Martian radiation: shielding, dose and storm shelter

Understand the invisible hazard: GCR, solar events and the Martian environment

On Mars, the most dangerous weather can be invisible

The sky may be clear, winds light and temperatures manageable while energetic charged particles pass through the habitat, ionize matter, upset electronics and damage cells. Radiation is difficult to picture because it does not look like dust, fire or decompression. There is no smell and usually no immediate symptom. A human Mars system nevertheless has to treat it as a continuously monitored environmental variable.

Two broad families dominate the discussion. GCR arrives mainly from outside the Solar System: protons, helium nuclei and a smaller but important population of very energetic heavy ions. It creates a persistent background that changes with solar activity and penetrates modest shielding. SPE, by contrast, is associated with solar activity and can increase particle flux sharply for hours or days. Protection strategy therefore differs: a compact heavily shielded refuge can strongly reduce the acute hazard from a major solar event, while the chronic GCR problem is much harder to eliminate.

Curiosity made this discussion far more concrete. Its Radiation Assessment Detector measured the environment during cruise to Mars and continued on the surface. That provides a rare bridge between interplanetary exposure inside a spacecraft and the Martian surface, where the planet blocks part of the sky and the atmosphere provides some shielding but there is no Earth-like global magnetic field or thick atmosphere.

This page therefore avoids the misleading idea of one universal “Mars radiation number.” Source, spectrum, energy, geometry, solar conditions, altitude, shielding, mission duration and biological metric all matter.

The operational lesson is that radiation cannot be delegated to a wall. Trajectory design, warning systems, shelter access, rover procedures, EVA scheduling, medical follow-up and settlement zoning all contribute to exposure. If any one of those functions is treated as someone else’s subsystem, the final dose budget can fail even when the shielding calculation itself is correct. Radiation protection is therefore a cross-cutting architecture discipline.

From cosmic rays to Curiosity: making an invisible hazard measurable

Martian radiation was not discovered by one rover. The physics of ionizing radiation, solar particles and cosmic rays developed over more than a century. What interplanetary exploration adds is the ability to measure the field encountered behind spacecraft shielding and then on another planet’s surface. The Radiation Assessment Detector on Mars Science Laboratory provided that unusually valuable continuity during cruise and after landing.

Radiation measurement is not a single counter. Instruments record particles and energy deposition; transport models must then consider spacecraft material, the Martian atmosphere, regolith and the secondary particles produced when energetic nuclei strike matter. Two quoted “dose” values are meaningful only when the quantity, geometry, time period and biological weighting are known.

This is why a reference work should treat historical numbers as observations, not universal constants. Galactic cosmic rays vary with solar modulation. Solar particle events are episodic. Atmospheric column depth changes with season and elevation. Local shielding varies from one room to another. Current NASA work still treats GCR as a particularly difficult long-duration driver while acute SPE exposure can be mitigated more effectively with local shielding, monitoring and procedures.

A settlement turns this physics into metrology: personal dosimeters, fixed monitors, individual dose histories, space-weather models, shielding maps and operational thresholds. The hazard becomes a measurable mission budget rather than an invisible fear.

Gray, sievert, LET and dose: use one language before making comparisons

Gray, sievert and LET: three concepts that prevent false comparisons

Absorbed dose measures ionizing energy deposited per unit mass. Its SI unit is the gray: 1 Gy = 1 joule per kilogram. That physical quantity is essential, but equal absorbed doses from photons, protons, neutrons and high-energy heavy ions do not necessarily have equal biological effect.

Biologically weighted quantities expressed in sieverts are therefore used to account for radiation type and, depending on the quantity, tissue sensitivity. A measured value in µGy/day cannot simply be converted into mSv/day without assumptions about the particle field and quality factors.

LET helps explain why heavy ions matter. High-LET radiation deposits substantial energy along a short track, producing dense ionization. HZE ions in GCR can create complex biological damage. Shielding complicates the field further because a high-energy primary can strike a nucleus in the shielding and generate fragments or secondary neutrons through spallation.

That is why more thickness does not automatically mean a proportional reduction in all radiation risk. Additional mass can be highly effective against many SPE protons, while GCR requires transport modeling that includes secondaries and composition. NASA’s Spaceflight Human-System Standard therefore links radiation-environment definition to both system design and operational constraints.

