Terraforming a world is a breathtaking task, one often thought about in relation to making Mars into a benign environment for human settlers. But there are less challenging alternatives for providing shelter to sustain a colony. As Robert Zubrin explains in the essay below, ice-covered lakes are an option that can offer needed resources while protecting colonists from radiation. The founder of the Mars Society and author of several books and numerous papers, Zubrin is the originator of the Mars Direct concept, which envisions exploration using current and near-term technologies. We’ve examined many of his ideas on interstellar flight, including magsail braking and the nuclear salt water rocket concept, in these pages. Now president of Pioneer Astronautics, Zubrin’s latest book is The Case for Space: How the Revolution in Spaceflight Opens Up a Future of Limitless Possibility, recently published by Prometheus Books.

by Robert Zubrin


This paper examines the possibilities of establishing Martian settlements beneath the surface of ice-covered lakes. It is shown that such settlements offer many advantages, including the ability to rapidly engineer very large volumes of pressurized space, comprehensive radiation protection, highly efficient power generation, temperature regulation, copious resource availability, outdoor recreation, and the creation of a vibrant local biosphere supporting both the nutritional and aesthetic needs of a growing human population.


The surface of Mars offers many challenges to human settlement. Atmospheric pressure is only about 1 percent that of Earth, imposing a necessity for pressurized habits, making spacesuits necessary for outdoor activity, and providing less than optimum shielding against cosmic radiation. For these reasons some have proposed creating large subsurface structures, comparable to city subway systems, to provide pressurized well-shielded volumes for human habitation [1]. The civil engineering challenges of constructing such systems, however, are quite formidable. Moreover, food for such settlements would have to be grown in greenhouses, limiting potential acreage, and imposing either huge power requirements if placed underground, or the necessity of building large transparent pressurized structures on the surface. Water is available on the Martian surface as either ice or permafrost. These materials can be mined and the product transported to the base, but the logistics of doing so, while greatly superior to anything possible on the Moon, are considerably less convenient than the direct access to liquid water available to nearly all human settlements on Earth. While daytime temperatures are acceptably close to 0 C, nighttime temperatures drop to -90 C, imposing issues on machinery and surface greenhouses. Yet despite the cold night temperatures, the efficiency of nuclear power is impaired by the necessity of rejecting waste heat to a near-vacuum environment.

All of these difficulties could readily be solved by terraforming the planet [2]. However, that is an enormous project whose vast scale will require an already-existing Martian civilization of considerable size and industrial power to be seriously undertaken. For this reason, some have proposed the idea of “para terraforming,” [3] that is, roofing over a more limited region of the Red Planet, such as the Valles Marineris, and terraforming just that part. But building such a roof would itself be a much larger engineering project than any yet done in human history.

There are, however, locations on Mars that have already been roofed over. These are the planet’s numerous ice-filled craters.

Making Lakes on Mars

Earth’s Arctic and Antarctic regions feature numerous permanently ice covered or “sub glacial” lakes [4]. These lakes have been shown to support active microbial and planktonic ecosystems.

Most sub Arctic and high latitude temperate lakes are ice-covered in winter, but many members of their aquatic communities remain highly active, a fact well-known to ice fishermen.

Could there be comparable ice-covered lakes on Mars?

At the moment, it appears that there are not. The ESA Mars Express orbiter has detected highly-saline liquid water deep underground on Mars using ground penetrating radar, and such environments are of great interest for scientific sampling via drilling. But to be of use for settlement, we need ice-covered lakes that are directly accessible from the surface. There are plenty of ice-filled craters on Mars. These are not lakes, however, as while composed of nearly pure water ice, they are frozen top to bottom. But might this shortcoming be correctable?

I believe so. Let us examine the problem by considering an example.

Korolev is an ice-filled impact crater in the Mare Boreum quadrangle of Mars, located at 73° north latitude and 165° east longitude (Fig. 1). It is 81.4 kilometers in diameter and contains about 2,200 cubic kilometers of water ice, similar in volume to Great Bear Lake in northern Canada. Why not use a nuclear reactor to melt the water under the ice to create a huge ice-covered lake?

Fig. 1. Korolev Crater could provide a home for sublake city on Mars. Photo by ESA/DLR.

Let’s do the math. Melting ice at 0 C requires 334 kJ/kg. We will need to supply this plus another 200 kJ/kg, assuming that the ice’s initial temperature is -100 C, for 534 kJ/kg in all. Ice has a density of 0.92 kg/liter, so melting 1 cubic kilometer of ice would require 4.9 x 1017 J, or 15.6 GW-years of energy. A 1 GWe nuclear power plant on Earth requires about 3 GWt of thermal power generation. This would also be true in the case of a power plant located adjacent to Korolev, since it would be using the ice water it was creating in the crater as an excellent heat rejection medium. With the aid of 5 such installations, using both their waste heat and the dissipation from their electric power systems, we could melt a cubic kilometer of ice every year.

