Pricing Search Start free trial
// EVENT · TRAFFIC DISRUPTION CLOSED · confirmed receiver-coverage loss at PMO, 17–30 Aug — the ledger has balanced on every readable day since 31 Aug (five settled days, mean 213 movements, −4% vs the ~222 baseline; departures ~110/day against a ~104 norm); no further updates unless a new signal appears Data as of 8 Sep 2026, 00:47 UTC

Palermo (PMO) Traffic 'Drop' — a −41% Anomaly That Sits Entirely on One Side of the Ledger: 122 Arrivals, 9 Departures, August 2026

← All airspace events

Observed traffic at Palermo Falcone–Borsellino (PMO) settled at 131 movements on Monday 17 August — 41% below its ~222 seven-day average — and tripped our traffic-anomaly detection, one day after Mount Etna, 173 km east, began a renewed eruptive phase. The cross-checks say the volcano is in the frame but not in the mechanism: the entire drop sits on one side of the ledger — 122 arrivals (normal for a Monday) against 9 departures (−91% vs the prior-Monday average of ~104) — and a real schedule cannot land 122 aircraft and launch 9. The pattern held on 18 August (118 arrivals, 5 departures), then broke on 19 August, when departures returned to 105 against 64 arrivals, and again on 21 August (102 departures against 85 arrivals) as coverage recovered; every day this page has flagged as a collection gap in our own feed has since backfilled — the 20th to 155, the 22nd to 205, the 26th to a balanced 201, the 27th to a balanced 208 (96 departures, 112 arrivals) and the 29th to 152 — alongside the 23rd at 217, the 24th at 166 and 28 August balanced at 180 (89 departures, 91 arrivals) on a fully healthy collection day. The one-sided shape still recurred, and we said so: fully on 25 August (12 departures against 129 arrivals) on a day our feed collected normally elsewhere and a second receiver network read Palermo low too, and partially on 29–30 August (38 and 47 departures) — the 30th on a day both instruments read Palermo short. 30 August is the last time it happened. Every readable day since has balanced: 31 August 181 movements (94 departures against 87 arrivals), 1 September 261 (133 against 128), 2 September 230 (105 against 125), 3 September 155 (92 against 63) and 5 September 239 (125 against 114) — a five-day mean of 213 movements, −4% against the ~222 baseline, with departures back at ~110 a day against the ~104 prior-Monday norm. Both flags earlier revisions had to raise are now withdrawn: 1 September and 2 September were rewritten downward by refetches running during worldwide under-collections in our own feed, and both have finalised balanced. Two days are excluded rather than read — 4 and 6 September, when our feed collected only ~36k and ~18k movements worldwide against a ~230–265k norm while the second network collected normally and read Palermo at 203 and 180; 7 September has not been collected in our feed at all. This page is closed on that recovery. Etna's ash plume moved southeast, away from Palermo on the island's northwest coast; no closure signal registered for the airport; its weather station kept its half-hourly cadence in unbroken VFR conditions; GNSS interference stayed in the low band (7-day ratio 0.001); and a second crowdsourced ADS-B aggregate we cross-check against lost the Palermo area almost completely on the same timeline (321 → 110 → 1 movements, 15–17 Aug) while seeing Rome and Milan normally, then regained the area (160 on the 19th, 99 on the 20th), in step with our own feed's 19 August recovery. We publish this as a receiver-coverage anomaly with a volcano in the picture, honestly labeled — because a −41% at a Sicilian airport during an Etna eruption is exactly the number that becomes a phantom headline.
// Flying to or from Palermo? The short version
  • This page is closed. Palermo airport showed no disruption signal in any layer we track — the −41% 'drop' on 17 August was confined to observed departures (9, vs 122 normal arrivals; the 18th repeated it, 5 vs 118), the fingerprint of receiver-coverage loss rather than of an airport in trouble — and on 19 August departures returned to 105 against 64 arrivals, with a fully balanced 21 August (102 vs 85) as coverage recovered. Coverage lapsed again on 25 August (12 departures against 129 arrivals) and ran departure-light on the 29th and 30th (38 and 47 departures). Since then the ledger has balanced on every readable day: 31 August (94 vs 87), 1 September (133 vs 128), 2 September (105 vs 125), 3 September (92 vs 63) and 5 September (125 vs 114) — a five-day mean of 213 movements, −4% against the ~222 baseline.
  • Etna's renewed eruptive phase (from 16 August) sent ash southeast — away from Palermo, which sits 173 km northwest of the volcano; no ash-related closure registered for PMO. The Catania side of the island is the side to watch, and it is tracked separately.
  • When departure-side coverage drops we cannot measure PMO's true outbound traffic — observed counts are a floor, not a total. Every day we flagged as a collection gap has since backfilled: the 20th, 22nd, 26th, 27th and 29th of August to 155, 205, 201, 208 and 152, and 1 and 2 September — both briefly rewritten downward by refetches during worldwide under-collections in our own feed — to a balanced 261 and 230. Two days remain excluded rather than read (4 and 6 September, when our feed under-collected worldwide while a second receiver network read Palermo normally at 203 and 180), and 7 September has not been collected in our feed. For a specific flight, check your airline as always.

