Assessment of wind energy resource potential for future human missions to Mars

Energy sustainability and redundancy for surface habitats, life support systems and scientific instrumentation represent one of the highest-priority issues for future crewed missions to Mars. However, power sources utilized for the current class of robotic missions to Mars may be potentially dangerous near human surface habitats (for example, nuclear) or lack stability on diurnal or seasonal timescales (for example, solar) that cannot be easily compensated for by power storage. Here, we evaluate the power potential for wind turbines as an alternative energy resource on the Mars surface. Using a state-of-the-art Mars global climate model, we analyse the total planetary Martian wind potential and calculate its spatial and temporal variability. We find that wind speeds at some proposed landing sites are sufficiently fast to provide a stand-alone or complementary energy source to solar or nuclear power. While several regions show promising wind energy resource potential, other regions of scientific interest can be discarded based on the natural solar and wind energy potential alone. We demonstrate that wind energy compensates for diurnal and seasonal reductions in solar power particularly in regions of scientific merit in the midlatitudes and during regional dust storms. Critically, proposed turbines stabilize power production when combined with solar arrays, increasing the percent time that power exceeds estimated mission requirements from ~40% for solar arrays alone to greater than 60–90% across a broad fraction of the Mars surface. We encourage additional study aimed at advancing wind turbine technology to operate efficiently under Mars conditions and to extract more power from Mars winds. Wind power can be an oft-neglected source of energy for future human exploration missions on Mars, especially coupled with solar power. Modelling shows that solar and wind energy can fully power such missions for more than half of the Martian year for ten regions of interest identified by NASA. Another 13 promising sites are identified.

