Flight Radiation logoFLIGHTRADIATION

Corrections

We publish every factual correction we make to a public page or a report template. If a correction affects a previously delivered report, we re-issue the corrected report to the buyer free of charge.

2026

Reviewed 4 September 2026 · Report errors to [email protected]

4 September 2026 · The accuracy figure we published was measured the wrong way

What was wrong. We published a mean absolute deviation of 11% against the FAA's 27 published route doses. Our flight profile has a fitted parameter, and it had been fitted against those same 27 flights, so that figure told you how well the model fits data it had already seen. It is not what you want to know.

What it actually is. Refitting on 26 flights and scoring the 27th, in turn, the deviation is 13.3% when you give us the cruise altitude and flight time and 17% when we estimate them from the route. We had also only ever measured the first case, while most customers supply neither. Both are now published, separately, on the validation page, and reports carry the band that applies to each flight.

4 September 2026 · The uncertainty band cited a paper that does not say what we said

We stated that Meier et al. (2018) found CARI-7A "within 3 to 8%" of measurement. That phrase is not in the paper. It reports per-point deviations from -8% to +11%, compares ambient dose equivalent and absorbed dose rather than the effective dose we report, and concludes the upper limit of the model's uncertainty is about 20%. We had also combined that figure with the in-sample residual above, which was not a meaningful thing to do. The band is now measured directly and CARI-7's own distance from measurement is stated separately.

4 September 2026 · The 6 mSv figure was attributed to the FAA, including on the paid report

Pages across this site, and the PDF itself, described a "FAA aircrew action level" of 6 mSv a year. Advisory Circular 120-61B contains no such figure. It recommends 20 mSv a year averaged over five years, no more than 50 mSv in any single year, and no more than 0.5 mSv in any month once a pregnancy is known. The 6 mSv is the level at which Directive 2013/59/Euratom requires airlines to manage crew rosters. Separately, the 1 mSv gestational figure was attributed to the same Advisory Circular; it is ICRP Publication 103.

4 September 2026 · Historical flights used the present-day magnetic field

Cosmic ray shielding depends on Earth's magnetic field, which drifts. CARI-7 ships six grids spanning 1965 to 2010 and interpolates between them by flight date; we shipped only the 2010 grid. A flight dated 2000 could be out by up to 0.72 GV of cutoff rigidity, and one dated 1970 by 2.53 GV. All six grids are now shipped and blended the way CARI-7 does it, matching it to 0.005 GV across dates from 1960 to 2026.

4 September 2026 · Smaller engine corrections

Round trips were computed as twice the outbound leg; the flight profile is not symmetric about the track, so the two legs differ by up to 6% and the return is now computed separately. The flight-time estimate treated 27 minutes of taxi as time in the air, inflating short sectors by about a third; it is now fitted to the FAA's own published air times. Climb and descent ground speed now matches CARI-7's own half-cruise assumption. The profile had four fitted parameters where the data can only identify one, so three are now stated as assumptions rather than dressed up as fits.

4 September 2026 · The dose model was overstating by about a factor of two

What was wrong. Until this date the site computed dose from a hand-fitted curve of dose rate against geographic latitude, pinned to solar-minimum conditions. Checked against the 27 route doses the FAA published in Advisory Circular 120-61B, which the FAA computed with CARI-7, that curve came out between 1.75 and 2.38 times too high on every single route, a mean of 1.99 times.

Why. Two independent faults. The curve had no solar-cycle term at all, so it reported solar-minimum values year round, which alone runs about 55% high on polar routes in 2026. It was also keyed to geographic latitude, which does not govern dose: what matters is geomagnetic cutoff rigidity, and at 40 degrees north that is about 3.6 GV over the Atlantic against about 9.0 GV over Asia, a difference of nearly a factor of two at identical latitude.

What changed. The engine now interpolates a table generated by running the FAA's own CARI-7 v4.1 build across cutoff rigidity, atmospheric depth and heliocentric potential. Cutoffs come from the Shea and Smart IGRF 2010 grid distributed with CARI-7, and solar modulation from the FAA's published monthly series, so the month you flew now changes the answer. Given the same flight profile the engine reproduces CARI-7 to within 1%, and it reproduces the FAA's published route table to a mean absolute deviation of 11% with no systematic bias. The full route-by-route comparison is on the validation page.

