Subsystem 07 of 12 · Radiation
Lunar Habitat ACT, the Autonomous Construction Testbed·Radiation·Concept definition, revision A·August 2026·See it in the tour

A shelter first, and two metres of soil later

There is no magnetosphere and no atmosphere. Two separate threats, on completely different timescales, and only one of them can be planned for. In the built increment the modules stand unshielded on a compacted regolith skirt, so the routine dose is the open surface figure, about 380 mSv a year, which is about 190 mSv over a 180 day rotation. A water jacketed shelter under the core deck takes the storm, and rotation length and extravehicular activity (EVA) limits carry the rest. The two metres of regolith that takes the routine dose under the career limit comes in a later increment.

380 mSva year unshielded, about 190 per 180 day rotation
12 g/cm²storm shelter, the only shielding built
2 mregolith planned, a later increment
30 minsolar particle event warning
Contents
  1. Two threats, not one
  2. The two metre target
  3. The storm shelter
  4. Dose budget
  5. The cupola, and whether it is worth it
  6. Materials and the secondary problem
  7. What is not yet known
  8. Increment plan, shelter first
  9. Sources and further reading
Section 01

Two threats, not one

Galactic cosmic rays (GCR) and solar particle events (SPE) are usually discussed together and should not be. They differ in energy, in timescale, and in what can be done about them.

Galactic cosmic rays are a continuous, low intensity flux of very high energy nuclei from outside the solar system. They are not stoppable in any practical sense: a few centimetres of shielding makes things worse, because the primary shatters into a shower of secondaries. Only mass in bulk helps, and even then the return diminishes.

Solar particle events are sudden, intense bursts of mostly protons from the Sun. They are far lower energy and therefore genuinely stoppable, but they arrive with very little warning and can deliver a career dose in hours.

PropertyGalactic cosmic raysSolar particle events
SourceOutside the solar systemThe Sun, flares and coronal mass ejections
EnergyUp to 10¹⁰ MeV, GeV typical10 to 500 MeV typical
TimescaleContinuous, always presentHours, a few times a solar cycle
Annual dose, unshieldedabout 380 mSv0 to 1,000+ mSv in a single event
ShieldablePoorly, and secondaries make it worseYes, effectively
CountermeasureBulk regolith, and time limitsA shelter, and warning
Section 02

The two metre target

Shielding effectiveness is measured in areal density, grams per square centimetre, not in thickness. Regolith is about 1.5 g/cm³ loose near the surface and closer to 1.8 at depth or compacted, so two metres of it is roughly 300 g/cm² or a little more. That is a great deal of shielding by any standard.

The unshielded figure needs a word, because it is quoted three ways. The one measurement is Chang'e 4's Lunar Lander Neutron and Dosimetry instrument, which recorded 1,369 microsieverts a day on the far side in 2019, at solar minimum when galactic cosmic rays are strongest: about 500 mSv a year as dose equivalent in a small detector. A body shields itself, so the effective dose to a person is lower; Matthiä and Berger's 2024 transport calculation puts it near 306 mSv a year for the same conditions, and Reitz, Berger and Matthiä estimated about 380 mSv a year at solar minimum and 110 at solar maximum in 2012. This document uses 380, the solar minimum estimate, as its design value, and it is the figure every table below starts from.

The curve below is the reason for that number and not a larger one. GCR dose falls steeply at first, then flattens, and past about 250 g/cm² the secondaries produced in the shield itself start to offset further gains. Two metres sits just past the knee.

The transport calculations published since put the knee in a less flattering place, and the chart should be read with them in mind. Matthiä and Berger find that the first 20 g/cm² of material cuts the effective dose rate by about a quarter, that secondaries then hold it nearly flat until well past 100 g/cm², and that 180 g/cm² of regolith gives about 190 mSv a year against 306 unshielded, with neutrons by then about half of the dose equivalent. Horst and colleagues' 2022 Monte Carlo study gets below about 100 mSv a year only with 400 g/cm² and a polyethylene liner inside it. The 126 mSv the chart shows at two metres is therefore a design estimate at the optimistic end of the published range, and the number to carry until a transport calculation is run for this hull and this cover lies somewhere between 126 and 190.

None of it is in place yet. The base is shown at its early increment: each module stands on a compacted regolith skirt to about a fifth of its height, which anchors the hull and protects its foot but is not radiation shielding. So the modules sit at the left edge of the chart below, at the unshielded figure, and bulk regolith over the vessels is the increment two job in section 08.

0 100 200 300 400 0 0.5 1 1.5 2 2.5 3 Regolith depth, metres Dose, mSv per year Galactic cosmic rays SPE, worst case SPE off scale: 1000 mSv at zero depth Career limit
Solar particles are essentially gone by one metre. Galactic cosmic rays are not, and never will be: two metres, once placed, takes 380 mSv down to about 126 in this design estimate, which brings a 180 day rotation to about 78 mSv all in and lets a crew member fly several rotations inside NASA's 600 mSv career limit. Without it, at 380, a rotation costs about 200 mSv and a career is three rotations. That difference is what the cover buys. In the built increment the modules sit at the left edge of this chart, at 380. The SPE curve starts at 1000 mSv at zero depth, above the top of this scale, and is clipped there.
Section 03

The storm shelter

Two metres over the habitat will handle the routine case once it is placed. Bulk regolith does not handle a large solar particle event either, because during one the dose rate inside a normally shielded module can still be significant, and because the crew may be outside when it starts. In the built increment the shelter is the only shielding on the base, so it matters more, not less.