Energy sustainability and redundancy for surface habitats, life support systems and scientific instrumentation represent one of the highest-priority issues for future crewed missions to Mars. However, power sources utilized for the current class of robotic missions to Mars may be potentially dangerous near human surface habitats (for example, nuclear) or lack stability on diurnal or seasonal timescales (for example, solar) that cannot be easily compensated for by power storage. Here, we evaluate the power potential for wind turbines as an alternative energy resource on the Mars surface. Using a state-of-the-art Mars global climate model, we analyse the total planetary Martian wind potential and calculate its spatial and temporal variability. We find that wind speeds at some proposed landing sites are sufficiently fast to provide a stand-alone or complementary energy source to solar or nuclear power. While several regions show promising wind energy resource potential, other regions of scientific interest can be discarded based on the natural solar and wind energy potential alone. We demonstrate that wind energy compensates for diurnal and seasonal reductions in solar power particularly in regions of scientific merit in the midlatitudes and during regional dust storms. Critically, proposed turbines stabilize power production when combined with solar arrays, increasing the percent time that power exceeds estimated mission requirements from ~40% for solar arrays alone to greater than 60-90% across a broad fraction of the Mars surface. We encourage additional study aimed at advancing wind turbine technology to operate efficiently under Mars conditions and to extract more power from Mars winds.
Future crewed missions to Mars will require sustained sources of energy, including solar, nuclear and wind. Site selection and risk assessment strategies must critically assess the available energy resources on both long-term and shorter diurnal and seasonal timescales. The driving principle behind site selection for human missions to Mars thus far has centred around physical resource allocation and, in particular, the surface or near-surface availability of water ice and the distribution of volcanic lava tubes that can serve as long-term habitats. Mandates to 'follow the water' and 'find shelter' direct attention to regions with evidence of surface liquid water as in recurring slope lineae in the Valles Marineris, Mawrth Vallis and midlatitudes [1][2][3] , close to large near-subsurface ice deposits as in the Northern Hemisphere polar and Article https://doi.org/10.1038/s41550-022-01851-4 the use of wind power as an additional energy resource for future human missions to Mars.
Wind energy has historically been disregarded as an alternative energy resource due to Mars' low atmospheric density, which reduces the force associated with winds of all magnitudes by approximately 99% compared with Earth. A recent study by ref. 14 examined the wind power potential associated with winds at the Viking Lander 2 site during the 1977b global dust storm. They concluded that low energetic yields from simple turbines would rule out wind energy as a primary or even secondary backup power resource. However, site selection for missions with the most reliable wind measurements (for example, the Viking and Insight landers) have prioritized quiescent environments with low wind speeds that would otherwise complicate landing operations or contribute to instrument noise 15,16 . To evaluate the impact of higher wind speeds, for example along crated rims or at higher altitudes, other studies have assessed power production by a fixed wind speed distribution bounded by theorized Martian minimum and maximum wind speeds [16][17][18][19][20][21] . However, static wind fields do not capture critical diurnal and seasonal variability that would augment the local power potential or complement time-varying solar energy. Without a current or forthcoming global network of surface wind speed measurements, global climate models (as used in this research) provide the best opportunity to assess the impact on potential wind power yields of varying topography, surface thermal properties, seasonal and diurnal solar heating and dust opacity.
However, regions of scientific interest or with the greatest diversity of physical resources may not overlap with regions with the highest energy production potential. The very characteristics that make the current class of proposed landing sites or regions attractive, including their geology, theorized mineralogical history and availability of local resources, often limit traditional energy resources. Solar energy requires that a sufficient proportion of solar radiation reaches the planetary surface on a per-Mars day (sol) basis. Available solar power varies with time of day, season and latitude. If potential landing sites move poleward, where water ice deposits are more readily available, the seasonal variability of solar energy increases and the need for a secondary energy resource is amplified. During dust events, atmospheric optical depths (a measure of the attenuation of light transmitted through the atmosphere) can reach prohibitively high values >10 in visible wavebands 11 and greatly reduce solar energy on short to multi-sol timescales 12 . Dust accumulation on solar cells will further decrease efficiency (by ~0.2-2% per sol 13 based on lander and rover histories). Solar energy is further limited by time of day; solar cells are not operational during nighttime hours, while mission and life support systems remain critical. Remediation of seasonal and shorter-term, multi-sol power outages with nuclear power has its own risks, such as the safety of nuclear devices near human settlements and long-term waste disposal. If mission plans prioritize local physical resources and scientific interest above theoretical solar energy yields, and if power storage is limited, engineering multiple redundant energy sources will be critical. In this research we propose  a Martian application. At the same time, a greater number of options for turbine types (for example, vertical-axis and airborne or balloon turbines), hub heights and sizes are available. These options provide flexibility depending on the energetic requirements for individual missions and engineering considerations for transportation and in situ assembly. It is beyond the scope of this work to assess the ultimate feasibility of turbine transport, in situ construction and operations on Mars; rather, we intend this article to serve as a proof of concept. We encourage future work on this topic, including in-depth engineering studies that reduce turbine weight and maximize efficiency. We explore specific engineering uncertainties related to transport and operation on Mars in the Supplementary Information. In this paper, we use a state-of-the-art Mars global climate model (GCM) to assess the planetary wind power potential and its short-term and seasonal variability using two basic metrics: the wind power density (WPD) and the power generated by specific analogue turbines. The WPD represents the maximum possible power production per unit area for an idealized or 100% efficient turbine. The WPD has the added benefit of being technology agnostic and therefore remains useful as turbine technology continues to advance. Additionally, the WPD is the most straightforward tool to help compare wind and solar energy without complicating considerations of array or turbine size and efficiency. We also calculate the power return from a medium-scale base turbine, the Enercon E33, and three additional turbines ranging from microscale to industry scale. We compare the turbine power generation alone and in combination with a theoretical solar array to the hypothesized power requirements for a crewed mission to Mars. We evaluate the surface distribution and seasonality of these metrics and identify potential regions of interest for future landing sites.

WPD demonstrates the maximum available energy
Wind power (W m −2 ) is proportional to atmospheric density ρ (kg m −3 ) and wind speed (u) cubed ((m s −1 ) 3 ) (equation (1)). Mars has a thin atmosphere and, in general, weaker winds than on Earth. As a result, power return from any wind-based energy system will be proportionally less. For the same wind speed, the Mars WPD will be ~1% of the Earth WPD, due to the large difference in atmospheric density, as given by equation (1). Figure 1 shows the annual average WPD at 5, 30, 50 and 100 m for a non-dust storm year. In general, wind speeds increase with height, and therefore the power potential increases as the turbine hub height (altitude of the turbine blades) increases from a global average of 6 W m −2 at 5 m to 55.7 W m −2 at 100 m. Wind turbines on Earth have hub heights as high as 100-150 m; however, engineering or transport restrictions will likely limit the maximum hub height on Mars. We therefore choose a moderate hub height altitude of 50 m, an average hub height for medium-scale turbines. For power calculations that use turbine-specific power curves (Supplementary Data 1-4), we vary hub height from 5 to 100 m to match individual turbine specifications: At all altitudes, baroclinic wave activity drives high wind speeds associated with the winter hemisphere polar vortex and the descending branch of the seasonal Hadley circulation. Baroclinic waves travel towards the equator and summer hemisphere through north/south topographic channels. As a result, the WPD is highest along regions with large topographic gradients, such as along crater rims and throughout the volcanic highlands. Wind power is similarly enhanced in regions with high thermal variability. During Northern Hemisphere winter, winds blow from cooler surface ice deposits to warm regolith. This effect, analogous to a 'sea breeze', may be particularly important at proposed high-latitude sites adjacent to seasonal ice deposits. In several locations, the annual average wind power exceeds available solar power by up to 3.4 times (Fig. 1e). While the 50 m Mars annual average WPD is low (typically less than 100 W m −2 , global average 31 W m −2 ), on short timescales and in some locations, the instantaneous WPD is greater than 1,000 W m −2 for ~10% of the Mars year and can exceed 3,000 W m −2 (<1% of year, Supplementary Table 1). Large values must be considered in the context of overall resource stability, which is used as a criterion for the identification of potential sites of interest later in this study.
The WPD represents the upper bound of available power; sites that are discarded based on the available WPD should be eliminated regardless of potential turbine type.

