Global remote-sensing data for Mars are analyzed to obtain a simple, self-consistent model for the surface year. The data sets discussed include radar cross-section measurements at several wavelengths, radio whole-disk thermal emission observations at two wavelengths, the global distribution of thermal inertia, deviations of diurnal temperatures from those of a homogeneous model, and thermal spectral estimates of surface rock abundance and of the thermal inertia of the nonrocky component of the surface. The data sets which most constrain the interpretation are the rock abundance map and the correlation of thermal inertia with radar cross section; these require the rock abundance to not vary significantly from place to place and simultaneously require the density and thermal inertia of the fines to vary in a consistent manner. The simplest model which can explain all of the data involves a global case-hardened crust. (''duricrust'') which varies spatially in its degree of formation. In general, low-thermal-inertia regions have a poorly-developed crust and high-inertia regions have a well-developed crust there are, however, regions that consist of coarse particles and which do not fit this model (e.g., Chryse). This model is consistent with the ages of low-inertia regions and with aeolian mechanisms for their development. The duricrust is thought to form via the mobilization of salt ions within a layer of water adsorbed within the regolith, and its formation may be associated with the exchange of water between the regolith and atmosphere which occurs on the 105- and 106-year time scale.