Absorbed dose, equivalent dose and biological risk should never be collapsed into one number without stating the conversion. A gray is one joule of ionizing energy deposited per kilogram. A sievert applies radiation-weighting concepts because a sparse track of energy deposition and a dense heavy-ion track can produce different biological consequences even when absorbed energy is equal. LET adds another layer by describing how densely energy is deposited along a particle track. These distinctions are not academic vocabulary; they determine how measurements are interpreted and how shielding trades are compared.

For Mars design, the safest practice is therefore to state exactly which quantity is being quoted: particle flux, absorbed dose, dose equivalent, organ dose or effective dose, and over what time interval. Mixing daily surface dose with a mission-integrated effective-dose limit produces apparent precision while comparing different things.

Build a dose budget by integrating time, not by collecting isolated numbers

If mission phases have different dose rates, the simplest useful model is a time integral:

Take a deliberately illustrative scenario close to historical RAD-era orders of magnitude, not a mission prediction: 1.8 mSv/day for 180 days outbound, 0.7 mSv/day for 500 days on the surface, then 1.8 mSv/day for 180 days inbound. Outbound = 1.8 × 180 = 324 mSv. Surface = 0.7 × 500 = 350 mSv. Return = 324 mSv. Total = 998 mSv, approximately 1 Sv.

If transit falls to 120 days each way while every other assumption is frozen, transit becomes 1.8 × 120 × 2 = 432 mSv; surface remains 350 mSv; total becomes 782 mSv. The difference is 216 mSv. This does not prove that faster transit is always preferable—propulsion, mass, reliability and trajectory change too—but it shows why time itself can be a radiation countermeasure.

A flight design would require much more: spacecraft geometry, distributed materials, water, food, equipment, particle spectra, individual dosimetry, biological uncertainty and margins. The professional object is a versioned mission model, not one universal “Mars dose.”

A mission dose budget works like any other engineering budget: define phases, durations, environments and uncertainty. Transit outbound, surface stay, EVA, storm-shelter periods and return transit are not interchangeable. If a crew spends 180 days in one radiation field and 500 days in another, the arithmetic has to preserve those durations. The result should then be carried as a range rather than a single immortal number because solar conditions and shielding geometry vary.

The same discipline applies within a settlement. Sleeping quarters may sit behind more areal density than laboratories; rovers may provide less protection than the habitat; an EVA may be postponed when space-weather conditions deteriorate. Personal dosimetry then closes the loop between design assumptions and the dose actually accumulated by each individual.

What Curiosity measured and the limits of extrapolating to humans

What Curiosity measured — and what the measurements do not promise

During the rover’s early surface campaign, RAD observed a background dominated by galactic cosmic rays with an absorbed dose rate around 210 µGy/day under the conditions then measured. NASA also documented day-night differences related to atmospheric shielding, seasonal variation and Forbush decreases. NASA Science — Radiation Measurements on Mars.

Using RAD cruise and surface data, NASA published an illustrative Mars-mission comparison in which roughly 180 days outbound, 500 days on the surface and 180 days return produced an order of magnitude near 1 sievert, with comparable contributions from the three phases. This 2013 reference graphic is neither a future mission prediction nor a current exposure limit. It remains useful as a scale-setting example. NASA source.

More recent NASA work continues to separate the hazards. A 2024 review, The Martian Radiation Environment and Human Health Risks, emphasizes that real-time dosimetry and shielding are useful against acute solar-particle exposure, while GCR remains deeply penetrating. A 2025 NASA chapter on The Space Radiation Environment likewise describes dedicated storm shelters as an important SPE countermeasure.

Solar cycle matters too. Conditions during one measurement campaign do not represent every year of a future mission. Mission timing changes the trade without removing the hazard.

Design shielding: water, polymers, regolith and secondary particles

Water, polyethylene and regolith: shielding composition matters

Mass along the particle path matters, but composition matters too. Hydrogen-rich materials such as water and polyethylene are attractive because hydrogen can slow many protons and neutrons without introducing as many heavy nuclei as dense metals. Metals remain structurally essential; protection is therefore a systems problem involving many layers and already-required commodities.

Martian regolith has one extraordinary advantage: it does not have to be launched from Earth. Engineers can move tonnes locally, although those tonnes still require excavators, power, wear parts and construction time. Areal density is a useful way to think. A bulk density of 1,500 kg/m³ and a one-metre layer corresponds to about 1,500 kg/m²; two metres correspond to about 3,000 kg/m².

GCR performance is not linear with thickness because secondary particles are created. NASA’s Safe Human Expeditions Beyond Low Earth Orbit discusses the continuing challenge of passive GCR shielding and the more tractable design of SPE storm shelters.

