Search

We evaluate the vertical turbulent-kinetic-energy (TKE) mixing scheme of the NEMO-SI3 ocean–sea-ice model in sea-ice-covered regions of the Arctic Ocean. Specifically, we assess the parameters involved in TKE mixed-layer-penetration (MLP) parameterization. This ad hoc parameterization aims to capture processes that impact the ocean surface boundary layer, such as near-inertial oscillations, ocean swells, and waves, which are often not well represented in the default TKE scheme. We evaluate this parameterization for the first time in three regions of the Arctic Ocean: the Makarov, Eurasian, and Canada basins. We demonstrate the strong effect of the scaling parameter that accounts for the presence of sea ice. Our results confirm that TKE MLP must be scaled down below sea ice to avoid unrealistically deep mixed layers. The other parameters evaluated are the percentage of energy penetrating below the mixed layer and the length scale of its decay with depth. All these parameters affect mixed-layer depth and its seasonal cycle, surface temperature, and salinity, as well as underlying stratification. Shallow mixed layers are associated with stronger stratification and fresh surface anomalies, and deeper mixed layers correspond to weaker stratification and salty surface anomalies. Notably, we observe significant impacts on sea-ice thickness across the Arctic Ocean in two scenarios: when the scaling parameter due to sea ice is absent and when the TKE mixed-layer-penetration process vanishes. In the former case, we observe an increase of several meters in mixed-layer depth, along with a reduction in sea-ice thickness ranging from 30 to 40 cm, reflecting the impact of stronger mixing. Conversely, in the latter case, we notice that a shallower mixed layer is accompanied by a moderate increase in sea-ice thickness, ranging from 10 to 20 cm, as expected from weaker mixing. Additionally, interannual variability suggests that experiments incorporating a scaling parameter based on sea-ice concentration display an increased mixed-layer depth during periods of reduced sea ice, which is consistent with observed trends. These findings underscore the influence of enhanced ocean mixing, through specific parameterizations, on the physical properties of the upper ocean and sea ice.

Understanding the variability of Antarctic sea ice is still a challenge. After decades of modest growth, an unprecedented minimum in the sea ice extent (SIE) was registered in summer 2017, and, following years of anomalously low SIE, a new record was established in early 2022. These two memorable minima have received great attention as single cases, but a comprehensive analysis of summer SIE minima is currently lacking. Indeed, other similar events are present in the observational record, although they are minor compared to the most recent ones, and a full analysis of all summer SIE minima is essential to separate potential common drivers from event-specific dynamics in order to ultimately improve our understanding of the Antarctic sea ice and climate variability. In this work, we examine sea ice and atmospheric conditions during and before all summer SIE minima over the satellite period up to 2022. We use observations and reanalysis data and compare our main findings with results from an ocean–sea ice model (NEMO–LIM) driven by prescribed atmospheric fields from ERA5. Examining SIE and sea ice concentration (SIC) anomalies, we find that the main contributors to the summer minima are the Ross and Weddell sectors. However, the two regions play different roles, and the variability of the Ross Sea explains most of the minima, with typical negative SIE anomalies about twice as large as the ones in the Weddell Sea. Furthermore, the distribution of SIC anomalies is also different: in the Weddell Sea, they exhibit a dipolar structure, with increased SIC next to the continent and decreased SIC at the sea ice margin, while the Ross Sea displays a more homogenous decrease. We also examine the role of wintertime sea ice conditions before the summer SIE minima and find mixed results depending on the period: the winter conditions are relevant in the most recent events, after 2017, but they are marginal for previous years. Next, we consider the influence of the atmosphere on the SIE minima, which is shown to play a major role: after analyzing the anomalous atmospheric circulation during the preceding spring, we find that different large-scale anomalies can lead to similar regional prevailing winds that drive the summer minima. Specifically, the SIE minima are generally associated with dominant northwesterly anomalous winds in the Weddell Sea, while a southwesterly anomalous flow prevails in the Ross Sea. Finally, we investigate the relative contribution of dynamic (e.g., ice transport) and thermodynamic (e.g., local melting) processes to the summer minima. Our results indicate that the exceptional sea ice loss in both the Ross and Weddell sectors is dominated at the large scale by thermodynamic processes, while dynamics are also present but with a minor role.

Atmospheric reanalyses are valuable datasets for driving ocean–sea ice general circulation models and for proposing multidecadal reconstructions of the ocean–sea ice system in polar regions. However, these reanalyses exhibit biases in these regions. It was previously found that the representation of Arctic and Antarctic sea ice in models participating in the Ocean Model Intercomparison Project Phase 2 (OMIP2, using the updated Japanese 55-year atmospheric reanalysis, JRA55-do) was significantly more realistic than in OMIP1 (forced by the atmospheric state from the Coordinated Ocean-ice Reference Experiments version 2, CORE-II). To understand why, we study the sea ice concentration budget and its relations to surface heat and momentum fluxes as well as the connections between the simulated ice drift and the ice concentration, the ice thickness and the wind stress in a subset of three models (CMCC-CM2-SR5, MRI-ESM2-0 and NorESM2-LM). These three models are representative of the ensemble and are the only ones to provide the surface fluxes and the tendencies of ice concentrations attributed to dynamic and thermodynamic processes required for the ice concentration budget analysis. The sea ice simulations of two other models (EC-Earth3 and MIROC6) forced by both CORE-II and JRA55-do reanalysis are also included in the analysis. It is found that negative summer biases in high-ice-concentration regions and positive biases in the Canadian Arctic Archipelago (CAA) and central Weddell Sea (CWS) regions are reduced from OMIP1 to OMIP2 due to surface heat flux changes. Net shortwave radiation fluxes provide key improvements in the Arctic interior, CAA and CWS regions. There is also an influence of improved surface wind stress in OMIP2 giving better winter Antarctic ice concentration and the Arctic ice drift magnitude simulations near the ice edge. The ice velocity direction simulations in the Beaufort Gyre and the Pacific and Atlantic sectors of the Southern Ocean in OMIP2 are also improved owing to surface wind stress changes. This study provides clues on how improved atmospheric reanalysis products influence sea ice simulations. Our findings suggest that attention should be paid to the radiation fluxes and winds in atmospheric reanalyses in polar regions.

