[1] The Martian cryosphere is defined as that region of the crust where the temperature remains continuously below the freezing point of water. Previous estimates of its present thickness have ranged from ∼2.3-4.7 km at the equator to ∼6.5-12.5 km at the poles. Here we revisit these calculations, review some of the assumptions on which they were based, and investigate the effects of several parameters, not previously considered, on the cryosphere's thermal evolution and extent. These include astronomically driven climate change, the temperature-dependent thermal properties of an ice-rich crust, the potential presence of gas hydrate and perchlorate-saturated groundwater, and consideration of recent lower estimates of present-day global heat flow (which suggest a mean value roughly half that previously thought, effectively doubling the potential thickness of frozen ground). The implications of these findings for the continued survival of subpermafrost groundwater and its potential detection by the MARSIS radar sounder onboard Mars Express are then discussed. Although our estimates of the maximum potential thickness of the cryosphere have significantly increased, consideration of the likely range and spatial variability of crustal heat flow and thermal properties, in combination with the potential presence of potent freezing point depressing salts, may result in substantial local variations in cryosphere thickness. The locations that appear best suited for the detection of groundwater are those that combine low latitude (minimizing the thickness of frozen ground) and low elevation (minimizing the depth to a water table in hydrostatic equilibrium). Preliminary results from a MARSIS investigation of one such area are discussed.

[1] Radar detection of subsurface ice on Mars has been widely debated in part because the dielectric signature of ice, as deduced from the dielectric constant, can be confused with dry-silicate-rich materials. To identify the ice dielectric signature, it is crucial to estimate the imaginary part of the dielectric permittivity inferred from the dielectric attenuation after removing the scattering loss. Unfortunately, the latter remains poorly quantified at both Mars Advanced Radar for Subsurface and Ionospheric Sounding (MARSIS) and shallow subsurface radar SHARAD frequencies. To address this ambiguity, we conducted multiple-frequency ground-penetrating radar and resistivity investigations in well-characterized temperate permafrost in Fairbanks, Alaska. The area shows several geomorphologic similarities to midlatitude and high-latitude terrains on Mars. This approach allowed us to quantify the dielectric and scattering losses in temperate permafrost over the 10 to 1000 MHz frequency band. At 20 MHz, our results suggest an average dielectric loss rate of 0.25 ± 0.03 dB/m, whereas the corresponding average scattering loss rate is 0.94 ± 0.37 dB/m. The scattering loss was found to represent ∼69% of the total signal attenuation. Considering this result and the study by Heggy et al. (2006a) in volcanic environments, we revised the interpretation of the attenuation coefficient calculated from SHARAD data over the Deuteronilus Mensae region and Amazonis Planitia; we then used the reevaluated dielectric loss to estimate the imaginary part of the dielectric permittivity. Our results suggest that even if Deuteronilus Mensae deposits and the Vastitas Borealis Formation may have similar dielectric constants, their imaginary parts are different. This implies that the two regions have different bulk compositions, with the former being ice-rich sediments and the latter being nonconsolidated volcanic deposits.