Hydrogen-rich materials are attractive for many high-energy particle environments because hydrogen nuclei are efficient at slowing protons and can produce a different secondary-particle spectrum from high-atomic-number metals. Water and polyethylene therefore appear frequently in Mars shielding studies. Regolith is attractive for another reason: it is already on Mars and can provide large areal density without being launched from Earth.

But composition and geometry still matter. A thick wall can generate secondary neutrons and nuclear fragments when energetic primaries interact with nuclei. The design question is not ‘which material blocks radiation?’ but ‘what particle field exists behind this exact stack of materials, thicknesses and angles?’ That requires transport modelling and measurement, not a universal centimetres-to-sieverts conversion.

More material does not always buy the same amount of protection

For a simple X-ray beam, the intuition “thicker is better” is useful. Space radiation is more complicated. High-energy GCR nuclei lose energy in shielding but can also fragment and generate secondary particles, including neutrons. The material and thickness therefore reshape the field behind the wall rather than merely subtracting particles.

Hydrogen-rich materials such as water and some polymers are attractive because their nuclear composition can reduce certain secondary-production penalties compared with heavier materials. That does not mean a water tank stops galactic cosmic rays. At high energies the residual field remains difficult, and very large passive mass eventually brings diminishing practical returns. Mission design therefore combines material selection, transit time, solar monitoring, storm shelters and medical risk management.

Regolith offers a different advantage: it is already on Mars. Moving hundreds of tonnes locally may be far easier than launching the same mass from Earth. But useful thickness depends on real density, composition, geometry and the protection objective. A cover that reduces radiation can also complicate inspection, add structural load and make access to leaks more difficult.

That is why engineers often think in areal density, such as g/cm², rather than metres of unspecified wall. Two layers of equal thickness but different density do not place the same mass in the particle path, and two materials with the same areal density may generate different secondary fields.

The wall can create particles too: understanding radiation transport

Space shielding is counter-intuitive because matter is not merely an obstacle. When a very energetic proton or heavy nucleus strikes a material, it can ionize atoms, fragment nuclei and generate a cascade of secondary particles. Neutrons are particularly important because they carry no electric charge and interact differently from charged particles. The field behind a wall is therefore the result of both the incoming radiation and what that radiation creates while crossing the structure.

Radiation teams consequently use transport codes such as HZETRN, GEANT, FLUKA and other specialized tools instead of a single attenuation rule. These models follow particle interactions and secondary spectra through a specified geometry. They must be compared with measurements; Curiosity RAD data have been used precisely to benchmark several model families for the Martian surface. Differences among models identify where particle species, energies or geometries remain difficult rather than implying that nothing is known.

This is also why an “optimal thickness” cannot be copied from one paper to every habitat. Aluminium, water, polyethylene and regolith do not create identical secondary fields. A real shell combines structure, insulation, equipment, storage and local shielding. The relevant object is the complete mass distribution around the crew, not one material in isolation.

At city scale, radiation modelling can become geographic. Every room has a different mass distribution around it: a water tank on one side, a workshop on another, overburden above, an access tunnel nearby. A dose map can then guide function placement. Sleeping quarters and medical spaces occupy better-protected zones; transit corridors can accept a different exposure; storm shelters are identified before an event.

The chain of evidence should remain explicit: external environment → transport model → real geometry → measured or calculated dose → estimated biological effect → operational rule. Skipping a link is how a useful scientific number turns into false certainty. A reader should always be able to tell whether a value is measured, simulated, required by a standard or introduced as a transparent teaching scenario.

Radiation transport also links shielding to geometry. A slab calculation does not automatically describe a real habitat with doors, windows, equipment racks, water tanks and penetrations. Particles arrive from many directions, scatter, fragment and encounter regions with different areal density. Monte Carlo transport models and local dosimeters are therefore complementary: the model predicts where weak spots should be, while measurements reveal whether the built configuration behaves as expected.

This matters when a settlement changes. Moving a water tank, opening a new corridor or adding a surface workshop changes local shielding. Radiation configuration should therefore be managed like any other safety configuration, with drawings, material records and dose surveys updated when the physical layout changes. The shield is not only the wall specified at design review; it is the mass distribution that exists around people on that day.

The storm shelter: architecture and a timed procedure

The storm shelter: turn consumables into shielding for the hours that matter

An SPE refuge does not need to be the whole habitat. It needs enough protected volume and life support for the dangerous interval. That opens an elegant design strategy: arrange water, food and other hydrogen-rich supplies around a compact central refuge so commodities perform both logistics and shielding functions.