The seasonal cycle of the Antarctic sea ice extent is strongly asymmetric, with a relatively slow increase after the summer minimum followed by a more rapid decrease after the winter maximum. This cycle is intimately linked to the seasonal cycle of the insolation received at the top of the atmosphere, but sea ice processes as well as the exchanges with the atmosphere and ocean may also play a role. To quantify these contributions, a series of idealized sensitivity experiments have been performed with an eddy-permitting (1/4∘) NEMO-LIM3 (Nucleus for European Modelling of the Ocean–Louvain-la-Neuve sea ice model version 3) Southern Ocean configuration, including a representation of ice shelf cavities, in which the model was either driven by an atmospheric reanalysis or coupled to the COSMO-CLM2 regional atmospheric model. In those experiments, sea ice thermodynamics and dynamics as well as the exchanges with the ocean and atmosphere are strongly perturbed. This perturbation is achieved by modifying snow and ice thermal conductivities, the vertical mixing in the ocean top layers, the effect of freshwater uptake and release upon sea ice growth and melt, ice dynamics, and surface albedo. We find that the evolution of sea ice extent during the ice advance season is largely independent of the direct effect of the perturbation and appears thus mainly controlled by initial state in summer and subsequent insolation changes. In contrast, the melting rate varies strongly between the experiments during the retreat, in particular if the surface albedo or sea ice transport are modified, demonstrating a strong contribution of those elements to the evolution of ice coverage through spring and summer. As with the advance phase, the retreat is also influenced by conditions at the beginning of the melt season in September. Atmospheric feedbacks enhance the model winter ice extent response to any of the perturbed processes, and the enhancement is strongest when the albedo is modified. The response of sea ice volume and extent to changes in entrainment of subsurface warm waters to the ocean surface is also greatly amplified by the coupling with the atmosphere.

The retreat of Arctic sea ice is frequently considered to be a possible driver of changes in climate extremes in the Arctic and possibly down to mid-latitudes. However, it remains unclear how the atmosphere will respond to a near-total retreat of summer Arctic sea ice, a reality that might occur in the foreseeable future. This study explores this question by conducting sensitivity experiments with two global coupled climate models run at two different horizontal resolutions to investigate the change in temperature and precipitation extremes during summer over peripheral Arctic regions following a sudden reduction in summer Arctic sea ice cover. An increase in frequency and persistence of maximum surface air temperature is found in all peripheral Arctic regions during the summer, when sea ice loss occurs. For each 1×106 km2 of Arctic sea ice extent reduction, the absolute frequency of days exceeding the surface air temperature of the climatological 90th percentile increases by ∼ 4 % over the Svalbard area, and the duration of warm spells increases by ∼ 1 d per month over the same region. Furthermore, we find that the 10th percentile of surface daily air temperature increases more than the 90th percentile, leading to a weakened diurnal cycle of surface air temperature. Finally, an increase in extreme precipitation, which is less robust than the increase in extreme temperatures, is found in all regions in summer. These findings suggest that a sudden retreat of summer Arctic sea ice clearly impacts the extremes in maximum surface air temperature and precipitation over the peripheral Arctic regions with the largest influence over inhabited islands such as Svalbard or northern Canada. Nonetheless, even with a large sea ice reduction in regions close to the North Pole, the local precipitation response is relatively small compared to internal climate variability.

The Earth system model EC-Earth3 for contributions to CMIP6 is documented here, with its flexible coupling framework, major model configurations, a methodology for ensuring the simulations are comparable across different high-performance computing (HPC) systems, and with the physical performance of base configurations over the historical period. The variety of possible configurations and sub-models reflects the broad interests in the EC-Earth community. EC-Earth3 key performance metrics demonstrate physical behavior and biases well within the frame known from recent CMIP models. With improved physical and dynamic features, new Earth system model (ESM) components, community tools, and largely improved physical performance compared to the CMIP5 version, EC-Earth3 represents a clear step forward for the only European community ESM. We demonstrate here that EC-Earth3 is suited for a range of tasks in CMIP6 and beyond.

We introduce PARASO, a novel five-component fully coupled regional climate model over an Antarctic circumpolar domain covering the full Southern Ocean. The state-of-the-art models used are the fast Elementary Thermomechanical Ice Sheet model (f.ETISh) v1.7 (ice sheet), the Nucleus for European Modelling of the Ocean (NEMO) v3.6 (ocean), the Louvain-la-Neuve sea-ice model (LIM) v3.6 (sea ice), the COnsortium for Small-scale MOdeling (COSMO) model v5.0 (atmosphere) and its CLimate Mode (CLM) v4.5 (land), which are here run at a horizontal resolution close to 1/4∘. One key feature of this tool resides in a novel two-way coupling interface for representing ocean–ice-sheet interactions, through explicitly resolved ice-shelf cavities. The impact of atmospheric processes on the Antarctic ice sheet is also conveyed through computed COSMO-CLM–f.ETISh surface mass exchange. In this technical paper, we briefly introduce each model's configuration and document the developments that were carried out in order to establish PARASO. The new offline-based NEMO–f.ETISh coupling interface is thoroughly described. Our developments also include a new surface tiling approach to combine open-ocean and sea-ice-covered cells within COSMO, which was required to make this model relevant in the context of coupled simulations in polar regions. We present results from a 2000–2001 coupled 2-year experiment. PARASO is numerically stable and fully operational. The 2-year simulation conducted without fine tuning of the model reproduced the main expected features, although remaining systematic biases provide perspectives for further adjustment and development.

