Temporal and Spatial Variability of Lunar Hydration As Observed by the Deep Impact Spacecraft

See allHide authors and affiliations

Science  23 Oct 2009:
Vol. 326, Issue 5952, pp. 565-568
DOI: 10.1126/science.1179788


The Moon is generally anhydrous, yet the Deep Impact spacecraft found the entire surface to be hydrated during some portions of the day. Hydroxyl (OH) and water (H2O) absorptions in the near infrared were strongest near the North Pole and are consistent with <0.5 weight percent H2O. Hydration varied with temperature, rather than cumulative solar radiation, but no inherent absorptivity differences with composition were observed. However, comparisons between data collected 1 week (a quarter lunar day) apart show a dynamic process with diurnal changes in hydration that were greater for mare basalts (~70%) than for highlands (~50%). This hydration loss and return to a steady state occurred entirely between local morning and evening, requiring a ready daytime source of water-group ions, which is consistent with a solar wind origin.

EPOXI, NASA’s extended mission for the Deep Impact spacecraft, is targeted to fly by comet 103P/Hartley 2 in November 2010. En route to the comet, the spacecraft has had numerous close approaches to the Earth-Moon system and has observed the Moon as a calibration source, particularly for the 1.05- to 4.5-μm near-infrared spectrometer (1). The Moon Mineralogy Mapper spectrometer team (M3, onboard India’s Chandrayaan-1 spacecraft) learned of our planned June 2009 calibrations and, given our spectral range, requested confirmation of their results (2, 3). We observed the Moon twice in June 2009, 1 week (one-quarter of a lunar day) apart, when the spacecraft was over the northern polar regions. Nearside equatorial regions were also observed in December 2007 (table S1). Together, these data enable us to explore the distribution of surficial OH/H2O as a function of temperature, latitude, time of day, and composition.

The spectral range of the Deep Impact High-Resolution Instrument–infrared spectrometer (HRI-IR) includes the entire span of the broad 3-μm hydration feature, while independently constraining thermal emissions that dominate at 4 to 4.5 μm (4). In our analyses, we removed these thermal contributions (5) to reveal the full shape and structure of the hydration feature from 2.7 to 3.6 μm (Fig. 1). The observed hydration features show a maximum absorption at 2.81 μm, consistent with stretching vibrations of surface or structural hydroxyls (6, 7). A local minimum was also observed at ~2.95 μm, and in some spectra an additional feature was present at ~3.14 μm, which is commonly observed for hydrated minerals (6, 8, 9) and on Mars (1012). The 2.95-μm absorption can be attributed to asymmetric and symmetric stretching modes of the H2O molecule (6) or to OH bonds, depending on the bonding cations or bond energies. The first overtone of the H2O bending vibration commonly occurs near 3.1 μm and would be diagnostic of H2O. However, because of calibration uncertainties, this absorption is not distinctly resolved in our data. The overall shape of this lunar hydration feature is similar to hydration features observed on Mars (11, 12) and on certain classes of asteroids (13), both of which are attributed to OH- and/or H2O-bearing phases.

Fig. 1

High signal-to-noise reflectance spectra of the average response along a chord of the Moon acquired as Deep Impact scanned rapidly across the Moon. (A) Location of the equatorial chord (purple arrow) acquired on 29 December 2007 over a 750-nm Clementine basemap (15° grid). Arrow width indicates size of spectrometer slit. (B) Locations of two chords acquired on 2 June 2009 at mid-latitudes toward the morning (cyan arrow) and evening (blue arrow) sides. (C) Spectra of the three chords as compared to laboratory data of two lunar soils (14259, red; 62231, orange). Continua over the 3-μm region (dashed) reveal absorptions due to hydration (shaded regions) similar to hydration features in the laboratory data (some, if not all, of which is terrestrial in origin). (D) Continuum-removed spectra of the 3-μm regions of the Deep Impact chord spectra. The three major OH and H2O absorptions near 2.8, 2.95, and 3.14 μm are indicated (dotted red lines).

