Arctic sea ice cover data from spaceborne SAR by deep learning

Mapping Intimacies ◽

10.5194/essd-2020-332 ◽

2020 ◽

Author(s):

Yi-Ran Wang ◽

Xiao-Ming Li

Keyword(s):

Deep Learning ◽

High Resolution ◽

Sea Ice ◽

Spatial Resolution ◽

Ice Cover ◽

Arctic Sea Ice ◽

The Arctic ◽

Absolute Difference ◽

Sea Ice Concentration ◽

Ice Concentration

Abstract. Widely used sea ice concentration and sea ice cover in polar regions are derived mainly from spaceborne microwave radiometer and scatterometer data, and the typical spatial resolution of these products ranges from several to dozens of kilometers. Due to dramatic changes in polar sea ice, high-resolution sea ice cover data are drawing increasing attention for polar navigation, environmental research, and offshore operations. In this paper, we focused on developing an approach for deriving a high-resolution sea ice cover product for the Arctic using Sentinel-1 (S1) dual-polarization (horizontal-horizontal, HH, and horizontal-vertical, HV) data in extra wide swath (EW) mode. The approach for discriminating sea ice from open water by synthetic aperture radar (SAR) data is based on a modified U-Net architecture, a deep learning network. By employing an integrated stacking model to combine multiple U-Net classifiers with diverse specializations, sea ice segmentation is achieved with superior accuracy over any individual classifier. We applied the proposed approach to over 28,000 S1 EW images acquired in 2019 to obtain sea ice cover products in a high spatial resolution of 400 m. By converting the S1-derived sea ice cover to concentration and then compared with Advanced Microwave Scanning Radiometer 2 (AMSR2) sea ice concentration data, showing an average absolute difference of 5.55 % with seasonal fluctuations. A direct comparison with Interactive Multisensor Snow and Ice Mapping System (IMS) daily sea ice cover data achieves an average accuracy of 93.98 %. These results show that the developed S1-derived sea ice cover results are comparable to the AMSR and IMS data in terms of overall accuracy but superior to these data in presenting detailed sea ice cover information, particularly in the marginal ice zone (MIZ). Data are available at: https://doi.org/10.11922/sciencedb.00273 (Wang and Li, 2020).

Download Full-text

Arctic sea ice cover data from spaceborne synthetic aperture radar by deep learning

Earth System Science Data ◽

10.5194/essd-13-2723-2021 ◽

2021 ◽

Vol 13 (6) ◽

pp. 2723-2742

Author(s):

Yi-Ran Wang ◽

Xiao-Ming Li

Keyword(s):

Deep Learning ◽

High Resolution ◽

Sea Ice ◽

Synthetic Aperture Radar ◽

Spatial Resolution ◽

Ice Cover ◽

Synthetic Aperture ◽

Sea Ice Concentration ◽

Ice Concentration ◽

Aperture Radar

Abstract. Widely used sea ice concentration and sea ice cover in polar regions are derived mainly from spaceborne microwave radiometer and scatterometer data, and the typical spatial resolution of these products ranges from several to dozens of kilometers. Due to dramatic changes in polar sea ice, high-resolution sea ice cover data are drawing increasing attention for polar navigation, environmental research, and offshore operations. In this paper, we focused on developing an approach for deriving a high-resolution sea ice cover product for the Arctic using Sentinel-1 (S1) dual-polarization (horizontal-horizontal, HH, and horizontal-vertical, HV) data in extra wide swath (EW) mode. The approach for discriminating sea ice from open water by synthetic aperture radar (SAR) data is based on a modified U-Net architecture, a deep learning network. By employing an integrated stacking model to combine multiple U-Net classifiers with diverse specializations, sea ice segmentation is achieved with superior accuracy over any individual classifier. We applied the proposed approach to over 28 000 S1 EW images acquired in 2019 to obtain sea ice cover products in a high spatial resolution of 400 m. The validation by 96 cases of visual interpretation results shows an overall accuracy of 96.10 %. The S1-derived sea ice cover was converted to concentration and then compared with Advanced Microwave Scanning Radiometer 2 (AMSR2) sea ice concentration data, showing an average absolute difference of 5.55 % with seasonal fluctuations. A direct comparison with Interactive Multisensor Snow and Ice Mapping System (IMS) daily sea ice cover data achieves an average accuracy of 93.98 %. These results show that the developed S1-derived sea ice cover results are comparable to the AMSR and IMS data in terms of overall accuracy but superior to these data in presenting detailed sea ice cover information, particularly in the marginal ice zone (MIZ). Data are available at https://doi.org/10.11922/sciencedb.00273 (Wang and Li, 2020).