Advisory timeline — how active is this episode

This chart deliberately shows only departures — because that is where the entire anomaly lives. Palermo's observed arrivals on 17 August were normal: 122, against a prior-Monday average of ~103. Its observed departures were not: 9, against a prior-Monday average of ~104 — a −91% collapse on one side of the ledger while the other side held. Through 8–14 August the network logged 136–173 departures a day at PMO; the count softened over the post-Ferragosto weekend (105 on the 15th, 90 on the 16th) and then fell off a cliff: 9 observed departures on the 17th, 5 on the 18th, while arrivals stayed normal both days — then rebounding to 105 on the 19th, 70 on the 20th (a day that first read as a whole-feed collection gap before late records backfilled it) and 102 on the 21st as receiver coverage returned — and staying two-sided through 84 on the 24th. Then the one-sided shape recurred: 12 observed departures on 25 August against 129 arrivals — on a day our feed collected normally everywhere else (Rome 982 movements, Milan 717) and the second crowdsourced network also read Palermo low (77, against ~200–280 on its normal days). 26 August then ran balanced again: 95 departures against 106 arrivals. 27 August, which this page plotted at 0 in an earlier revision, after a refetch run during a worldwide under-collection wiped it, has finalised back at a balanced 96 departures against 112 arrivals — vindicating the gap label rather than the zero. 28 August likewise finalised balanced on a healthy collection day: 89 departures against 91 arrivals. Then two departure-light days, each with its own caveat: 38 departures against 114 arrivals on the 29th (a day whose worldwide collection has since backfilled to a normal ~238k, so the deficit stands on its own — though the second receiver network read Palermo balanced at 280 that day), and 47 against 134 on the 30th (our feed read Rome 1,070 and Milan 702 normally, so that deficit is local to Palermo, and the second network read Palermo low too). 31 August, which this page plotted at 59 against 42 in the previous revision and disclaimed as our own worldwide under-collection, has finalised balanced at 94 departures against 87 arrivals — that day's worldwide count backfilled to a normal ~241k (Rome 1,066, Milan 754), so the disclaimer is gone and the day reads clean. 1 September did the same thing: the bar an earlier revision plotted at 48 and refused to read finalised at 133 departures against 128 arrivals — 261 movements — because that day's worldwide count settled at a normal ~248k (Rome 989, Milan 721). 2 September has now closed the same loop, and it is the last correction this page makes: the previous revision had to set that day aside at 23 departures against 31 arrivals, because a refetch running during a worldwide under-collection had rewritten its rollup downward; it has finalised at 105 departures against 125 arrivals — 230 movements — on a worldwide count that settled at a normal ~250k (Rome 971, Milan 750). 92 departures against 63 arrivals on 3 September (155 movements, ~247k worldwide; Rome 1,013, Milan 760) and 125 against 114 on 5 September (239 movements, ~225k worldwide; Rome 924, Milan 649) close the chart with the departure side the fuller of the two on both days. The one low bar left is 5 departures on 4 September, marked here as a partial day: our own feed collected only ~36k movements worldwide that day (Rome 31, Milan 49) while the second receiver network collected normally and read Palermo at 203 — so that bar measures our collection, not Palermo. A real airport cannot make departures vanish and reappear like this — every aircraft that lands must eventually take off again — but a receiver network that loses and regains its close-in, low-altitude reach can, because a departure is heard first on the runway and in initial climb, the shortest-range signal there is:

173
8 Aug
147
9 Aug
136
10 Aug
138
11 Aug
171
12 Aug
167
13 Aug
161
14 Aug
105
15 Aug
90
16 Aug
9
17 Aug
5
18 Aug
105
19 Aug
70
20 Aug
102
21 Aug
108
22 Aug
114
23 Aug
84
24 Aug
12
25 Aug
95
26 Aug
96
27 Aug
89
28 Aug
38
29 Aug
47
30 Aug
94
31 Aug
133
1 Sep
105
2 Sep
92
3 Sep
5*
4 Sep
125
5 Sep

Fig. 1 — Observed departures per day at Palermo Falcone–Borsellino (PMO), 8 Aug – 5 Sep 2026 · * partial day · Data as of 8 Sep 2026, 00:47 UTC · Source: FlySafe monitoring (crowdsourced ADS-B receiver network).

