Stormy water on Mars: The distribution and saturation of atmospheric water during the dusty season

See allHide authors and affiliations

Science  17 Jan 2020:
Vol. 367, Issue 6475, pp. 297-300
DOI: 10.1126/science.aay9522

Water reaches Mars' upper atmosphere

Mars once hosted abundant water on its surface but subsequently lost most of it to space. Small amounts of water vapor are still present in the atmosphere, which can escape if they reach sufficiently high altitudes. Fedorova et al. used data from the ExoMars Trace Gas Orbiter spacecraft to determine the distribution of water in Mars' atmosphere and investigate how it varies over seasons. Water vapor is sometimes heavily saturated, and its distribution is affected by the planet's large dust storms. Water can efficiently reach the upper atmosphere when Mars is in the warmest part of its orbit, and this behavior may have controlled the overall rate at which Mars lost its water.

Science, this issue p. 297


The loss of water from Mars to space is thought to result from the transport of water to the upper atmosphere, where it is dissociated to hydrogen and escapes the planet. Recent observations have suggested large, rapid seasonal intrusions of water into the upper atmosphere, boosting the hydrogen abundance. We use the Atmospheric Chemistry Suite on the ExoMars Trace Gas Orbiter to characterize the water distribution by altitude. Water profiles during the 2018–2019 southern spring and summer stormy seasons show that high-altitude water is preferentially supplied close to perihelion, and supersaturation occurs even when clouds are present. This implies that the potential for water to escape from Mars is higher than previously thought.

Mars once harbored an active hydrological cycle, as demonstrated by geological features on its surface, but it no longer holds the quantity of water required to produce such geological imprints (1, 2). The planet’s bulk inventory of water amounts to a global equivalent layer (GEL) of ~30 m, mostly contained in its polar ice caps (2). This is less than 10% of the water that once flowed on the surface (1). Mars’ enhanced concentration of heavy water (semiheavy water five or more times the terrestrial standard) (35), strengthens the hypothesis that most of Mars’ primordial water has escaped over time.

Water in the atmosphere is a negligible component of the planet’s total water inventory, being equivalent to a global layer 10-μm thick, but nevertheless regulates the dissipation of water over time. Most martian water has been lost to space because its decomposition products (atomic hydrogen and oxygen) reach the upper atmosphere, where they can acquire sufficient thermal energy to overcome the low gravity of Mars (which is about one-third that of Earth’s). Water decomposition is theorized to follow a complex reaction chain involving the recombination of H atoms into H2 on a time scale of centuries (68), buffering any short-term hydrogen abundance variations. This mechanism has been challenged by observations showing that freshly produced H atoms can reach the exosphere (the uppermost layer where the atmosphere thins out and exchanges matter with interplanetary space) on a monthly time scale (9, 10). The observed short-term variability of the hydrogen atoms populating the exosphere could be caused by direct deposition of water molecules at altitudes high enough to expose them to sunlight, which subsequently triggers a rapid enhancement of hydrogen atoms in the exosphere (1113).

Testing this hypothesis requires characterizing the mechanisms contributing to upward water propagation through large-scale atmospheric circulation. One such mechanism is the cold trap imposed (as on Earth) by water ice cloud formation at low altitude, subsequent to water condensation. The condensation is predicted to occur whenever the partial pressure of water vapor exceeds saturation. The vapor pressure law causes the cold trap efficiency to depend heavily on temperature, which eventually limits the amount of water that can be transported to higher altitudes (1416).