Dual use does not remove systems engineering. Concentrating every water tank around one room can create a common-cause vulnerability if a local leak, fire or impact damages both the refuge and the reserve. The shelter needs independent ventilation, CO₂ control, oxygen, water, lighting, communications, power, dosimetry and waste-management provisions for the planned occupancy time.

Operations matter as much as material. The crew must know who orders sheltering, which radiation instrument triggers the decision and how long it takes every crewmember to reach the protected volume. Mars communication delay means the decision must be possible locally. A slightly less massive refuge reached in two minutes may be operationally safer than a theoretically ideal one that takes forty minutes to configure.

The storm shelter can exploit mass that the mission already needs. Water tanks, food stores and selected equipment can be arranged around a small refuge so that logistics become shielding. The arrangement must remain maintainable: the crew still needs access to valves, inventory and fire boundaries, and a leak in one tank must not defeat both water supply and radiation protection at once.

A credible refuge has independent life support for the intended occupancy time. Oxygen, CO₂ removal, ventilation, power, communications, dosimetry, drinking water, sanitation and medical provisions all belong in the requirement. The shelter is therefore a small emergency habitat embedded inside the larger habitat, not simply a closet with thick walls.

A shelter that is never drilled is not yet an operational shelter

Emergency protection has a time dimension. The crew must recognize an alert, stop or safe current activities, move into the protected volume and establish a stable life-support configuration. Drills reveal bottlenecks that radiation-transport models cannot: blocked access, missing dosimeters, a medical patient who cannot move quickly, a valve that takes too long to configure or communications equipment stored outside the refuge.

Readiness therefore has measurable criteria: maximum time to shelter, minimum oxygen and CO₂-control duration, independent power endurance, water available inside, dosimetry status and communication capability. These are engineering requirements derived from operations, not merely human-factors decoration.

A storm shelter is a room — and a timed procedure

A storm shelter is useful only if every person can reach it in time and remain there while critical life-support functions continue. It needs breathable air, carbon-dioxide removal, water, communications, minimum power and medical capability. The design must work in a degraded habitat, not only in a nominal computer model.

Its geometry can exploit mass that already has to fly: water tanks, food, consumables, batteries and equipment can surround a smaller protected volume. This integration is especially attractive during transit, when every kilogram has launch and propulsion cost. On Mars, moved regolith can add local mass, provided construction has been completed before the shelter is credited as an operational capability.

Operations begin before an alert. Crews need clear triggers, accountability procedures, EVA-abort rules, dosimeter handling, spacecraft configuration steps and authority to act without waiting for Earth. Mars communication delays make a shelter that depends on real-time terrestrial approval fundamentally unsafe.

Drills should measure the time from warning to protected configuration. A blocked hatch, badly stowed cargo or unavailable suit can invalidate a beautiful paper design. As with fire safety, a shelter that has never been exercised is not yet a mature capability.

Transit, surface, EVA and space weather: manage dose through time

The spacecraft and the surface habitat do not see the same sky

In interplanetary space, radiation arrives from nearly all directions and the crew depends mostly on vehicle mass. On Mars, the planet blocks the lower half-space and the atmosphere adds angle-, altitude- and pressure-dependent shielding. The ground also creates secondary particles. A transit dose rate therefore cannot simply be copied into a surface model.

Site altitude and atmospheric column can influence the environment, but radiation is only one site-selection driver. Ice, landing safety, power, thermal conditions, science, communications and mobility must be traded together. See Choosing a site and deploying a Mars base.

Heavy surface infrastructure can eventually make shielding easier than during cruise because berms, trenches and burial are possible. The first crew, however, arrives before the mature city. Initial protection must therefore be landed from Earth or constructed robotically before crew arrival.

Transit is exposed to deep-space radiation with only vehicle shielding, while the Martian surface gains partial protection from the planet beneath the crew and a thin atmosphere above them. That difference does not make the surface ‘safe’; it changes the angular and energy distribution of the field and adds secondaries produced in atmosphere and ground. An architecture that shortens transit can therefore reduce one part of the exposure while leaving a long surface campaign that still needs shielding and operational controls.

An EVA has a radiation budget too

Outside the habitat, the crew leaves part of its shielding behind. A mobile spacesuit cannot reproduce a refuge surrounded by tonnes of water and regolith. EVA planning can therefore incorporate current particle environment, space-weather forecasts and accumulated dose.

