Drones Have Closed Taiwanese Airports 23 Times, and Airports May Only 'Enforce Against'

🌏 中文版

The previous post read every Taiwanese provision that might authorise counter-drone action and concluded that the law authorises an outcome and no means — and that the airport is the cell squeezed from three sides: the weakest verb, soft kill named as the prohibited thing, hard kill meaning gunfire beside a runway.

That was an inference from the text. This post goes looking for what has actually happened at the airports.

What turned up is more concrete than expected: the Legislative Yuan's evaluation of the CAA's FY2021 budget attaches an itemised list the agency supplied itself.

About the number in the title. This post originally had only the first tranche of data (FY2019 to end-August 2020: six closures, 300 minutes), and the title said so. After the two later evaluation reports were added on 2026-08-09, the public series runs to 23 occurrences from FY2019 to end-August 2023, and the title now reflects that. The narrative below keeps its original order: the first six itemised rows, then the later additions.

Six closures, 300 minutes, 5,929 passengers

The source is the Legislative Yuan Budget Center's Evaluation Report on the FY2021 Unit Budgets of the Taiwan Transportation Safety Board and Ministry of Transportation Agencies (CAA, Central Weather Bureau, Institute of Transportation, Directorate General of Highways), Table 2, sourced to "data provided by the CAA on 14 September 2020". Verbatim:

The total row is mine; the original table has none. It comes to exactly 300 minutes — five hours.

Three things worth pulling out:

One, Songshan accounts for four of the six. Those four total 232 minutes, or 77% of all closure time. Songshan sits inside Taipei City, with dense housing and riverside parks a few hundred metres beyond each runway threshold — that is the common-sense explanation for why it is the main victim, but note that it is an explanation, not a finding; this post does no geographic analysis to test it.

Two, on 4 August 2019 both Songshan and Taoyuan closed on the same day. The table does not say whether the two were connected and I could not find out. But that single day accounts for 128 minutes of closure, 23 flights and 3,419 passengers — the worst day of the six.

Three, the Taichung closure lasted 8 minutes and affected no flights. That row matters because it shows closure time is not the same as damage — and also that once detected, you close anyway. The cost of an 8-minute closure is low, but the decision is identical: stop movements.

All six 'response' cells are the same sentence, copy-pasted

The original table has a "response" column too. Six incidents, six rows, word for word identical:

On notification that a drone was being operated unlawfully around the airport, the airport station, acting under the Civil Aviation Act, proceeded to the scene with the Aviation Police Bureau to investigate, and requested assistance from relevant agencies such as the municipal or county police and 〔the military〕.

(The last agency name is not legible in that PDF's text layer, so it is generalised here as 〔the military〕; the rest is verbatim.)

This is not lazy formatting — that sentence is Article 99-13(6) of the Civil Aviation Act in plain words:

A drone operating without permission within the announced distance around an airport or airfield shall be enforced against by the airport's or airfield's operator or manager together with the Aviation Police Bureau; where necessary, assistance may be requested from relevant agencies.

The provision says "enforce against, together with the Aviation Police"; the response to all six incidents was "go to the scene with the Aviation Police and investigate." What an airport can do and what the law wrote for it are the same thing — and the previous post took that verb apart: managers of prohibited airspace and government agency premises get "take appropriate measures to stop or remove"; commercial ports got it too in 2023; airports did not.

Put that against the timescale. The closures ran 8 to 68 minutes. Getting people to the scene, investigating, and identifying the operator takes longer than that. Across all six incidents, nothing in the official response procedure was directed at the drone in the air during the minutes the aircraft were stopped.

The CAA told five airports to buy jammers in 2020

The same table's "CAA remedial actions" column records three things, verbatim:

1. Adopted the Standard Operating Procedure for Response to and Investigation of Drone Interference at Airports, combining airport stations, the Aviation Police, local police and the military into a regional joint-defence mechanism.
2. To prevent drones entering the announced distance around airport stations … (1) In 2018 the CAA commissioned the National Chung-Shan Institute of Science and Technology to conduct an "Overall Evaluation of Drone Countermeasures" … Taipei Airport Station and NCSIST are now contracted to run the Songshan Airport Drone Countermeasure Pilot Deployment Project (project period 2020 to 2025). (2) Kaohsiung Airport Station … will set requirement specifications by December 2020 and complete procurement by December 2022. (3) The CAA has written to the higher-traffic Kaohsiung, Taichung, Taitung, Kinmen and Magong stations instructing them to procure jammers.

