The Odds of Finding Water on Mars: Mapping the Subsurface Ice That Could Make Human Exploration Possible
Where would you land on Mars if the water had to be underneath you? It is not an idle question. Subsurface water ice on Mars is not merely a scientific curiosity — it is, potentially, the single most important resource determining whether human beings can ever set foot on the Red Planet and return safely. Ice excavated from the Martian subsurface is drinking water, yes, but it is also a source of breathable oxygen through electrolysis, and critically, it is the raw feedstock for manufacturing rocket propellant directly on the surface rather than hauling it at enormous cost from Earth.
The mathematics here are stark and compelling. Manufacture your return fuel where you land — a concept engineers call in situ resource utilisation (ISRU) — and the mass of the entire mission drops enormously. Rockets are, at their core, machines for lifting fuel to burn more fuel, and every kilogram of propellant you no longer need to launch from Earth cascades into massive savings across the entire mission architecture. This is the difference between a plausible crewed Mars mission and an implausible one, between a programme that fits within a realistic budget and one that does not.
"In situ resource utilisation isn't just an engineering nicety — it is the economic and logistical cornerstone upon which any sustainable human presence on Mars must be built."
The Geography Problem: Why Mars Hides Its Ice in All the Wrong Places
Unfortunately, Martian geography is not cooperative. Mars keeps the bulk of its accessible ice concentrated at its polar ice caps, where sunlight arrives at an oblique angle for much of the year and solar panels operate at a fraction of their potential output. The Martian poles are energy deserts as much as they are ice reservoirs, which makes them deeply problematic as landing sites for crewed missions dependent on solar power for life support, heating, and ISRU operations.
The equatorial and near-equatorial regions — where sunlight is plentiful, temperatures are comparatively moderate, and landing is easiest — present the opposite problem. There, the ice is long gone from the accessible near-surface, driven away by a Martian atmosphere so thin (roughly 1% of Earth's atmospheric pressure at sea level) that surface water ice does not melt in the conventional sense. Instead, it sublimates — transitioning directly from solid to vapour — vanishing into the atmosphere without ever passing through a liquid phase. This thermodynamic reality was dramatically illustrated by NASA's Phoenix lander, which in 2008 scooped up buried ice at a high-latitude landing site and actually filmed chunks of it disappearing over the course of just a few days of exposure to the Martian environment. The sublimation was the confirmation: salt, the alternative explanation, would not have vanished that way.
Closer to the equator, any ice that once existed near the surface was likely eliminated hundreds of millions of years ago as Mars lost its thick early atmosphere. If ice persists at all in the mid-latitudes, it is buried — protected beneath an insulating layer of dust and regolith. The deeper it is buried, the harder it is to find from orbit, and the harder it is for future astronauts to excavate.
The Mid-Latitude Mystery: Where the Maps Fail Us Most
The scientific and engineering imperative thus converges on Mars's mid-latitudes — roughly between 30° and 60° in both hemispheres — as the most promising territory for a crewed landing site that balances solar energy access with the potential for accessible subsurface ice. And this, frustratingly, is precisely where existing orbital maps have been least reliable and most contradictory.
Two landmark papers recently published in the Planetary Science Journal, led by Hanna Sizemore and Samuel Courville of the Planetary Science Institute, tackle this mapping problem from complementary directions. Both studies draw on data and methodology from the Subsurface Water Ice Mapping (SWIM) project, a collaborative effort to synthesise orbital observations into the most comprehensive and reliable map of near-surface ice deposits yet attempted.
The SWIM project works by reading buried ice indirectly, exploiting a fundamental principle of thermal physics: different subsurface compositions hold and release heat at different rates. Soil overlying a layer of ice warms and cools differently from soil underlain by dust or rock. Sensitive thermal infrared instruments aboard orbiters — including those on Mars Global Surveyor and Mars Reconnaissance Orbiter — have been recording these thermal rhythms across Martian days and seasons for well over a decade, building up a rich dataset that scientists can invert to infer what lies beneath.
Sizemore's Approach: The Power of Disagreement
Sizemore's contribution addresses a fundamental weakness in how scientists had previously used these thermal datasets. Different instruments, different computational models, and variations in atmospheric dust loading — a dustier atmosphere absorbs and scatters solar radiation differently from a clear one — all pull the inferred ice-depth answer in different directions. A single instrument reading a single dataset on a single day can mislead.