Download Full-text

Ice Concentration Retrieval from the Analysis of Microwaves: A New Methodology Designed for the Copernicus Imaging Microwave Radiometer

Remote Sensing ◽

10.3390/rs12071060 ◽

2020 ◽

Vol 12 (7) ◽

pp. 1060 ◽

Cited By ~ 1

Author(s):

Lise Kilic ◽

Catherine Prigent ◽

Filipe Aires ◽

Georg Heygster ◽

Victor Pellet ◽

...

Keyword(s):

Sea Ice ◽

Spatial Resolution ◽

Microwave Radiometer ◽

Optimal Estimation ◽

Arctic Sea Ice ◽

The Arctic ◽

Special Focus ◽

Polar Regions ◽

Sea Ice Concentration ◽

Ice Concentration

Over the last 25 years, the Arctic sea ice has seen its extent decline dramatically. Passive microwave observations, with their ability to penetrate clouds and their independency to sunlight, have been used to provide sea ice concentration (SIC) measurements since the 1970s. The Copernicus Imaging Microwave Radiometer (CIMR) is a high priority candidate mission within the European Copernicus Expansion program, with a special focus on the observation of the polar regions. It will observe at 6.9 and 10.65 GHz with 15 km spatial resolution, and at 18.7 and 36.5 GHz with 5 km spatial resolution. SIC algorithms are based on empirical methods, using the difference in radiometric signatures between the ocean and sea ice. Up to now, the existing algorithms have been limited in the number of channels they use. In this study, we proposed a new SIC algorithm called Ice Concentration REtrieval from the Analysis of Microwaves (IceCREAM). It can accommodate a large range of channels, and it is based on the optimal estimation. Linear relationships between the satellite measurements and the SIC are derived from the Round Robin Data Package of the sea ice Climate Change Initiative. The 6 and 10 GHz channels are very sensitive to the sea ice presence, whereas the 18 and 36 GHz channels have a better spatial resolution. A data fusion method is proposed to combine these two estimations. Therefore, IceCREAM will provide SIC estimates with the good accuracy of the 6+10GHz combination, and the high spatial resolution of the 18+36GHz combination.

Download Full-text

Deep learning for monthly Arctic sea ice concentration prediction

10.5194/egusphere-egu2020-15481 ◽

2020 ◽

Author(s):

Tom Andersson ◽

Fruzsina Agocs ◽

Scott Hosking ◽

María Pérez-Ortiz ◽

Brooks Paige ◽

...

Keyword(s):

Climate Change ◽

Global Warming ◽

Deep Learning ◽

Sea Ice ◽

Arctic Sea Ice ◽

The Arctic ◽

Prediction System ◽

Sea Ice Concentration ◽

Arctic Sea ◽

Ice Concentration

Over recent decades, the Arctic has warmed faster than any region on Earth. The rapid decline in Arctic sea ice extent (SIE) is often highlighted as a key indicator of anthropogenic climate change. Changes in sea ice disrupt Arctic wildlife and indigenous communities, and influence weather patterns as far as the mid-latitudes. Furthermore, melting sea ice attenuates the albedo effect by replacing the white, reflective ice with dark, heat-absorbing melt ponds and open sea, increasing the Sun&#8217;s radiative heat input to the Arctic and amplifying global warming through a positive feedback loop. Thus, the reliable prediction of sea ice under a changing climate is of both regional and global importance. However, Arctic sea ice presents severe modelling challenges due to its complex coupled interactions with the ocean and atmosphere, leading to high levels of uncertainty in numerical sea ice forecasts.Deep learning (a subset of machine learning) is a family of algorithms that use multiple nonlinear processing layers to extract increasingly high-level features from raw input data. Recent advances in deep learning techniques have enabled widespread success in diverse areas where significant volumes of data are available, such as image recognition, genetics, and online recommendation systems. Despite this success, and the presence of large climate datasets, applications of deep learning in climate science have been scarce until recent years. For example, few studies have posed the prediction of Arctic sea ice in a deep learning framework. We investigate the potential of a fully data-driven, neural network sea ice prediction system based on satellite observations of the Arctic. In particular, we use inputs of monthly-averaged sea ice concentration (SIC) maps since 1979 from the National Snow and Ice Data Centre, as well as climatological variables (such as surface pressure and temperature) from the European Centre for Medium-Range Weather Forecasts reanalysis (ERA5) dataset. Past deep learning-based Arctic sea ice prediction systems tend to overestimate sea ice in recent years - we investigate the potential to learn the non-stationarity induced by climate change with the inclusion of multi-decade global warming indicators (such as average Arctic air temperature). We train the networks to predict SIC maps one month into the future, evaluating network prediction uncertainty by ensembling independent networks with different random weight initialisations. Our model accounts for seasonal variations in the drivers of sea ice by controlling for the month of the year being predicted. We benchmark our prediction system against persistence, linear extrapolation and autoregressive models, as well as September minimum SIE predictions from submissions to the Sea Ice Prediction Network's Sea Ice Outlook. Performance is evaluated quantitatively using the root mean square error and qualitatively by analysing maps of prediction error and uncertainty.

