The southern cratered highlands of Mars contain a large population of flat-floored, rimless craters which have previously been interpreted to have formed by aeolian mantling or flood volcanism. Neither of these geologic processes accurately explains the observed morphology or the crater statistics. Geologic mapping in the Amenthes and Tyrrhena region indicates that craters with this morphology occur on undulating intercrater materials near the dichotomy boundary and on more rugged materials farther into the highlands containing numerous ancient valley networks. Cumulative size-frequency distribution curves indicate ages of N(5)=790 (early to middle Noachian) and N(5)=540 (middle Noachian) for the cratered plateau and cratered highland materials, respectively, opposite the observed stratigraphic relations. For crater diameters >16 km the population of impact craters is consistent with stratigraphy, but the population of smaller craters in the region indicates that importance of resurfacing. Superposed, fresh craters indicate a resurfacing event that ceased at N(5)=200--250 (late Noachian to early Hesperian). Crater counts divided into 5¿ latitudinal bins show an increase in the number of craters between the 8- and 50-km-diameter range with increasing northerly latitudes, suggesting that the resurfacing was not a single event.

Statistical modelling of an erosive event capable of removing the continuous ejecta deposits from the craters, eroding them to reduce the apparent diameter, and simultaneously burying smaller eroded craters explains both the morphology of the flat-floored, rimless craters, and their population distribution. Matching the slope of the modelled curves with that of the cratered highland materials suggests that up to 1400 m of erosion may have taken place in the highlands. Some of this eroded material probably remained locally, burying the smaller craters. Later stripping of this material indicated by exposed layering and inverted topography has exhumed the smaller craters in the cratered plateau material resulting in the apparent age discrepancy determined from crater counts. Up to 1.0 km of material may have been removed by this later erosive event to explain the low density of ancient valley networks in the cratered plateau materials.