Wind power variability
Temporal averaging used to produce the annual average WPD masks the predicted resource potential in regions with large seasonal and/or diurnal variability. Figure 2 and Supplementary Video 1 show the wind and solar power density at the Mars cardinal seasons. We draw particular attention to the potential solar and wind energy yields in the polar and midlatitudes in the solstitial winter hemispheres. When solar energy is seasonally reduced (Fig. 2a,c), wind energy represents an important energy backup. Fortuitously, when wind energy is potentially most important, wind energy yields are also at their highest-for example, as along the Hellas (42.74° S, 70.5° E) and Argyre (49° S, 318° E) impact basins at the solar longitude 90° (L s ~90°, time of the northern summer solstice) and northwards of approximately 40° N latitude at L s ~270° (northern winter solstice). In these regions, the seasonal average wind power regularly exceeds 100 W m −2 , while solar power is reduced to values below 25-50 W m −2 . To demonstrate this seasonal cycling between solar and wind power, we plot the wind in 5° longitude bands and zonal average solar power versus time at several latitudes ( Supplementary  Fig. 1). Wind power more than compensates for reduced solar power in the winter hemisphere. Reductions in the Southern Hemisphere solar power between L s ~220° and 240° are associated with regional storm activity ( Supplementary Fig. 1d,e).
During global dust storm years, higher atmospheric dust levels are positively correlated with greater wind power production, as we show with explicit calculations summarized in Fig. 3. At the time of the most intense dust storm activity (L s = 260°-280°), wind power in Mars Year (MY) 28 (a global dust storm year) exceeds MY24 (a typical, non-global dust storm year) wind power production by greater than 60 W m −2 on average and by up to 300 W m −2 . This elevated wind power occurs simultaneously with an approximate 25 W m −2 (22%) global decrease in solar power due to high atmospheric dust opacity. The extent to which turbines remain efficient if dust accumulates during local storms requires additional study 29 . The most fundamental limitation for solar power development is its diurnal variability. If excess energy cannot be easily stored, then power redundancies during nighttime hours will be required. Wind energy is particularly suited to address this problem because winds on Mars are generally fastest at dawn and dusk. In Fig. 4, we show the annual average day and night WPD and the ratio. Night is defined in our simulations as local time (LT) 17-8; one Mars hour is calculated by dividing the Mars sol into 24 equal segments of 3,696 s. Nighttime wind power exceeds the available daytime wind energy by up to a factor of 53.9 and on average by a factor of ~2. Supplementary Fig. 2 and Supplementary Video 2 show the annual average solar and wind power density in one-Mars-hour intervals. Solar power exceeds 140 W m −2 during daytime hours (~8-17 LT), while wind power maximizes at night (~17-8 LT). Wind power could be a valuable resource, particularly at night, during local and global dust events and seasonally in the midlatitudes and polar regions.

