Ballistic transport and solar wind hydroxylation as the mechanisms for diurnal variability of hydration signatures on the Moon

The presence of OH/H₂O on the lunar surface has been known since 2009 through observations of the 3-micrometer absorption feature diagnostic of OH and/or H₂O [1, 2, 3].  This hydration signature has been observed to vary across a lunar day, with similar signatures in the morning and evening, and the lowest abundance observed at local noon [4, 5]. Evidence for diurnal variability suggests that a component of the hydration strength is mobile, migrating in response to temperature variations of the surface. It has been proposed that localized, temporary shadows can provide refuge for these adsorbed particles into the daytime, due to surface roughness and topography [6]. The dependence of variability in band strength on illumination has been poorly explored. 

Previous modeling work [7] predict a much higher H₂O abundance (3 ×10⁵ molecules/cm³) than the ~0.6 molecules/cm³ measured by the neutral mass spectrometer (NMS) aboard the Lunar Atmosphere and Dust Environment Explorer (LADEE) as presented in [8]. However, Laferriere et al. (2025) [9] found that in accounting for the desorption from a rough surface, the fraction of the 3 μm absorption that is variable on diurnal timescale, the number of particles observable (reaching >20 km over the equatorial region on the dayside), and the fraction of molecules that are H₂O separate from OH, modeled exospheric abundances consistent with surface hydration are within an order of magnitude of the LADEE exospheric observations. However, consistent with that found previously with other similar models [10], [9] were not able to replicate the observed symmetry in diurnal variability. This is expected, as these modeling approaches do not account for an ongoing source mechanism at the lunar surface due to the solar wind. Combining our results of loss rates with the solar wind proton flux [11] and the predicted H ion to OH conversion rate [12] yields an OH production rate of 28–93 molecules/cm²/s that would be needed to reach equilibrium.

Thus, it is necessary to better constrain sourcing mechanism(s) to understand the diurnal behavior of hydration on the lunar surface. Previous observations of the 3 µm hydration feature are complicated by the lack of successful instrumentation designed specifically to study this feature. While the Moon Mineralogy Mapper (M³) aboard Chandrayaan-1 provides high spatial resolution, the spectrometer’s range ends at 3 µm which limits analysis of the OH/H₂O feature in detail and also makes thermal emission removal difficult. This has led to disagreements over interpretations of 3 μm band strength, ranging from no variability [13] to variability with temperature, latitude, and time of day [14, 15, 16]. The spectrometers used in [2, 3, 5] were limited in spatial scale (>30 km/pixel) when they observed during lunar flybys.

Chandrayaan-2’s Imaging Infrared Spectrometer (IIRS) observations present a new opportunity to study the variability of lunar hydration due to its spectral range from 0.9–5.3 μm and a spatial resolution ~80 m. Here, we explore the IIRS data to constrain the diurnal variability of hydration. As the level-1 (calibrated) data provided by ISRO has a correction applied which minimizes the 3 µm hydration feature we do not use this dataset for our analysis (Figure 1). This instrument is still relatively new, and does not have extensive calibration pipelines published compared to those produced for M³ [17], so we first produced our own calibration of the level-0 data (raw minus dark data).

We first apply a linearization (provided by ISRO) to account for the non-uniformity of the detector. Then, we convert the data from digital number to radiance. We exclude any pixels within our bad pixel map, which was determined by searching through the spectra by hand and includes the regions of the order sorting filters (Figure 2). We do not attempt to apply a radiance adjustment for the OSF regions or the edges of the spectrometer and instead choose to exclude these regions from our analysis. We use a resistant mean to smooth the spectral data. We apply a modified version of the methodology from [5] to remove thermal emission beyond ~3 µm, allowing us to determine the temperature per pixel. The two observations that we focus on here are of the same surface of the Moon at two local times of day (morning and evening, Figure 3), taken ~1.5 months (~1.5 lunar days) apart.

We also measured the 2 µm pyroxene absorption band to ensure that the abundance does not appear to evolve between the two observations. This would suggest that the optical geometry strongly affects the observations and thus the 3 µm variability would not be trustworthy. We find that the 3 µm feature is ubiquitous, with lower abundances in the morning (~2 hours after the morning terminator) than the evening (1 hour prior to evening terminator), as found previously.

 

References

[1] Clark (2009), Science, 326, 5952. [2] Pieters et al., (2009), Science, 326, 5952. [3] Sunshine et al., (2009), Science, 326, 5952.[4] Hendrix et al., (2019), GRL, 46, 5. [5] Laferriere et al., (2022), JGR: Planets, 127, 8. [6] Davidsson & Hosseini, (2021), MNRAS, 506, 3. [7] Smolka et al., (2023), Icarus, 397. [8] Benna et al., (2019), Nature Geoscience, 12, 5. [9] Laferriere et al., (2025), JGR: Planets, 130, 4. [10] Schorghofer (2014), GRL, 41, 14. [11] Zeller & Ronco., (1967), Icarus, 7, 1-3. [12] McLain et al., (2021), JGR: Planets, 126, 5. [13] Bandfield et al., (2018), Nature Geoscience, 11, 3. [14] Li & Milliken, (2017), Science Advances, 3, 9. [15] Grumpe et al., (2019), Icarus, 321. [16] Wohler et al., (2017), Science Advances, 3, 9. [17] Green et al. (2011), JGR: Planets, 116, E10.

 

This work is funded by a NASA LDAP under grant number 80NSSC23K1336. We acknowledge the use of data from the Chandrayaan-II, second lunar mission of the Indian Space Research Organization (ISRO), archived at the Indian Space Science Data Centre (ISSDC).

Figure 1: Radiance for Level-1 and our recalibrated data revealing difference around 3 µm feature.

Figure 2: Bad pixel map for the spectrometer.

 

Figure 3: Two observations covering the same surface, observed under different lighting conditions.