The Late Pliocene, 3.65–2.75 million years ago, is the most recent period in Earth's history that was warmer than the present. This makes it interesting for climatological research, because it provides a possible analogue for the near future. We used a coupled ice-sheet–climate model to simulate the behaviour of these systems during this period. We show that the warmest moment saw a sea-level rise of 8–14 m, with a CO₂ concentration of 320–400 ppmv.

An ice core from Taylor Glacier, Antarctica, spans a period ~ 70 000 years ago when Earth entered the last ice age. Chemical analyses of the ice and air bubbles allow for an independent determination of the ages of the ice and gas bubbles. The difference between the age of the ice and the bubbles at any given depth, called ∆age, is unusually high in the Taylor Glacier core compared to the Taylor Dome ice core situated to the south. This implies a dramatic accumulation gradient between the sites.

To reconstruct past winter climate, we studied ice wedges at two sites in the Yana Highlands, interior Yakutia (Russia), the most continental region of the Northern Hemisphere. Our ice wedges of the upper ice complex unit of the Batagay megaslump and a river terrace show much more depleted stable-isotope compositions than other study sites in coastal and central Yakutia, reflecting lower winter temperatures and a higher continentality of the study region during Marine Isotope Stages 3 and 1.

The 18th and early 19th centuries saw pronounced climatic variations with impacts on the environment and society. Although instrumental meteorological data for that period exist, only a small fraction has been the subject of research. This study provides an overview of early instrumental meteorological records in Switzerland resulting from an archive survey and demonstrates the great potential of such data. It is accompanied by the online publication of the imaged data series and metadata.

The mid-Holocene has been an excellent target for comparing models and data. This work shows that, over China, all the ocean–atmosphere general circulation models involved in PMIP3 show a very large discrepancy with pollen data reconstruction when comparing annual and seasonal temperature. It demonstrates that to reconcile models and data and to capture the signature of seasonal thermal response, it is necessary to integrate non-linear processes, particularly those related to vegetation changes.