Airports in the exposure zone

AirportSizeObserved signalFIR
Palermo Falcone–Borsellino (PMO) large — · 17–18 Aug: 122 and 118 arrivals vs 9 and 5 departures — one-sided loss; departures recovered to 105 on the 19th and 102 on the 21st, relapsed on the 25th (12 vs 129), departure-light again on the 29th–30th (38, 47), then balanced on every readable day since — 94/87, 133/128, 105/125, 92/63, 125/114 (31 Aug – 5 Sep); 4 and 6 Sep excluded as our own collection gaps LIRR
Trapani–Birgi (Vincenzo Florio) (TPS) medium — · parallel decline 50 → 24 → 18, 13–18 Aug LIRR
Comiso (CIY) medium — · decline 44 → 14, 13–17 Aug; 22 on the 18th LIRR
Catania–Fontanarossa (CTA) large — · renewed Etna phase on this side; our primary feed unreliable here since 11 Aug LIRR

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: half the ledger went missing, not half the traffic

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. Palermo Falcone–Borsellino's last 30 days:

Fig. 2 — Observed daily movements at PMO, 21 Jul – 5 Sept (latest complete UTC day). Data as of 8 Sep 2026, 00:47 UTC · Source: FlySafe monitoring (ADS-B receiver network).

  • Seven-day average around the key day: ~222 observed movements/day; the four prior Mondays averaged ~207 (180, 168, 219, 262). Peak of the window: 347 on Friday 7 August, the pre-Ferragosto travel wave.
  • Monday 17 Aug settled at 131 observed movements — −41% vs the 7-day average, −37% vs the Monday average. Composition: 122 arrivals, 9 departures. Detection first read the day at 131 (122 arrivals, 9 departures) and successive nightly refetches — flight records finalise over ~2–3 days — have held it there.
  • Arrivals alone were normal for a Monday: 122 vs a prior-Monday arrival average of ~103 (+18%). Departures alone were −91%: 9 vs ~104.
  • A schedule that lands 122 aircraft and launches 9 would stack more than a hundred airliners on Palermo's apron in a single day. It did not happen — the missing departures are missing from the measurement, not from the airport.
  • The pattern held a second day, then recovered: 18 August measures 123 movements at PMO — 118 arrivals, 5 departures, the same one-sided shape. Then, tellingly, 19 August turned the departure side back on: 105 departures against 64 arrivals (169 total), and 21 August ran a fully balanced ledger — 102 departures against 85 arrivals (187 total). The second crowdsourced network regained the same area over the 19th–20th (160 then 99 movements). Coverage returning is what restored the missing departures — the runways never changed.
  • Resolved gaps — every one of them: five days first registered as whole-feed collection gaps in our own feed, and every one has since backfilled as late-arriving records finalised (flight records settle over ~2–3 days): the 20th (logged at 14 movements at the time) to 155, the 22nd (7) to 205, the 26th (54) to a balanced 201 (95 departures, 106 arrivals), the 27th (7) to a balanced 208 (96 departures, 112 arrivals) and the 29th (7) to 152. Alongside them the 23rd finalised at 217 (114 departures, 103 arrivals), the 24th at 166 (84 against 82) and 28 August balanced at 180 (89 departures, 91 arrivals) on a fully healthy collection day. Not one flagged day has come back as a quiet airport.
  • The one-sided shape recurred in full on 25 August: that day finalised at 141 movements — 12 departures against 129 arrivals — and this time it is not a collection artefact of ours: our feed collected normally everywhere else on the 25th (Rome 982, Milan 717, both at their norms), and the second crowdsourced network independently read Palermo low the same day (77 movements, against ~200–280 on its normal days) while seeing Rome (863) and Milan (676) normally. Two instruments, the same dip, the same place: close-in receiver coverage at Palermo degraded again for a day, then recovered. It remains the sharpest such day since the 19th — but it is no longer the only one.
  • The departure side stayed unsteady for two more days — then the ledger balanced: 29 August finalised at 152 movements (38 departures against 114 arrivals) and 30 August at 181 (47 against 134) — the departure-light shape again, far milder than the 9-against-122 of the 17th, and the two days carry different weight. The 29th changed on us in the honest direction: the worldwide under-collection we once flagged has itself backfilled to a normal ~238k movements (Rome 967, Milan 685), so we cannot excuse that day as our own outage — though the second network read Palermo balanced at 280, so the two instruments simply disagree about it. On the 30th our feed collected the mainland normally (Rome 1,070, Milan 702, ~230k worldwide) and the second network also read Palermo low (124 against its ~200–280 norm): both instruments short, same place, same day. 31 August has now moved against our previous reading, and we correct it: the day finalised balanced at 181 movements — 94 departures against 87 arrivals, not the 59-against-42 we published, because the worldwide under-collection we disclaimed it for backfilled to a normal ~241k (Rome 1,066, Milan 754). The second network still reads Palermo low that day (92), so the disagreement between the instruments simply flipped sides.
  • 1 September came back, and we took back the flag we had put on it: an earlier revision published it at 100 (48 departures against 52 arrivals) and refused to read it, because a refetch had run while our feed under-collected worldwide. That day finalised at 261 movements — 133 departures against 128 arrivals — on a worldwide count that settled at a normal ~248k (Rome 989, Milan 721). The original reading was right and the rewrite was ours; the second network had it at 170 (85 against 85).
  • 2 September has now done exactly the same thing, and that is the last correction this page owes: the previous revision had to set the day aside at 54 (23 departures against 31 arrivals), because a later refetch ran while our feed was under-collecting worldwide and rewrote the whole day's rollup downward. It has finalised at 230 movements — 105 departures against 125 arrivals, on a worldwide count that settled at a normal ~250k (Rome 971, Milan 750) — close to the balanced 227 we had originally published, and matching the second network's own balanced 215 (123 against 92). Every day this page ever flagged as a gap has now come back, and not one came back as a quiet airport.
  • The last two readable days are clean: 3 September at 155 movements (92 departures against 63 arrivals) and 5 September at 239 (125 against 114), both on fully healthy collection days (~247k and ~225k worldwide; Rome 1,013 and 924, Milan 760 and 649) — the departure side the fuller of the two on each, the exact inverse of the shape this page opened on. The second network agreed closely both days (163, then 274). Two days in between are excluded rather than read: 4 September (PMO 8) and 6 September (PMO 9) are days our own feed under-collected worldwide — ~36k and ~18k movements against a ~230–265k norm, with Rome at 31 and 49 and Milan at 49 and 70 — while the second network collected normally on both (~138k and ~121k worldwide) and read Palermo at 203 (106 departures against 97 arrivals) and 180 (93 against 87). 7 September has not been collected in our feed at all and sits outside the chart above. Those are limits on our measurement, stated as such — not readings about the airport.
  • The number that closes this page: across the five settled, readable days since the last one-sided reading — 31 August, 1, 2, 3 and 5 September — Palermo ran 181, 261, 230, 155 and 239 observed movements, a mean of 213 a day — −4% against the ~222 seven-day baseline this event opened on. The departure side, which is where the whole anomaly lived, averaged ~110 departures a day against the ~104 prior-Monday norm, and against 9 and 5 on 17–18 August. The ledger balanced on every one of those days (94/87, 133/128, 105/125, 92/63, 125/114). The airport's own brief reads 5 September as a normal day against its ~205 baseline.
  • The clearest evidence is the oldest: departure-light days at Palermo pre-date the eruption in our feed — 22 July logged 5 departures against 115 arrivals, 24 July 26 against 104, 1 August 46 against 136 — on days with no volcano and no event anywhere in the frame. What recurs at this airport is the reach of the receiver network around it, not something happening on its runways.