Robotics can then be treated as a radiation countermeasure: repetitive exterior work can be automated so human exposure is reserved for tasks where human judgment or dexterity is worth the dose and operational risk.

Protection starts before shielding: watch the Sun and decide locally

An SPE shelter only helps if the crew knows when to enter it. A complete architecture therefore combines local detectors, space-weather forecasts, Earth updates, propagation models and pre-authorized procedures. Not every flare or coronal mass ejection produces a dangerous particle event at Mars; direction, particle energy and timing matter.

Earth–Mars delay demands local authority. The crew should not wait for detailed permission when mission dosimetry crosses a predefined threshold. Procedures identify who shelters, which EVA is cancelled, what equipment is shut down and how an injured crewmember is moved.

The analogy is terrestrial severe-weather warning, except that a Mars settlement cannot evacuate to another region. The refuge therefore belongs in routine architecture and training rather than in a forgotten emergency appendix.

Because Earth cannot command an emergency response instantly, Mars needs local space-weather awareness. Solar observations, onboard particle detectors, alert thresholds and pre-rehearsed actions should connect directly to operations. The decision chain has to answer practical questions: who has authority to terminate an EVA, how long does it take to reach shelter, what happens to a rover crew that cannot return immediately, and which autonomous systems continue exterior work while people are protected?

The solar cycle turns radiation into an operational planning problem

GCR intensity and solar-particle-event probability do not vary in the same way through the solar cycle. A mission architecture therefore cannot treat one year of measurements as a timeless constant. Long-duration planning needs a range of solar conditions and explicit uncertainty rather than a single “average Mars” environment.

Operationally, this means dose forecasts, current detector readings and shelter readiness belong in mission planning. Exterior maintenance can be postponed when the particle environment is unfavorable; robots can perform some tasks; sleeping areas can be located behind additional mass; water and food inventory can be arranged to improve shielding without making emergency access impossible.

Radiation protection is therefore not a single subsystem bought once before launch. It is a continuous interaction among trajectory, construction, logistics, forecasting, crew scheduling and medicine.

People and electronics: two vulnerabilities, two strategies

The same radiation threatens people and electronics in different ways

A proton or heavy ion can deposit energy in tissue and can also disturb a semiconductor. Single-event upsets may flip memory bits; more severe events can trigger latch-up, while total ionizing dose contributes to component aging. A settlement therefore manages biological dosimetry and electronic reliability at the same time.

Countermeasures differ. Computers can use ECC memory, redundancy, voting, rad-hard parts and autonomous reboot. Humans cannot. Conversely, electronics can be buried in a small heavily shielded vault while people need habitable volume and mobility. Protection level is therefore distributed by function and criticality.

See Electronics, components and sensors and the durability/FDIR chapter for the systems side of the same environment.

Electronics and humans respond through different failure mechanisms. A person accumulates biological risk through tissue exposure; electronics can suffer single-event upsets, latch-up, displacement damage or long-term total ionizing dose. Shielding that is acceptable for people may not be sufficient for a sensitive component, and rad-hard electronics do not make the crew safer. A resilient habitat therefore has separate radiation requirements for occupants, sensors, computers, power electronics and stored data.

From selected astronaut to a permanent civilian population

From selected astronauts to a civilian population: the problem changes

Exposure limits developed for medically selected professional astronauts on finite missions cannot simply be copied to a city containing children, pregnant people, older residents and inhabitants exposed for decades. For a permanent population, radiation policy moves beyond mission completion and must define what chronic risk, uncertainty and lifetime exposure a society is prepared to accept and reduce.

Biological uncertainty associated with GCR, especially heavy ions, makes the answer difficult. Cancer, cardiovascular, central-nervous-system and other risks remain research topics. NASA’s 2025 Moon to Mars Architecture Definition Document, Revision C still identifies radiation as a significant architecture driver and distinguishes relatively better-understood SPE shielding from the harder GCR problem.

A permanent settlement will probably distribute protection: heavily shielded sleeping or refuge zones, covered habitats, individual dosimetry, robotic execution of the most exposed tasks, solar monitoring and time accounting for exterior work. Surface activity need not become forbidden; instead, every exterior task should be planned with radiation exposure in mind.

A permanent settlement introduces people for whom the professional-astronaut model may be inappropriate: older adults, people with different medical histories and eventually children or pregnancies. Risk policy can no longer rely solely on selecting a small exceptionally healthy cohort for a finite mission. The settlement must decide what protection level is owed to ordinary residents, how individual dose history is recorded, who may accept higher occupational exposure and which activities are restricted. Those are engineering, medical and governance questions at the same time.