The Sea Ice Evaluation Tool (SITool) described in this paper is a performance metrics and diagnostics tool developed to evaluate the skill of Arctic and Antarctic model reconstructions of sea ice concentration, extent, edge location, drift, thickness, and snow depth. It is a Python-based software and consists of well-documented functions used to derive various sea ice metrics and diagnostics. Here, SITool version 1.0 (v1.0) is introduced and documented, and is then used to evaluate the performance of global sea ice reconstructions from nine models that provided sea ice output under the experimental protocols of the Coupled Model Intercomparison Project phase 6 (CMIP6) Ocean Model Intercomparison Project with two different atmospheric forcing datasets: the Coordinated Ocean-ice Reference Experiments version 2 (CORE-II) and the updated Japanese 55-year atmospheric reanalysis (JRA55-do). Two sets of observational references for the sea ice concentration, thickness, snow depth, and ice drift are systematically used to reflect the impact of observational uncertainty on model performance. Based on available model outputs and observational references, the ice concentration, extent, and edge location during 1980–2007, as well as the ice thickness, snow depth, and ice drift during 2003–2007 are evaluated. In general, model biases are larger than observational uncertainties, and model performance is primarily consistent compared to different observational references. By changing the atmospheric forcing from CORE-II to JRA55-do reanalysis data, the overall performance (mean state, interannual variability, and trend) of the simulated sea ice areal properties in both hemispheres, as well as the mean ice thickness simulation in the Antarctic, the mean snow depth, and ice drift simulations in both hemispheres are improved. The simulated sea ice areal properties are also improved in the model with higher spatial resolution. For the cross-metric analysis, there is no link between the performance in one variable and the performance in another. SITool is an open-access version-controlled software that can run on a wide range of CMIP6-compliant sea ice outputs. The current version of SITool (v1.0) is primarily developed to evaluate atmosphere-forced simulations and it could be eventually extended to fully coupled models.

A substantial part of Arctic climate predictability at interannual timescales stems from the knowledge of the initial sea ice conditions. Among all sea ice properties, its volume, which is a product of sea ice concentration (SIC) and thickness (SIT), is the most responsive parameter to climate change. However, the majority of climate prediction systems are only assimilating the observed SIC due to lack of long-term reliable global observation of SIT. In this study, the EC-Earth3 Climate Prediction System with anomaly initialization to ocean, SIC and SIT states is developed. In order to evaluate the regional benefits of specific initialized variables, three sets of retrospective ensemble prediction experiments are performed with different initialization strategies: ocean only; ocean plus SIC; and ocean plus SIC and SIT initialization. In the Atlantic Arctic, the Greenland–Iceland–Norway (GIN) and Barents seas are the two most skilful regions in SIC prediction for up to 5–6 lead years with ocean initialization; there are re-emerging skills for SIC in the Barents and Kara seas in lead years 7–9 coinciding with improved skills of sea surface temperature (SST), reflecting the impact of SIC initialization on ocean–atmosphere interactions for interannual-to-decadal timescales. For the year 2–9 average, the region with significant skill for SIT is confined to the central Arctic Ocean, covered by multi-year sea ice (CAO-MYI). Winter preconditioning with SIT initialization increases the skill for September SIC in the eastern Arctic (e.g. Kara, Laptev and East Siberian seas) and in turn improve the skill of air surface temperature locally and further expanded over land. SIT initialization outperforms the other initialization methods in improving SIT prediction in the Pacific Arctic (e.g. East Siberian and Beaufort seas) in the first few lead years. Our results suggest that as the climate warming continues and the central Arctic Ocean might become seasonal ice free in the future, the controlling mechanism for decadal predictability may thus shift from sea ice volume to ocean-driven processes.

We compare the mass budget of the Arctic sea ice for 15 models submitted to the latest Coupled Model Intercomparison Project (CMIP6), using new diagnostics that have not been available for previous model inter-comparisons. These diagnostics allow us to look beyond the standard metrics of ice cover and thickness to compare the processes of sea ice growth and loss in climate models in a more detailed way than has previously been possible. For the 1960–1989 multi-model mean, the dominant processes causing annual ice growth are basal growth and frazil ice formation, which both occur during the winter. The main processes by which ice is lost are basal melting, top melting and advection of ice out of the Arctic. The first two processes occur in summer, while the latter process is present all year. The sea ice budgets for individual models are strikingly similar overall in terms of the major processes causing ice growth and loss and in terms of the time of year during which each process is important. However, there are also some key differences between the models, and we have found a number of relationships between model formulation and components of the ice budget that hold for all or most of the CMIP6 models considered here. The relative amounts of frazil and basal ice formation vary between the models, and the amount of frazil ice formation is strongly dependent on the value chosen for the minimum frazil ice thickness. There are also differences in the relative amounts of top and basal melting, potentially dependent on how much shortwave radiation can penetrate through the sea ice into the ocean. For models with prognostic melt ponds, the choice of scheme may affect the amount of basal growth, basal melt and top melt, and the choice of thermodynamic scheme is important in determining the amount of basal growth and top melt. As the ice cover and mass decline during the 21st century, we see a shift in the timing of the top and basal melting in the multi-model mean, with more melt occurring earlier in the year and less melt later in the summer. The amount of basal growth reduces in the autumn, but it increases in the winter due to thinner sea ice over the course of the 21st century. Overall, extra ice loss in May–June and reduced ice growth in October–November are partially offset by reduced ice melt in August and increased ice growth in January–February. For the individual models, changes in the budget components vary considerably in terms of magnitude and timing of change. However, when the evolving budget terms are considered as a function of the changing ice state itself, behaviours common to all the models emerge, suggesting that the sea ice components of the models are fundamentally responding in a broadly consistent way to the warming climate. It is possible that this similarity in the model budgets may represent a lack of diversity in the model physics of the CMIP6 models considered here. The development of new observational datasets for validating the budget terms would help to clarify this.