Observed counts are a floor, not a total — and they are only meaningful while receiver coverage holds. The departure side of Palermo's ledger is where that assumption broke on 17–18 August and returned on the 19th; the cross-checks below are how we know. Live view: PMO airport brief.

What it means for flights

What fired, and why we publish it anyway. Our detection layer flagged PMO when the 17 August observed count came in at 131 movements against a seven-day average of ~222 — a −41% traffic anomaly at Sicily's second-busiest airport (successive nightly refetches, over the ~2–3 days flight records take to finalise, have held the day at 131), with the airport marked ash-exposed because a renewed eruptive phase at Etna had begun the evening before. That is precisely the combination that produces phantom headlines: a big negative number, a famous island, an erupting volcano. Automated multi-signal analysis then did what it always does — checked the anomaly against every other layer we track: closure signals, weather-reporting continuity, the ash advisories themselves, GNSS interference, neighboring airports, and the composition of the count. The layers do not corroborate an airport disruption; they corroborate a measurement problem. We publish the evidence for that verdict, as we did for the Monterrey coverage collapse and the Memphis Sunday false alarm — publishing observed counts honestly means publishing their failure modes too.

A ledger that cannot balance. The decisive fact is not the size of the drop but its shape. On 17 August our network attributed 122 arrivals to Palermo — a normal Monday (the four prior Mondays averaged ~103 arrivals) — and 9 departures, against a prior-Monday average of ~104. Aviation is a conserved system: over any full day, an airport's arrivals and departures balance to within a handful of positioning flights. Landing 122 and launching 9 would leave over a hundred airliners parked at a mid-size airport by midnight; nothing of the kind was reported anywhere, and the airport's own weather station kept reporting a routine operating day. The asymmetry has a mundane radio explanation, which we label as interpretation: a departing aircraft is heard first on the runway and in initial climb — the lowest-altitude, shortest-range phase of flight, audible only to close-in antennas — while an arriving aircraft descends along a predictable profile from within reach of distant receivers, so its track can still be attributed to the airport even when the last miles go unheard. Lose the close-in receivers and departures vanish from the ledger first. That is what the fingerprint of coverage loss looks like — and it is the same fingerprint we measured at Monterrey.