Korolev averages 500 m in depth, which is much deeper than we need. So rather than try to melt it all the way through, an optimized strategy might be to focus on coastal regions with an average depth of perhaps 40 meters. In that case, each cubic kilometer of ice melted would open 25 square kilometers of liquid lake for settlement. Alternatively, we could just choose a smaller crater with less depth, and melt the whole thing, except the ice cover at its top.

Housing in a Martian Lake

On Earth, 10 meters of water creates one atmosphere of pressure. Because Martian gravity is only 38 percent as great as that of Earth, 26 meters of water would be required to create the same pressure. But so much pressure is not necessary. With as little as 10 meters of water above, we would still have 0.38 bar of outside pressure, or 5.6 psi, allowing a 3 psi oxygen/2.6 psi nitrogen atmosphere comparable to that used on the Skylab space station. Reducing nitrogen atmospheric content in this way could also be advantageous because nitrogen is only a small minority constituent of the Martian atmosphere, making it harder to come by on Mars, and limiting the nitrogen fraction of breathing air would also facilitate traveling to lower pressure environments without fear of getting the bends. Ten meters of water above an underwater habitat would also provide shielding against cosmic rays equivalent to that provided by Earth’s atmosphere at sea level.

Construction of the habitats could be done using any the methods employed for underwater habitats on Earth. These include closed pressure vessels, like submarines, or open-bottom systems, like diving bells. The latter offer the advantage of minimizing structural mass since they have an interior pressure nearly equal to that of the surrounding environment, and direct easy access to the sea via their bottom doors, without any need for airlocks. Thus, while closed submarines are probably better for travel, as their occupants do not experience pressure changes with depth, open bottom habitats offer superior options for settlement. We will therefore focus our interest on the latter.

Consider an open-bottom settlement module consisting of a dome 100 m in diameter, whose peak is 4 meters below the surface and whose base in 16 meters below the surface. The dome thus has four decks, with 3 meters of head space for each. The dome is in tension, because all the air in it is all at a pressure of 9 psi, corresponding to the lake water pressure at its base, while the lake water pressure at its top is only about 2.2 psi, for an outward pressure on the dome material near the top of 6.8 psi. The dome has a radius of curvature of 110 m.

The required yield stress of the material composing a pressurized sphere is given by:

? = xPR/2t (1)

Where ? is the yield stress, P is the pressure, R is the radius, t is the dome thickness, and x is the safety factor. Let’s say the dome is made of steel with a yield stress of 100,000 psi and x=2. In that case, equation (1) says that:

100,000 = (6.8)(110)/t, or t= 0.0075 m = 7.5 mm.

The mass of the steel would be about 600 tons. That’s not to bad, for creating a habitat with about 30,000 square meters of living space.

If instead of using steel, we made a tent dome from spectra fabric, which has 4 times the strength of steel and 1/9th the density, the mass of the dome would only need to be about 17 tons. It would, however, need to be tied down around its circumference. Ballast weights of 90,000 tons of rocks could be used for this purpose. Otherwise the tie down lines could be anchored to stakes driven deep into the frozen ground under the lake.

An attractive alternative to these engineering methods for creating a dome out of manufactured materials could be to simply melt the dome out of the ice covering the lake itself. For example, let’s say the ice cover is 20 m thick, and we melt a dome into it that is 12 m tall, 100 m in diameter, and has a radius of curvature of 110 m. Filling this with an oxygen/nitrogen gas mixture would provide a habitat of equal size to that discussed above. The pressure under 20 m of ice (density = 0.92) is 0.7 bar, or 10.3 psi. The roof of the dome is under 8 m of ice, whose mass exerts of compressive pressure of 0.28 bar, or 4.1 psi, leaving a pressure difference of 6.2 psi to be held by the strength of the ice. The tensile strength of ice is about 150 psi, so sticking these values into equation (1) we find that the safety factor, x, at the dome’s thinnest point would be:

150 = x(6.2)(110)/[(8)(2)], or x = 3.52

This safety factor is more than adequate. Networks of domes of this size could be melted into the ice cover, linked by tunnels through the thick material at their bases. If domes with a much larger radius of curvature were desired, the ice could be greatly strengthened by freezing a spectra net into it.

The mass of ice melted to create each such dome is about 80,000 tons, requiring 1 MWt-year of energy to do the melting. It would also require about 90 tons of oxygen to fill the dome with gas. This could be generated via water electrolysis. Assuming 80% efficient electrolysis units, this would require 1950 GJ, or 62 kWe-year of electric power to produce. Such large habitation domes could therefore be constructed and filled with breathable gas well in advance of the creation of the lake using much more modest power sources.