Enercon E33 power and energy calculations
The degree to which power production will depart from the WPD or theoretical maximum depends on both the turbine efficiency and the area swept by its blades (equation (2)). Here, ρ is the atmospheric density, u is the wind speed at the turbine hub or blade height, A is the rotor swept area and C p is the wind speed-dependent power coefficient 30 . The efficiency has an upper bound of 59%, the Betz limit, which is further reduced by aerodynamic and engineering losses: To account for turbine-specific engineering with complex efficiency factors that shift energy production from equation (2) (for example, at different wind speeds), turbine documentation includes the measured power curve. The power curve provides the anticipated power return in wind speed bins at standard sea-level conditions (Supplementary Data 1-4). Power curves are unique to each turbine and consider variations of the power coefficient with wind speed, as well as threshold wind speeds required to initiate (cut in) and stop (cut off) energy production. Simulated wind speeds are corrected based on the relative air density and therefore wind-generated force before being used as inputs for power curves (equation (3)). This density adjustment is common practice for high-altitude sites on Earth and follows the International Electrotechnical Commission Standard 31 . While additional factors may need to be considered under Martian conditions (for example, blade dynamics in a low-Reynolds-number atmosphere), power curves represent a meaningful improvement over theoretical power yields that are highly generalizable but often inaccurate. For example, many turbines have higher operational efficiency at wind speeds below the rated wind speed 32 (Supplementary Data 3). Calculations using a constant efficiency factor (equation (2)) underpredict the power generation at low wind speeds and overpredict the power generation at high wind speeds near the cut-off threshold ( Supplementary Fig. 3, purple curve versus blue curve). Here, we use the Enercon E33 power curve, which is currently used at the Ross Island Wind Farm, an analogue site for present-day Mars 33 . Turbine specifications are summarized in Supplementary Table 2.
The turbine energetic yield quantifies the cumulative power produced over some amount of time (for example, per sol, seasonally or annually). Supplementary Fig. 4 shows the annual energy production (AEP), or the total power at each location summed over the duration of the Mars year. Values in regions with high wind potential exceed 1.6 GWh (2,589 kWh per sol), while the global average equals 0.169 GWh (959.96 kWh per sol). While this value is small compared to the average energy generated by an Earth turbine (~27,715 kWh per day 34 ), it compares favourably with expected energy production by solar arrays,   (4)), as well as estimates for human mission energy requirements (576-840 kWh per sol 36 ). We compare the global wind and solar AEP and approximate solar panel array dimensions required to generate the same total annual energy at each location. In regions with the highest wind AEP, solar arrays need to exceed 7,000 m 2 to match the highest AEP. To match the average AEP would require solar array sizes greater than 600 m 2 . Supplementary Fig. 5 shows the MY28 wind and solar AEP.
In the Supplementary Information, we explore the timing of power generation-for example, continuously or intermittently (turbine load duration). We also discuss the power output from four additional turbines, scaling from micro-to industry scale (alternative turbines).