The Etna question — a volcano in the frame, not in the mechanism. Etna began a renewed eruptive phase on 16 August at 16:20 UTC (VAAC Toulouse), with strong activity through 17 August, an ash cloud estimated around 7,000 m AMSL (~FL230) and the plume moving southeast — out over the Ionian side, away from Palermo on the island's northwest coast, 173 km from the vent. By the morning of 18 August the advisories read 'explosive activity is decreasing', with weak ash possible only near the crater. Could ash have caused Palermo's number anyway? The evidence says no: ash disruption suppresses both sides of a schedule (as it did at Catania in the 4–15 August episode — near-shutdown days across the board), not departures alone; no ash-related closure registered for PMO; and Palermo's arrivals actually ran above their Monday norm — if anything, consistent with its usual role as the western diversion field when the Catania side degrades, which we flag as interpretation, not measurement. The renewed eruption is real and it matters — on the other side of Sicily, where the ash went and where our measurement is honestly weakest: our primary movement feed has been unreliable for Catania since 11 August. That side of the story lives on the Etna event page and the CTA brief, and we will not use a broken feed to make claims about it in either direction.

The independent checks we ran. Palermo's weather station kept its normal half-hourly cadence throughout — 46–48 observations on each of 14–19 August and 45–48 a day from 21 August to 7 September (bar 30 on 31 August, a gap in our own polling rather than in the station's reporting), 1,394 over the last 30 days with IFR conditions 0% of the time, every single report VFR, no thunderstorm flags, and the latest METAR (8 Sep, 00:20 UTC) showing CAVOK — visibility 10 km or more with no significant cloud — a variable 2-knot wind and 25 °C. The GNSS-interference ratio for the area stayed in the low band across the 7-day window (0.001). No closure signal registered for the airport at any point: the only airspace signals recorded in the Rome FIR are Etna-related and FIR-level, on the island's opposite coast — the August run ended on the 18th, and the run that opened on 1 September closed on 7 September at 23:59 UTC, having tracked a series of weak eruptive episodes at Etna. Etna itself is still active as this page closes: VAAC Toulouse's latest advisory (7 Sep, 21:56 UTC) reports a weak eruption ongoing from 18:48 UTC that day with the plume moving east, and no quantitative ash graphic issued because of the low intensity of the event — away from Palermo on the island's northwest coast in every advisory of the episode. All of it is labeled as FIR-level, not as Palermo facts. A second crowdsourced ADS-B aggregate we cross-check against lost the Palermo area on the same timeline — 321 movements on 15 August, 110 on the 16th, 1 on the 17th — while seeing Rome (843) and Milan (729) normally the same day, then regaining the area (160 on the 19th, 99 on the 20th) in step with our own 19 August recovery, and running mostly normally since — 199 movements on the 23rd, 190 on the 24th, 77 on the 25th, 215 on the 26th, 250 on the 27th, 252 on the 28th, 280 on the 29th, 124 on the 30th, 92 on the 31st, 170 on 1 September, 215 on the 2nd, 163 on the 3rd, 203 on the 4th, 274 on the 5th and 180 on the 6th. Four things in that run do the work: the 25th, where that network's own dip lines up with the one-sided 12-against-129 day in our feed and makes a second, independent case for a repeat coverage lapse at Palermo; the 27th and 29th, where its balanced 250 and 280 were why we labeled our feed's momentary 7 at PMO on those days a collection gap rather than a quiet airport — a call both days vindicated by backfilling to 208 and 152; the 30th and 31st, where its own 124 and 92 sit well below its ~200–280 norm — the 30th on the same day our feed logged 47 departures against 134 arrivals, so both instruments read Palermo short at once, and the 31st on a day our feed has since backfilled to a balanced 94 against 87, so the two instruments end up disagreeing with the sides reversed; and 1–6 September, where it read Palermo balanced throughout (85/85, 123/92, 95/68, 106/97, 137/137, 93/87) — matching our own backfilled 1 and 2 September and our clean 3 and 5 September, and standing as the only reading we have for 4 and 6 September, the two days our own feed under-collected worldwide. We label that check honestly: crowdsourced networks share volunteer receiver hosts, so it corroborates that reception died in the area and returned, not anything about operations. And the two feeds disagreeing about the same day at the same airport — 131 in ours, 1 in theirs on the 17th — is itself evidence: an airport has only one reality, and two instruments reading it differently means the instruments, not the runways, are the story. One more honest nuance: our raw position-snapshot layer still heard low-altitude traffic near Palermo on 17 August (97 low-altitude reports, down from a ~155/day average but far from zero, with en-route overflights steady) — so unlike Monterrey this looks like partial close-in degradation rather than a total blackout, which fits a count that fell 41% instead of collapsing to nothing.