The shelter is a volume under the habitation core deck, surrounded by the water and waste tanks the base has to carry anyway. Water is an excellent proton shield per unit mass, and putting the tanks around the shelter costs nothing that was not already in the mass budget. The whole crew fits, uncomfortably, for the duration of an event.

12 g/cm²additional shielding
4crew capacity
90 stime to seal
72 hprovisioned duration
The twenty minute rule

The first protons of a large event can reach the surface within about thirty minutes of the flare, and that is the warning the shelter is sized to; the forecast services being built for Artemis aim to give hours to a day of notice from the state of the Sun, which is what EVA planning uses. Thirty minutes sounds like plenty until you are 800 metres out at the excavation face in a suit. The rule is that no crew member operates further from an airlock than they can return from in twenty minutes, leaving ten for ingress and for the cupola shutter to close.

Section 04

Dose budget

The table is the design case, with two metres of regolith in place. In the built increment the modules are unshielded except for the storm shelter, so the inside rate is the open surface rate, the 0.043 mSv an hour of the EVA row, and the routine dose is about 380 mSv a year: 380 divided by 365 is about 1.04 mSv a day, so a 180 day rotation collects about 190 mSv before the transit dose. With the transit dose that is about 200 mSv a rotation, inside the 250 mSv per rotation budget this design uses, but a third of NASA's 600 mSv career limit spent in one stay. The budget is chosen against that standard, NASA STD 3001, which sets the career limit at 600 mSv effective dose and also caps the dose to blood forming organs at 250 milligray equivalent in any 30 days and 500 in a year, the short term case the shelter exists for. The hull's own structure, around 20 g/cm², is not credited, which is conservative: the 2024 transport calculation finds the first 20 g/cm² of material cuts the effective dose rate by about a quarter before secondaries flatten the curve. Until the cover exists the dose is managed by the shelter, by the 180 day rotation and by the EVA hours below, which are a budget and not an estimate.

ExposureDose rateHours per yearAnnual doseNote
Inside, shieldeddesign case, under 2 m regolith0.0144 mSv/h6,57094.6 mSv75 percent of the crew's time
Inside the cupolaunshielded aperture0.036 mSv/h36513.1 mSvOne hour a day, and worth it
EVA, suitedextravehicular activity, no bulk shielding0.043 mSv/h91239.2 mSv2.5 hours a day of surface work
In the shelterduring events0.002 mSv/h480.1 mSvTwo events a year, 24 hours each
Transit, Earth to Mooneach way1.2 mSv/day1929.6 mSvFour days each way, twice
Total per 180 day rotation78.3 mSvAgainst the 250 mSv per rotation budget

The EVA contribution is the one that constrains operations. Surface work is 12 percent of the crew's time and 42 percent of their dose, which is the argument for autonomy doing more of the construction: every hour a machine works outside is an hour a person does not.

Section 05

The cupola, and whether it is worth it

Once the cover is placed the cupola will be the only unshielded volume on the base, and it costs 13 mSv a year, about 17 percent of the crew's exposure, for an hour a day. In the built increment every module shares that exposure, so the trade here is about the design case. Purely on a dose budget it is indefensible.

It is defended anyway, on evidence. Every long duration isolation study, from Antarctic winter over to Mir to the station, reports the same thing: a window is among the most valued features of the environment and its absence is repeatedly cited in crew debriefs as a stressor. The glazing is cerium doped to resist browning, four panes deep, and behind a shutter that closes in ninety seconds.

The trade is stated here rather than buried because it is a real one and somebody will eventually ask why the safest option was not taken.

Section 06

Materials and the secondary problem

Not all shielding is equal per unit mass. Hydrogen rich materials are better, because a proton scattering off a hydrogen nucleus of similar mass transfers energy efficiently, while a proton hitting a heavy nucleus produces a shower of secondary neutrons and gamma rays that can be worse than the primary.

MaterialRelative effectiveness per kgAvailable hereUsed for
Liquid hydrogenthe theoretical best1.00NoNot carried
WaterH₂O0.73Yes, already carriedThe storm shelter jacket
PolyethyleneCH₂0.71Shipped onlySleep station liners
Regolithsilicates0.42Unlimited, on siteBulk shielding over every module, a later increment
Aluminiumstructure0.38Structural onlyNot used as shielding

Regolith is not the best shielding material. It is the one that is already there in unlimited quantity, and 300 g/cm² of a mediocre shield beats 20 g/cm² of an excellent one that had to be launched.

