Continuous monitoring of the polar regions by satellites has shown that sea ice extent (SIE) in the Antarctic has increased slightly since 1979. By contrast, climate model simulations including all major anthropogenic and natural climate influences simulate an average decrease in SIE since 1979. Here we take a longer view and assess the consistency of observed and simulated changes in Antarctic SIE using recently recovered satellite‐based estimates of Antarctic SIE for September 1964 and May–July 1966, hence extending the current observational record from 35 to 50 years. While there is evidence of inconsistency between observed trends in Antarctic SIE and those simulated since 1979, particularly in models with realistic interannual variability, the observed trends since the mid‐1960s fall within the 5–95% range of simulated trends. Thus, our results broadly support the hypothesis that the recent increase in Antarctic SIE is due to internal variability, though the reasons for the inconsistency in simulated and observed changes since 1979 remain to be determined.

Updated observational data sets without climatological infilling show that there was an increase in sea ice concentration in the eastern Arctic between 1950 and 1975, contrary to earlier climatology infilled observational data sets that show weak interannual variations during that time period. We here present climate model simulations showing that this observed sea ice concentration increase was primarily a consequence of cooling induced by increasing anthropogenic aerosols and natural forcing. Indeed, sulphur dioxide emissions, which lead to the formation of sulphate aerosols, peaked around 1980 causing a sharp increase in the burden of sulphate between the 1950s and 1970s; but since 1980, the burden has dropped. Our climate model simulations show that the cooling contribution of aerosols offset the warming effect of increasing greenhouse gases over the midtwentieth century resulting in the expansion of the Arctic sea ice cover. These results challenge the perception that Arctic sea ice extent was unperturbed by human influence until the 1970s, suggesting instead that it exhibited earlier forced multidecadal variations, with implications for our understanding of impacts and adaptation in human and natural Arctic systems.

[1] We report observations of NO nightglow with the Spectroscopy for the Investigation of the Characteristics of the Atmosphere of Mars (SPICAM) experiment on board the Mars Express (MEx) spacecraft. NO molecules emit an ultraviolet photon when N and O atoms (produced at high altitude in the thermosphere) recombine. Therefore, this emission is a tracer of the atmospheric dynamics in the lower thermosphere where O and N atoms are produced, and below, in the altitude region 50-100 km where the emission is detected. A new retrieval method has been developed to analyze the measurements from this instrument in the stellar occultation mode without slit and retrieve the absolute brightness of the emission. We present the results from the processing of more than 2000 orbits, providing the first global latitude-season distribution of the emission, established over three Martian years. The results are globally consistent with previously available measurements of dedicated limb nightglow obtained during the first Martian year of MEx (MY27). We compared the ensemble of both data sets with the predictions of the Laboratoire de Météorologie Dynamique Mars General Circulation Model (LMD-MGCM), with the addition of the full chemistry of N atoms. We find an overall agreement between the observed and modeled airglow intensities, but discrepancies are also found. The frequency and magnitude of the NO airglow observations show important asymmetries between the Northern and the Southern Hemispheres. There is no detection of emission near the poles during equinox conditions, while the model predicts that it should be most intense because of a circulation with two descending branches at the poles.

Using a large initial condition ensemble of climate model simulations, we examine the impact of volcanic activity on Arctic sea ice cover from 1960 to 2005, a period that includes three very large tropical eruptions. Ensemble averaging across simulations with natural (volcanic and solar) forcings alone reduces noise due to internal variability to show a decade of increased Arctic sea extent (of up to half a million square kilometers) following each of the Mount Agung (1963), Mount El Chichón (1982), and Mount Pinatubo (1991) eruptions. A similar impact is seen when averaging over a large ensemble of simulations with natural and all‐known anthropogenic forcings. We show that the volcanic response in sea ice cover is sensitive to preeruption temperature, with warmer conditions before an eruption being associated with a larger than average response. Finally, a detection and attribution analysis using second‐generation Canadian Earth System Model (CanESM2) did not identify a significant response in the observations, while finding no evidence of inconsistency between observations and CanESM2 since regression coefficients were consistent with unity. A similar detection and attribution analysis using the somewhat stronger volcanic response from the simulations in the average of the CMIP5 models did identify a detectable natural forcing response in four observational sea ice extent data sets.