A permanent population breaks the “selected astronaut” risk model

Early crews will probably be highly selected and medically monitored for defined missions. A permanent population is different. What is acceptable exposure for people who may live decades on Mars, for workers who spend far more time outside, for older residents, pregnant people and eventually children? Direct human evidence for a lifetime in the Martian radiation environment does not exist.

The answer cannot simply be a professional-astronaut career rule scaled upward. A settlement would need individual dosimetry, longitudinal medicine, work rules, heavily protected communal zones and transparent methods for allocating higher-exposure tasks. A person performing frequent EVA accumulates a different risk history from a person working in shielded underground infrastructure.

This becomes a civic issue as well as a medical one. What minimum shielding is required for housing? Can an employer assign more outside work to someone with lower cumulative dose? How should uncertainty enter occupational health, insurance or family planning? Engineering cannot answer these questions alone.

A rigorous reference should therefore distinguish acute high-dose effects, probabilistic long-term risks estimated from models and terrestrial/space data, and the much less certain consequences of a lifetime combining partial gravity with chronic space radiation. Stating uncertainty is part of scientific accuracy.

Shielding becomes urban planning: deep functions, work zones and logistics

Radiation protection becomes urban planning: sleep deep, work nearer the surface

A first base may stack water bags around bunks. A city needs geometry. Not every room needs identical shielding: a corridor occupied for minutes, an airlock and a bedroom occupied eight hours per day have different time-integrated exposure. Radiation-aware planning puts long-duration functions in the best-protected zones and uses lighter zones for brief access or equipment.

“Heavy cores” could contain sleeping quarters, medical care, storm shelter, biological archives and critical control functions. Water and stores can be arranged around them. Tunnels can receive progressively more overburden as the settlement matures. The same geometry may also improve thermal stability and protection from some surface hazards.

Maintenance is the counterweight. Deeply buried pressure vessels are difficult to inspect. A leak beneath metres of fill requires access routes, sensors and excavation capability. Shielding must not imprison the pressure shell. Galleries, removable layers and robotic access paths should be designed from the beginning.

At one thousand residents, radiation therefore influences neighbourhood depth, EVA distances, hospital location, water storage, shelter access and mobility between districts. Protection has stopped being a component; it has become a property of the urban fabric.

Decide under uncertainty: no magic thickness, only justified margins

Why “two metres of regolith” must never become a magic formula

Popular summaries favor round thicknesses. Radiation transport does not. Attenuation depends on density, elemental composition, particle spectrum, geometry and the biological quantity of interest. Two metres of low-density material do not have the same areal density as two metres of dense material, and equal mass of different materials can create different secondary fields.

Good practice therefore states areal mass and uses particle-transport models or experimental data for the specific spectrum. Regolith can be extremely valuable because it is local, but no simple thickness makes the GCR problem disappear.

Uncertainty must be designed into the answer

Radiation risk combines measured particle fields with models of transport and biological response. Each layer carries uncertainty. A settlement should therefore preserve raw dosimetry, model version, shielding configuration and individual exposure history so future science can update earlier estimates. A margin is meaningful only when the uncertainty it covers is named. This is especially important for children and long residence times, where Earth has no direct epidemiological equivalent to decades in the Martian radiation environment.

Uncertainty belongs in the architecture rather than in a footnote. Instrument calibration, solar-cycle variability, biological quality factors, future crew demographics and the exact geometry of shielding all contribute. A defensible design therefore publishes assumptions, sensitivity cases and margin. No credible model can know a 20-year cancer risk to several decimal places; the useful question is which design decisions remain robust across plausible ranges.

The same discipline should be applied to long-term health interpretation. Dose limits written for professional astronauts, organ-specific models, age and sex effects, and future evidence can change the policy without changing the underlying particle field. A settlement should therefore retain raw dosimetry and exposure history in forms that can be reinterpreted as medical knowledge evolves, instead of preserving only one risk score calculated years earlier.

That archive is a scientific asset as well as a medical one.

Operational note. A radiation-protection decision is only as good as the distinctions it preserves: absorbed dose is not the same quantity as dose equivalent, chronic galactic-cosmic-ray exposure is not the same event as a solar-particle storm, and more shielding is not automatically better for every particle spectrum. A safe design therefore links dosimetry, material choice, occupancy time, warning, shelter access and uncertainty. The aim is not to reduce the subject to one annual number, but to show which exposure can be avoided, which must be monitored and what action the crew can take when conditions change.