Rapid dynamic changes of ocean-terminating outlet glaciers of the Greenland ice sheet are related to atmospheric and oceanic warming but the detailed link to external forcing is not well understood. Observations show high variability in dynamic changes and are mainly limited to the past three decades with dense satellite observations. Here we compile a long-term record of dynamic changes of Eqip Sermia Glacier, West Greenland. Starting in 1912, we combine historical measurements of terminus positions, ice-surface elevation and flow velocity together with more recent in-situ and remote-sensing observations. In the 20th century, the glacier underwent small variations in terminus position and flow speed. Between 2000 and 2003, the terminus retreated substantially, but stabilized thereafter. In 2009, the northern terminus lobe started to retreat very rapidly; the southern lobe collapsed in 2013. The present terminus position, which has retreated by 4 km since 1920, is unprecedented in the historical record. Flow velocities were relatively stable until 2010. The recent acceleration reached threefold velocities in 2014 and rapidly affected the whole terminus region up to 15 km inland. Comparison with forcings from the atmosphere and the ocean over the past century shows that no dominant cause can be identified, and that local effects of bed geometry modulate the glacier response.

Numerous ice-sheet reconstructions of the last glacial cycle have been proposed, however due to limited geological evidence, reconstructing older Northern Hemisphere ice sheets remains a difficult exercise. Here we focus on the penultimate glacial maximum (PGM; ~140 ka BP) over the Northern Hemisphere. While some evidence of the PGM Eurasian ice-sheet extent were found, this is not the case for the corresponding Laurentide ice sheet. To improve the glaciological reconstructions of the PGM Northern Hemisphere ice sheets, we explore the parameter space of ice-sheet model uncertainties and carry out numerous univariate ice-sheet steady-state sensitivity simulations. We use two PGM climate simulations to force the ice-sheet model, differing in the prescribed Laurentide ice topography (small and large). The simulated Northern Hemisphere ice volume ranges from 124.7 to 152 m SLE when using the climate accounting for a small Laurentide ice sheet, which is compatible with global sea-level reconstructions of this period (−92 to −150 m). Conversely, using the climate simulation with a Laurentide ice sheet comparable in size to that of the last glacial maximum results in too large ice volumes. Changes in basal drag provide the upper bound ice volume of our experiments, whereas changes in the distribution of ice streams provide the lower bound.