Ocean–sea-ice coupled models constrained by various observations provide different ice thickness estimates in the Antarctic. We evaluate contemporary monthly ice thickness from four reanalyses in the Weddell Sea: the German contribution of the project Estimating the Circulation and Climate of the Ocean Version 2 (GECCO2), the Southern Ocean State Estimate (SOSE), the Ensemble Kalman Filter system based on the Nucleus for European Modelling of the Ocean (NEMO-EnKF) and the Global Ice–Ocean Modeling and Assimilation System (GIOMAS). The evaluation is performed against reference satellite and in situ observations from ICESat-1, Envisat, upward-looking sonars and visual ship-based sea-ice observations. Compared with ICESat-1, NEMO-EnKF has the highest correlation coefficient (CC) of 0.54 and lowest root mean square error (RMSE) of 0.44 m. Compared with in situ observations, SOSE has the highest CC of 0.77 and lowest RMSE of 0.72 m. All reanalyses underestimate ice thickness near the coast of the western Weddell Sea with respect to ICESat-1 and in situ observations even though these observational estimates may be biased low. GECCO2 and NEMO-EnKF reproduce the seasonal variation in first-year ice thickness reasonably well in the eastern Weddell Sea. In contrast, GIOMAS ice thickness performs best in the central Weddell Sea, while SOSE ice thickness agrees most with the observations from the southern coast of the Weddell Sea. In addition, only NEMO-EnKF can reproduce the seasonal evolution of the large-scale spatial distribution of ice thickness, characterized by the thick ice shifting from the southwestern and western Weddell Sea in summer to the western and northwestern Weddell Sea in spring. We infer that the thick ice distribution is correlated with its better simulation of northward ice motion in the western Weddell Sea. These results demonstrate the possibilities and limitations of using current sea-ice reanalysis for understanding the recent variability of sea-ice volume in the Antarctic.

This study assesses the impact of different sea ice thickness distribution (ITD) discretizations on the sea ice concentration (SIC) variability in ocean stand-alone NEMO3.6–LIM3 simulations. Three ITD discretizations with different numbers of sea ice thickness categories and boundaries are evaluated against three different satellite products (hereafter referred to as “data”). Typical model and data interannual SIC variability is characterized by K-means clustering both in the Arctic and Antarctica between 1979 and 2014. We focus on two seasons, winter (January–March) and summer (August–October), in which correlation coefficients across clusters in individual months are largest. In the Arctic, clusters are computed before and after detrending the series with a second-degree polynomial to separate interannual from longer-term variability. The analysis shows that, before detrending, winter clusters reflect the SIC response to large-scale atmospheric variability at both poles, while summer clusters capture the negative and positive trends in Arctic and Antarctic SIC, respectively. After detrending, Arctic clusters reflect the SIC response to interannual atmospheric variability predominantly. The cluster analysis is complemented with a model–data comparison of the sea ice extent and SIC anomaly patterns. The single-category discretization shows the worst model–data agreement in the Arctic summer before detrending, related to a misrepresentation of the long-term melting trend. Similarly, increasing the number of thin categories reduces model–data agreement in the Arctic, due to a poor representation of the summer melting trend and an overly large winter sea ice volume associated with a net increase in basal ice growth. In contrast, more thin categories improve model realism in Antarctica, and more thick ones improve it in central Arctic regions with very thick ice. In all the analyses we nonetheless identify no optimal discretization. Our results thus suggest that no clear benefit in the representation of SIC variability is obtained from increasing the number of sea ice thickness categories beyond the current standard with five categories in NEMO3.6–LIM3.

We use model simulations from the CESM1-CAM5-BGC-LE dataset to characterise the Arctic sea ice thickness internal variability both spatially and temporally. These properties, and their stationarity, are investigated in three different contexts: (1) constant pre-industrial, (2) historical and (3) projected conditions. Spatial modes of variability show highly stationary patterns regardless of the forcing and mean state. A temporal analysis reveals two peaks of significant variability, and despite a non-stationarity on short timescales, they remain more or less stable until the first half of the 21st century, where they start to change once summer ice-free events occur, after 2050.

This work evaluates the statistical predictability of the Arctic sea ice volume (SIV) anomaly – here defined as the detrended and deseasonalized SIV – on the interannual timescale. To do so, we made use of six datasets, from three different atmosphere–ocean general circulation models, with two different horizontal grid resolutions each. Based on these datasets, we have developed a statistical empirical model which in turn was used to test the performance of different predictor variables, as well as to identify optimal locations from where the SIV anomaly could be better reconstructed and/or predicted. We tested the hypothesis that an ideal sampling strategy characterized by only a few optimal sampling locations can provide in situ data for statistically reproducing and/or predicting the SIV interannual variability. The results showed that, apart from the SIV itself, the sea ice thickness is the best predictor variable, although total sea ice area, sea ice concentration, sea surface temperature, and sea ice drift can also contribute to improving the prediction skill. The prediction skill can be enhanced further by combining several predictors into the statistical model. Applying the statistical model with predictor data from four well-placed locations is sufficient for reconstructing about 70 % of the SIV anomaly variance. As suggested by the results, the four first best locations are placed at the transition Chukchi Sea–central Arctic–Beaufort Sea (79.5∘ N, 158.0∘ W), near the North Pole (88.5∘ N, 40.0∘ E), at the transition central Arctic–Laptev Sea (81.5∘ N, 107.0∘ E), and offshore the Canadian Archipelago (82.5∘ N, 109.0∘ W), in this respective order. Adding further to six well-placed locations, which explain about 80 % of the SIV anomaly variance, the statistical predictability does not substantially improve taking into account that 10 locations explain about 84 % of that variance. An improved model horizontal resolution allows a better trained statistical model so that the reconstructed values better approach the original SIV anomaly. On the other hand, if we inspect the interannual variability, the predictors provided by numerical models with lower horizontal resolution perform better when reconstructing the original SIV variability. We believe that this study provides recommendations for the ongoing and upcoming observational initiatives, in terms of an Arctic optimal observing design, for studying and predicting not only the SIV values but also its interannual variability.

