Where Would the First Mars Colony Land? The Shortlist

No first-colony landing site is confirmed in the sources discussed here. NASA’s 2015 workshop received 47 proposed exploration zones. A separate 2021 study with SpaceX and JPL authors retained seven sites for further study: three in Arcadia Planitia, two in Phlegra Montes and two in Erebus Montes.

Every plan for a first Mars colony hits the same wall before it reaches habitats, greenhouses or power budgets. No single place on Mars satisfies all the requirements at once.

Water ice lies toward the poles. Sunlight and survivable temperatures lie toward the equator, and air thick enough to slow a descending spacecraft lies at low elevation. A landing site is what you get after trading those against each other, which is why the same handful of names keeps recurring.

Where does NASA say a first crew should land?

NASA’s First Landing Site/Exploration Zone Workshop for Human Missions to the Surface of Mars ran at the Lunar and Planetary Institute in Houston on 27–30 October 2015. The workshop statement records 47 exploration zones proposed in 45 abstracts; 44 abstracts were presented. These were proposals, not approved landing sites.

The workshop introduced a unit of planning that still shapes the argument. An Exploration Zone is a collection of Regions of Interest sitting within roughly 100 kilometres of one central landing site. That radius encodes an assumption about how far a crew can practically travel and return. Pick a landing point and you are really picking everything inside a 100 km reach: the science, the ice, the construction material.

No site was selected then, and none has been selected since. Those 47 proposals were a call for ideas rather than a contest with a winner.

Why not just land at the equator?

Because two of the three constraints point that way and the most important one does not.

Mars has roughly one percent of Earth’s surface air pressure, so lower ground gives a descending spacecraft more atmosphere to brake against. InSight’s published site-selection criteria required an elevation of −2.5 km or lower relative to the reference level. That is a vehicle-specific limit, not a universal threshold for a crewed lander.

Power, temperature and terrain impose separate constraints. InSight’s final latitude band was 3°N–5°N. Its certified site met slopes below 15° at 2 m and 84 m scales and rock abundance below 10%. These are InSight requirements; the 2021 Starship study instead used a reference slope limit below 5° over 10 m.

One number is worth holding on to. InSight's ballistic entry gave a landing ellipse of 130 by 27 kilometres. You do not choose a point on Mars. You choose an area, and you have to be willing to live anywhere inside it.

So the equator wins on sunlight, warmth and, in the low basins, on air. It loses on the resource that decides whether a base can grow at all: reachable water.

Where is the ice shallow enough to matter?

This is the question the last decade of orbital work has been answering. The Subsurface Water Ice Mapping project (SWIM), run through the Planetary Science Institute for NASA's Mars Exploration Program, combines orbital data into ice consistency maps at about 3 km per pixel across three depth bands: 0 to 1 metre, 1 to 5 metres, and deeper than 5 metres.

The 2015 workshop considered sites between 50°S and 50°N. SWIM’s later synthesis found positive ice-consistency values extending into the 20°–30° latitude bands, but those values are remote-sensing indicators, not proof of shallow, mineable ice at every location. Morgan and colleagues’ Nature Astronomy paper, published on 8 February 2021, highlighted Arcadia Planitia and Deuteronilus Mensae.

Phoenix provides direct sampling evidence. It landed at 68.22°N and dug into shallow water ice. JPL announced the disappearing ice chunks in Dodo-Goldilocks on 20 June 2008, using images from sols 20 and 24 (15 and 19 June). On 31 July, a separate announcement reported water identified by the Thermal and Evolved-Gas Analyzer in a soil sample. These are two different confirmations.

Bright material vanishing from the Dodo-Goldilocks trench over four sols: sublimation, and so water ice rather than salt. Credit: NASA/JPL-Caltech/University of Arizona/Texas A&M University. Source: NASA image PIA10906

Phoenix proved the ice exists but sat far too near the pole to be a settlement candidate. Its solar power faded with the northern winter and the mission ended. The useful ground is further south, and fresh impacts have been quietly mapping it: over two decades the HiRISE and Context cameras on Mars Reconnaissance Orbiter catalogued hundreds of new craters, and some at mid to high latitudes punched straight into buried ice. One 13-metre crater in Arcadia Planitia exposed ice in both its cavity and its ejecta, with a dark blast zone reaching almost 850 metres from the centre. That crater is the image at the top of this article.

Later orbital work has kept pointing the same way. Successive SWIM mapping rounds have continued to support extensive shallow ground ice across Arcadia rather than walking the interpretation back.