These spectra were obtained as part of our calibration measurements by rapidly scanning the spectrometer slit lengthwise across the Moon such that every pixel along the slit crossed the lunar disk along the same chord. This resulted in a spectral average of all areas along that chord. These calibration measurements were obtained in full-resolution mode with 1024 spectral channels [versus the nominal 512 (1)], and each was acquired multiple times. Thus, these data have the highest signal-to-noise ratio and the highest spectral resolution of all our lunar observations. Three different lunar chords were measured (Fig. 1). The first chord (purple in Fig. 1A) was obtained on 29 December 2007 and lies on the evening side of the lunar disk, roughly parallel to the equator at 6°S to 11°S. The other two chords were obtained on 2 June 2009 when the spacecraft was looking down on the northern hemisphere (blue and cyan in Fig. 1B) and cover mid-latitude regions (~20°N to 60°N), with one on the morning side and the other on the evening side.

The three chord-averaged spectra all show well-defined absorption features in the 3-μm region, with minima at ~2.81 μm and asymmetric increases in reflectance toward longer wavelengths consistent with OH and/or H2O as seen in laboratory spectra of lunar samples (in which at least some of the hydration is terrestrial in origin). The strongest absorption occurs in the mid-latitude evening chord (blue) from June 2009. The mid-latitude morning chord (cyan) is weaker, but still relatively strong, whereas the December 2007 equatorial chord (purple) has the weakest absorption. As illustrated in Figs. 1, C and D, we fit a continuum across the feature (near 2.6 and 3.6 μm) such that the depth of the 2.8-μm OH absorption and the equivalent width (or integrated band depth) from 2.6 to 3.6 μm can be measured (14). This results in 2.8-μm band depths of 12, 9, and 6% and equivalent widths of 0.06, 0.03, and 0.01 μm for the 2 June evening, morning, and December 2007 chords, respectively.

In addition to the integrated chord measurements, we collected three sets of spatially resolved imaging scans of the lunar surface at 512 wavelengths across the spectral range of the instrument. These scans were acquired by moving the spacecraft in a direction perpendicular to the slit of the spectrometer at a rate of one slit width per integration. To minimize saturation, we imaged the Moon using the 64 central pixels along the slit, which are covered by an antisaturation filter with the shortest possible integration time [0.72 s (1)]. A small area along the equator was imaged in December 2007, and the northern polar regions were scanned on both 2 June and 9 June 2009. Due to rotation of the Moon, the 90° of longitude that overlap between the June data sets are observed at two different local times of day (Fig. 2). The thermal contribution was removed from these data with the same method used for the chords (5), to produce temperature maps (Fig. 2, C and G) and, after subsequent removal of the spectral continuum over the 3-μm absorption, band depth maps (Fig. 2, D and H).

Fig. 2

Two sets of observations over the North Pole separated by one-quarter of a lunar day. Hydration is detected on both dates at all observable locations (>10°N). (A to D) Data acquired on 2 June 2009 at ~80 km/pixel. (A) Clementine basemap (15° grid) of observed area. (B) The 1.2-μm Deep Impact albedo image. (C) Corresponding temperature map (in K) derived from >4 μm spectra. (D) Corresponding map of the strength of the continuum-removed 2.8-μm hydration feature. (E to H) Data acquired on 9 June 2009 at ~60 km/pixel; images as in (A) to (D).

In the June 2009 polar data, hydration is detected in all pixels and thus at all latitudes greater than 10°N. The band depths also exhibit a clear dependence on temperature (i.e., insolation) and therefore on distance from the subsolar point (Fig. 2, D and H). Polar regions above 70°N have only minimal differences in band depth and, independent of local composition, exhibit the strongest absorptions. However, band strength is not solely dependent on latitude. All regions along the terminator (both morning and evening) were relatively cold (~280 K) and contained similar OH and H2O abundances, yet they span latitudes from 10°N to 90°N. Thus, spatial and temporal variations in volatile abundance appear to be more strongly controlled by the instantaneous, rather than cumulative, solar insolation.

