Understanding the evolution of volatiles on Mars requires understanding the processes which are currently acting to cause exchange between the various reservoirs on annual and longer timescales. On the seasonal timescale, exchange of water can occur between the atmosphere and reservoirs of ice in the polar caps and of adsorbed water in the near-surface regolith covering the remainder of the planet. This exchange is driven by the seasonally-varying insolation and its consequent effects on the surface and subsurface temperatures and on the advance and retreat of the predominantly-CO2 polar caps. On a longer timescale, exchange can occur between these same reservoirs, and is driven by the changing annual insolation patterns which result from the 105-year timescale variations in Mars' orbital elements (predominantly the orbital obliquity). Observations of the seasonal water cycle and its variations from year to year from the Viking spacecraft and from Earth provide clues as to the importance of the various reservoirs and provide boundary conditions against which models of the various processes can be compared.
The water vapor content of the Mars atmosphere was measured from the Viking Orbiter Mars Atmospheric Water Detectors (MAWD) for a period of more than one Martian year, from June, 1976, through April. 1979, and the results are presented here. The data reduction incorporates spatial and seasonal variations in surface pressure, and supplements earlier published versions of less-complete data. Column abundances vary between zero and about 100 precipitable microns (pr µm), depending on location and season, while the entire global abundance varies seasonally between an equivalent of about 1 and 2 km3 of ice. The first appearance of vapor at non-polar latitudes as northern summer approaches, and the drop in abundance at mid-latitudes as summer ends, both strongly imply the existence of a seasonal reservoir for water within the regolith. There appear to be no net annual sources away from the poles that contribute significant amounts of water. However, the strong annual gradient of vapor from north to south implies a net annual flow of vapor toward the south; this southward flow may be balanced in part by a northward flow during the global dust storms, by transport in the form of clouds or adsorbed onto dust grains, or during other years. The perennially-cold nature of the south-polar residual cap, along with the relatively large summertime vapor abundances over the cap, implies a net annual condensation of vapor onto the cap. Estimates are made of the southward transport, and are consistent with the movement of ice being important in the formation and evolution of the polar layered terrain, and with the formation of the individual layers at the rate of one per obliquity cycle (105 years).
The global distribution of the annual average abundance of vapor is found to correlate well with Martian topography, as might be expected for a uniform constant atmospheric mixing ratio. If this topographic effect is divided out, the resulting residual map correlates with maps of surface albedo and thermal inertia; this correlation may be related to the control exerted by the surface and subsurface temperatures on the adsorption/desorption process and on the atmospheric temperature profile and, hence, the vapor holding capacity of the atmosphere.
The vertical distribution of vapor within the atmosphere is inferred through comparison of the observed water vapor abundances with measurements of atmospheric temperatures. In order to not saturate, the vapor must be confined to the lowermost 1 to 3 scale heights (~ 10-30 km), with this height varying with location and season. Near-surface water vapor can condense out overnight and form a morning fog; estimates of the optical thickness of the resulting fog are made, and they agree well with observations of diurnal variations of opacity due to fog formation.
Previous Earth-based near-infrared observations are re-interpreted here; they show that water ice condenses out onto the seasonal polar caps, but not during midday near the equator. Earth-based observations of the vapor column abundance are compared with the Viking MAWD results, and indicate that the seasonal cycle shows a remarkable repeatability, except during 1969 when large vapor abundances were present during southern summer. This difference is explained by postulating that all of the CO2 had sublimed off of the south residual cap that year, exposing the underlying water cap which would subsequently sublime and produce large amounts of atmospheric vapor; the rate and amount of CO2 sublimation may depend on the degree of dust storm activity each year and hence on the different thermal loads placed on the cap.
The possible processes for producing seasonal changes in the atmospheric vapor abundances have been modeled in order to infer the relative importance of each process in the seasonal cycle. The equilibrium between water vapor and water adsorbed onto the regolith grains is sufficiently temperature-dependent that seasonal surface temperature variations are capable of driving a large exchange of water between the atmosphere and subsurface. For the likely range of regolith properties, this exchange is found to be from 10-150% of the observed seasonal change in atmospheric abundance; the differences between this exchange and the observed behavior result from transport of vapor due to the atmospheric circulation. Due to the latitudinal gradient of atmospheric vapor, there will also be a gradient of adsorbed water, with the south regolith containing much less water than that in the north; this gradient in the regolith will result independent of the vapor diffusivity in the regolith, as the near-surface water will be able to equilibrate on some timescale.
Models have been constructed which include regolith exchange, polar cap formation, and atmospheric transport. Comparison of the model results with the vapor observations and with other data regarding the physical nature of the surface allows constraints to be placed on the relative importance of each process. The models are capable of satisfactorily explaining the gross features of the observed behavior using plausible values for the regolith and atmosphere mixing terms. In the region between the polar caps, the regolith contributes as much water to the seasonal cycle of vapor as does transport in from the more-poleward regions, to within a factor of two. Globally, 10-50% of the seasonal cycle of vapor results from exchange of water with the regolith, about 40% results from the behavior of the residual caps, and the remainder is due to exchange of water with the seasonal caps. It is difficult to determine the relative importance of the processes more precisely than this because both regolith and polar cap exchange of water act in the same direction, producing the largest vapor abundance during the local summer. The system is ultimately regulated on the seasonal timescale by the polar caps, as the time to reach equilibrium between the atmosphere and regolith or between the polar atmosphere and the global atmosphere is much longer than the time for the polar caps to equilibrate with the local atmosphere. This same behavior will bold for longer timescales, with the polar caps being in equilibrium with the insolation as it changes on the obliquity timescale, and the atmosphere and regolith following along.