What has SpaceX publicly studied?

A study list, not a destination. At the 2021 Lunar and Planetary Science Conference, a paper titled SpaceX Starship Landing Sites on Mars set out seven sites selected for further study. The affiliations behind it are the interesting part: JPL and SpaceX appear side by side, together with the University of Arizona, the Planetary Science Institute, Purdue, Brown and NASA Ames. A company and a NASA centre were working the siting problem together, in the open literature.

The seven comprise three Arcadia sites (AP-1, AP-9, AP-8), two Phlegra sites (PM-1, PM-7) and two Erebus sites (EM-16, EM-15). The study selected four primary sites, PM-1, AP-1, AP-9 and EM-16, without ranking them; the secondary sites were AP-8, EM-15 and PM-7 in that order. Table 1 spans 35.23°N–40.75°N and elevations of −3.9 to −2.3 km. These are screening results, not demonstrated landing or mining capability.

Orbital images document the wider search. HiRISE image ESP_060416_2195, acquired on 17 June 2019, is centred at 39.083°N, 189.793°E and labelled a candidate Starship landing site in the Arcadia region. Its centre is not one of the seven coordinates in the 2021 table; an image of a candidate area should not be confused with the later downselection.

Western Arcadia Planitia near 39°N, 179°E. The concentric layers and slumped material inside this crater are the kind of feature that points to surface or subsurface ice. Credit: NASA/JPL/Arizona State University/Cornell University. Source: NASA image PIA04263

A candidate is not a commitment. The 2021 abstract describes 22 prospective sites before narrowing them to seven; a larger number can therefore refer to an earlier stage of the same study. Its site list is a dated research result, not evidence that NASA or SpaceX has approved a crewed destination.

How does the shortlist compare?

The names that recur, and what each one is actually offering:

• Arcadia Planitia, near 40°N. Three of the seven retained Starship study sites: AP-1, AP-9 and AP-8. Low elevation, comparatively smooth terrain and multiple orbital indicators of buried ice make these research candidates; the study does not rank Arcadia as the chosen region.

• Phlegra Montes and Erebus Montes. Two retained sites in each region, associated with glacier-like deposits. Their slopes, rocks and access to ice must be evaluated site by site.

• Deuteronilus Mensae. A strong ice-resource candidate in SWIM, with glacier-like deposits. But the 2021 Starship study found the available images too rocky to identify suitable landing surfaces there. Evidence for ice is not evidence for a safe touchdown.

• Jezero Crater, 18.44°N, 77.45°E, floor around 2,600 metres below the reference level. Where Perseverance works. Superb science, an ancient river delta, and no particular claim to accessible ice.

• Oxia Planum, 18.28°N, 335.37°E, its clay unit running from about 2,600 to 3,100 metres below datum. The target for ESA's Rosalind Franklin rover, chosen for ancient clays rather than resources.

• Utopia Planitia, around 48°N at Viking 2’s landing site. Viking 2 landed in 1976 and remains the second-most-northerly successful surface mission after Phoenix. This is a historical example, not one of the seven retained Starship sites.

The comparison separates two questions: where to search for ancient environments, and where to obtain usable water while landing safely. The seven retained Starship candidates span roughly 35°N–41°N, rather than all lying near or north of 39°N. No latitude alone proves a site suitable for a base.

What do the robotic landings already tell us?

Our interactive Mars map helps compare landing regions, but a map marker is not always a confirmed hardware location. The useful distinction is between measured landing positions, approximate historical estimates and targets for missions that have not landed. Interactive Mars map.

Among the missions that operated on the surface, Viking 2 and Phoenix are the northern exceptions: roughly 48°N and 68°N. All six rovers that have driven on Mars landed within about 26° of the equator, from Opportunity at 1.95°S to Zhurong at 25.07°N. Mars 3’s often-quoted 45°S is an approximate predicted location; JPL’s 2013 report did not confirm the candidate hardware.

The 35°N–41°N candidate belt still lacks a rover investigation of the proposed ice resources. Mars 2, Mars 6 and Mars Polar Lander have no confirmed hardware locations in the sources cited here. Beagle 2 is different: orbital images identified its partially deployed lander in 2015, showing that it reached the surface intact but could not communicate.

The mission record behind those pins is set out in our timeline of Mars missions, and what a base would have to build once it arrives is covered in our guide to colonizing Mars.