The footprint is regional — and it is labeled as ours, not the airports'. Observed totals at two other western-and-southern Sicilian fields declined in parallel — Trapani–Birgi from 50 movements on 13 August to 24 on the 17th and 18 on the 18th, Comiso from 44 to 14 (with a partial return to 22 on the 18th) — while Lampedusa, on its own island with its own receivers, held steady (31–47 movements through 17 August) and every mainland airport we track ran normally (Rome Fiumicino 1,052 on 17 August, Milan Malpensa 783). A decline that follows receiver geography rather than airline geography is a statement about coverage. We flag one honest ambiguity in it: Comiso sits on the island's southeast, under the side of Sicily the renewed Etna plume actually moved toward, so a real operational component there cannot be excluded from our data — that is a FIR-level observation about our own visibility, labeled as such, not a claim about operations at any of these airports.

What we cannot say, and what changes this page. With coverage lapsing intermittently, we still cannot treat Palermo's outbound counts as complete on every day — observed counts are a floor, not a total, and we will not declare the airport 'operating normally' on evidence we do not have; the operating picture on any given day belongs to the airport, the airlines and the Italian authorities. We also flag our own blind spots plainly: 18 August measures 123 movements at PMO (118 arrivals, 5 departures), a second one-sided day; 19 August then turned the departure side back on — 105 departures against 64 arrivals (169 total), the first firm sign of coverage returning; 20 August first read as a whole-feed collection gap (PMO logged 14, with Rome and Milan simultaneously dropping to 35 and 44) but has since backfilled to 155 as late records landed; 21 August ran a fully balanced ledger — 102 departures against 85 arrivals; 22 August likewise backfilled from an initial 7 to 205; 23 August finalised at 217 (114 departures, 103 arrivals), 24 August backfilled to 166 (84 against 82), 26 August — flagged here as a gap at 54 — backfilled to a balanced 201 (95 against 106); 27 August, which a refetch run during a worldwide under-collection had momentarily rewritten to 7, has finalised back at a balanced 208 (96 departures, 112 arrivals); 28 August finalised balanced at 180 (89 against 91) on a fully healthy collection day; and 29 August, first flagged at 7, has backfilled to 152. Every gap we flagged came back, and none of them came back as a quiet airport. The recovery is holding but it is not unbroken: 25 August finalised at 141 with 12 departures against 129 arrivals — the one-sided shape again, on a day our feed collected normally everywhere else (Rome 982, Milan 717) and the second receiver network independently read Palermo low (77 movements against ~200–280 on its normal days), so we read it as a repeat coverage lapse rather than an airport event; it is the sharpest such day since the 19th, and no longer the only one. 29 and 30 August both finalised departure-light — 38 departures against 114 arrivals (152 total) and 47 against 134 (181) — and they do not carry the same weight. The 29th moved against our earlier reading and we correct it here: the worldwide under-collection we cited for that day has itself backfilled to a normal ~238k movements (Rome 967, Milan 685), so the day no longer excuses itself as our outage — but the second network read Palermo balanced at 280, so the two instruments disagree and we leave it unresolved. On the 30th our feed collected the mainland normally (Rome 1,070, Milan 702, ~230k worldwide), which leaves the Palermo deficit standing on its own — and the second network read Palermo at 124 that day, itself under its ~200–280 norm. 31 August has now moved against us in the other direction, and we correct that too: we published it at 101 movements (59 departures against 42) and disclaimed it, because our feed was then under-collecting worldwide; that day has since backfilled to a normal ~241k globally (Rome 1,066, Milan 754) and PMO to a balanced 181 — 94 departures against 87 arrivals. The second network still reads Palermo at 92 that day, so the disagreement between the instruments has simply changed sides. 1 and 2 September are the two corrections that run the other way, and we make them plainly: earlier revisions had to set both days aside, because refetches running during worldwide under-collections in our own feed had rewritten their rollups down — 1 September to 100 (48 departures against 52 arrivals), 2 September to 54 (23 against 31). Both have finalised balanced: 1 September at 261 movements (133 departures against 128 arrivals) on a worldwide count that settled at a normal ~248k (Rome 989, Milan 721), and 2 September at 230 (105 against 125) on ~250k (Rome 971, Milan 750) — close to the balanced 227 we had originally published for it. The second network read both balanced too (170 and 215). Both flags are withdrawn. 3 and 5 September are clean days: 155 movements (92 departures against 63 arrivals) and 239 (125 against 114), on fully healthy collection days (~247k and ~225k worldwide; Rome 1,013 and 924, Milan 760 and 649) — the departure side the fuller of the two on each, the exact inverse of the shape that opened this page — with the second network agreeing closely (163, then 274). 4 and 6 September are excluded rather than read: our feed collected only ~36k and ~18k movements worldwide on those days (Rome 31 and 49, Milan 49 and 70) against a ~230–265k norm, so its PMO counts of 8 and 9 measure our collection, not the airport; the second network collected normally both days and read Palermo at 203 and 180. 7 September has not been collected in our feed at all. We would rather publish a day moving on us in either direction than quietly settle on whichever reading suits the story — and for the same reason we note that departure-light days at PMO pre-date the eruption entirely: 22 July logged 5 departures against 115 arrivals, 24 July 26 against 104, 1 August 46 against 136. Our primary movement source has also been unreliable for Catania since 11 August. The condition we set for closing this page — coverage staying restored, counts back near the norm and the ledger balanced for a clear run of days — has now been met, and the closing entry below states the number it was met on. Automated multi-signal analysis keeps re-checking every layer — movements, advisories, closures, weather continuity, interference — every three hours, and this page carries the standing lesson: before believing a drop, check which side of the ledger it lives on.

