Search for: All records

Revolutionary advances in artificial intelligence (AI) in the past decade have brought transformative innovation across science and engineering disciplines. In the field of Arctic science, we have witnessed an increasing trend in the adoption of AI, especially deep learning, to support the analysis of Arctic big data and facilitate new discoveries. In this paper, we provide a comprehensive review of the applications of deep learning in sea ice remote sensing domains, focusing on problems such as sea ice lead detection, thickness estimation, sea ice concentration and extent forecasting, motion detection, and sea ice type classification. In addition to discussing these applications, we also summarize technological advances that provide customized deep learning solutions, including new loss functions and learning strategies to better understand sea ice dynamics. To promote the growth of this exciting interdisciplinary field, we further explore several research areas where the Arctic sea ice community can benefit from cutting-edge AI technology. These areas include improving multimodal deep learning capabilities, enhancing model accuracy in measuring prediction uncertainty, better leveraging AI foundation models, and deepening integration with physics-based models. We hope that this paper can serve as a cornerstone in the progress of Arctic sea ice research using AI and inspire further advances in this field. 

Abstract The exceptional atmospheric conditions that have accelerated Greenland Ice Sheet mass loss in recent decades have been repeatedly recognized as a possible dynamical response to Arctic amplification. Here, we present evidence of two potentially synergistic mechanisms linking high-latitude warming to the observed increase in Greenland blocking. Consistent with a prominent hypothesis associating Arctic amplification and persistent weather extremes, we show that the summer atmospheric circulation over the North Atlantic has become wavier and link this wavier flow to more prevalent Greenland blocking. While a concomitant decline in terrestrial snow cover has likely contributed to this mechanism by further amplifying warming at high latitudes, we also show that there is a direct stationary Rossby wave response to low spring North American snow cover that enforces an anomalous anticyclone over Greenland, thus helping to anchor the ridge over Greenland in this wavier atmospheric state. 

This dataset contains output from a prescribed model experiment conducted to investigate the impact of snow cover loss over North America on summer atmospheric circulation. We utilized the National Center for Atmospheric Research’s Community Earth System Model version 2.2 to complete a 10-year control simulation. We then modified the land-surface restart files for May 1st of each year of the control period by reducing the snow cover over North America to zero. Using these modified files, we then completed a reduced snow simulation by rerunning three-month simulations from May through July for each of the ten years. This dataset contains both the 10-year control simulation as well as the May–July “no-snow” simulations for each year. More details about the experimental setup and example output can be found in the following publication: Preece, J.R., Mote, T.L., Cohen, J. et al. Summer atmospheric circulation over Greenland in response to Arctic amplification and diminished spring snow cover. Nat Commun 14, 3759 (2023). https://doi.org/10.1038/s41467-023-39466-6 

Abstract. Surface mass loss from the Greenland ice sheet (GrIS) hasaccelerated over the past decades, mainly due to enhanced surface meltingand liquid water runoff in response to atmospheric warming. A large portionof runoff from the GrIS originates from exposure of the darker bare ice inthe ablation zone when the overlying snow melts, where surface albedo playsa critical role in modulating the energy available for melting. In thisregard, it is imperative to understand the processes governing albedovariability to accurately project future mass loss from the GrIS. Bare-icealbedo is spatially and temporally variable and contingent on non-linearfeedbacks and the presence of light-absorbing constituents. An assessment ofmodels aiming at simulating albedo variability and associated impacts onmeltwater production is crucial for improving our understanding of theprocesses governing these feedbacks and, in turn, surface mass loss fromGreenland. Here, we report the results of a comparison of the bare-iceextent and albedo simulated by the regional climate model ModèleAtmosphérique Régional (MAR) with satellite imagery from theModerate Resolution Imaging Spectroradiometer (MODIS) for the GrIS below70∘ N. Our findings suggest that MAR overestimates bare-ice albedoby 22.8 % on average in this area during the 2000–2021 period with respectto the estimates obtained from MODIS. Using an energy balance model toparameterize meltwater production, we find this bare-ice albedo bias canlead to an underestimation of total meltwater production from the bare-icezone below 70∘ N of 42.8 % during the summers of 2000–2021. 

Abstract. Understanding the role of atmospheric circulation anomalies on the surfacemass balance of the Greenland ice sheet (GrIS) is fundamental for improvingestimates of its current and future contributions to sea level rise. Here,we show, using a combination of remote sensing observations, regionalclimate model outputs, reanalysis data, and artificial neural networks, thatunprecedented atmospheric conditions (1948–2019) occurring in the summerof 2019 over Greenland promoted new record or close-to-record values ofsurfacemass balance (SMB), runoff, and snowfall. Specifically, runoff in 2019 ranked second withinthe 1948–2019 period (after 2012) and first in terms of surface massbalance negative anomaly for the hydrological year 1 September 2018–31 August 2019. The summer of 2019 was characterized by an exceptionalpersistence of anticyclonic conditions that, in conjunction with low albedoassociated with reduced snowfall in summer, enhanced the melt–albedofeedback by promoting the absorption of solar radiation and favoredadvection of warm, moist air along the western portion of the ice sheettowards the north, where the surface melt has been the highest since 1948.The analysis of the frequency of daily 500 hPa geopotential heights obtainedfrom artificial neural networks shows that the total number of days with thefive most frequent atmospheric patterns that characterized the summer of2019 was 5 standard deviations above the 1981–2010 mean, confirming theexceptional nature of the 2019 season over Greenland. 