Potential ROI for future human missions to Mars
Wind energy will only be useful if it can be produced in regions that are attractive for future human missions. Considerable work has already been done to identify potential human landing sites based on geology, resource potential and engineering limitations. The NASA Human Landing Site Study (HLS2) identified 50 potential regions of interest (ROI) based on these criteria. The study did not consider the energy availability of each region beyond coarse latitude restrictions to avoid polar night or excessive terrain shading that would limit agricultural development.
Numerous studies have additionally assessed the energy requirements for human missions of varied size and duration. The most comprehensive of these studies 36 calculates a 24-35 kW surface system power requirement to support a six-crew, 500-sol mission. Red squares in Fig. 5 (Supplementary Data 5) show locations where the diurnal average wind power produced by the Enercon E33 turbine in MY24 exceeds this limit at all simulated times. At these locations, wind energy is theoretically sufficient to power the entire mission on its own. Any human mission to Mars will include multiple redundant energy sources, including solar and nuclear. We similarly assess regions of the planet where our base turbine could power different portions of the mission-for example, the surface habitat and life support systems (12.1-17.258 kW) or scientific instrumentation (2.2 kW) 36 . These locations are well suited for wind energy development, but need  further study to assess their scientific merit, resource accessibility and engineering or landing site limitations. The major advantage of wind energy is its availability as a complement to solar during global dust storms, seasonally in the winter mid-to high latitudes or at night (Supplementary Fig. 6). At winter solstices, wind energy is a valuable resource particularly near the poles and midlatitudes. The increase in locations with high wind energy is in part due to the reduced time period over which the energy resource is required to stay stable (here, ~50 sols or 20° L s compared with Fig. 5, which covers the entire Mars year). Elevated energy levels in the winter hemisphere correspond to seasonally enhanced wind speeds. Many sites at these latitudes have been dismissed due to solar energetic limitations during polar night. We demonstrate that if the availability of water outweighs other challenges, wind energy could act as the dominant energy source when solar energy is seasonally depleted, opening a large fraction of the polar landscape to human exploration. Similarly, wind energy resources maximize at night when solar energy is at its minimum.
To demonstrate the combined resource potential from solar and wind, we show in Fig. 6a-c the percentage of time over the Mars year that power generation exceeds 24 kW based on production from the Enercon E33 turbine and a theoretical solar array. We calculate solar power using a solar panel efficiency factor of 0.2, at the high end of the range for commercial technology 35 , not accounting for reductions in efficiency due to surface dust accumulation, which can exceed 89% (ref. 37 ). Simulated solar arrays have a total area of 2,500 m 2 based on the 2017 NASA Small Business Innovation Research solicitation 38 , but may be smaller 32,39 (Supplementary Fig. 7). Solar arrays of this size generate the requisite power about 40% of the time in a broad zonal band from ~60° S to 60° N. This makes sense because solar power is unavailable at night (~50% of the time) and is further reduced in the winter hemisphere and during local dust events. The Enercon E33 generates power with equivalent or greater stability (>24 kW for 30%-50% of the year), but peaks are more highly localized. Combined percentages can exceed 90%, which indicates that wind power is most active when solar power is reduced, rather than generating useful but redundant power at the same times of the day and year.
Based on these metrics, we identify favourable locations for wind energy development on its own and as a complement to seasonally and diurnally varying solar output. Of the 50 ROI identified in the HLS2, a single Enercon E33 turbine at Deuteronilus Mensae 2 (35° N, 23° E), Protonilus Mensae (38° N, 48° E) and Ismenius Lacus (29° N, 17° E) could produce 24 kW more than 35% of the year (and >75% of the year when combined with solar power) and therefore represent the most attractive sites from a wind resource perspective. A further seven sites generate substantial wind energy seasonally and contribute more than 50% of the total power generation either in the winter hemisphere or in the dusty Southern Hemisphere at L s = 270°. Supplementary Table  4 highlights ROIs from the HLS2 that represent the best candidates for wind energy development. We note that if wind power is used to maintain power for scientific instrumentation (2.2 kW) 36 , only ten ROI are excluded by a 50% seasonal wind power time requirement. We also identify several new broad ROI for human exploration that generate the mission energy requirements >40% of the time (or >65% of the time with solar) annually or seasonally, and therefore compensate for reduced solar output. Findings are shown in Fig. 6d.

Conclusion
Wind energy represents a valuable but previously dismissed energy resource for future human missions to Mars, which will be useful as a complementary energy source to solar power. Wind is particularly valuable when solar power is reduced, such as at nighttime, during local or large-scale dust events and at higher latitudes that are attractive due to their proximity to water resources but that experience large seasonal variability in solar power. GCM studies that capture the wind field's seasonal and diurnal variability enable a comprehensive probing of the global wind resource potential and aid in the identification or exclusion of potential landing sites for future human missions. We find that wind power represents a stable, sustained energy resource across large portions of the Mars surface. Based on wind energy analysis, we identify 13 new ROI for human exploration of Mars and highlight ten of the 50 previously identified HLS2 ROI that have good wind resource potential. The best candidates are Deuteronilus Mensae 2 (35° N, 23° E), Protonilus Mensae (38° N, 48° E) and Ismenius Lacus (29° N, 17° E).

Summary of global climate model set-up
We use the new NASA Ames Mars GCM, which couples the National Oceanic and Atmospheric Administration Geophysical Fluid Dynamics Laboratory cubed-sphere finite volume dynamical core with physics packages from the NASA Ames Legacy model described in refs. 40,41 . Simulations are performed with 1 × 1-degree resolution and 30 vertical atmospheric levels between 724 and 0.35 Pa. The vertical pressure grid and corresponding altitudes above the surface are shown in Supplementary Table 4. The model uses topographic, thermal inertia and albedo maps from the Mars Orbiter Laser Altimeter and Viking and Mars Global Surveyor Thermal Emission Spectrometer observations, respectively. The surface roughness is fixed to 0.01 m (ref. 40 ). To assess both wind and solar energy resources during a nominal background dust cycle (MY24) and during a global storm (MY28), we use dust opacity maps that match the observed dust optical depths from MY24 and MY28 ( Supplementary Fig. 8 (ref. 42 )). The model has been previously validated by comparison with Mars Climate Sounder atmospheric temperature and surface pressure observations (described in ref. 43 ).
Near-surface winds are strongly influenced by the local environment, including small-scale topography, surface thermal properties and short-term variations to the atmospheric thermal profile and dust heating. As a result, one cannot use simple analytic wind speed profiles to assess the available power. Many of these processes occur below the resolution of the GCM and are best captured with very high-resolution mesoscale models. However, we can broadly evaluate model performance by comparison with observations at landing sites with relatively flat topography, which demonstrates the model's ability to simulate large-scale forcing and winds. Simulated wind speeds reproduce the observed annual wind cycle at the Insight Lander (4.5° N, 135° E), the most recent and comprehensive set of wind observations (Supplementary Fig. 9 (ref. 44 )). For regions with more complex topography, as at Jezero Crater, winds follow trends predicted by other state-of-the-art Mars GCMs 45 . Without a more comprehensive data set of observed wind speeds on Mars, error estimation for simulated winds is non-trivial. However, since the wind power scales with the wind speed cubed, the potential error on the predicted wind power can be large. For example, a 10% error on the simulated winds translates to a 33% error on the wind power density. This further motivates the use of mesoscale models for in-depth characterization and energy analysis of potential ROI. Future missions that provide a network of meteorological stations would further improve our ability to more carefully test the model.