The Earth System Model Evaluation Tool (ESMValTool) is a community diagnostics and performance metrics tool designed to improve comprehensive and routine evaluation of Earth system models (ESMs) participating in the Coupled Model Intercomparison Project (CMIP). It has undergone rapid development since the first release in 2016 and is now a well-tested tool that provides end-to-end provenance tracking to ensure reproducibility. It consists of (1) an easy-to-install, well-documented Python package providing the core functionalities (ESMValCore) that performs common preprocessing operations and (2) a diagnostic part that includes tailored diagnostics and performance metrics for specific scientific applications. Here we describe large-scale diagnostics of the second major release of the tool that supports the evaluation of ESMs participating in CMIP Phase 6 (CMIP6). ESMValTool v2.0 includes a large collection of diagnostics and performance metrics for atmospheric, oceanic, and terrestrial variables for the mean state, trends, and variability. ESMValTool v2.0 also successfully reproduces figures from the evaluation and projections chapters of the Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5) and incorporates updates from targeted analysis packages, such as the NCAR Climate Variability Diagnostics Package for the evaluation of modes of variability, the Thermodynamic Diagnostic Tool (TheDiaTo) to evaluate the energetics of the climate system, as well as parts of AutoAssess that contains a mix of top–down performance metrics. The tool has been fully integrated into the Earth System Grid Federation (ESGF) infrastructure at the Deutsches Klimarechenzentrum (DKRZ) to provide evaluation results from CMIP6 model simulations shortly after the output is published to the CMIP archive. A result browser has been implemented that enables advanced monitoring of the evaluation results by a broad user community at much faster timescales than what was possible in CMIP5.

Most Earth system models (ESMs) are running under different high-performance computing (HPC) environments. This has several advantages, from allowing different groups to work with the same tool in parallel to leveraging the burden of ensemble climate simulations, but it also offers alternative solutions in the case of shutdown (expected or not) of any of the environments. However, for obvious scientific reasons, it is critical to ensure that ESMs provide identical results under changes in computing environment. While strict bit-for-bit reproducibility is not always guaranteed with ESMs, it is desirable that results obtained under one computing environment are at least statistically indistinguishable from those obtained under another environment, which we term a “replicability” condition following the metrology nomenclature. Here, we develop a protocol to assess the replicability of the EC-Earth ESM. Using two versions of EC-Earth, we present one case of non-replicability and one case of replicability. The non-replicable case occurs with the older version of the model and likely finds its origin in the treatment of river runoff along Antarctic coasts. By contrast, the more recent version of the model provides replicable results. The methodology presented here has been adopted as a standard test by the EC-Earth consortium (27 institutions in Europe) to evaluate the replicability of any new model version across platforms, including for CMIP6 experiments. To a larger extent, it can be used to assess whether other ESMs can safely be ported from one HPC environment to another for studying climate-related questions. Our results and experience with this work suggest that the default assumption should be that ESMs are not replicable under changes in the HPC environment, until proven otherwise.

The ice thickness distribution (ITD) is one of the core constituents of modern sea ice models. The ITD accounts for the unresolved spatial variability of sea ice thickness within each model grid cell. While there is a general consensus on the added physical realism brought by the ITD, how to discretize it remains an open question. Here, we use the ocean–sea ice general circulation model, Nucleus for European Modelling of the Ocean (NEMO) version 3.6 and Louvain-la-Neuve sea Ice Model (LIM) version 3 (NEMO3.6-LIM3), forced by atmospheric reanalyses to test how the ITD discretization (number of ice thickness categories, positions of the category boundaries) impacts the simulated mean Arctic and Antarctic sea ice states. We find that winter ice volumes in both hemispheres increase with the number of categories and attribute that increase to a net enhancement of basal ice growth rates. The range of simulated mean winter volumes in the various experiments amounts to ∼30 % and ∼10 % of the reference values (run with five categories) in the Arctic and Antarctic, respectively. This suggests that the way the ITD is discretized has a significant influence on the model mean state, all other things being equal. We also find that the existence of a thick category with lower bounds at ∼4 and ∼2 m for the Arctic and Antarctic, respectively, is a prerequisite for allowing the storage of deformed ice and therefore for fostering thermodynamic growth in thinner categories. Our analysis finally suggests that increasing the resolution of the ITD without changing the lower limit of the upper category results in small but not negligible variations of ice volume and extent. Our study proposes for the first time a bi-polar process-based explanation of the origin of mean sea ice state changes when the ITD discretization is modified. The sensitivity experiments conducted in this study, based on one model, emphasize that the choice of category positions, especially of thickest categories, has a primary influence on the simulated mean sea ice states while the number of categories and resolution have only a secondary influence. It is also found that the current default discretization of the NEMO3.6-LIM3 model is sufficient for large-scale present-day climate applications. In all cases, the role of the ITD discretization on the simulated mean sea ice state has to be appreciated relative to other influences (parameter uncertainty, forcing uncertainty, internal climate variability).