Download Full-text

Short-Term Daily Prediction of Sea Ice Concentration Based on Deep Learning of Gradient Loss Function

Frontiers in Marine Science ◽

10.3389/fmars.2021.736429 ◽

2021 ◽

Vol 8 ◽

Author(s):

Quanhong Liu ◽

Ren Zhang ◽

Yangjun Wang ◽

Hengqian Yan ◽

Mei Hong

Keyword(s):

Deep Learning ◽

Sea Ice ◽

Loss Function ◽

Model Performance ◽

Arctic Sea Ice ◽

The Arctic ◽

Short Term ◽

Sea Ice Concentration ◽

Ice Concentration ◽

The Impact

The navigability potential of the Northeast Passage has gradually emerged with the melting of Arctic sea ice. For the purpose of navigation safety in the Arctic area, a reliable daily sea ice concentration (SIC) prediction result is required. As the mature application of deep learning technique in short-term prediction of other fields (atmosphere, ocean, and hurricane, etc.), a new model was proposed for daily SIC prediction by selecting multiple factors, adopting gradient loss function (Grad-loss) and incorporating an improved predictive recurrent neural network (PredRNN++). Three control experiments are designed to test the impact of these three improvements for model performance with multiple indicators. Results show that the proposed model has best prediction skill in our experiments by taking physical process and local SIC variation into consideration, which can continuously predict daily SIC for up to 9 days.

Download Full-text

Cryospheric Impacts of Soviet River Diversion Schemes

Annals of Glaciology ◽

10.3189/1984aog5-1-61-68 ◽

1984 ◽

Vol 5 ◽

pp. 61-68 ◽

Cited By ~ 4

Author(s):

T. Holt ◽

P. M. Kelly ◽

B. S. G. Cherry

Keyword(s):

Sea Ice ◽

Arctic Ocean ◽

River Discharge ◽

Large Scale ◽

Arctic Sea Ice ◽

The Arctic ◽

Sea Ice Concentration ◽

Ice Concentration ◽

The Arctic Ocean ◽

The Impact

Soviet plans to divert water from rivers flowing into the Arctic Ocean have led to research into the impact of a reduction in discharge on Arctic sea ice. We consider the mechanisms by which discharge reductions might affect sea-ice cover and then test various hypotheses related to these mechanisms. We find several large areas over which sea-ice concentration correlates significantly with variations in river discharge, supporting two particular hypotheses. The first hypothesis concerns the area where the initial impacts are likely to which is the Kara Sea. Reduced riverflow is associated occur, with decreased sea-ice concentration in October, at the time of ice formation. This is believed to be the result of decreased freshening of the surface layer. The second hypothesis concerns possible effects on the large-scale current system of the Arctic Ocean and, in particular, on the inflow of Atlantic and Pacific water. These effects occur as a result of changes in the strength of northward-flowing gradient currents associated with variations in river discharge. Although it is still not certain that substantial transfers of riverflow will take place, it is concluded that the possibility of significant cryospheric effects and, hence, large-scale climate impact should not be neglected.

Download Full-text

Changes of the Arctic marginal ice zone during the satellite era

The Cryosphere ◽

10.5194/tc-14-1971-2020 ◽

2020 ◽

Vol 14 (6) ◽

pp. 1971-1984 ◽

Cited By ~ 1

Author(s):

Rebecca J. Rolph ◽

Daniel L. Feltham ◽

David Schröder

Keyword(s):

Sea Ice ◽

Arctic Sea Ice ◽

Satellite Observations ◽

The Arctic ◽

Upper And Lower Bounds ◽

Sea Ice Concentration ◽

Sea Ice Extent ◽

Ice Concentration ◽

Marginal Ice Zone ◽

Definition Of

Abstract. Many studies have shown a decrease in Arctic sea ice extent. It does not logically follow, however, that the extent of the marginal ice zone (MIZ), here defined as the area of the ocean with ice concentrations from 15 % to 80 %, is also changing. Changes in the MIZ extent has implications for the level of atmospheric and ocean heat and gas exchange in the area of partially ice-covered ocean and for the extent of habitat for organisms that rely on the MIZ, from primary producers like sea ice algae to seals and birds. Here, we present, for the first time, an analysis of satellite observations of pan-Arctic averaged MIZ extent. We find no trend in the MIZ extent over the last 40 years from observations. Our results indicate that the constancy of the MIZ extent is the result of an observed increase in width of the MIZ being compensated for by a decrease in the perimeter of the MIZ as it moves further north. We present simulations from a coupled sea ice–ocean mixed layer model using a prognostic floe size distribution, which we find is consistent with, but poorly constrained by, existing satellite observations of pan-Arctic MIZ extent. We provide seasonal upper and lower bounds on MIZ extent based on the four satellite-derived sea ice concentration datasets used. We find a large and significant increase (>50 %) in the August and September MIZ fraction (MIZ extent divided by sea ice extent) for the Bootstrap and OSI-450 observational datasets, which can be attributed to the reduction in total sea ice extent. Given the results of this study, we suggest that references to “rapid changes” in the MIZ should remain cautious and provide a specific and clear definition of both the MIZ itself and also the property of the MIZ that is changing.

