 on 5 Sep against its own 30-day median of ~995) — while the two Istanbul fields were flat on the floor. A decline that draws a line around the Istanbul area specifically, and spares airports fed by other receivers, is a statement about coverage, not about LTBB operations. The FIR-level caveat we published on 29 August has resolved in the same direction, and we close it plainly: our observed en-route aircraft count for the whole Istanbul FIR did soften with the airports — a daily average of ~292 aircraft on 16 August falling to ~233–252 across 17–25 August — but it climbed back and has stayed back, reading 254 on 28 August, 270 on the 29th, 256 on the 30th, 246 on the 31st, 234 on 1 September, 241 on the 2nd, 244 on the 3rd, 262 on the 4th, 272 on the 5th and 279 on the 6th — its post-cliff high, against a trailing average of ~253. Worth naming: that count held its full 95–96 snapshots a day straight through 6 September, including the day our flight-detection feed collected almost nothing — the starred gaps on the chart above are in that feed, not in this one, which is itself a small proof that the two are independent. So airspace-wide visibility of LTBB was back at its own norm about a week before the terminal-area count at IST followed, which is exactly the sequence a loss and recovery of close-in receivers around one field produces — and it is a statement about our visibility, not a claim about how much traffic the FIR actually carried.
One honest ambiguity: the GNSS reading — now closed by the instrument that raised it. When we first published this page we flagged a possible rise in the GNSS-interference ratio for the Istanbul FIR around 17 August and refused to call it a jamming surge, because the aircraft sample the ratio is built from collapsed on the same day. The recovery has now settled the question, and it settled it against the earlier hint. The instrument that produced the apparent surge reversed it exactly as its sample came back. The denser of our two telemetry reads stepped from a FIR ratio of 0.058–0.085 on the five full days before the cliff (sample ~83–89 aircraft per snapshot) to 0.145–0.179 on every day from 17 August to 3 September (sample ~52–60) — and then, as the coverage returned, fell straight back through it: 0.106 on 4 September at ~71 aircraft, 0.085 on the 5th at ~68, 0.065 on the 6th at ~77, and 0.077 so far on the 7th at ~73 — back inside its own pre-17-August band, on a full 48 snapshots a day throughout. A ratio that rises when the sample thins and falls when the sample returns is measuring the sample, not Istanbul. The area-wide interference measurement, built independently across all 141 map cells of the FIR, agrees and never moved at the cliff: 0.016–0.074 across the window, 0.038 on the latest measured day (6 September), and a 7-day figure for LTBB of 0.057 — moderate, and mildly up from 0.044 the previous seven days, with no step at either the 17 August collapse or the 4–5 September recovery. The sparser telemetry read says the same thing from the other end: its per-snapshot aircraft count fell from ~33–37 to ~11–14, and the only high readings it produced — 0.10 on 30 August, 0.22 on the 31st, 0.67 on 1 September and 0.13 on the 2nd — rest on 2, 2, 1 and 2 snapshots respectively, seen by as few as 3 aircraft; that collector has produced no rows at all since 2 September, so it contributes nothing to the close and we say so. We read no GNSS event over Istanbul across this window, and we would rather print the correction than leave the earlier hint standing.
About the low-confidence closure label on LTBB — what it still says at close. On 27 August our consensus labeler opened a low-confidence (0.18) closure candidate for the Istanbul FIR. It is still open as we close this brief, last updated 7 September, and we publish it here rather than let a reader find it elsewhere and think we hid it. It rests on exactly two signals, and both are readable. The first is our own traffic anomaly — a thin one: a single snapshot at 00:45 UTC on 27 August that read zero aircraft over the whole FIR, sitting between readings of 77 and 84 fifteen minutes either side. Our live-closure tracker resolved that flag as unconfirmed at 02:30 UTC the same morning, and the day it fired on has since settled as a normal collection day. The same artefact recurred once on 4 September at 12:15 UTC — a lone zero between ordinary readings — and opened no new candidate; there has been no third occurrence since, and no new closure candidate has been opened for LTBB at any point in this window. The second signal is three NOTAMs, and the set has kept rotating as items expire — which is worth showing, because it demonstrates that the label is not tracking anything that happened on 17 August. The SAR exercise in the Izmir TMA that the set cited a week ago has expired and dropped out. Today the same 0.18 candidate cites LTFJ-B3199/26 — one taxiway (TWY C) closed at Sabiha Gökçen since 12 August, to 28 October — a permanent danger area near Osmaneli, surface to 3,000 ft, in force since September 2025, and one further item that is no longer in the active NOTAM set at all. Alongside them, LTFM-B3129/26 and LTFM-B3130/26 still close one runway (RWY 09) and part of one taxiway at İstanbul Airport, both since 7 August and both running to 15 September. Every one of these predates the 17 August cliff, none closes an airport, and a rotating cast of taxiway, apron and danger-area items is ordinary paperwork at hubs of this size. Nothing here is an airport or airspace closure, so the framing of this page does not change — but the label exists in our data, it is open at close, and this is what is in it.