Her team's solution was to stop trusting any single measurement and instead compare three independently produced ice-distribution maps, looking carefully at where they converge and where they diverge. Where multiple independent analyses agree that ice is present — or absent — that consensus carries real scientific weight. The disagreements, however, are arguably more valuable still. Regions where the maps conflict are precisely the areas our understanding is weakest, the ground where orbital data is ambiguous and model assumptions matter most.
- High agreement, ice predicted: Likely safe targets for robotic prospectors and eventual human missions.
- High agreement, no ice predicted: Regions to deprioritise for ISRU-dependent missions.
- Low agreement, conflicting signals: Priority targets for future high-resolution investigation — excellent places to send a robot, but dangerous places to commit a crewed mission.
This framework transforms a patchwork of competing maps into a structured research agenda, identifying not just where we think ice might be, but where we most urgently need better data.
Courville's Approach: From Confidence to Probability
Courville's paper does something more subtle and, arguably, more operationally important. It converts qualitative scientific confidence into quantitative probability estimates — a shift that may seem technical but has profound implications for mission planning.
There is a fundamental difference between saying "two techniques detected ice here and one did not" — a statement about the performance of our instruments and methods — and saying "there is a 64% probability of accessible ice at this location" — a statement about the actual state of Mars. The former is a scientific footnote. The latter is the kind of number a mission architect can feed into a risk analysis, a landing site selection committee can weigh against competing priorities, and a spacecraft engineer can use to design an excavation system with appropriate margins.
Probabilistic mapping of this kind mirrors approaches used extensively in terrestrial resource exploration — oil and gas prospecting, groundwater surveying, mineral exploration — where the economics of drilling demand quantified risk assessments rather than qualitative judgements. Applying the same rigour to Mars ice mapping represents a meaningful maturation of the field, moving planetary science closer to the engineering-ready outputs that an actual crewed mission programme will require.
The Critical Gap: One to Five Metres Below the Surface
Both papers, taken together, represent the most sophisticated attempt yet to answer the question of where Mars hides its ice. And yet they share a sobering limitation — one that the authors are candid about — which defines the frontier of current knowledge.
The thermal infrared technique that underpins the SWIM project is sensitive only to the uppermost metre of the Martian subsurface. It cannot see deeper. Ground-penetrating radar, carried by instruments such as SHARAD aboard Mars Reconnaissance Orbiter, can probe to depths of five metres and beyond — but it operates best at low frequencies optimised for deep penetration and loses sensitivity in the critical near-surface zone. The result is an instrumental blind spot:
- 0–1 metre depth: Covered by thermal infrared orbital instruments.
- 1–5 metre depth: Not reliably covered by any current orbital instrument.
- 5+ metres depth: Accessible to orbital radar, with limitations.
That 1–5 metre range is precisely the depth a crew of astronauts would realistically be able to excavate using near-term technology. It is the zone that matters most for ISRU operations. And it is currently invisible to us from orbit. Closing this gap would require a high-frequency ground-penetrating radar — an instrument type that has never yet been flown to Mars — capable of resolving structure in the shallow subsurface with sufficient sensitivity to detect ice layers.
Until such an instrument makes the journey to Mars, the maps produced by Sizemore, Courville, and the SWIM team represent the best water atlas of Mars humanity has ever produced. They are imperfect, incomplete, and honest about both qualities — which is precisely what makes them valuable. In the long road toward putting boots on Martian soil, knowing precisely what we do not know is progress.
What Comes Next
The implications of this research extend well beyond academic interest. Space agencies and private spaceflight companies actively planning crewed Mars missions — including NASA's Moon to Mars programme — must eventually commit to specific landing sites years or even decades before a mission launches. Landing site selection is irreversible, and the consequences of choosing a site without accessible water ice could be catastrophic for crew survival and mission success.
The probabilistic framework developed by Courville's team, combined with Sizemore's map of regions requiring urgent further investigation, provides a structured scientific basis for prioritising future robotic missions. A well-targeted small robotic lander or rover equipped with a high-frequency radar, or even direct drilling capability, deployed to the most uncertain mid-latitude regions identified in these studies, could dramatically sharpen our knowledge before any human mission is committed.
Mars has water. The question has shifted from whether to where, and from where to how confident are we, and at what depth. Science is narrowing the uncertainty, one probabilistic percentage point at a time.