What the data showed — closing entry. This page is closed as of 8 September 2026, 00:47 UTC. The signal that opened it — a one-sided ledger at Palermo — has not appeared for more than a week, and the number that closes it is the recovery: across the five settled, readable days from 31 August to 5 September, PMO ran 181, 261, 230, 155 and 239 observed movements — a mean of 213 a day, or −4% against the ~222 seven-day baseline the anomaly was measured on. The departure side, which is where the entire anomaly lived, came back with it: 94, 133, 105, 92 and 125 departures on those days — a mean of ~110 against the ~104 prior-Monday norm, and against the 9 and 5 of 17–18 August that started this. Every one of those days balanced against its own arrivals (87, 128, 125, 63, 114). Against the anomaly window that is the whole argument in one line: 9 departures against 122 arrivals on 17 August, 5 against 118 on the 18th and 12 against 129 on 25 August, versus a ledger that has not tipped since 30 August. What the episode measured was our own coverage, not Palermo's traffic. Two independent instruments back that: our raw position layer still heard low-altitude traffic near the airport on the worst day (97 reports against a ~155/day mean), which is partial close-in degradation rather than a blackout, and the second crowdsourced network lost the same area on the same timeline (321 → 110 → 1 movements, 15–17 August) while reading Rome and Milan normally, then regained it in step with our own 19 August rebound and has read Palermo inside its own band every day of the recovery (170, 215, 163, 203, 274, 180 across 1–6 September). Every day this page ever flagged as a collection gap in our feed has since backfilled, and not one came back as a quiet airport — including 1 and 2 September, which earlier revisions of this page had to set aside and which finalised at a balanced 261 and 230. Two days are still excluded by our own gap rule rather than read: 4 and 6 September, when our feed under-collected worldwide (~36k and ~18k movements against a ~230–265k norm) while the second network collected normally; 7 September has not been collected in our feed at all. Nothing in the other layers ever corroborated a disruption: no closure signal registered for PMO at any point, the airport's weather station kept its half-hourly cadence with every report VFR and IFR conditions 0% of the last 30 days (latest METAR 8 September, 00:20 UTC: CAVOK, variable 2 kt, 25 °C), and the GNSS-interference ratio for the area held in the low band at 0.001. Etna remains active on the other side of the island as this page closes — VAAC Toulouse's latest advisory (7 September, 21:56 UTC) reports a weak ongoing eruption with the plume moving east and too little intensity for a quantitative ash graphic — and the FIR-level volcano run recorded in the Rome FIR from 1 September closed on 7 September at 23:59 UTC. That side of Sicily is tracked on the Etna event page, where our own measurement is honestly weakest: our primary movement feed has been unreliable for Catania since 11 August, and we will not use a broken feed to make claims in either direction. What we do not claim here is that Palermo was 'operating normally' on any given day — the operating picture belongs to the airport, the airlines and the Italian authorities; what we can say is that our instrument lost the airport's departures and got them back, and we publish both halves of that. No further updates unless a new signal appears. Live view: PMO airport brief.

Passenger questions, answered from the data

Is Palermo airport closed or disrupted?