Calculation of wind power density and turbine-specific power output
Diagnostic variables including atmospheric temperature and zonal and meridional wind fields are calculated every 15 min and averaged over 5 sols in one-Mars-hour (3,696 s) increments to capture diurnal variability, and then interpolated to altitude above the surface to assess power outputs at hub heights between 5 and 100 m above the surface. The wind power density and turbine power and energy calculations use the wind speed (u) = (u 2 , where u z is the zonal wind and u m is the meridional wind. One cannot use simple analytic wind speed profiles to assess the available power. Because winds are so tightly connected to varying atmospheric conditions (temperature, dust heating, surface topography and thermal properties, planetary waves and thermal tides) that can vary on short timescales, a GCM is the best Article https://doi.org/10.1038/s41550-022-01851-4 way to evaluate the potential WPD across the planetary surface and its seasonal and diurnal variability.
To calculate power outputs from individual turbines (Supplementary Table 2), we use industry power curves, which measure the turbine power output versus wind speed at standard atmospheric temperature and pressure ( Supplementary Data 1-4). Power curves account for varying turbine efficiency with wind speed and other turbine-specific engineering factors that impact the expected power output. Estimations of available power using the power curve are therefore more accurate than assessments using a theoretical turbine that assume a constant efficiency factor.
Due to Mars' low-density atmosphere, winds are less forceful (and therefore generate less power) than winds of the same speed on Earth. To account for this difference, simulated wind speeds are first scaled based on the relative atmospheric densities of Mars and Earth using a simple wind speed transformation for high-altitude sites on Earth (equation (3) (refs. 29,31 )). The adjusted wind speed represents the Earth wind speed that would generate an equivalent force as the simulated wind on Mars and is used as the input for the power curve. So, for example, a ~15 m s −1 wind on Mars generates the same equivalent force as a 3.9 m s −1 wind on Earth and therefore would produce 9 kW based on the power curve for a 330 kW turbine. In the same manner, the turbine cut-in wind speed-the wind speed at which power production beginsis shifted to higher wind speeds on Mars than on Earth. Assuming the average atmospheric densities (0.2 kg m −3 for Mars and 1.125 kg m −3 for Earth), the density-adjusted cut-in wind speed for the Enercon E33 wind turbine would be ~4.45 m s −1 on Mars compared to 2.5 m s −1 on Earth. The global annual average wind speed distribution and the adjusted wind speed distribution are shown in Supplementary Fig. 3. The adjusted wind speeds are used as inputs for the various power curves (Supplementary Fig. 3, solid lines). In equation (3), ρ earth = 1.125 kg m −3 . Because Mars' atmospheric density (ρ mars ) varies dramatically with location as well as with time of year, the wind speed adjustment is performed at the highest output frequency (5-sol, hourly average) and uses simulated density fields that vary by location and atmospheric level.
Our base turbine is the Enercon E33 330 kW turbine 46 . Additional turbines discussed in the Supplementary Information scale from small-scale, household-sized turbines (20 kW) 47 to industrial turbines (5 MW) 48 . We also include an analysis of a microturbine, the Aeolos-V 300 W turbine 49 . Wind turbine power curves, engineering specifications (hub height, weight, blade radius) and cut-in, rated and cut-off wind speeds are provided by technical documentation.