The ocean–sea ice reanalyses are one of the main sources of Arctic sea ice thickness data both in terms of spatial and temporal resolution, since observations are still sparse in time and space. In this work, we first aim at comparing how the sea ice thickness from an ensemble of 14 reanalyses compares with different sources of observations, such as moored upward-looking sonars, submarines, airbornes, satellites, and ice boreholes. Second, based on the same reanalyses, we intend to characterize the timescales (persistence) and length scales of sea ice thickness anomalies. We investigate whether data assimilation of sea ice concentration by the reanalyses impacts the realism of sea ice thickness as well as its respective timescales and length scales. The results suggest that reanalyses with sea ice data assimilation do not necessarily perform better in terms of sea ice thickness compared with the reanalyses which do not assimilate sea ice concentration. However, data assimilation has a clear impact on the timescales and length scales: reanalyses built with sea ice data assimilation present shorter timescales and length scales. The mean timescales and length scales for reanalyses with data assimilation vary from 2.5 to 5.0 months and 337.0 to 732.5 km, respectively, while reanalyses with no data assimilation are characterized by values from 4.9 to 7.8 months and 846.7 to 935.7 km, respectively.

The recent Arctic sea ice reduction comes with an increase in the ice-free season duration, with comparable contributions of earlier ice retreat and later advance. CMIP5 models all project that the trend towards later advance should progressively exceed and ultimately double the trend towards earlier retreat, causing the ice-free season to shift into autumn. We show that such a shift is a basic feature of the thermodynamic response of seasonal ice to warming. The detailed analysis of an idealised thermodynamic ice–ocean model stresses the role of two seasonal amplifying feedbacks. The summer feedback generates a 1.6-day-later advance in response to a 1-day-earlier retreat. The underlying physics are the property of the upper ocean to absorb solar radiation more efficiently than it can release heat right before ice advance. The winter feedback is comparatively weak, prompting a 0.3-day-earlier retreat in response to a 1-day shift towards later advance. This is because a shorter growth season implies thinner ice, which subsequently melts away faster. However, the winter feedback is dampened by the relatively long ice growth period and by the inverse relationship between ice growth rate and thickness. At inter-annual timescales, the thermodynamic response of ice seasonality to warming is obscured by inter-annual variability. Nevertheless, in the long term, because all feedback mechanisms relate to basic and stable elements of the Arctic climate system, there is little inter-model uncertainty on the projected long-term shift into autumn of the ice-free season.

A data assimilation method capable of constraining the sea ice of an Earth system model in a dynamically consistent manner has the potential to enhance the accuracy of climate reconstructions and predictions. Finding such a method is challenging because the sea ice dynamics is highly non-linear, and sea ice variables are strongly non-Gaussian distributed and tightly coupled to the rest of the Earth system – particularly thermodynamically with the ocean. We investigate key practical implementations for assimilating sea ice concentration – the predominant source of observations in polar regions – with the Norwegian Climate Prediction Model that combines the Norwegian Earth System Model with the Ensemble Kalman Filter. The performances of the different configurations are investigated by conducting 10-year reanalyses in a perfect model framework. First, we find that with a flow-dependent assimilation method, strongly coupled ocean–sea ice assimilation outperforms weakly coupled (sea ice only) assimilation. An attempt to prescribe the covariance between the ocean temperature and the sea ice concentration performed poorly. Extending the ocean updates below the mixed layer is slightly beneficial for the Arctic hydrography. Second, we find that solving the analysis for the multicategory instead of the aggregated ice state variables greatly reduces the errors in the ice state. Updating the ice volumes induces a weak drift in the bias for the thick ice category that relates to the postprocessing of unphysical thicknesses. Preserving the ice thicknesses for each category during the assimilation mitigates the drift without degrading the performance. The robustness and reliability of the optimal setting is demonstrated for a 20-year reanalysis. The error of sea ice concentration reduces by 50% (65%), sea ice thickness by 25% (35%), sea surface temperature by 33% (23%) and sea surface salinity by 11% (25%) in the Arctic (Antarctic) compared to a reference run without assimilation.

Sea ice cover and thickness have substantially decreased in the Arctic Ocean since the beginning of the satellite era. As a result, sea ice strength has been reduced, allowing more deformation and fracturing and leading to increased sea ice drift speed. We use the version 3.6 of the global ocean–sea ice NEMO-LIM model (Nucleus for European Modelling of the Ocean coupled to the Louvain-la-Neuve sea Ice Model), satellite, buoy and submarine observations, as well as reanalysis data over the period from 1979 to 2013 to study these relationships. Overall, the model agrees well with observations in terms of sea ice extent, concentration and thickness. The seasonal cycle of sea ice drift speed is reasonably well reproduced by the model. NEMO-LIM3.6 is able to capture the relationships between the seasonal cycles of sea ice drift speed, concentration and thickness, with higher drift speed for both lower concentration and lower thickness, in agreement with observations. Model experiments are carried out to test the sensitivity of Arctic sea ice drift speed, thickness and concentration to changes in sea ice strength parameter P*. These show that higher values of P* generally lead to lower sea ice deformation and lower sea ice thickness, and that no single value of P* is the best option for reproducing the observed drift speed and thickness. The methodology proposed in this analysis provides a benchmark for a further model intercomparison related to the relationships between sea ice drift speed and strength, which is especially relevant in the context of the upcoming Coupled Model Intercomparison Project 6 (CMIP6).