What the data showed — the closing number. This brief closes on the recovery of our own coverage, not on a claim about Istanbul’s traffic. The closing measurement: across the eighteen measured days from 17 August to 3 September, IST’s observed count ran at a median of 12 movements a day — 0.8% of its 1,499/day pre-cliff baseline, bottoming at a straight zero on 26 August, itself a normal collection day. On 5 September, the first clean day after the coverage returned, it read 428 (185 departures, 243 arrivals): 36× the post-cliff median, and 29% of baseline. Sabiha Gökçen tracks it one-for-one: a median of 244 a day through the same window against its ~993/day baseline, then 485 (188 departures, 297 arrivals) on 5 September, with departures clearing 22 in a day for the first time since 16 August. Four independent layers turned at the same time, which is what makes this a coverage story rather than an operational one: the one-sided gaps closed at both fields on the same two days; our own count overtook the second, unrelated ADS-B network (556 v 504 on 4 September, 428 v 355 on the 5th) after weeks of reading up to fifty times lower; the denser GNSS telemetry ratio fell from 0.145–0.179 back to 0.065–0.106 in lockstep with its aircraft sample recovering from ~52–60 to ~68–77 per snapshot; and the FIR-wide en-route count, which had returned first, sat at a post-cliff high of 279 on 6 September against a ~253 trailing average. Meanwhile the layers that would have shown a real disruption never moved at all: no airport or airspace closure at any point, 48 METAR observations a day at IST without a single missed cycle from 17 August to 6 September, and Antalya running at its own median on the same feed throughout. In one line: a nineteen-day loss of crowdsourced receiver coverage over one metropolitan area, measured end to end, corroborated as a measurement failure by every cross-check we have, and reversed at both airports over the same forty-eight hours — with no disruption found at either field on any day. What we never obtained is the cause on the receiver side: we do not know why the Istanbul-area coverage degraded on 17 August or what restored it, and that gap goes into the archive with the brief.
What we cannot say, and what would reopen this page. Even with coverage back, we cannot tell you what IST or SAW actually flew on any day in this window — observed counts are a floor, not a total, and we will not translate a coverage hole into a claim that either airport was quiet, disrupted, or normal; the operating picture on any given day belongs to the airports, the airlines and the Turkish authorities. That applies to the closing number as much as to the collapse: 5 September’s 428 movements at IST are 29% of the pre-cliff baseline, so our coverage of the field is demonstrably back but not demonstrably whole, and the sensible reading is that we can hear IST again — not that we have measured it. Nor do we know what caused the receiver loss or the return; we never found that, and we are not going to invent it. The blind spots we carried through this brief stand at close. The last two or three days of this feed are provisional in both directions, and this page corrected itself repeatedly in public: 2 September moved 10 → 6 → 10, and its “~17% under median, read it as a floor” caveat is withdrawn — the day settled clean; 3 September moved 19 → 79 → 95 and its split from 53/26 to 67/28; 4 September moved 210 → 556 and stays starred anyway, because network-wide collection that day was ~33% under median. 6 September is a near-total collection gap on our side (~93% under median) and is excluded from every claim here — the recovery rests on 5 September alone, one clean day, which is why we describe it as coverage returning rather than as a trend. Sabiha Gökçen has no weather feed in our data, so the METAR-continuity check ran on IST only; the second-network cross-check is meaningful for IST but not for SAW, which that network barely sees at all (0–3 movements a day there on almost every day of the window, with two unexplained spikes of 20 and 81); our sparser GNSS collector produced no rows at all after 2 September; and the low-confidence closure candidate on LTBB is still open in our data, resting on the readable signals set out above. Two things would reopen this page: a genuine airport or airspace closure for Istanbul appearing retrospectively in any layer — not the partial-facility NOTAMs described above, which close a runway, taxiways and an apron but not either field — which would make this a disruption event and change the framing; or a relapse, our Istanbul-area counts falling back toward the floor, which would turn a one-off coverage loss into a pattern and is the fastest thing to check. Automated multi-signal analysis keeps re-checking every layer — movements, closures, weather continuity, interference, the second network — every three hours, and the IST brief keeps updating from the same cycle; this page stops here unless one of those two things appears. The standing lesson it leaves behind: when two airports drop on the same day with opposite gaps, suspect the receivers before the runways — and when they come back on the same day with both gaps closing, that is the receivers confirming it.