We investigate these processes using occultations of the Sun by the martian atmosphere (henceforth, solar occultations), where the vertical distributions of gases and particles can be directly observed. We used the Atmospheric Chemistry Suite (ACS) (17) on the ExoMars Trace Gas Orbiter (TGO) spacecraft. ACS is an assembly of three infrared spectrometers that together provide continuous spectral coverage from 0.7 to 17 μm, with a spectral resolving power ranging from 10,000 to 50,000. Our dataset was assembled by performing solar occultations with the near-infrared (NIR), mid-infrared (MIR), and thermal infrared in honor of professor V. I. Moroz (TIRVIM) channels of ACS. The NIR channel (0.7 to 1.7 μm) encompasses absorption bands of CO2, H2O, CO, and O2, diagnostic of their molecular concentrations over altitudes of 5 to 100 km, with a vertical resolution of 1 to 3 km. TIRVIM (2 to 17 μm) provides simultaneous information on dust and water ice particle abundance.

We retrieved the volume fraction of water (i.e., its mixing ratio) and temperature using established methods (1820), including the joint extraction of CO2 and H2O molecular abundances from the 1.57- and 1.38-μm absorption bands, respectively. Spectra were fitted with a spectroscopic model at all altitudes below 100 km, and the profiles of gaseous components were subsequently retrieved using an iterative algorithm (21). Figure 1A shows an example of model outputs fitted to the spectra, along with the resulting vertical water vapor profile. The sensitivity to H2O depends on altitude, as it is a strong function of both the total number of molecules along the line of sight and the signal-to-noise ratio (SNR). SNR decreases exponentially with increasing atmospheric opacity along the line of sight, which is dominated by suspended dust aerosols and icy particles. Sensitivity to water vapor reaches 0.1 parts per million by volume (ppmv) at low altitudes in clear atmospheric conditions and is typically better than 1 ppmv between 10 and 75 km, rising up to ~20 ppmv at 100 km (17, 21).

Fig. 1 Example ACS spectra and retrieved profiles.

(A) ACS NIR spectra measured during orbit 2580 (Ls = 197.8°, 47.27°N, 85.2°W, local time 17:27) at three example altitudes, labeled in each panel. Synthetic models (blue curves) fitted to the data (red dots) account for the water content, CO2 number density, and atmospheric temperature. The residuals are shown with gray lines. (B) Corresponding retrieved profiles of the H2O mixing ratio, temperature, and saturation ratio derived from them. The profiles of the mass of aerosol particles (ice, blue circles; dust, red dots) per cubic centimeter obtained during the same orbit from ACS TIRVIM data are also shown. All altitudes are above areoid (equipotential surface for Mars, the analog of geoid for Earth). See (21) for details of the model fitting and retrieval procedures. The error bars or shaded regions, corresponding to 1σ confidence, are also explained in (21).

ACS NIR resolves the spectral structure of the CO2 rotational band, providing simultaneous temperature and pressure parameters self-consistently (21). This simultaneity allows us to evaluate the local water vapor saturation state (Fig. 1B), a necessary parameter to estimate how much water can pass through the condensation level and reach the upper atmosphere. For most occultations, the NIR water vapor and temperature profiles can also be evaluated against aerosol profiles from ACS TIRVIM or ACS MIR data (Fig. 1B and fig. S1) (21).

The Sun-synchronous near-polar orbit of the TGO allows us to survey water vapor and aerosol vertical distributions on a global scale. The TGO performs two occultations on each 2-hour orbit. On average, ACS NIR accomplished nine occultation observations per sol (martian day) in both hemispheres [except during five periods of ~15 to 20 days each, around solar longitudes Ls 175°, 205°, 270°, 330°, and 355° (22)]. This produced a dataset of ~1700 occultations between April 2018 (Ls 163° MY34, where MY stands for martian year) and March 2019 (Ls 356° MY34). A summary of our results is shown in Fig. 2.

Fig. 2 Derived atmospheric properties during the dusty season of MY34.

Each panel shows the data value in color, plotted as functions of Ls and altitude, with the northern hemisphere on the left and the southern hemisphere on the right. (A) Distribution of the ACS-NIR solar occultation observations, showing morning (red) and evening (blue) events. (B) Atmospheric temperature. (C) Water vapor mixing ratio. (D) Saturation ratio of water vapor; the blue regions correspond to an undersaturated state (i.e., saturation ratio < 1). (E) Water ice (blue) and dust (brown) aerosol extinctions.

