The aim of this study is to bring new elements concerning the astronomical theory of paleoclimates using two tools : glaciological dating of polar ice cores using an inverse method and conceptual models of paleoclimate. We show how an inverse method applied to the glaciological dating model enables us to obtain an optimal chronology and to analyse its uncertainties. It is not possible to obtain a chronology for Vostok that satisfies all chronological constraints, probably because of our poor knowledge of the conditions along the ice flow line, upstream of Vostok. The chronologies for Dome C and Dome F are coherent with the constraints, meaning that the simple glaciological model used in the case of a dome is adequate. The Dome F case allows us to show that the Devils Hole record is probably biased by local conditions, and suggest that the phase relationship between insolation and Antarctic temperature is roughly constant. We also confirm that thinning is more important in the upper part of the ice sheet for a dome than for the rest of the flow line. Moreover, we suggest that estimates of the accumulation during glacial periods based on saturation vapour pressure and spatial isotope−surface temperature relationships are overestimated by ~30%. The conceptual model we developed yields a description of the sea level variations during the Quaternary as a response to the insolation forcing. We show that sea level variations can be explained with a conceptual model forced by insolation changes. In particular, we justify why the most important sea level transitions occurred when insolation changes were the weakest. Moreover, we show that small insolation−climate phase variations during deglaciations are compatibles with the astronomical theory of paleoclimate.