What we still do not know

The honest gaps, because they are larger than the settled parts:

• No agency or company has selected a landing site for a first crewed mission, let alone a permanent base.

• Ice depth and purity at candidate sites remain dependent on remote-sensing interpretation. Phoenix sampled ice much farther north; fresh ice-exposing impacts provide additional local checks from orbit, not a mining survey of the seven sites.

• The trade between ice access and solar power has no agreed answer. SWIM frames a sweet spot at the southern edge of the ice belt; that is a research framing, not a decision.

• The 2021 list is a research snapshot. A later site claim needs its own primary source; neither the age of this study nor a different candidate count can settle it.

What would settle it is ground truth: a precursor mission that lands in the candidate band and measures how deep the ice really is, how clean, and how hard to dig. Until then the shortlist stays a shortlist. The surrounding questions have their own answers on this site, including how long the journey takes, what it would cost and whether the radiation is survivable.

Common questions

Has a landing site for the first Mars colony been chosen?

No. NASA gathered 47 Exploration Zone proposals in 2015 and picked none of them. The 2021 SpaceX and JPL paper narrowed to seven sites for further study, which is a research shortlist, not a commitment.

Why does Arcadia Planitia come up so often?

The cited studies identify low-elevation plains, comparatively smooth surfaces and evidence consistent with buried ice around 40°N. Those attributes justify further study, not a claim that ice extraction, year-round power or safe Starship landings have been demonstrated there.

Could a colony land at the poles, where there is more ice?

Ice gets easier there and everything else gets harder. Sunlight weakens, temperatures drop, and the caps are seasonal carbon dioxide over water. Phoenix reached 68.22°N in 2008 and lost power once northern winter arrived.

How large an area does a Mars landing site actually cover?

Larger than most people picture. An Exploration Zone spans roughly 100 kilometres around its central point, and InSight's ballistic entry gave an ellipse of 130 by 27 kilometres. The whole ellipse must be survivable, not just its centre.

How deep is the ice at these sites?

SWIM maps three bands: 0 to 1 metre, 1 to 5 metres, deeper than 5 metres. Fresh craters in Arcadia have exposed ice near the top of that range. Phoenix, much further north, hit ice about 5 centimetres down. Nobody has dug in the candidate belt itself.

Is Jezero Crater a candidate for a colony?

Jezero was chosen for Perseverance’s search for ancient habitable environments and possible biosignatures, not as a proven ice mine for a settlement. Its landing site is at 18.44°N. It is not among the seven sites retained by the cited 2021 Starship study; that does not rule out every possible future base architecture.

What would move a site from candidate to chosen?

Ground measurements could establish ice depth, purity and excavation difficulty, alongside slopes and rocks at vehicle scale. Better orbital observations can still narrow the uncertainty; the 2021 study explicitly calls for further assessment before human missions or persistent habitation.

Sources

• First Landing Site/Exploration Zone Workshop for Human Missions to the Surface of Mars, Lunar and Planetary Institute, Houston, 27 to 30 October 2015.

• Golombek et al., Selection of the InSight Landing Site, Space Science Reviews, on elevation, latitude, slope and rock-abundance constraints.

• Subsurface Water Ice Mapping (SWIM) project, Planetary Science Institute, for NASA's Mars Exploration Program.

• Morgan et al., Availability of subsurface water-ice resources in the northern mid-latitudes of Mars, Nature Astronomy, 2021.

• Putzig, Morgan et al., Ice Resource Mapping on Mars, chapter in Handbook for Space Resources, Springer, 2023 — the SWIM project's own review of how its ice-resource mapping developed.

• Golombek, Williams, Wooster, McEwen et al., SpaceX Starship Landing Sites on Mars, 52nd Lunar and Planetary Science Conference, 2021.

• NASA — NASA Spacecraft Confirms Martian Water, Mission Extended (2008-07-31)

• Candidate Landing Site for SpaceX Starship in Arcadia Region, ESP_060416_2195, HiRISE, University of Arizona.

• JPL — NASA Phoenix Mars Lander Confirms Frozen Water (2008-06-20)

• JPL — NASA Mars Orbiter Images May Show 1971 Soviet Lander (2013-04-11)

• ESA — Beagle-2 lander found on Mars (2015-01-16)

Correction review: 8 November 2026. The 2021 study’s regional counts, publication dates and the distinction between predicted and confirmed landing locations were corrected. Earlier mission observations retain their original dates.