Abstract. Light transmission into bare glacial ice affects surfaceenergy balance, biophotochemistry, and light detection and ranging (lidar)laser elevation measurements but has not previously been reported for theGreenland Ice Sheet. We present measurements of spectral transmittance at350–900 nm in bare glacial ice collected at a field site in the westernGreenland ablation zone (67.15∘ N, 50.02∘ W). Empirical irradianceattenuation coefficients at 350–750 nm are ∼ 0.9–8.0 m−1 for ice at 12–124 cm depth. The absorption minimum is at∼ 390–397 nm, in agreement with snow transmissionmeasurements in Antarctica and optical mapping of deep ice at the SouthPole. From 350–530 nm, our empirical attenuation coefficients are nearly1 order of magnitude larger than theoretical values for optically pureice. The estimated absorption coefficient at 400 nm suggests the ice volumecontained a light-absorbing particle concentration equivalent to∼ 1–2 parts per billion (ppb) of black carbon, which is similar topre-industrial values found in remote polar snow. The equivalent mineraldust concentration is ∼ 300–600 ppb, which is similar to values forNorthern Hemisphere warm periods with low aeolian activity inferred from icecores. For a layer of quasi-granular white ice (weathering crust)extending from the surface to ∼ 10 cm depth, attenuationcoefficients are 1.5 to 4 times larger than for deeper bubbly ice. Owing tohigher attenuation in this layer of near-surface granular ice, opticalpenetration depth at 532 nm is 14 cm (20 %) lower than asymptoticattenuation lengths for optically pure bubbly ice. In addition to thetraditional concept of light scattering on air bubbles, our results implythat the granular near-surface ice microstructure of weathering crust isan important control on radiative transfer in bare ice on the Greenland IceSheet ablation zone, and we provide new values of flux attenuation,absorption, and scattering coefficients to support model development andvalidation. 

Abstract We investigate wintertime extreme sea ice loss events on synoptic to subseasonal time scales over the Barents–Kara Sea, where the largest sea ice variability is located. Consistent with previous studies, extreme sea ice loss events are associated with moisture intrusions over the Barents–Kara Sea, which are driven by the large-scale atmospheric circulation. In addition to the role of downward longwave radiation associated with moisture intrusions, which is emphasized by previous studies, our analysis shows that strong turbulent heat fluxes are associated with extreme sea ice melting events, with both turbulent sensible and latent heat fluxes contributing, although turbulent sensible heat fluxes dominate. Our analysis also shows that these events are connected to tropical convective anomalies. A dipole pattern of convective anomalies with enhanced convection over the Maritime Continent and suppressed convection over the central to eastern Pacific is consistently detected about 6–10 days prior to extreme sea ice loss events. This pattern is associated with either the Madden–Julian oscillation (MJO) or El Niño–Southern Oscillation (ENSO). Composites show that extreme sea ice loss events are connected to tropical convection via Rossby wave propagation in the midlatitudes. However, tropical convective anomalies alone are not sufficient to trigger extreme sea ice loss events, suggesting that extratropical variability likely modulates the connection between tropical convection and extreme sea ice loss events. 

The SUMup database is a compilation of surface mass balance (SMB), subsurface temperature and density measurements from the Greenland and Antarctic ice sheets. This 2023 release contains 4 490 442 data points: 1 778 540 SMB measurements, 2 706 413 density measurements and 5 489 subsurface temperature measurements. This is respectively 1 477 132, 420 825 and 4 715 additional observations of SMB, density and temperature compared to the 2022 release. This new release provides not only snow accumulation on ice sheets, like its predecessors, but all types of SMB measurements, including from ablation areas. On the other hand, snow depth on sea ice is discontinued, but can still be found in the previous releases. The data files are provided in both CSV and NetCDF format and contain, for each measurement, the following metadata: latitude, longitude, elevation, timestamp, method, reference of the data source and, when applicable, the name of the measurement group it belongs to (core name for SMB, profile name for density, station name for temperature). Data users are encouraged to cite all the original data sources that are being used. Issues about this release as well as suggestions of datasets to be added in next releases can be done on a dedicated user forum: https://github.com/SUMup-database/SUMup-data-suggestion/issues. Example scripts to use the SUMup 2023 files are made available on our script repository: https://github.com/SUMup-database/SUMup-example-scripts.