Search

Sea ice type classification is of great significance for the exploration of waterways, fisheries, and offshore operations in the Arctic. However, to date, there is no multiple remote sensing method to detect sea ice type in the Arctic. This study develops a multiple sea ice type algorithm using the HaiYang-2B Scatterometer (HY-2B SCA). First, the parameters most applicable to classify sea ice type are selected through feature extraction, and a stacking model is established for the first time, which integrates decision tree and image segmentation algorithms. Finally, multiple sea ice types are classified in the Arctic, comprising Nilas, Young Ice, First Year Ice, Old Ice, and Fast Ice. Comparing the results with the Ocean and Sea Ice Satellite Application Facility (OSI-SAF) Sea Ice Type dataset (SIT) indicates that the sea ice type classified by HY-2B SCA (Stacking-HY2B) is similar to OSI-SAF SIT with regard to the changing trends in extent of sea ice. We use the Copernicus Marine Environment Monitoring Service (CMEMS) high-resolution sea ice type data and EM-Bird ice thickness data to validate the result, and accuracies of 87% and 88% are obtained, respectively. This indicates that the algorithm in this work is comparable with the performance of OSI-SAF dataset, while providing information of multiple sea ice types. [ABSTRACT FROM AUTHOR]

In this paper, a Bayesian sea ice detection algorithm is first used based on the HY-2A/SCAT data, and a backpropagation (BP) neural network is used to classify the Arctic sea ice type. During the implementation of the Bayesian sea ice detection algorithm, linear sea ice model parameters and the backscatter variance suitable for HY-2A/SCAT were proposed. The sea ice extent obtained by the Bayesian sea ice detection algorithm was projected on a 12.5 km grid sea ice map and validated by the Advanced Microwave Scanning Radiometer 2 (AMSR2) 15% sea ice concentration data. The sea ice extent obtained by the Bayesian sea ice detection algorithm was found to be in good agreement with that of the AMSR2 during the ice growth season. Meanwhile, the Bayesian sea ice detection algorithm gave a wider ice edge than the AMSR2 during the ice melting season. For the sea ice type classification, the BP neural network was used to classify the Arctic sea ice type (multi-year and first-year ice) from January to May and October to December in 2014. Comparison results between the HY-2A/SCAT sea ice type and Equal-Area Scalable Earth Grid (EASE-Grid) sea ice age data showed that the HY-2A/SCAT multi-year ice extent variation had the same trend as the EASE-Grid data. Classification errors, defined as the ratio of the mismatched sea ice type points between HY-2A/SCAT and EASE-Grid to the total sea ice points, were less than 12%, and the average classification error was 8.6% for the study period, which indicated that the BP neural network classification was a feasible algorithm for HY-2A/SCAT sea ice type classification. [ABSTRACT FROM AUTHOR]

Sea ice type classification is of great significance for the exploration of waterways, fisheries, and offshore operations in the Arctic. However, to date, there is no multiple remote sensing method to detect sea ice type in the Arctic. This study develops a multiple sea ice type algorithm using the HaiYang-2B Scatterometer (HY-2B SCA). First, the parameters most applicable to classify sea ice type are selected through feature extraction, and a stacking model is established for the first time, which integrates decision tree and image segmentation algorithms. Finally, multiple sea ice types are classified in the Arctic, comprising Nilas, Young Ice, First Year Ice, Old Ice, and Fast Ice. Comparing the results with the Ocean and Sea Ice Satellite Application Facility (OSI-SAF) Sea Ice Type dataset (SIT) indicates that the sea ice type classified by HY-2B SCA (Stacking-HY2B) is similar to OSI-SAF SIT with regard to the changing trends in extent of sea ice. We use the Copernicus Marine Environment Monitoring Service (CMEMS) high-resolution sea ice type data and EM-Bird ice thickness data to validate the result, and accuracies of 87% and 88% are obtained, respectively. This indicates that the algorithm in this work is comparable with the performance of OSI-SAF dataset, while providing information of multiple sea ice types.

Information on sea ice type is an important factor for deriving sea ice parameters from satellite remote sensing data, such as sea ice concentration, extent and thickness. In this study, sea ice in the Weddell Sea was classified by the histogram threshold (HT) method, the Spreen model (SM) method from satellite scatterometer data and the strong contrast (SC) method from radiometer data, and this information was compared with Antarctic Sea Ice Processes and Climate (ASPeCt) sea ice-type ship-based observations. The results show that all three methods can distinguish the multi-year (MY) ice and first-year (FY) ice using Ku-band scatterometer data and radiometer data during the ice growth season, while C-band scatterometer data are not suitable for MY ice and FY ice discrimination using HT and SM methods. The SM model has a smaller MY ice classification extent than the HT method from scatterometer data. The classification accuracy of the SM method is the higher compared to ship-based observations. It can be concluded that the SM method is a promising method for discriminating MY ice from FY ice. These results provide a reference for further retrieval of long-term sea ice-type information for the whole of Antarctica. [ABSTRACT FROM AUTHOR]

A new algorithm for classification of sea ice types on Sentinel-1 Synthetic Aperture Radar (SAR) data using a convolutional neural network (CNN) is presented. The CNN is trained on reference ice charts produced by human experts and compared with an existing machine learning algorithm based on texture features and random forest classifier. The CNN is trained on two datasets in 2018 and 2020 for retrieval of four classes: ice free, young ice, first-year ice and old ice. The accuracy of our classification is 90.5% for the 2018-dataset and 91.6% for the 2020-dataset. The uncertainty is a bit higher for young ice (85%/76% accuracy in 2018/2020) and first-year ice (86%/84% accuracy in 2018/2020). Our algorithm outperforms the existing random forest product for each ice type. It has also proved to be more efficient in computing time and less sensitive to the noise in SAR data. The code is publicly available.