Who was affected. Every report issued before this date. If you bought one, email [email protected] and we will recompute and reissue it at no charge, or refund you, whichever you prefer.

4 September 2026 · Pages described capabilities the software did not have

Several pages stated that supplying flight dates would have them cross-checked against NOAA's solar particle event record, and that reports contained a "polar-route attribution metric" giving the share of dose received above 60 degrees. Neither existed. The order form did not even collect a date. Those claims have been removed. The form now does collect a date, and it is used to select the solar conditions for that month, but solar particle events are still not modeled and the report says so.

4 September 2026 · Methodology page overstated what the reports contained

The methodology page claimed that heliocentric potential tables were refreshed monthly, that every report listed the model version and the potential table month, and that a reader could reproduce every number in the FAA's public CARI-7A web tool. None was true at the time. The page has been rewritten. Reports now do carry a provenance table listing the model version, the cutoff grid, the solar months used and the uncertainty band. Separately, the FAA's own public web calculator has been unavailable for some time, so the reproduction route is the downloadable software.

4 September 2026 · Pregnancy comparison was scoped wrongly

The report compared an annual dose total against the 1 mSv figure that applies across a whole pregnancy, and applied a limit written for declared pregnant radiation workers to ordinary passengers without saying so. It now compares the itinerary as listed, states plainly that 1 mSv is an occupational protection limit rather than a clinical threshold, and cites the ACOG Committee Opinion 656 reference of 50 mGy alongside it.

4 September 2026 · Several numerical claims in the guides were wrong

Corrected: climbing 6,000 feet was said to "roughly double" the dose rate (it raises it about 46%); polar transatlantic routes were said to carry "20 to 40% more" dose per hour than equatorial ones (about three times); ten polar transatlantic round trips a year were given as 4 to 6 mSv (about 1.1 mSv); the 1 mSv pregnancy figure was given as 30 to 50 hours of transatlantic flying (about 127 hours). A statistic attributed to a sample report "Subject A" was removed: there was no such report and the figure should never have been published.

4 September 2026 · A data study was withdrawn

An unlinked page publishing dose figures for 20 named routes has been taken down rather than corrected. It used a chest X-ray dose five times smaller than the rest of the site, was not run through the engine the site actually sells, disagreed with its own route count, and presented London to Sydney as a nonstop service, which does not exist. If it returns it will be recomputed from scratch.

4 September 2026 · Airport coverage and smaller fixes

Airport coverage went from 104 hand-listed hubs to the 4,133 airports with an IATA code and scheduled service, so secondary airports are no longer refused at checkout. A "child" passenger profile was removed: it was collected and then ignored, producing output identical to an adult. The natural background figure, which appeared on the site as 0.39, 3, 6.2 and 0.62 mSv per year in different places, is now consistently 3 mSv per year for total US background, citing NCRP Report 160 rather than NCRP 184, which covers medical imaging.

How to report an error

Email [email protected] with the page URL, the specific claim that is wrong, and a primary source citation if you have one. We aim to acknowledge within two business days and to log the correction within a week. Speed matters more for numeric errors than for typographical ones, so we triage accordingly.

What counts as a correction

  • A numeric value that does not match its cited source.
  • A misattributed source, for example a quote attributed to ICRP-103 that is actually from ICRP-132.
  • An out-of-date publication number (e.g. citing FAA AC 120-61A after 120-61B was issued).
  • A factually incorrect explanation of how the CARI-7 model works.

What does NOT count as a correction

  • Editorial choices about which sources to feature (we are happy to discuss them, but they are not errors).
  • Differences of opinion on the appropriateness of a given dose limit for a given context. We report what ICRP, FAA, NCRP say, and we are not the standards body.
  • Typos, broken images, or layout issues. Please report these as bug reports rather than corrections; they go through the same email.

Reviewed 4 September 2026