Point (3) is the line to remember. Set it beside the timeline from the previous post:

The competent authority instructed its own airports in 2020 to buy a class of device that had no public-duty import path at the time, and whose use prohibition remains untouched today — a prohibition whose text names "flight safety" as precisely the thing that must not be affected.

I did not check whether those five stations subsequently bought anything, how many, or whether any has been used. That is this post's largest blank, and it is restated below.

Note also the project period: Songshan's countermeasure pilot deployment runs 2020 to 2025 — six years. All six closures happened before that project started, or in its first year.

Update: the later years are in, and the count did not fall

Update, 2026-08-09. This post originally listed "no closure list after August 2020" as the gap it most needed to fill, noting that "the same database publishes a budget evaluation report every year." That was right — the later reports do continue the series. Here it is. Filling it did not change the direction of the third conclusion; it strengthened it.

Two continuations:

• FY2022 unit budget evaluation report (October 2022, sourced to "CAA data of 14 September 2022"): 13 occurrences from 2020 to end-August 2022
• Civil Aviation Operating Fund FY2024 budget evaluation report (October 2023, sourced to "CAA data of 11 September 2023"): 12 occurrences from 2022 to end-August 2023

By year (my compilation; neither report presents this table):

A trough of 2 in 2021, then 9 in 2022 — four and a half times. The FY2022 report's own reading: "7 occurrences by the end of August 2022, considerably more than in fiscal 2020 and 2021."

One confounder must be flagged: 2021 was Taiwan's worst pandemic year and its lowest for traffic. How much of that trough is fewer flights rather than fewer drones cannot be computed — neither report gives a denominator, and the FY2022 report itself attributes the subsequent rise to "the gradual relaxation of border controls." So the accurate statement is: the count did not fall as the chapter took effect and equipment was installed, and 2022 is the peak of the public series.

The same agency reported the same events two different ways

Laying the overlapping stretch (January–August 2022) side by side turns up three things that do not line up. This deserves its own section, because it determines how finely these numbers can be used.

The third row needs spelling out: the FY2021 report covers "FY2019 to end-August 2020", and the 2020-01-10 Magong event falls inside that window, yet it does not appear in that report's Table 2. It only surfaces in the later FY2022 report.

None of the three discrepancies is large, but they all point the same way: this is one agency reporting the same events to the legislature in different years, and it does not agree with itself. So these figures support trend and order-of-magnitude claims and should not be used for incident-level statistical inference — which is why the by-year table above deliberately stops at the count.

Did those five airports buy the jammers? Yes — and not only those five

This post's second declared gap was "no check on whether those five stations actually bought anything." The FY2024 report says so, verbatim:

5. All domestic airports: handheld jammers have been procured, investigated through the regional joint-defence mechanism, and used to counter the drone as the situation requires, so as to reduce as quickly as possible the impact of drone incursions on aircraft movements.

The same report records the division of countermeasure deployment: Songshan is contracted to NCSIST (the stated reason being that "there are national defence and military installations and other critical facilities around the airport"); Kaohsiung contracted a domestic vendor for a detection system, "with a counter-jamming system to be added later"; Taoyuan is phasing a system in, due for completion in 2024; and Taichung, Chiayi, Tainan, Hualien and Penghu — joint civil-military airports — are covered by the military's own deployment, due end-2024.

This turns the question in the "who may bring it down" post from hypothetical into current fact. That post concluded: you may buy it, import it and hold it lawfully, and no provision says you may pull the trigger. It can now be put more concretely:

• Every domestic airport already holds handheld jammers, and the report explicitly says they are used to "counter the drone as the situation requires" — not held in reserve, but listed as a response measure.
• The lawful public-duty import path the NCC opened dates from 3 February 2025.
• And Article 67(1) — "the use of RF equipment shall not interfere with lawful communications or affect flight safety" — is still unamended.

One sharp piece of corroboration in passing: the FY2024 report quotes the National Audit Office's FY2022 audit report on the Songshan system —

A drone countermeasure system was installed at Taipei International Airport to build a mechanism for detecting and countering drone incursions; however, acceptance testing specifications were not properly established to ensure the capability to detect and locate multiple simultaneous drones, which should be reviewed and remedied.

"The specification has to be able to verify itself" — the National Audit Office said it again in its own words. This series met another version of the same disease in the production-ramp post (the Army's NT$980m fully-terminated contract). The place a counter-drone procurement most often fails is not selection; it is an acceptance clause that cannot state how the thing proves it worked.