Download Full-text

Gridded Arctic sea ice concentration reconstruction for the first half of the 20th century based on different proxy data

10.5194/egusphere-egu21-3106 ◽

2021 ◽

Author(s):

Vladimir Semenov ◽

Tatiana Matveeva

Keyword(s):

Sea Ice ◽

20Th Century ◽

Climate Model ◽

Surface Air Temperature ◽

Arctic Sea Ice ◽

The Arctic ◽

Sea Ice Concentration ◽

The North ◽

Arctic Sea ◽

Ice Concentration

Global warming in the recent decades has been accompanied by a rapid recline of the Arctic sea ice area most pronounced in summer (10% per decade). To understand the relative contribution of external forcing and natural variability to the modern and future sea ice area changes, it is necessary to evaluate a range of long-term variations of the Arctic sea ice area in the period before a significant increase in anthropogenic emissions of greenhouse gases into the atmosphere. Available observational data on the spatiotemporal dynamics of Arctic sea ice until 1950s are characterized by significant gaps and uncertainties. In the recent years, there have appeared several reconstructions of the early 20th century Arctic sea ice area that filled the gaps by analogue methods or utilized combined empirical data and climate model&#8217;s output. All of them resulted in a stronger that earlier believed negative sea ice area anomaly in the 1940s concurrent with the early 20th century warming (ETCW) peak. In this study, we reconstruct the monthly average gridded sea ice concentration (SIC) in the first half of the 20th century using the relationship between the spatiotemporal features of SIC variability, surface air temperature over the Northern Hemisphere extratropical continents, sea surface temperature in the North Atlantic and North Pacific, and sea level pressure. In agreement with a few previous results, our reconstructed data also show a significant negative anomaly of the Arctic sea ice area in the middle of the 20th century, however with some 15% to 30% stronger amplitude, about 1.5 million km2 in September and 0.7 million km2 in March. The reconstruction demonstrates a good agreement with regional Arctic sea ice area data when available and suggests that ETWC in the Arctic has been accompanied by a concurrent sea ice area decline of a magnitude that have been exceeded only in the beginning of the 21st century.

Download Full-text

Convolutional Neural Networks for Sea Ice Concentration Charting for Maritime Navigation in the Arctic

10.5194/egusphere-egu21-15193 ◽

2021 ◽

Author(s):

Andreas Stokholm ◽

Leif Pedersen ◽

René Forsberg ◽

Sine Hvidegaard

Keyword(s):

Sea Ice ◽

Arctic Sea Ice ◽

The Arctic ◽

Arctic Seas ◽

Sea Ice Concentration ◽

Data Set ◽

Architecture Model ◽

Radar Images ◽

North East ◽

Ice Concentration

In recent years the Arctic has seen renewed political and economic interest, increased maritime traffic and desire for improved sea ice navigational tools. Despite a rise in digital technology, maps of sea ice concentration used for Arctic maritime operations are still today created by humans manually interpreting radar images. This process is slow with low map release frequency, uncertainties up to 20 % and discrepancies up to 60 %. Utilizing emerging AI Convolutional Neural Network (CNN) semantic image segmentation techniques to automate this process is drastically changing navigation in the Arctic seas, with better resolution, accuracy, release frequency and coverage. Automatic Arctic sea ice products may contribute to enabling the disruptive Northern Sea Route connecting North East Asia to Europe via the Arctic oceans.The AI4Arctic/ASIP V2 data set, that combines 466 Sentinel-1 HH and HV SAR images from Greenland, Passive Microwave Radiometry from the AMSR2 instrument, and an equivalent sea ice concentration chart produced by ice analysts at the Danish Meteorological Institute, have been used to train a CNN U-Net Architecture model. The model shows robust capabilities in producing highly detailed sea ice concentration maps with open water, intermediate sea ice concentrations as well as full sea ice cover, which resemble those created by professional sea ice analysts. Often cited obstacles in automatic sea ice concentration models are wind-roughened sea ambiguities resembling sea ice. Final inference scenes show robustness towards such ambiguities.

Download Full-text