During MY34 southern spring and summer, the atmosphere was affected by two large-scale dust storms. The first, storm 2018A (Ls 188° to 250°) (23), enshrouded the planet globally [a global dust storm (GDS)], and the second, a recurrent large regional storm designated C (24, 25) occurred later in the southern summer (Ls 320° to 335°).

Several prominent features are visible in Fig. 2. Both hemispheres exhibit relatively high water vapor mixing ratios (VMRs) throughout the southern spring-summer season. The southern hemisphere is distinctly wetter, with VMR exceeding 50 ppmv in the 50-to-100-km altitude range, while above 40 to 50 km, the northern hemisphere exhibits a gradual decline in VMR after Ls 210°. Water vapor propagates recurrently up to 100 km in the south, around Ls 200° (during the GDS), 270°, and 290°. In the north, this only occurred during the GDS (see also fig. S2). This migration of water vapor to high altitudes does not correlate with an increase in temperature, as evident around Ls 195° in the north, where water vapor propagates higher than altitudes where the temperature rises. In the southern hemisphere, this is also evident around perihelion, when Mars orbits closest to the Sun. This implies that water vapor at high altitudes is primarily controlled by the large-scale motion of the martian atmosphere, particularly the upward branch of the cross-hemispheric atmospheric circulation cell, also known as the Hadley cell.

The saturation state in each hemisphere is shown in Fig. 2D, along with the temperature (Fig. 2B), water vapor (Fig. 2C), and aerosol concentrations (Fig. 2E). Large saturation ratios (1 to 10) are present in both hemispheres below 30 km, before and during the early phase of the GDS (Ls < 190°). ACS was observing the latitudes poleward of 60° at that time, implying that at least a third of the global atmosphere between 5 and 30 km was supersaturated. This is reminiscent of a previously observed supersaturation phenomenon (19) yet extends over a greater vertical range. Similar features can be seen in both hemispheres during this period—thick water ice clouds in a supersaturated atmosphere—suggesting that cloud formation does not limit the atmospheric saturation even when dust aerosols, on which ice crystals can form, are present. The clearing of dust particles carried by falling ice crystals (known as scavenging) is therefore not the sole reason for the existence of supersaturation on Mars, as was previously proposed (19).

During the following season, the northern hemisphere exhibits saturation ratios of >10 after Ls 330°, perhaps beginning around Ls 315°, before the early phase of the C storm. The southern hemisphere exhibits a more prevalent supersaturation at the same time, in the form of a discrete supersaturated area stretching between 15- and 40-km altitude, with unsaturated air beneath it extending to the pole. The same feature reforms soon after the C storm but slowly vanishes by the time of the spring equinox (Ls 360°).

In the 80-to-100-km part of the profiles shown in Fig. 2D, a supersaturated layer seems to persist throughout the observing period. Although the saturation state is generally less reliable in that altitude range (fig. S1), this layer is distinctly observed at Ls 190° to 200° in the northern hemisphere (Fig. 1B), and water ice clouds are also observed. The existence of such high-altitude supersaturation indicates efficient ascent of water to the upper atmosphere. Other regions of saturation are observed intermittently in both hemispheres at 50 to 60 km. The temperature is erratic in this region owing to atmospheric waves that generate higher temporal variability, causing fluctuations of >20 K over a few weeks. Because this region was previously filled with humid air during the GDS, cloud formation is enhanced. The clouds form in a similar configuration to the lower troposphere, as discussed above, characterized by the simultaneous presence of clouds and supersaturation. An alternative version of Fig. 2, showing the averaged VMR and saturation ratio profiles binned into three narrow Ls intervals, is shown in fig. S3.