The imaging scan acquired in December 2007 contains 64 by 64 pixels straddling the equator on the central nearside (~11°S to ~10°N) (Fig. 3). The warmer (sunward) third of the region lacked 3-μm absorptions and thus contained no detectable hydration. However, a weak 3-μm feature was evident toward the evening terminator (beginning at solar incidence angles of >65°; <~330 K) with a 2.8-μm band depth of ~4%. These magnitudes are consistent with the chord measurement acquired contemporaneously at similar latitudes (Fig. 1). As with the polar observations, this equatorial scan indicates that instantaneous temperature, not latitude, is the dominant factor determining OH and H2O abundance.

Fig. 3

Equatorial observations acquired on 27 December 2007 at 10 km/pixel. (A) Clementine basemap (15° grid) of observed area (yellow box). (B) The 1.2-μm Deep Impact albedo image (5° grid). Solid lines are the equator and prime meridian, which intersect in Sinus Medii. (C) Corresponding temperature map (in K) derived from >4-μm spectra. (D) The average spectrum for most of the image lacks a hydration feature (e.g., red). However, regions nearer the evening terminator have weak, but distinct, 3-μm absorptions (blue). The red spectrum is scaled to match the reflectance of the blue spectrum. See boxes in (C) for locations.

Due to the rotation of the Moon, the morning terminator contained basaltic maria on 2 June and anorthositic highlands on 9 June. On both days, spectra along the terminator exhibit strong 3-μm absorptions at all imaged latitudes, from 10°N to the North Pole, with the strongest absorptions at the pole (Fig. 2, D and H). The 2.8-μm OH band depth values range from 9 to 11% for maria and 10 to 12% for highlands. These ranges, as well as comparisons of continuum-removed spectra of mare (Fig. 4A) and highland (Fig. 4B) units at similar latitudes along the terminator, show variations of only 1 to 2% band depth. We therefore observe no statistically significant compositional differences in maximum OH or H2O content at our resolution of tens of kilometers.

Fig. 4

Comparisons of mare and highland terrains as a function of latitude and time of day as observed over a quarter of a lunar day (90° rotation) from 2 and 9 June 2009. (A) Maria ranging from 16°N to 54°N along the morning terminator on 2 June have the same continuum-removed band depth and overall shape. Locations as indicated on (E). (B) Same as (A) for highland units ranging from 15°N to 80°N on 9 June. (C) The same location in eastern Mare Imbrium (“M”) that is observed in the morning (orange) on 2 June rotates to noon (black; red circle) on 9 June with a 70% decrease in continuum-removed band depth and a distinct change in shape. (D) The highland unit south of Mare Humboldtianum (“H”) was observed at noon (black) on 2 June and in the evening (gray) on 9 June with only a 50% change in band depth. By evening, both mare (brown) and highland (gray) terrains return to their morning abundances (orange and blue, respectively). (E) Locations for terrains whose spectra are plotted on (A) to (D), on 2 June (top) and 9 June (bottom). The mare and highland units marked with “M” and “H” are visible in both data sets. All spectra are averages of 3 by 3 pixel areas.

