Simulating abundances of stable water isotopologues, that is, molecules differing in their isotopic composition, within climate models allows for comparisons with proxy data and, thus, for testing hypotheses about past climate and validating climate models under varying climatic conditions. However, many models are run without explicitly simulating water isotopologues. We investigate the possibility of replacing the explicit physics-based simulation of oxygen isotopic composition in precipitation using machine learning methods. These methods estimate isotopic composition at each time step for given fields of surface temperature and precipitation amount. We implement convolutional neural networks (CNNs) based on the successful UNet architecture and test whether a spherical network architecture outperforms the naive approach of treating Earth’s latitude-longitude grid as a flat image. Conducting a case study on a last millennium run with the iHadCM3 climate model, we find that roughly 40% of the temporal variance in the isotopic composition is explained by the emulations on interannual and monthly timescale, with spatially varying emulation quality. The tested CNNs outperform simple baseline models such as random forest and pixel-wise linear regression substantially. A modified version of the standard UNet architecture for flat images yields results that are as good as the predictions by the spherical CNN. Variations in the implementation of isotopes between climate models likely contribute to an observed deterioration of emulation results when testing on data obtained from different climate models than the one used for training. Future work toward stable water-isotope emulation might focus on achieving robust climate–oxygen isotope relationships or exploring the set of possible predictor variables.

Studies agree on a significant global mean sea level rise in the 20th century and its recent 21st century acceleration in the satellite record. At regional scale, the evolution of sea level probability distributions is often assumed to be dominated by changes in the mean. However, a quantification of changes in distributional shapes in a changing climate is currently missing. To this end, we propose a novel framework quantifying significant changes in probability distributions from time series data. The framework first quantifies linear trends in quantiles through quantile regression. Quantile slopes are then projected onto a set of four orthogonal polynomials quantifying how such changes can be explained by independent shifts in the first four statistical moments. The framework proposed is theoretically founded, general, and can be applied to any climate observable with close-to-linear changes in distributions. We focus on observations and a coupled climate model (GFDL-CM4). In the historical period, trends in coastal daily sea level have been driven mainly by changes in the mean and can therefore be explained by a shift of the distribution with no change in shape. In the modeled world, robust changes in higher order moments emerge with increasing $ {\mathrm{CO}}_2 $ concentration. Such changes are driven in part by ocean circulation alone and get amplified by sea level pressure fluctuations, with possible consequences for sea level extremes attribution studies.

Twelve climate models and observations are used to attribute the global mean surface temperature (GMST) changes from 1900 to 2014 to external climate forcings. The external forcings are decomposed into the effects of the well-mixed greenhouse gas concentration variation, the effects of anthropogenic aerosol concentration changes, and the effects of natural forcings. First, a convolutional neural network (CNN) is trained to estimate the simulated historical GMST from single-forcing experiments using outputs from the multi-model ensemble. We then use this CNN to solve the attribution problem using an original variational inversion approach. The variational inversion is first validated using historical climate simulations as pseudo-observations. Then we perform an inversion from observations. This provides a distribution of the GMST resulting from the three forcings. For 2014, inversions estimate that the greenhouse gases changes are responsible for a GMST anomaly within [0.8 $ {}^{\circ } $ C,1.9 $ {}^{\circ } $ C], while anthropogenic aerosols and natural forcings anomalies are within [−0.7 $ {}^{\circ } $ C,−0.1 $ {}^{\circ } $ C] and [−0.1 $ {}^{\circ } $ C,0.3 $ {}^{\circ } $ C], respectively. The method designed here can be adapted and extended to attribute the changes of other variables or to focus on the regional scale.

Supraglacial lakes and rivers dominate the storage and transport of meltwater on the southwest Greenland Ice Sheet (GrIS) surface. Despite functioning as interconnected hydrologic networks, supraglacial lakes and rivers are commonly studied as independent features, resulting in an incomplete understanding of their collective impact on meltwater storage and routing. We use Landsat 8 satellite imagery to assess the seasonal evolution of supraglacial lakes and rivers on the southwest GrIS during the 2015 melt season. Remotely sensed meltwater areas and volumes are compared with surface runoff simulations from three climate models (MERRA-2, MAR 3.6 and RACMO 2.3), and with in situ observations of proglacial discharge in the Watson River. We find: (1) at elevations >1600 m, 21% of supraglacial lakes and 28% of supraglacial rivers drain into moulins, signifying the presence of high-elevation surface-to-bed meltwater connections even during a colder-than-average melt season; (2) while supraglacial lakes dominate instantaneous surface meltwater storage, supraglacial rivers dominate total surface meltwater area and discharge; (3) the combined surface area of supraglacial lakes and rivers is strongly correlated with modeled surface runoff; and (4) of the three models examined here, MERRA-2 runoff yields the highest overall correlation with observed proglacial discharge in the Watson River.