Our analysis constrains the mechanism controlling water propagation from the lower to the upper atmosphere. While dust controls upward propagation of water during the 2018 GDS and the 2019 C storm, water vapor can effectively and persistently reach the upper atmosphere around perihelion in the southern hemisphere. This coincides with the seasonal intensification of the Hadley circulation that reaches its peak around Ls 240° and is active until Ls 290° (fig. S2).

Large portions of the atmosphere are in a state of supersaturation, complementing previous observations (19, 26). Unconstrained by saturation, the water vapor globally penetrates through the cloud level, regardless of the dust distribution, facilitating the loss of water to space. Because supersaturation is observed concomitantly with dust or ice particles, we conclude that condensation does not efficiently prevent water vapor from becoming supersaturated, even when seeds for condensation exist. We speculate that this may be due to rapid drops in temperature and/or rises in water concentration, which occur faster than condensation can keep up with.

Our results also show that water access to high altitude is affected by the seasonal changes around perihelion. Although planetary-scale dust storms appear in this period, those irregular events have a lesser impact than does seasonal change, which we suggest is the major atmospheric regulator for water. The seasonal recurrence and duration of the perihelion climate dominate the intermittent and short-lived effects of nonperihelion storms. Because perihelion coincides with the most intense period of the Hadley circulation, whose upwelling region is theoretically located in the southern tropical latitudes (2730), and with the warmest period of the year, the perihelion season has likely governed the escape of water to space over geological time scales.

Supplementary Materials

Materials and Methods

Figs. S1 to S14

Tables S1 to S3

References (3149)

References and Notes

  1. Materials and methods are available as supplementary materials.
  2. The solar longitude Ls is the Mars-Sun angle, measured from the northern hemisphere spring equinox, which is defined as Ls = 0°. The northern winter solstice is at Ls = 270°. Perihelion, where Mars orbits closest to the Sun, occurs at Ls = 251°.
Acknowledgments: ExoMars is a joint space mission of the European Space Agency (ESA) and Roscosmos. The ACS experiment is led by the Space Research Institute (IKI) in Moscow, assisted by LATMOS in France. We thank the numerous people responsible for designing, building, testing, launching, communicating with, and operating the spacecraft and science instruments of the TGO. Funding: The project was funded by Roscosmos and Centre National d’Etudes Spatiales (CNES). The science operations of ACS are funded by Roscosmos and ESA. Authors affiliated with IKI acknowledge funding from the Russian government under grant number 14.W03.31.0017 and contract number 0120.0 602993 (0028-2014-0004). Authors affiliated with the University of Oxford acknowledge funding from the U.K. Space Agency under grants ST/R001502/1 and ST/P001572/1. Authors affiliated with LATMOS acknowledge funding from CNES and Centre National de la Recherche Scientifique (CNRS). K.S.O. acknowledges the Natural Sciences and Engineering Research Council of Canada grant PDF-516895–2018. Author contributions: A.A.F., F.M., and O.K. conceived of the study, collected input from the other authors, and wrote the paper. The ACS observations and raw dataset were prepared by A.T., A.V.G., A.S., A.P., and N.K. A.A.F. calibrated the NIR ACS data and analyzed the profiles (assisted by A.T. and J.-L.B.). N.I.I. provided TIRVIM calibrated data. A.T., K.S.O., and L.B. provided MIR calibrated data. D.B. provided the ACS MIR aerosol extinction profiles. M.L. provided retrieval of aerosol properties using TIRVIM, NIR, and MIR datasets. J.A. and P.G.J.I. provided the MIR temperatures for validations. S.K. validated the MCS dataset. F.M., F.L., F.F., E.M., A.M., and C.F.W. provided expertise on the chemistry, circulation, and microphysics (assisted by J.-L.B.). All authors contributed to the preparation of the manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: The ACS data are available from ESA’s Planetary Science Archive at!Table%20View/ACS=instrument; we used the level 2 data (21). The temperature, H2O, and aerosol extinction profiles retrieved from the ACS measurements and analyzed in this article are available at, including the list of orbits we used.

Stay Connected to Science

Navigate This Article