A time-dependent theory for the evolution of water on Mars is presented. Using this theory and invoking a large number of observational constraints, I argue that these constraints require that a large reservoir of water exists in the Martian crust at depths shallow enough to interact strongly with the atmosphere. The constraints include the abundance of atmospheric water vapor, escape fluxes of hydrogen and deuterium, D͞H ratios in the atmosphere and in hydrous minerals found in one Martian meteorite, alteration of minerals in other meteorites, and fluvial features on the Martian surface. These results are consonant with visual evidence for recent groundwater seepage obtained by the Mars Global Surveyor satellite.
Given the low temperature and pressure at the surface of Mars, the existence of liquid water at shallow depths was considered once to be impossible at the present epoch. Nevertheless, the Mars Orbiter camera on the Mars Global Surveyor spacecraft apparently has produced the first visual evidence of recent transport of a fluid, presumably water, from the interior to the surface of Mars (1). These observations actually are the culmination of a series of discoveries strongly suggesting the presence in the Martian crust of liquid water that interacts efficiently with the atmosphere. This evidence involves the presence of modified minerals in Martian meteorite samples that require the intervention of liquid water (2-4) and, in particular, measurements of the D͞H ratios in hydrous minerals created hundreds of millions of years ago (5) that can be virtually identical with the atmospheric D͞H ratio at the present time (6). † It has been argued that this surprising result requires efficient exchange of water between the atmosphere and an underground reservoir containing much more water than the atmosphere, such that the amount of hydrogen and deuterium lost by escape from space over the past several hundred millions of years is small compared with that in the underground (or polar-cap) reservoirs (7,8). On the other hand, to account for the amount of isotopic fractionation in atmospheric water vapor, very large escape fluxes at very early times and a water supply hundreds of meters deep, sufficient to create fluvial features, were required. In this paper, a time-dependent treatment of hydrogen and deuterium evolution, developed for a study of the evolution of water on Venus (9), will be applied to the Martian problem. A model that explains the evolution of Martian water adequately must be able to account for the near equality of D͞H ratio in contemporaneous water vapor and water of hydration in minerals formed many hundreds of million years ago, the presence of crustal liquid water capable of modifying other crystals in Martian meteorite samples, today's atmospheric D͞H ratio, and the presence of enough water at early times to account for surface fluvial features. It is desirable also that the model be insensitive to the relative efficiency of D and H escape and the D͞H ratio in primitive Martian water. I shall...