Comparison with solar power
Solar power for a theoretical solar array is estimated using equation (4) (ref. 17 ), where A is the total area of the solar panel array = 2,500 m 2 (ref. 38 ), ϕ is the solar power density or solar irradiance incident on a horizontal surface (W m −2 ) and C p is the panel efficiency factor. For this study, C p = 0.2. The solar irradiance is calculated within the model and varies based on the cosine of the solar zenith angle, the total atmospheric column and absorption and scattering by atmospheric gas and aerosols 41 . Future studies may consider inclined panels that would increase the incident solar flux 50 . The current work does not include depletions in solar array efficiency due to dust accumulation. Taking into account the proposed size of arrays (2,500 m 2 in ref. 30 ) and the complexity and resource expenditure for astronaut excursions outside of the surface habitat, we believe periodic dust clearing could represent a considerable time sink. However, it is outside the scope of this paper to speculate on the cost or relative ease of specific tasks.

Energetic yield
The turbine energetic yield is calculated by summing the turbine hourly power production, P (kW), over a specified duration of time. We include a multiplicative factor in equation (5) to convert from the model output 5-sol average to the total power production. For the annual energy production, the power is summed over the Mars year.

Turbine load duration
Neither the WPD nor the AEP quantify the intermittency of power production that will be critical for supporting long-term human missions. For example, two sites with the same AEP may generate energy very differently: slowly over the entire year or in high-magnitude but sporadic bursts. To better understand the stability of wind resources over the Martian year, we calculate the power duration curves at proposed landing sites. The power duration curve measures the percentage of time at which a turbine is functioning at some percentage of its rated power. Most turbines operate at varying capacity throughout the year. The optimal operational capacity will vary with each turbine, site and mission. For example, using a larger turbine with a higher-rated power that operates at a lower capacity may be more desirable and ultimately generate more power than using a small turbine that operates near capacity. Therefore, when evaluating power duration in this work, we focus on the distribution of energy production (for example, smooth or stochastic) rather than the exact capacity factor.
We show the global average annual and seasonal power duration curves as well as the power duration curves at three potential landing sites in Supplementary Fig. 10. The global average load duration ( Supplementary Fig. 10a) is generally low, operating at less than ~2-4% capacity for 50% of the time. For the Enercon E33 wind turbine, this translates to an average operational power of approximately 10 kW. As expected, load durations vary substantially with season and location. At Protonilus Mensae (38° N, 48° E), a proposed landing site in the Northern Hemisphere midlatitudes, the capacity factor exceeds 20% for ~22% of the year and increases to above 40% capacity for 9.66% and 6.56% of the time during Northern Hemisphere spring and fall equinoxes (L s = 0° and 180°, respectively), when solar power yields are reduced. Seasonal curves show the time average over 20° L s or approximately 30-50 sols. Power duration curves suggest that wind power generation occurs at low levels but consistently rather than sporadically over the Mars year. This is critical because energy storage capabilities on Mars are still limited. Consistent, low-power generation makes wind energy more valuable than substantial energy production in short, sporadic bursts.

Alternative turbines
Wind turbine power production will vary with turbine specifications, including turbine hub height, blade area and engineering specifications, that determine cut-in, cut-off and rated wind speeds, as well as the wind speed-dependent power coefficient. To demonstrate this variability, we include analysis of three additional turbines (see Supplementary  Table 2 for a description of turbine hub height and rotor diameter). The Jacobs 31-20 turbine is the smallest turbine considered here 47 . Its primary use is single-family, local residences or small-occupancy power needs. At the other extreme, we also show the power return of an industry-standard 5 MW turbine used in offshore wind farms on Earth to power large industries or population centres 48 . The largest turbines on Earth now produce up to 15 MW (ref. 51 ). Each turbine has different advantages and disadvantages for operation at an interplanetary scale. For example, the National Renewable Energy Laboratory 5 MW turbine generates the highest theoretical power by a large margin; however, it is also the largest and most difficult to transport, set up and maintain on Mars.
Supplementary Table 5 lists the global average AEP for each wind turbine. As expected, smaller turbines with lower hub heights and lesser-rated power produce less energy. Similarly, an industrial-scale Article https://doi.org/10.1038/s41550-022-01851-4 turbine such as the National Renewable Energy Laboratory 5 MW turbine produces large amounts of energy nearly uniformly across the entire Mars surface (Supplementary Fig. 11). The smallest turbines are best suited to act as a backup power resource for solar arrays, while larger turbines could act as independent power sources. The most important takeaway is that even a small turbine will produce enough energy to power some portion of a human mission to Mars.
In addition to large industrial or household wind turbines, microturbines could potentially act as auxiliary power for scientific instruments or to recharge batteries. For example, we consider the power required to maintain operation of the Opportunity Rover during the 2018 (MY34) global dust storm. On 6 July 2018, the solar flux incident on the rover's panels fell below the threshold required to maintain communication with the rover (22 Wh) (~0.86 W) 52 . Solar panel power generation remained reduced for at least two months following the storm and recovery was ultimately impossible. This provides a unique test for microwind turbines as power failsafes for solar-powered missions. By contrast with crater landing sites for the MSL Curiosity and Mars 2020 Perseverance rovers, the topographic environment of Meridiani Planum lends confidence to the model's ability to simulate winds with reasonable accuracy.
The Aeolos-V turbine was selected based on its moderate weight and low cut-in wind speed. In theory, it could be mounted directly on the rover, providing a hub height of approximately 5 m. We simulated the 5 m power output of the Aeolos-V turbine at the mission end date (L s = 190.4°, MY34 dust map) and found a 5-sol diurnally resolved average power production of 207 Wh ( Supplementary Fig. 12). This amount of energy would be sufficient to maintain the rover clock (but not drive) through the most intense period of the dust storm or until solar power could be restored.