Compressive habitation structures can be created under ice that are much larger still. This is so because ice has 92 percent the density of water, so that if a 50 meters deep column of ice beneath the lake’s ice surface were melted, it would yield a column of water 42 meters deep and 8 meters of void, which could be filled with air.

So, let’s say we had an ice crater, section of an ice crater, or even a glacier 5 km in radius and 70 meters or more deep. We melt a section of it starting 20 m under the top of the ice and going down 50 m. As noted, this would create a headroom space 4 m thick above the water. The ice above this void would have a weight of 7 psi, so we would fill the void with an oxygen/nitrogen gas mixture with a pressure of 6.999 psi. This would negate almost all the weight to leave the ice roof in an extremely mild state of compression. (Mild compression is preferred to mild tension, because the compressive strength of ice is about 1500 psi – ten times the tensile strength.) Under such circumstances the radius of curvature of the overhanging surface could be unlimited. As a result, a pressurized and amply shielded habitable region of 78 square kilometers would be created. Habitats could be placed on rafts or houseboats on this indoor lake, or an ice shelf formed to provide a solid floor for conventional buildings over much of it.

The total amount of water that would need to be melted to create this indoor lake city would be 4 cubic kilometers. This could be done in about 4 years by our proposed 5 GWe power system. Further heating would continue to expand the habitable region laterally over time. If the lake were deep, so that there was ice beneath the water column, it would gradually melt, increasing the headroom over the settlement as well.

Terraforming the Lake

The living environment of the sublake Mars settlement need not be limited to the interior of the air-filled habitats. By melting the ice, we are creating the potential for a vibrant surrounding aquatic biosphere, which could be readily visited by Mars colonists wearing ordinary wet suits and SCUBA gear.

The lake is being melted using hot water produced by the heat rejection of onshore or floating nuclear reactors. If the heat is rejected near the bottom of the lake, forceful upwelling will occur, powerfully fertilizing the lake water with mineral nutrients.

Assuming that the ice cover is reduced to less than 30 meters, there will be enough natural light during daytime to support phytoplankton growth, as has been observed in the Earth’s Arctic ocean [5]. The lake’s primary biological productivity could be greatly augmented, however, by the addition of artificial light.

The Arctic ocean exhibits high biological activity as far north as 75 N, where the sea receives an average day/night year-round solar illumination of about 50 W/m2. If we take this as our standard, then each GW of our available electric power could be used to illuminate 20 square kilometers of lake. Combined with the mineral-rich water produced by thermal upwelling, and artificial delivery of CO2 from the Martian atmosphere as required, this illumination could serve to create an extremely productive biosphere in the waters surrounding the settlement.

The first organisms to be released into the lake should be photosynthetic phytoplankton and other algae, including macroscopic forms such as kelp. These would serve to oxygenate the water. Once that is done, animals could be released, starting with zooplankton, with a wide range of aquatic macrofauna, potentially including sponges, corals, worms, mollusks, arthropods, and fish coming next. Penguins and sea otters could follow.

As the lake continues to grow, its cities would multiply, giving birth to a new branch of human civilization, supported by and supporting a lively new biosphere on a new world.


We find that the best places to settle Mars could be under water. By creating lakes beneath the surface of ice-covered craters, we can create miniature worlds, providing acceptable pressure, temperature, radiation protection, voluminous living space, and everything else needed for life and civilization. The sublake cities of Mars could serve as bases for the exploration and development of the Red Planet, providing homes within which new nations can be born and grow in size, technological ability, and industrial capacity, until such time as they can wield sufficient power to go forth and take on the challenge of terraforming Mars itself.


1. Frank Crossman, editor, Mars Colonies: Plans for Settling the Red Planet, The Mars Society, Polaris Books, 2019

2. Robert Zubrin with Richard Wagner, The Case for Mars: The Planet to Settle the Red Planet and Why We Must, Simon and Schuster, NY, 1996, 2011.

3. Richard S. Taylor, “Paraterraforming: The Worldhouse Concept,” Journal of the British Interplanetary Society, vol. 45, no. 8, Aug. 1992, p. 341-352.

4. Sub Glacial Lake, Wikipedia, https://en.wikipedia.org/wiki/Subglacial_lake#Biology accessed May 15, 2020.

5. Kevin Arrigo, et al, “Massive Phytoplankton Blooms Under Sea Ice,” Science, Vol. 336, page 1408, June 15, 2012 https://www2.whoi.edu/staff/hsosik/wp-content/uploads/sites/11/2017/03/Arrigo_etal_Science2012.pdf. Accessed May 15, 2020.