The money column: NT$2.7m budgeted for a year of fines, against a NT$300,000 floor

Table 1 of the same report is the CAA's FY2021 drone-related revenue budget, in thousands of NT dollars:

Two computable things:

First, 71% of what this agency takes in from drones is not fines. Review plus licence fees are 6,565 of 9,265 thousand, or 70.9%; fines are 2,700 thousand, or 29.1%. On drones this is a licensing agency, not a penalty agency.

Second, the whole year's fine budget is NT$2.7m. Article 118-1 fines an incursion into prohibited airspace or the announced airport zone NT$300,000 to NT$1.5m. NT$2.7m divided by the NT$300,000 floor is nine cases.

A cleaner figure (update 2026-08-09): the FY2022 evaluation report opens by noting that the CAA's FY2022 budget put fines at NT$11.7m, "including NT$4.5m of penalty revenue for releasing objects hazardous to flight safety around airports". That line is specific to the airport perimeter, which makes it a better fit than the FY2021 figure covering all drone violations. NT$4.5m divided by the NT$300,000 floor is 15 cases — still a budget line rather than receipts, but with a much cleaner denominator, and 15 against the 9 actual closures in 2022 is a consistent order of magnitude.

Being honest about this: the Act also has Article 118-2, fining failure to register and similar violations NT$30,000 to NT$150,000, and those cases are far more numerous. So NT$2.7m cannot simply be read as "nine expected airport-grade violations". Both readings hold: if the NT$2.7m is mostly Article 118-1, that is about nine cases a year; if it is mostly Article 118-2, then Article 118-1 cases barely register in the budget at all. Both readings point the same way: airport-grade penalties are not the main event in the regulator's own budget.

The penalty targets "the owner or the operator", and the base was unmeasured

Article 118-1 penalises "the owner or the operator of the drone". To fine a person, you first have to know whose drone it was.

Table 3 of the same report, registrations as of 27 August 2020:

And the Budget Center's assessment:

However, the Administration has had no control or statistical data on the actual number of drones, and therefore cannot ascertain whether all owners have registered as required.

Those three facts interlock into the real shape of the airport cell: the airport is authorised only to enforce against → enforcement requires catching a person → catching a person requires identification → identification runs on registration, and the registration base was unmeasured. Every link in that chain points after the fact. None of them points at the minutes when the aircraft are stopped.

Three conclusions

1. Closure time is a directly readable cost, and the government has already computed it. The 300 minutes, 47 flights and 5,929 passengers do not need estimating — they are in a public Legislative Yuan document. Anyone assessing whether an airport counter-drone system is worth its price should start from this table, not from a vendor's threat scenario.
2. A "regional joint-defence mechanism" is an organisational answer, not a temporal one. Combining airport stations, Aviation Police, local police and the military resolves who should turn up; the closures ran 8 to 68 minutes, and joint defence addresses who comes, not how fast. This is the same disease the counter-drone post described as an authorisation-level problem: no org chart, however complete, fills in an action the statute never wrote.
3. Procurement ran ahead of the rules by five years. The CAA instructed five airport stations to buy jammers in 2020; the NCC opened the import path in 2025; the use prohibition is still unamended. This is not anyone's oversight — it is Taiwan's normal ordering for new equipment: buy first, work out whether you may use it later. So the buyer's question is not "is this a good unit" but "which provision applies when I pull the trigger."

What this post does not answer, and whether those gaps could refute it

• No closure list after August 2020. Filled (2026-08-09), see the two sections above: the FY2022 and FY2024 reports carry the series to end-August 2023, the count did not fall, and 2022 is the peak — so the third conclusion is strengthened, not softened. Still missing: anything after September 2023. I have not worked through the FY2026 report, and that is the period that would actually test the Songshan pilot, since the project runs 2020–2025.
• No check on whether those five stations actually bought or used jammers. Filled (2026-08-09): the FY2024 report states that "all domestic airports have procured handheld jammers … and counter the drone as the situation requires". Still missing: how many, at what price, how often used, and how Article 67(1) is handled when they are — none of which the reports say, and the last is the one worth knowing.
• A new gap: the two reports disagree about the same events (Taichung 2022-07-26 at 6 vs 1 flights; the Kaohsiung event dated 2022-07-26 vs 2022-09-04; the 2020-01-10 Magong event missing from the FY2021 report). I did not ask the CAA which version is right, nor find a third document to cross-check. Could this refute the post? Not the trend or the magnitude, but it bounds how these numbers may be used — not for incident-level inference.
• I did not read the airport interference SOP itself. There is no public version; I have only the evaluation report's one-sentence paraphrase. Could this refute the post? Possibly: if the SOP specifies concrete removal measures, then "all six response cells are the same sentence" is a coincidence of summary-level reporting. But note that an SOP is an administrative rule and cannot create an authority the Civil Aviation Act withheld from airports — so even a detailed SOP leaves the previous post's textual problem intact.
• No geographic analysis behind "Songshan is urban, hence most affected." That sentence is flagged in the text as a common-sense explanation, not a conclusion.

