Abstract. Ice sheets provide exceptional archives of past changes in polar climate, regional environment and global atmospheric composition. The oldest dated deep ice core drilled in Antarctica has been retrieved at EPICA Dome C (EDC), reaching ∼ 800 000 years. Obtaining an older paleoclimatic record from Antarctica is one of the greatest challenges of the ice core community. Here, we use internal isochrones, identified from airborne radar coupled to ice-flow modelling to estimate the age of basal ice along transects in the Dome C area. Three glaciological properties are inferred from isochrones: surface accumulation rate, geothermal flux and the exponent of the Lliboutry velocity profile. We find that old ice (> 1.5 Myr, 1.5 million years) likely exists in two regions: one ∼ 40 km south-west of Dome C along the ice divide to Vostok, close to a secondary dome that we name "Little Dome C" (LDC), and a second region named "North Patch" (NP) located 10-30 km north-east of Dome C, in a region where the geothermal flux is apparently relatively low. Our work demonstrates the value of combining radar observations with ice flow modelling to accurately represent the true nature of ice flow, and understand the formation of ice-sheet architecture, in the centre of large ice sheets.

Abstract. Basal melt rate is the most important physical quantity to be evaluated when looking for an old-ice drilling site, and it depends to a great extent on the geothermal flux (GF), which is poorly known under the East Antarctic ice sheet. Given that wet bedrock has higher reflectivity than dry bedrock, the wetness of the ice-bed interface can be assessed using radar echoes from the bedrock. But, since basal conditions depend on heat transfer forced by climate but lagged by the thick ice, the basal ice may currently be frozen whereas in the past it was generally melting. For that reason, the risk of bias between present and past conditions has to be evaluated. The objective of this study is to assess which locations in the Dome C area could have been protected from basal melting at any time in the past, which requires evaluating GF. We used an inverse approach to retrieve GF from radarinferred distribution of wet and dry beds. A 1-D heat model is run over the last 800 ka to constrain the value of GF by assessing a critical ice thickness, i.e. the minimum ice thickness that would allow the present local distribution of basal melting. A regional map of the GF was then inferred over a 80 km × 130 km area, with a N-S gradient and with values ranging from 48 to 60 mW m −2 . The forward model was then emulated by a polynomial function to compute a timeaveraged value of the spatially variable basal melt rate over the region. Three main subregions appear to be free of basal melting, two because of a thin overlying ice and one, north of Dome C, because of a low GF.