Engineering uncertainties and alternatives
Additional study is required to assess engineering restrictions on the transport and construction of turbines on Mars, as well as the operational efficiency of turbines subject to Mars' extreme environment. We provide a brief discussion here of potential engineering caveats that could impact potential power return and merit close consideration for future studies. First, transport of turbines to Mars may require substantial weight reduction. While the turbine nacelle would likely need to be constructed on Earth, blades are typically composed of fibreglass and therefore could potentially be fabricated in situ, substantially reducing the overall transported size and weight. Fabric-covered blades have also been proposed 53 . However, in situ manufacturing and assembly would require additional equipment, including, for example, blade moulds or frames and large-scale excavators or cranes. Some turbine components, including the tower, rotor hub, gearbox and frame, have been constructed out of aluminium alloy versus the traditional steel 54,55 . The use of alternative turbine types (for example, airborne or vertical-axis) or placing turbines on rovers or crater rims would reduce the overall height of towers and could therefore further reduce the total weight 56 . Tower foundations are typically composed of concrete with footing depths up to 3.5 m (ref. 57 ). While concrete alternatives including altered Mars regolith could be used 58 , excavation equipment would likely also need to be transported. However, drilling equipment could be used for multiple purposes, including accessing subsurface water ice deposits at similar depths. Site planning for missions with multiple turbines would need to evaluate optimal turbine spacing, including the effects of wake-induced efficiency degradation. On Earth, the general rule of thumb is to space turbines greater than seven rotor diameters apart, so for the Enercon E33, this would equate to approximately 234 m. On Mars, spacing would likely be altered due to Mars' thinner atmosphere and low Reynolds numbers.
A major outstanding question for the application of this work is the operational efficiency and technical viability of turbines subject to Mars' environment. For example, we propose that wind power would be particularly valuable as a complement to solar power at night and in the winter hemisphere midlatitudes and polar regions (as at Antarctic and Arctic analogue sites 28 ). However, as temperatures fall, some fraction of generated power may need to be utilized to warm turbine electronics. Heating mechanisms or alternative turbine blade coatings that reduce icing must also be considered 59,60 . Similarly, various factors will impact blade rotation dynamics and therefore power output, including the operation of electronics, lubricants and rotating parts at very low temperatures, dust accumulation inside and abrasion of turbine mechanics particularly during local dust storms 29,61 and density differences across the turbine blade length. We encourage future testing of both currently available technology and/or newly developed, specialized turbines in Mars wind tunnels or environmental chambers, as in refs. [18][19][20][21][62][63][64] , including generating Mars-specific power curves that account for Mars' reduced gravity, low atmospheric density and low Reynolds numbers. For example, additional study is required to understand the changing lift and drag on turbine blades in laminar or transitional airflow regimes associated with low-Reynolds-number atmospheres 21 . Given the success of previous missions to Mars, including instrumentation measuring local meteorology at night (for example, the TWINS instrument on Insight) and the recent successful deployment of the Mars helicopter, Ingenuity, we are optimistic about the engineering prospects. To illustrate the potential operating environment for wind turbines on Mars, Supplementary Fig. 13 shows the temperature, dust optical depth, density and wind fields at Deuteronilus Mensae 2 (35° N, 23° E).

Data availability
The data that support the findings of this study are available on Zenodo (https://doi.org/10.5281/zenodo.7246689). Source data are provided with this paper.