How LogAloft works out your radiation dose

Every flight in LogAloft gets a cosmic radiation dose in microsieverts. This article explains where that number comes from, step by step, so you can judge it for yourself. If you want the basics first, start with what cosmic radiation is and how much a flight gives on 14 common routes.

What the number is

The dose LogAloft records is effective dose from galactic cosmic rays, received along the path the aircraft flew or would most likely fly. Effective dose weights each kind of radiation and each organ by the harm it can do, using the weighting of ICRP Publication 103, the current international standard. Because every source is converted the same way, effective doses from a flight, an X-ray and a CT scan are in the same unit and add up.

It is a calculation. Nothing on board is measured or read. The model works out how many particles reach the aircraft at each point of the flight and what dose they would give a standard human body.

In Settings you can show two other quantities from the same calculation. ICRP 60 effective dose, the 1990 weighting some airlines still report, comes out about 26 to 33% higher at cruise, mostly because it counts neutrons more. H*(10), the quantity onboard dosimeters are calibrated to read, runs about 4 to 17% above ICRP 103. Your logbook record always stays in ICRP 103.

The model: CARI-7A

LogAloft uses CARI-7A, version 4.2.0, from the FAA's Civil Aerospace Medical Institute. The FAA describes it as a program that computes "the effective dose of galactic cosmic radiation" received on an aircraft. It follows cosmic rays from outside the solar system, lets the Sun and Earth's magnetic field filter them, and tracks the particle showers they set off in the air.

Running CARI-7A for every minute of every flight would be slow, so it was run ahead of time on a grid of 3,168 points:

• 24 levels of geomagnetic cutoff rigidity, from 0 to about 17.6 GV;
• 11 levels of solar modulation, from 228 to 1,872 MV;
• 12 altitudes, from 10,000 to 45,000 ft.

A flight's dose is the dose rate read from that grid once a minute along the path, multiplied by the time, and added up. Between grid points the rate is interpolated, and between two altitudes it is interpolated on a logarithmic scale, because the rate grows almost exponentially as the air above thins. Checked against CARI-7A run directly at 1,488 points, the grid agrees on average and is within 1.5% everywhere.

Below 10,000 ft, a curve taken from CARI-7A runs at polar and equatorial latitudes scales the rate down. At sea level it is about 16% of the 10,000 ft rate at high latitudes and about 20% near the equator. Climb and descent therefore count, but they add little.

No shielding by the aircraft is applied, which is how airlines and regulators work too. The structure and the cabin's contents lower the dose a little, so leaving them out errs on the high side.

The four inputs

Cutoff rigidity is the shield Earth's magnetic field gives at a place. Near the equator only very energetic particles get through; near the magnetic poles almost all do. The field drifts slowly, so LogAloft blends the two map years on either side of the flight's date and, after 2010, holds the last map, as CARI-7A does.

Solar modulation is how hard the Sun's wind pushes cosmic rays away. The FAA publishes a monthly heliocentric potential in megavolts for use with CARI, with "a lag of up to one month." For months it hasn't published yet, LogAloft uses a nowcast from the Oulu neutron monitor, fitted to the FAA table from 2010 to 2026. Its error is about 24 MV, or about 1% in dose. The effect of the Sun is large. With everything else held the same, New York–London comes out at about 41 µSv in October 2009, at the last solar minimum (220 MV), and about 24 µSv in January 2026, a month of high activity (1,185 MV).

Where the route and the levels come from

Dose depends most on how long you spend at cruise, how high and how far from the equator. LogAloft uses the best path it has, and each flight says which one it used.

The track flown. On Pro, about 30 minutes after landing the app fetches the positions and altitudes recorded by flight-tracking networks, for flights within the last 10 days, and the track replaces the estimate. You can also attach a GPS or avionics file. Ocean crossings often have gaps with no positions. If you typed your cruise levels, any gap over 20 minutes at cruise gets your level for that time. Otherwise a long gap at cruise is treated as level flight with one level change, placed where the track shows the aircraft was climbing or descending, or halfway when it doesn't.

The estimate. For a flight not yet flown, or with no track, LogAloft plans the path a dispatcher would plan for that day's winds, using the airways, the North Atlantic or Pacific tracks, the filed route when the flight number has one, the path the same airline has flown between those airports in recent weeks, and the airspace closed to that airline on that day. Air time is the block time less 21 minutes for taxi. The climb is set at 35,000 ft in 25 minutes and the descent at 35,000 ft in 30, so a lower cruise level takes less time to reach.

Step climbs. The estimate uses the filed cruise level when the flight number has one, with step climbs typical of the aircraft type. With nothing to go on, it cruises at FL350 and climbs 2,000 ft at 2.4 hours after takeoff, then every 3 hours, never above FL400 and only while more than 1.2 hours remain.

The levels you type. Crews can enter their levels in the flight, either by time ("FL350, FL370 at 3:00", hours after takeoff) or by place ("FL350, FL370 at 30W", as on an oceanic crossing). Those win over the estimate.

The levels matter more than most people expect. For New York–London, 6.8 hours in the air under July 2026 solar conditions, LogAloft gives:

On 20 long-haul flights flown in September 2026, the planned path gave a dose within 3% of the dose along the real track on average, and within 9% when the times and levels also had to be predicted. Across 42 flights of all lengths, the estimate made before departure was off by 3 µSv on average.

Solar storms

The logbook record is galactic cosmic radiation only. Solar storms are an optional layer, off by default, shown on their own line on each flight and added to totals only if you ask. Solar storms and your flight explains the space weather itself; here is what LogAloft does with it.