A better understanding of the role of sea ice for the changing climate of our planet is the central aim of the diagnostic Coupled Model Intercomparison Project 6 (CMIP6)-endorsed Sea-Ice Model Intercomparison Project (SIMIP). To reach this aim, SIMIP requests sea-ice-related variables from climate-model simulations that allow for a better understanding and, ultimately, improvement of biases and errors in sea-ice simulations with large-scale climate models. This then allows us to better understand to what degree CMIP6 model simulations relate to reality, thus improving our confidence in answering sea-ice-related questions based on these simulations. Furthermore, the SIMIP protocol provides a standard for sea-ice model output that will streamline and hence simplify the analysis of the simulated sea-ice evolution in research projects independent of CMIP. To reach its aims, SIMIP provides a structured list of model output that allows for an examination of the three main budgets that govern the evolution of sea ice, namely the heat budget, the momentum budget, and the mass budget. In this contribution, we explain the aims of SIMIP in more detail and outline how its design allows us to answer some of the most pressing questions that sea ice still poses to the international climate-research community.

Almost all global climate models and Earth system models that participated in the Coupled Model Intercomparison Project 5 (CMIP5) show strong declines in Arctic sea ice extent and volume under the highest forcing scenario of the representative concentration pathways (RCPs) through 2100, including a transition from perennial to seasonal ice cover. Extended RCP simulations through 2300 were completed for a~subset of models, and here we examine the time evolution of Arctic sea ice in these simulations. In RCP2.6, the summer Arctic sea ice extent increases compared to its minimum following the peak radiative forcing in 2044 in all nine models. RCP4.5 demonstrates continued summer Arctic sea ice decline after the forcing stabilizes due to continued warming on longer timescales. Based on the analysis of these two scenarios, we suggest that Arctic summer sea ice extent could begin to recover if and when radiative forcing from greenhouse gas concentrations were to decrease. In RCP8.5 the Arctic Ocean reaches annually ice-free conditions in seven of nine models. The ensemble of simulations completed under the extended RCPs provide insight into the global temperature increase at which sea ice disappears in the Arctic and the reversibility of declines in seasonal sea ice extent.

Observations over the last 30 yr have shown that the sea ice extent in the Southern Ocean has slightly increased since 1979. Mechanisms responsible for this positive trend have not been well established yet. In this study we tackle two related issues: is the observed positive trend compatible with the internal variability of the system, and do the models agree with what we know about the observed internal variability? For that purpose, we analyse the evolution of sea ice around the Antarctic simulated by 24 different general circulation models involved in the 5th Coupled Model Intercomparison Project (CMIP5), using both historical and hindcast experiments. Our analyses show that CMIP5 models respond to the forcing, including the one induced by stratospheric ozone depletion, by reducing the sea ice cover in the Southern Ocean. Some simulations display an increase in sea ice extent similar to the observed one. According to models, the observed positive trend is compatible with internal variability. However, models strongly overestimate the variance of sea ice extent and the initialization methods currently used in models do not improve systematically the simulated trends in sea ice extent. On the basis of those results, a critical role of the internal variability in the observed increase of sea ice extent in the Southern Ocean could not be ruled out, but current models results appear inadequate to test more precisely this hypothesis.

Short-term and decadal sea-ice prediction systems need a realistic initial state, generally obtained using ice–ocean model simulations with data assimilation. However, only sea-ice concentration and velocity data are currently assimilated. In this work, an ensemble Kalman filter system is used to assimilate observed ice concentration and freeboard (i.e. thickness of emerged) data into a global coupled ocean–sea-ice model. The impact and effectiveness of our data assimilation system is assessed in two steps: firstly, through the use of synthetic data (i.e. model-generated data), and secondly, through the assimilation of real satellite data. While ice concentrations are available daily, freeboard data used in this study are only available during six one-month periods spread over 2005–2007. Our results show that the simulated Arctic and Antarctic sea-ice extents are improved by the assimilation of synthetic ice concentration data. Assimilation of synthetic ice freeboard data improves the simulated sea-ice thickness field. Using real ice concentration data enhances the model realism in both hemispheres. Assimilation of ice concentration data significantly improves the total hemispheric sea-ice extent all year long, especially in summer. Combining the assimilation of ice freeboard and concentration data leads to better ice thickness, but does not further improve the ice extent. Moreover, the improvements in sea-ice thickness due to the assimilation of ice freeboard remain visible well beyond the assimilation periods.

We examine the recent (1979–2010) and future (2011–2100) characteristics of the summer Arctic sea ice cover as simulated by 29 Earth system and general circulation models from the Coupled Model Intercomparison Project, phase 5 (CMIP5). As was the case with CMIP3, a large intermodel spread persists in the simulated summer sea ice losses over the 21st century for a given forcing scenario. The 1979–2010 sea ice extent, thickness distribution and volume characteristics of each CMIP5 model are discussed as potential constraints on the September sea ice extent (SSIE) projections. Our results suggest first that the future changes in SSIE with respect to the 1979–2010 model SSIE are related in a complicated manner to the initial 1979–2010 sea ice model characteristics, due to the large diversity of the CMIP5 population: at a given time, some models are in an ice-free state while others are still on the track of ice loss. However, in phase plane plots (that do not consider the time as an independent variable), we show that the transition towards ice-free conditions is actually occurring in a very similar manner for all models. We also find that the year at which SSIE drops below a certain threshold is likely to be constrained by the present-day sea ice properties. In a second step, using several adequate 1979–2010 sea ice metrics, we effectively reduce the uncertainty as to when the Arctic could become nearly ice-free in summertime, the interval [2041, 2060] being our best estimate for a high climate forcing scenario.