Section 07

What is not yet known

In the built increment the modules are unshielded except for the storm shelter, so the routine dose is the open surface figure, about 380 mSv a year and about 200 mSv per 180 day rotation with the transit: inside the 250 mSv rotation budget, but a third of the 600 mSv career limit in one stay, with no margin for a large event outside the shelter. The open surface figure itself is an estimate; the one measurement, Chang'e 4's, reads higher as dose equivalent and lower as effective dose once the body's self shielding is counted, and the solar cycle moves it by a factor of three. What two metres of regolith buys is the largest uncertainty in this document: the chart's 126 mSv a year is at the optimistic end of the transport calculations, which give 190 mSv at 180 g/cm² and reach 100 only with 400 g/cm² and a hydrogen rich liner, because secondary neutrons from the shield itself offset more of the gain than the simple curve allows. No transport calculation has been run for this hull, this cover and this ground. The cupola rate in the dose table, 0.036 mSv an hour, and the storm shelter's 12 g/cm² jacket are design allocations, not calculations for the glazing, shutter and tank geometry as drawn. A large solar particle event can still deliver a significant dose inside a normally shielded module, which is why the shelter exists. The cupola costs 13 mSv a year in the design case and is defended on crew evidence rather than on dose.

Section 08

Increment plan, shelter first

Increment 1, the shelter and the skirt

The modules land with their own structure only, around 20 g/cm², and a compacted regolith skirt is built against the foot of each hull. The storm shelter arrives inside the core, so its water jacket is plumbed and provisioned in this increment; until it is, a large solar particle event means an emergency return. The routine dose is the open surface figure, about 190 mSv per 180 day rotation, so EVA hours are held to the budget in section 04, and at about 200 mSv a stay a crew member's career under NASA's 600 mSv limit is three rotations. This is a real constraint on the first rotations and is stated as such. It is the increment the site shows.

Increment 2, two metres over everything

The excavator places bulk regolith over every pressure vessel. This is the step that takes the routine dose to 126 mSv a year and gives the 180 day rotation its margin.

Increment 3, dosimetry and prediction

Personal dosimeters logged continuously, and a tie into the feed from the solar wind monitors at the Sun Earth L1 point so warnings arrive automatically rather than by voice call from Earth.

Sources

Sources and further reading

Every figure in this document traces to one of these. Agency documents and peer reviewed work first, reporting only where it is the primary record.

Zhang et al., First measurements of the radiation dose on the lunar surface, Science Advances 6, eaaz1334, 2020 https://www.science.org/doi/10.1126/sciadv.aaz1334Chang'e 4 LND: 1,369 microsieverts a day on the open surface at the 2019 solar minimum, the one measurement
Reitz, Berger and Matthiä, Radiation exposure in the moon environment, Planetary and Space Science 74, 2012 https://doi.org/10.1016/j.pss.2012.07.014About 380 mSv a year from galactic cosmic rays at solar minimum and 110 at solar maximum; the 380 design value
Matthiä and Berger, Radiation exposure and shielding effects on the lunar surface, Space Weather 22, 2024 https://doi.org/10.1029/2024SW004095Effective dose about 306 mSv a year unshielded and 190 behind 180 g/cm² of regolith; the first 20 g/cm² cuts about a quarter; neutrons about half of the dose equivalent at depth
Horst et al., Thick shielding against galactic cosmic radiation: a Monte Carlo study with focus on the role of secondary neutrons, Life Sciences in Space Research 33, 2022 https://www.sciencedirect.com/science/article/abs/pii/S2214552422000141Below about 100 mSv a year needs 400 g/cm² with a polyethylene liner
He, A review of lunar regolith radiation shielding using in situ resource utilisation, Journal of Radiological Protection 46, 2026 iopscience.iop.orgThe survey of transport results the depth discussion in section 02 draws on
NASA STD 3001, Volume 1, Revision C, Crew health, 2023 nasa.govCareer limit 600 mSv effective dose; blood forming organ limits of 250 mGy equivalent in 30 days and 500 in a year
Bell and Bannova, Lunar habitat micrometeoroid and radiation shielding: options, applications and assessments, Journal of Aerospace Engineering 24, 2011 ascelibrary.orgThe 0.993 five year no penetration goal and the trade offs of regolith cover
Heiken, Vaniman and French, Lunar Sourcebook, chapter 9, physical properties of the lunar surface, 1991 https://www.lpi.usra.edu/publications/books/lunar_sourcebook/Regolith bulk density about 1.5 g/cm³ near the surface rising toward 1.8 at depth, behind the 300 g/cm² for two metres
Schwadron et al., Update on the worsening particle radiation environment observed by CRaTER and implications for future human deep space exploration, Space Weather 16, 2018 https://doi.org/10.1002/2017SW001803Lunar Reconnaissance Orbiter CRaTER: the measured near Moon environment and its rise as solar activity weakens
University of Michigan, providing the Artemis mission with solar radiation forecasts, April 2026 news.engin.umich.eduForecasts up to 24 hours ahead; the particles themselves arrive within minutes to an hour of an eruption