Geomagnetic storms weaken the magnetic shield for a day or two. When GFZ Potsdam's Kp index is 5 or more, LogAloft lowers the cutoff rigidity the way CARI-7A does it. The cutoff is multiplied by a factor that runs from 0.89 at 0.74 GV down to 0.53 at 20 GV, so places near the equator lose the most shielding. In LogAloft's model at FL350 under July 2026 solar conditions, that raises the dose rate by about 55–60% at the equator (about 1 µSv/h more), and by a varying amount at mid-latitudes because the shield there depends on longitude as well as latitude: at 45°N about 1% over North America, about 22% over Europe and about 31% over East Asia. Above about 60°N it changes little, 2% or less, since there was little shield to lose. Over a whole New York–London flight the storm adds only a few percent.

Solar particle events. With the storm setting on, LogAloft reads NASA's NAIRAS model every hour. NAIRAS computes dose at aviation altitudes from galactic and solar particles. An hour counts as a storm hour when the polar dose at 20 km is 25% or more above the quiet level of the days before. LogAloft keeps the excess over that quiet level and adds it along any flight that flew through it. The record starts with the ground level enhancement of November 11, 2025. For a ground level enhancement in progress, the flight lookup also shows an estimate from polar neutron monitors by the SiGLE method, which Lantos and Fuller calibrated on dosimeters carried aboard Concorde.

How sure the numbers are

Treat a single flight's galactic dose as good to about ±20%. The inputs are few and well known (where you were, how high, the month), and the radiation field changes slowly.

The storm add-on is much less certain, and LogAloft labels it ±50%. On past events, models differed from one another by 30 to 50%, and for a single event the error can approach a factor of two. A neutron monitor on the ground counts what reaches sea level, while the dose at FL350 depends on the storm's energy spectrum and direction, which a count rate can't tell.

Validation

The most careful published check of aviation dose models, by Meier and colleagues in 2018, compared CARI-7A, PANDOCA and NAIRAS with reference measurements of H*(10) made on board at known positions, altitudes and dates. At those six points LogAloft was within 5% of the measurements on average and 10% at most.

Against the reference dose rates of ICRU Report 84, a fit to about 20,000 onboard measurements, LogAloft is 4% below on average across 162 points. Against 63 flights measured by Iberia and Spain's CIEMAT in 2001 to 2004, many of them to South America, it came out 6% below, every flight within the ±30% Spain's regulator requires. The method page gives every comparison with its tables.

What the numbers mean

A two-view chest X-ray, front and side, gives about 0.1 mSv (100 µSv), the figure in the catalog by Mettler and colleagues in Radiology. A New York–London flight at mid-2026 solar conditions is about 30 µSv, roughly a third of one. LogAloft shows each flight and each year in chest X-rays for that reason.

For risk, radiation protection uses the ICRP's estimate. ICRP Publication 103 puts the combined detriment from cancer and heritable effects at "around 5% per Sv". By that estimate, a career of 30 years at 3 mSv a year, 90 mSv in all, adds about a 1 in 220 lifetime chance of dying of cancer. The estimate assumes risk grows in step with dose with no threshold, and it carries three limits worth knowing. Cosmic radiation at cruise is largely neutrons and high-energy particles, unlike medical X-rays. It is spread evenly through the body. And the risk at low doses is uncertain, which is why the figure is an estimate for protection, not a forecast for one person.

The cancer study published in August 2026 shows what crew epidemiology can and can't say.

Limits of the model

• Your seat isn't modeled. The dose is for the aircraft's position with no shielding. The real dose in a cabin is a little lower.
• The estimate is only as good as its levels. An estimate assumes the default step climbs unless you or the filed route say otherwise. Type your levels, or let the flown track replace it.
• The magnetic field after 2010 is held at the 2010 map, as in CARI-7A. The field drifts slowly, but it does drift.
• Recent months use a nowcast until the FAA publishes the month, at about 1% in dose.
• Storm doses are rough, as described above, which is why they never enter the record unless you add them.
• Effective dose describes a standard body, not yours.

What LogAloft is not

LogAloft is not a dosimeter, and nothing is measured on your flight. It is not a regulatory dose record either. In the EU, where crew doses are liable to be above 1 mSv a year, Council Directive 2013/59/Euratom, Article 35, requires the operator to assess them, take them into account in rostering and tell crew their individual doses. That record is your airline's. In the US no rule requires one at all. LogAloft gives you your own record, worked out the same way the FAA's tools do it, to keep and compare.

Nor is it medical advice. What a dose means for your health, and especially for a pregnancy, is a question for your doctor or aviation medical examiner.

Sources

1. CARI-7 and CARI-7A, FAA Civil Aerospace Medical Institute
2. Recent and historic heliocentric potentials, FAA Civil Aerospace Medical Institute
3. ICRP Publication 103: The 2007 Recommendations of the International Commission on Radiological Protection (free extract), Annals of the ICRP 37(2–4), 2007
4. First steps toward the verification of models for the assessment of the radiation exposure at aviation altitudes during quiet space weather conditions (Meier, Copeland, Matthiä, Mertens and Schennetten), Space Weather 16(9), 2018
5. Effective doses in radiology and diagnostic nuclear medicine: a catalog (Mettler, Huda, Yoshizumi and Mahesh), Radiology 248(1), 2008
6. History of the solar particle event radiation doses on-board aeroplanes using a semi-empirical model and Concorde measurements (Lantos and Fuller), Radiation Protection Dosimetry 104(3), 2003
7. Prototype development of an operational global aircraft radiation exposure nowcast (NAIRAS; Mertens et al.), NASA Technical Reports Server, 2010
8. Council Directive 2013/59/Euratom, Article 35, legislation.gov.uk
9. Method and validation, LogAloft