No signal in the layers we track indicates a disruption: no closure registered for PMO, the airport's weather station reporting on its normal half-hourly cadence in VFR conditions, GNSS interference in the low band, and observed arrivals normal for a Monday (122 vs a ~103 average). The −41% figure comes entirely from the departure side of the count — 9 observed departures against a ~104 Monday average — which matches the signature of receiver-coverage loss, not a schedule collapse; on 19 August departures returned to 105 against 64 arrivals, and 21 August ran a balanced 102 vs 85 as coverage recovered. That coverage was still intermittent late in August — 29 and 30 August again read departure-light (38 and 47 departures) — but 30 August was the last such day: 31 August ran a balanced 94 against 87, and 1, 2, 3 and 5 September ran 133 against 128, 105 against 125, 92 against 63 and 125 against 114, a five-day mean of 213 movements or −4% against the ~222 baseline. This page is closed on that recovery. Departure-light days are a limit on what we can measure, not a signal about the airport. For the status of a specific flight, check your airline or the airport's own channels.

Is the Etna eruption affecting Palermo flights?

Etna began a renewed eruptive phase on 16 August 2026 (VAAC Toulouse), with ash estimated to about 7,000 m and the plume moving southeast — away from Palermo, which sits 173 km northwest of the volcano. No ash-related closure registered for PMO, and Palermo's observed arrivals ran above their Monday norm on 17 August (122 vs a ~103 average). Ash disruption suppresses arrivals and departures together; Palermo's anomaly is departures-only, which points at measurement, not ash. The side of Sicily the plume moved toward is covered on the Etna event page.

Did flights at Palermo really drop 41%?

The observed count dropped 41% — from a ~222 seven-day average to a settled 131 on 17 August (detection read the day at 131 and successive refetches have held it there). But observed counts are a floor, not a total, and the composition shows the drop is not real traffic: arrivals were normal (122, vs ~103 on an average Monday) while departures collapsed to 9 — and the 18th repeated the pattern with 118 arrivals against 5 departures. An airport that lands 122 aircraft and launches 9 would accumulate over a hundred aircraft on its apron in a day — that did not happen, and on 19 August departures returned to 105 against 64 arrivals, with a fully balanced 21 August (102 vs 85), as coverage recovered — lapsing once more on 25 August (12 departures against 129 arrivals), running balanced again on the 26th to 28th (95, 96 and 89 departures against 106, 112 and 91 arrivals), turning departure-light once more on the 29th and 30th (38 and 47) and then balancing on every readable day since — the 31st (94 against 87), 1 September (133 against 128), 2 September (105 against 125), 3 September (92 against 63) and 5 September (125 against 114), a five-day mean of 213 movements or −4% against the ~222 baseline. 4 and 6 September are excluded rather than read, because our own feed under-collected worldwide on both. Every cross-check (unbroken weather reporting, no closure signal, a second receiver network losing the same area while Rome and Milan stayed visible, then recovering in step) points to lost close-in receiver coverage, not to flights stopping.

Is it safe to fly to Sicily right now?

We publish measured signals, not safety verdicts — go/no-go decisions belong to airlines, airports and the aviation authorities, who work from complete operational data, not from crowdsourced tracking. Etna's renewed activity is handled the way ash always is: regulators and dispatchers work from the VAAC advisories, closing or rerouting around identified ash. Nothing in the layers we track indicates a hazard at Palermo; what degraded there, on the evidence, is public visibility of departing traffic. 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.

// Copy-paste quote

“Observed traffic at Palermo fell 41% below its weekly average on 17 August — the day after Etna's renewed eruption — but the entire drop sat on the departure side of the ledger: 122 arrivals, normal for a Monday, against 9 departures, while the ash plume moved southeast away from the city and the airport's weather station kept reporting a routine VFR day — the signature of receiver-coverage loss rather than a disruption, confirmed when departures returned to 105 against 64 arrivals on 19 August and a balanced 102 against 85 on the 21st as coverage recovered, according to flight-monitoring service FlySafe.zone.”

Chart: observed traffic at PMO
PNG · 1600px · attribution baked in
Observed daily movements at PMO — light version preview
Chart: advisory cadence
PNG · 1600px · attribution baked in
Observed departures per day at Palermo Falcone–Borsellino (PMO), 8 Aug – 5 Sep 2026 — light version preview
↓ Data behind the charts — CSV CC BY 4.0 · dates + daily values, ready for your own chart
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.

No newsletters — we only reply to this request. Press page
// Cite this
"FlySafe.zone — Palermo (PMO) Traffic 'Drop' — a −41% Anomaly That Sits Entirely on One Side of the Ledger: 122 Arrivals, 9 Departures, August 2026", https://flysafe.zone/events/pmo-traffic-drop-2026-08/, data as of 8 Sep 2026, 00:47 UTC — free to cite with a link. Observed-traffic figures are FlySafe’s own ADS-B-derived counts (floor, not total). Observed movement counts and position-report aggregates from the crowdsourced ADS-B receiver network (a floor, not a total); volcanic-ash context from published VAAC advisories (Toulouse); weather-reporting continuity from published METARs. Method: /methodology/ · Press: /press/.

Related coverage