The two June 2009 imaging scans also allow us to compare the same locations on the Moon at different local times and thus under different insolation conditions. For example, eastern Mare Imbrium is at the morning terminator (relatively cold) on 2 June and near local noon on 9 June (Fig. 4E). Although the same area cannot be tracked from morning through evening, comparison of similar terrain types (highlands or maria) (Fig. 4, C to E) shows that all surfaces begin the day with comparable band depths (~11%), lose water as they move off the morning terminator, have a minimum absorption near noon (Fig. 4, C and D, black), and return to their morning values by evening. However, the highlands consistently retain more water (~50% of the morning band depth value) at their minimum near noon than do the maria (~30% of the morning band depth value). These differences in rates of hydration loss between maria and highlands presumably reflect differences in their mineralogy or grain size. Nonetheless, their common steady-state band depth indicates that there is no inherent difference in total OH or H2O adsorptivity with composition. As seen in the continuum-removed spectra (Figs. 1 and 4), the overall shape of the hydration feature also changed with decreasing strength; absorptions beyond 3 μm decreased more than the 2.8-μm OH feature. This nonlinear change with wavelength argues against geometric or photometric effects. Instead, it may represent wide variability in OH bond energies due to different cations. More likely, the change in the shape indicates that the absorptions beyond 3 μm are attributable to H2O, which is lost more readily than the relatively strongly bonded OH.

The strongest 3-μm absorptions in the Deep Impact data occurred at the North Pole, where the 2.8-μm OH band depth was 14% and the 3-μm equivalent width was 0.07 μm. Such high band depth values are surprising for an object as dark as the Moon because previous studies have shown that dark absorbing materials can cause a nonlinear decrease in the strength of hydration features (1517). Using radiative transfer theory to account for these multiple scattering effects (18) and trends derived from laboratory data of synthetic OH- and H2O-bearing basaltic glasses (9), we estimate the water content at ~0.3 weight percent (wt %) (5). Similar values are obtained with trends derived from laboratory data of other hydrated phases [e.g., clay minerals, zeolites (9)], suggesting that the maximum volatile content of the bulk lunar soil is <0.5 wt %, regardless of the actual composition of the hydrated phase(s).

The Deep Impact observations of the Moon not only unequivocally confirm the presence of OH or H2O on the lunar surface, but also reveal that the entire lunar surface is hydrated during at least some portions of the lunar day. Furthermore, the temporal variations observed over the lunar diurnal cycle reveal a dynamic hydration process driven by solar radiation. Our results suggest that it is likely that all areas of the Moon, regardless of composition or location, exhibit similar maximum hydration during the night. The short time scale of loss and recovery of OH or H2O, a cycle that completes entirely within daylight hours, and the relatively low absolute abundances suggest that lunar hydration is a surface phenomenon. The rapid photodissociation of H2O and the return to steady-state abundances during the day require a ready source, which would be consistent with a solar wind origin, as supported by the recent discovery of an extended source of water-group ions in the inner heliosphere (19). Over time, this daily hydration and dehydration process may lead to a migration of OH and H+ toward the poles and their accumulation in permanently shadowed regions of the Moon (20). Such hydration via solar wind is expected to occur throughout the inner solar system on all airless bodies with oxygen-bearing minerals on their surfaces.

Supporting Online Material

Materials and Methods

Table S1


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. The position of this band depends in part on the cation attached to the hydroxyl. Although we cannot yet definitively identify the specific cation, the chemical composition of the Moon suggests it is likely Si, Al, Fe, Mg, Ca, or Na.
  3. The depth of the 2.8-μm band is calculated relative to the continuum. Equivalent width is the width in wavelength (μm) of a 100% absorption of the same total intensity as the overall absorption feature. It is calculated as the integrated area of the absorption feature below the continuum.
  4. EPOXI, the Deep Impact extended mission, is supported by the NASA Discovery Program under contract NNM07AA99C to the University of Maryland. The Deep Impact spacecraft and HRI-IR instrument were built by Ball Aerospace. We gratefully acknowledge the efforts and support of the EPOXI project team at the Jet Propulsion Laboratory in acquiring the data presented here and the continuing work of members of the EPOXI science team in improving the instrument calibration. Laboratory spectra of lunar soils were acquired with the NASA/Keck RELAB, a multi-user facility supported by NASA grant NNG06GJ31G. We also thank R. Clark for discussions on the spectral interpretations of the 3-μm feature and L. McFadden for comments that improved this manuscript.
View Abstract

Navigate This Article