How to check the numbers

Every figure comes from Tables 1, 2 and 3 of one PDF, so it can be checked directly:

Legislative Yuan Budget Center
Evaluation Report on FY2021 Unit Budgets (TTSB and MOTC agencies)
https://www.ly.gov.tw/Pages/ashx/File.ashx?FilePath=~/File/Attach/201989/File_276856.pdf

Table 1 = FY2021 drone revenue (fines 2,700 / review 4,150 / licence 2,415, NT$ thousand)
Table 2 = six closures, FY2019 to end-Aug 2020 (source: CAA data of 2020-09-14)
Table 3 = 39,379 registrations as of 2020-08-27, split three ways

Later series (added 2026-08-09):
https://www.ly.gov.tw/Pages/Detail.aspx?nodeid=45904&pid=224510
  FY2022 unit budget evaluation report → 13 occurrences, 2020 to end-Aug 2022
  (CAA data of 2022-09-14)
https://www.ly.gov.tw/Pages/Detail.aspx?nodeid=46508&pid=233504
  Civil Aviation Operating Fund FY2024 report → 12 occurrences, 2022 to end-Aug 2023
  (CAA data of 2023-09-11; includes the handheld-jammer statement and the audit finding)

The totals (300 minutes, 47 flights, 5,929 passengers, 70.9% / 29.1%) are my additions; the original has none. All six addends are in the table above and can be re-summed.

References

Primary: Legislative Yuan and agency documents

• Legislative Yuan Budget Center — Evaluation Report on the FY2021 Unit Budgets of the TTSB and Ministry of Transportation agencies (Table 2, the six itemised closures, response column, and CAA remedial actions including the letter instructing five stations to buy jammers; Table 1 revenue; Table 3 registrations)
• Legislative Yuan Budget Center — FY2022 unit budget evaluation report, item 2 (13 itemised occurrences from 2020 to end-Aug 2022; NT$4.5m of the FY2022 NT$11.7m fine line attributed to airport-perimeter violations; "jammers procured at the higher-traffic airports")
• Legislative Yuan Budget Center — Civil Aviation Operating Fund FY2024 budget evaluation report, item 3 (12 itemised occurrences from 2022 to end-Aug 2023; per-airport countermeasure deployment; "all domestic airports have procured handheld jammers … and counter the drone as the situation requires"; quotes the National Audit Office's FY2022 criticism of the Songshan acceptance-testing specification)
• CAA — announced prohibited/restricted airspace and the distance around airports under Article 99-13(1), as map data (the range is announced as a geodata file; the announcement itself states no distance figure)
• CAA — drone section (drone chapter promulgated 2018-04-25, in force 2020-03-31)

Primary: statutory text

• Civil Aviation Act, Article 99-13 (para 1, the CAA announces the airport range; para 6, airports "enforce against together with the Aviation Police Bureau")
• Civil Aviation Act, Article 118-1 (NT$300,000–1.5m, licence revocation, confiscation)
• Civil Aviation Act, Article 118-2 (registration and related violations, NT$30,000–150,000)
• Telecommunications Management Act, Article 67 (use shall not interfere with lawful communications or affect flight safety)
• Regulations Governing Manufacture, Import and Reporting of Controlled Telecom RF Equipment, Article 3 (reissued 2025-02-03; Class 1 includes drone-restricting equipment)

On this site

• Who May Bring a Drone Down? Taiwan's Law Authorises the Outcome and No Means of Achieving It
• Why Countering Drones Is Hard: Jamming Is Failing, and Taiwan's Problem Is Not Only Technical
• Police May Lawfully Fly a Drone, but That Is Not Authority to Film
• Taiwan's Drone Rules in Plain Language: What Must Be Registered, What Needs a Licence, What Gets You Fined
• Taiwan's Drone Industry Map