Two hindcast (1983–2007) simulations are performed with the global, ocean-sea ice models NEMO-LIM2 and NEMO-LIM3 driven by atmospheric reanalyses and climatologies. The two simulations differ only in their sea ice component, while all other elements of experimental design (resolution, initial conditions, atmospheric forcing) are kept identical. The main differences in the sea ice models lie in the formulation of the subgrid-scale ice thickness distribution, of the thermodynamic processes, of the sea ice salinity and of the sea ice rheology. To assess the differences in model skill over the period of investigation, we develop a set of metrics for both hemispheres, comparing the main sea ice variables (concentration, thickness and drift) to available observations and focusing on both mean state and seasonal to interannual variability. Based upon these metrics, we discuss the physical processes potentially responsible for the differences in model skill. In particular, we suggest that (i) a detailed representation of the ice thickness distribution increases the seasonal to interannual variability of ice extent, with spectacular improvement for the simulation of the recent observed summer Arctic sea ice retreats, (ii) the elastic-viscous-plastic rheology enhances the response of ice to wind stress, compared to the classical viscous-plastic approach, (iii) the grid formulation and the air-sea ice drag coefficient affect the simulated ice export through Fram Strait and the ice accumulation along the Canadian Archipelago, and (iv) both models show less skill in the Southern Ocean, probably due to the low quality of the reanalyses in this region and to the absence of important small-scale oceanic processes at the models' resolution (~1°).

Arctic sea-ice extent has strongly decreased since the beginning of satellite observations in the late 1970s. While several drivers are known to be implicated, their respective contribution is not fully understood. Here, we apply the Liang-Kleeman information flow method to five different large ensembles from the Coupled Model Intercomparison Project Phase 6 (CMIP6) over the 1970-2060 period to investigate the extent to which fluctuations in winter sea-ice volume, air temperature and ocean heat transport drive changes in subsequent summer Arctic sea-ice extent. This allows us to go beyond classical correlation analyses. Results show that air temperature is the most important controlling factor of summer sea-ice extent at interannual time scale, and that winter sea-ice volume and Atlantic Ocean heat transport play a secondary role. If we replace air temperature by net shortwave and downward longwave radiations, we find that the sum of influences from both radiations is almost similar to the air temperature influence, with the longwave radiation being dominant in driving changes in summer sea-ice extent. Finally, we find that the influence of air temperature is more prominent during periods of large sea-ice reduction and that this temperature influence has overall increased since 1970., (© 2024. The Author(s).)

Habitat degradation and climate change are globally acting as pivotal drivers of wildlife collapse, with mounting evidence that this erosion of biodiversity will accelerate in the following decades 1-3 . Here, we quantify the past, present and future ecological suitability of Europe for bumblebees, a threatened group of pollinators ranked among the highest contributors to crop production value in the northern hemisphere 4-8 . We demonstrate coherent declines of bumblebee populations since 1900 over most of Europe and identify future large-scale range contractions and species extirpations under all future climate and land use change scenarios. Around 38-76% of studied European bumblebee species currently classified as 'Least Concern' are projected to undergo losses of at least 30% of ecologically suitable territory by 2061-2080 compared to 2000-2014. All scenarios highlight that parts of Scandinavia will become potential refugia for European bumblebees; it is however uncertain whether these areas will remain clear of additional anthropogenic stressors not accounted for in present models. Our results underline the critical role of global change mitigation policies as effective levers to protect bumblebees from manmade transformation of the biosphere., (© 2023. The Author(s), under exclusive licence to Springer Nature Limited.)

Background: The COVID-19 pandemic is affecting nations globally, but with an impact exhibiting significant spatial and temporal variation at the sub-national level. Identifying and disentangling the drivers of resulting hospitalisation incidence at the local scale is key to predict, mitigate and manage epidemic surges, but also to develop targeted measures. However, this type of analysis is often not possible because of the lack of spatially-explicit health data and spatial uncertainties associated with infection., Methods: To overcome these limitations, we propose an analytical framework to investigate potential drivers of the spatio-temporal heterogeneity in COVID-19 hospitalisation incidence when data are only available at the hospital level. Specifically, the approach is based on the delimitation of hospital catchment areas, which allows analysing associations between hospitalisation incidence and spatial or temporal covariates. We illustrate and apply our analytical framework to Belgium, a country heavily impacted by two COVID-19 epidemic waves in 2020, both in terms of mortality and hospitalisation incidence., Results: Our spatial analyses reveal an association between the hospitalisation incidence and the local density of nursing home residents, which confirms the important impact of COVID-19 in elderly communities of Belgium. Our temporal analyses further indicate a pronounced seasonality in hospitalisation incidence associated with the seasonality of weather variables. Taking advantage of these associations, we discuss the feasibility of predictive models based on machine learning to predict future hospitalisation incidence., Conclusion: Our reproducible analytical workflow allows performing spatially-explicit analyses of data aggregated at the hospital level and can be used to explore potential drivers and dynamic of COVID-19 hospitalisation incidence at regional or national scales.

The concept of feedback is key in assessing whether a perturbation to a system is amplified or damped by mechanisms internal to the system. In polar regions, climate dynamics are controlled by both radiative and non-radiative interactions between the atmosphere, ocean, sea ice, ice sheets and land surfaces. Precisely quantifying polar feedbacks is required for a process-oriented evaluation of climate models, a clear understanding of the processes responsible for polar climate changes, and a reduction in uncertainty associated with model projections. This quantification can be performed using a simple and consistent approach that is valid for a wide range of feedbacks, offering the opportunity for more systematic feedback analyses and a better understanding of polar climate changes.

Observational estimates of the climate system are essential to monitoring and understanding ongoing climate change and to assessing the quality of climate models used to produce near- and long-term climate information. This study poses the dual and unconventional question: Can climate models be used to assess the quality of observational references? We show that this question not only rests on solid theoretical grounds but also offers insightful applications in practice. By comparing four observational products of sea surface temperature with a large multimodel climate forecast ensemble, we find compelling evidence that models systematically score better against the most recent, advanced, but also most independent product. These results call for generalized procedures of model-observation comparison and provide guidance for a more objective observational data set selection., (Copyright © 2016, American Association for the Advancement of Science.)