Peoria Loess deposited in western Iowa during the last glacial maximum (LGM) shows distinct geochemical and particle-size variations as a function of both depth and distance east of the Missouri River. Geochemical and particle-size data indicate that Peoria Loess in western Iowa probably had two sources: the Missouri River valley, and a source that lay to the west of the Missouri River. Both sources indicate that LGM paleowinds in western Iowa had a strong westerly component, similar to interpretations of previous workers. A compilation of loess studies in Iowa and elsewhere indicates that westerly winds were dominant during loess transport over much of the midcontinent south of the Laurentide ice sheet, which is not in agreement with paleowinds simulated by atmospheric general circulation models (AGCMs). AGCMs consistently generate a glacial anticyclone with easterly or northeasterly winds over the Laurentide ice sheet and the area to the south of it. Loess deposition in the midcontinent during the LGM may be a function of infrequent northwesterly winds that were unrelated to the presence of the glacial anticyclone.

Projections of ice-sheet mass balance require regional ocean warming projections derived from atmosphere-ocean general circulation models (AOGCMs). However, the coarse resolution of AOGCMs: (1) may lead to systematic or AOGCM-specific biases and (2) makes it difficult to identify relevant water masses. Here, we employ a large-scale metric of Antarctic Shelf Bottom Water (ASBW) to investigate circum-Antarctic temperature biases and warming projections in 19 different Coupled Model Intercomparison Project Phase 5 (CMIP5) AOGCMs forced with two different ‘representative concentration pathways’ (RCPs). For high-emissions RCP 8.5, the ensemble mean 21st century ASBW warming is 0.66, 0.74 and 0.58°C for the Amundsen, Ross and Weddell Seas (AS, RS and WS), respectively. RCP 2.6 ensemble mean projections are substantially lower: 0.21, 0.26, and 0.19°C. All distributions of regional ASBW warming are positively skewed; for RCP 8.5, four AOGCMs project warming of greater than 1.8°C in the RS. Across the ensemble, there is a strong, RCP-independent, correlation between WS and RS warming. AS warming is more closely linked to warming in the Southern Ocean. We discuss possible physical mechanisms underlying the spatial patterns of warming and highlight implications of these results on strategies for forcing ice-sheet mass balance projections.

In an effort to derive temperature-based criteria of habitability for multicellular life, we investigated the thermal limits of terrestrial poikilotherms, i.e. organisms whose body temperature and the functioning of all vital processes is directly affected by the ambient temperature. Multicellular poikilotherms are the most common and evolutionarily ancient form of complex life on earth. The thermal limits for the active metabolism and reproduction of multicellular poikilotherms on earth are approximately bracketed by the temperature interval 0°C ≤ T ≤ 50°C. The same interval applies to the photosynthetic production of oxygen, an essential ingredient of complex life, and for the generation of atmospheric biosignatures observable in exoplanets. Analysis of the main mechanisms responsible for the thermal thresholds of terrestrial life suggests that the same mechanisms would apply to other forms of chemical life. We therefore propose a habitability index for complex life, h050, representing the mean orbital fraction of planetary surface that satisfies the temperature limits 0°C ≤ T ≤ 50°C. With the aid of a climate model tailored for the calculation of the surface temperature of Earth-like planets, we calculated h050 as a function of planet insolation, S, and atmospheric columnar mass, Natm, for a few earth-like atmospheric compositions with trace levels of CO2. By displaying h050 as a function of S and Natm, we built up an atmospheric mass habitable zone (AMHZ) for complex life. At variance with the classic habitable zone, the inner edge of the complex life habitable zone is not affected by the uncertainties inherent to the calculation of the runaway greenhouse limit. The complex life habitable zone is significantly narrower than the habitable zone of dry planets. Our calculations illustrate how changes in ambient conditions dependent on S and Natm, such as temperature excursions and surface dose of secondary particles of cosmic rays, may influence the type of life potentially present at different epochs of planetary evolution inside the AMHZ.

Ongoing climate variations in the Arctic and Antarctic pose an apparent paradox. In contrast to the large warming and loss of sea ice in the Arctic in recent decades, Antarctic temperatures and sea ice show little change except for the Antarctic Peninsula. However, model simulations indicate that the Arctic changes have been shaped largely by low-frequency variations of the atmospheric circulation, superimposed on a greenhouse warming that is apparent in model simulations when ensemble averages smooth out the circulation-driven variability of the late 20th century. By contrast, the Antarctic changes of recent decades appear to be shaped by ozone depletion and an associated strengthening of the southern annular mode of the atmospheric circulation. While the signature of greenhouse-driven change is projected to emerge from the natural variability during the present century, the emergence of a statistically significant greenhouse signal may be slower than in other regions. Models suggest that feedbacks from retreating sea ice will make autumn and winter the seasons of the earliest emergence of the greenhouse signal in both Polar Regions. Priorities for enhanced robustness of the Antarctic climate simulations are the inclusion of ozone chemistry and the realistic simulation of water vapour over the Antarctic Ice Sheet.