Passenger questions, answered from the data
No — and no signal in the layers we track ever indicated a disruption. No airport or airspace closure was in force for Istanbul at any point in this window, the airport’s weather station kept reporting on its normal half-hourly cadence throughout in benign conditions, and an independent ADS-B network still saw hundreds of movements a day at IST — 765 on 28 August, 720 on the 29th, 701 on the 30th, 661 on the 31st, 604 on 1 September, 513 on the 2nd and 546 on the 3rd, inside its own pre-collapse band — while our own feed showed 12, 27, 7, 21, 22, 10 and 95. The only closure-shaped items in our data are partial-facility NOTAMs: one runway (RWY 09) and part of a taxiway closed at IST since 7 August, and taxiways and aprons closed at Sabiha Gökçen since 12 August — all predating the 17 August cliff and none closing an airport. The apparent ‘drop’ was our crowdsourced receiver network losing coverage of the Istanbul area from 17 August — Sabiha Gökçen collapsed on the same day with the opposite one-sided pattern, which no single operational cause can produce. That coverage returned on 4–5 September: on 5 September our IST count read 428 movements (185 departures, 243 arrivals) and Sabiha Gökçen 485 (188 / 297), with both one-sided gaps closed. For the status of a specific flight, check your airline or the airport’s own channels.
Because the cause is on the measurement side, not the airport side. İstanbul Airport (IST) and Sabiha Gökçen (SAW) are 64 km apart in the same Istanbul FIR and share crowdsourced ADS-B receiver coverage; when that coverage degraded on 17 August, both counts fell together. The tell is that the losses are opposite — IST’s observed arrivals nearly vanished while SAW’s observed departures did — which a real operational event (weather, closure, strike) cannot do to two airports at once, but a loss of different close-in receivers around each field can. The recovery pointed the same way and in the predicted order: our airspace-wide aircraft count for the Istanbul FIR came back first, then IST’s arrivals began returning on 3 September (28 against 67 departures), then both fields squared up on 4–5 September — IST 185 departures / 243 arrivals and SAW 188 / 297 on 5 September, the first day since 16 August that Sabiha Gökçen’s observed departures cleared 22. Different fields and different sides losing and regaining coverage at different rates is what receiver geography predicts; it is not what any single operational cause does.
The observed count dropped — the detector’s 22 August flag read −69% against its own trailing average of 223, and against IST’s genuine pre-17-August norm of ~1,500 movements a day the measured drop ran deeper: a median of 12 movements a day across 17 August–3 September (−99.2%), bottoming at a straight zero on 26 August. But observed counts are a floor, not a total, and the evidence says the traffic did not stop: IST’s weather station kept reporting a normal operating day throughout, a separate ADS-B network still saw 301–765 movements a day, no airport closure was in force, and Sabiha Gökçen collapsed on the same day with the opposite one-sided pattern. Every cross-check pointed to lost receiver coverage over the Istanbul area — and the count came back with the coverage on 4–5 September, reading 428 movements on 5 September, still 29% of the pre-cliff baseline, which is a statement about what we can hear rather than about what Istanbul flew.
We publish measured signals, not safety verdicts — go/no-go decisions belong to airlines, airports and the Turkish authorities, who work from complete operational data, not from crowdsourced tracking. Nothing in the layers we track indicated a hazard at either Istanbul airport at any point in this window; what degraded, on the evidence, was our own public visibility of their traffic, and that visibility returned on 4–5 September. For any specific flight, check your airline on the day of travel.
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.
“Observed traffic at İstanbul Airport ran at a median of 12 movements a day for eighteen days from 17 August, against a norm of about 1,500, but the drop was a measurement failure, not an airport one: Sabiha Gökçen, 64 km away in the same Istanbul airspace, collapsed on the exact same day with the opposite one-sided pattern, İstanbul’s weather station kept reporting a normal operating day throughout, and an independent aircraft-tracking network still saw hundreds of movements a day there — the signature of a regional receiver-coverage loss rather than a disruption. That coverage returned on 4–5 September: on 5 September the observed count at İstanbul read 428 movements, 185 departures against 243 arrivals, and Sabiha Gökçen 485, with the one-sided gaps closed at both airports on the same day and no disruption found at either field on any day of the window, 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.

