Global constraints on precipitation in Mars' ancient past

The paradigm of a warm, wet early Mars has been invoked periodically since the first observation of ancient flow channels and other potentially fluvial features on the surface of Mars in the 1970's. The mechanism of rainfall and surface runoff is appealing for the explanation of many such features on the Martian surface, but it remains problematical from a climate dynamical perspective for several reasons. The problem most frequently addressed is that of warmth. Given a faint young early Sun, and the currently observed volatile inventories, it has been challenging to conceive of an atmospheric composition that would have yielded sufficient greenhouse warming to lift temperatures above the freezing point for water. Much thicker carbon dioxide atmospheres, the role of infrared scattering carbon dioxide clouds, and sulphur compounds of volcanic origin have all been proposed with some degree of plausibility. However, while warmth has been the primary concern, it is only half of the problem. Here we use a modified version of the NCAR Community Atmosphere Model (CAM) to investigate climate mechanisms potentially relevant to ancient Mars. The model is not fully adapted to Mars conditions; instead, the topography of Mars is imposed at the model lower boundary, and oceans of various elevations are initialized. The goal is to examine how "wet" an ancient Mars could have been under the model ideal of ancient clement conditions. Using CAM in this "Mearth" configuration (Mars as Earth), we show that considerations of the geographical distribution of available surface water can dramatically change the aridity of a planet, even when the mean and variation of temperatures are Earth-like. These results suggest that the thermal state of the atmosphere and the abundance and geographical distribution of water need both to be considered when contemplating early Martian climatic states, and that they place equally important, orthogonal constraints on planetary aridity. Specifically, the distribution of rainfall predicted by such simulations is in rough agreement with the observed distribution of ancient valley networks, paleo-lake basins, and deltas.