Sea ice type is one of the most sensitive variables in Arctic ice monitoring and detailed information about it is essential for ice situation evaluation, vessel navigation, and climate prediction. Many machine-learning methods including deep learning can be employed for ice-type detection, and most classifiers tend to prefer different feature combinations. In order to find the optimal classifier–feature assembly (OCF) for sea ice classification, it is necessary to assess their performance differences. The objective of this letter is to make a recommendation for the OCF for sea ice classification using Cryosat-2 (CS-2) data. Six classifiers including convolutional neural network (CNN), Bayesian, $K$ nearest-neighbor (KNN), support vector machine (SVM), random forest (RF), and back propagation neural network (BPNN) were studied. CS-2 altimeter data of November 2015 and May 2016 in the whole Arctic were used. The overall accuracy was estimated using multivalidation to evaluate the performances of individual classifiers with different feature combinations. Overall, RF achieved a mean accuracy of 89.15%, followed by Bayesian, SVM, and BPNN (~86%), outperforming the worst (CNN and KNN) by 7%. Trailing-edge width (TeW) and leading-edge width (LeW) were the most important features, and feature combination of TeW, LeW, Sigma0, maximum of the returned power waveform (MAX), and pulse peakiness (PP) was the best choice. RF with feature combination of TeW, LeW, Sigma0, MAX, and PP was finally selected as the OCF for sea ice classification and the results that demonstrated this method achieved a mean accuracy of 91.45%, which outperformed the other state-of-art methods by 9%. [ABSTRACT FROM PUBLISHER]

A sea ice detection algorithm based on Fisher's linear discriminant analysis is developed to segment sea ice and open water for the Ku-band scatterometer onboard the China's Hai Yang 2A Satellite (HY-2A/SCAT). Residual classification errors are reduced through image erosion/dilation techniques and sea ice growth/retreat constraint methods. The arctic sea-ice-type classification is estimated via a time-dependent threshold derived from the annual backscatter trends based on previous HY-2A/SCAT derived sea ice extent. The extent and edge of the sea ice obtained in this study is compared with the Special Sensor Microwave Imager/Sounder (SSMIS) sea ice concentration data and the Sentinel-1 SAR imagery for verification, respectively. Meanwhile, the classified sea ice type is compared with a multi-sensor sea ice type product based on data from the Advanced Scatterometer (ASCAT) and SSMIS. Results show that HY-2A/SCAT is powerful in providing sea ice extent and type information, while differences in the sensitivities of active/passive products are found. In addition, HY-2A/SCAT derived sea ice products are also proved to be valuable complements for existing polar sea ice data products. [ABSTRACT FROM AUTHOR]

Based on radar penetrating measurements and analysis of sea ice in the Arctic Ocean, the potential of radar wave to measure sea ice thickness and map the morphology of the underside of sea ice is investigated. The results indicate that the radar wave can penetrate Arctic summer sea ice of over 6 m in thickness; and the propagation velocity of the radar wave in sea ice is in the range of 0.142 m · ns−1 to 0.154 m · ns−1. The radar images display the roughness and micro-relief variation of sea ice bottom surface. These features are closely related to sea ice types, which show that radar survey may be used to identify and classify ice types. Since radar images can simultaneously display the linear profile features of both the upper surface and the underside of sea ice, we use these images to quantify their actual linear length discrepancy. A new length factor is suggested in relation to the actual linear length discrepancy in linear profiles of sea ice, which may be useful in the further study of the area difference between the upper surface and bottom surface of sea ice. [ABSTRACT FROM AUTHOR]

A new algorithm for classification of sea ice types on Sentinel-1 Synthetic Aperture Radar (SAR) data using a convolutional neural network (CNN) is presented. The CNN is trained on reference ice charts produced by human experts and compared with an existing machine learning algorithm based on texture features and random forest classifier. The CNN is trained on two datasets in 2018 and 2020 for retrieval of four classes: ice free, young ice, first-year ice and old ice. The accuracy of our classification is 90.5% for the 2018-dataset and 91.6% for the 2020-dataset. The uncertainty is a bit higher for young ice (85%/76% accuracy in 2018/2020) and first-year ice (86%/84% accuracy in 2018/2020). Our algorithm outperforms the existing random forest product for each ice type. It has also proved to be more efficient in computing time and less sensitive to the noise in SAR data. The code is publicly available. [ABSTRACT FROM AUTHOR]