Classification Of Sea Ice Types Research Articles

A Deep Learning Method for Arctic Sea Ice Type Classification Based on Active-Passive Microwave Data

The Arctic-scale classification of sea ice type is important in the fields of macro-monitoring in the Arctic, climate change assessment, and long time-series interannual variability of sea ice. To date, a lack of deep learning-based algorithms for detecting Arctic sea ice types from microwave sensors is comparatively rare. In this study, a sea ice type classification algorithm based on the U-Net-CBAM deep learning semantic segmentation network was developed by carrying out research on sea ice type classification algorithms in the Arctic through the integrated use of ASCAT scatterometer and AMSR2 radiometer data. The results of the study show that the proposed method achieves an impressive average accuracy of 94.1% in extracting sea ice categories using AARI ice maps. The results of the algorithm show a high degree of agreement with the OSI-SAF sea ice type daily products.

Multiple Sea Ice Type Retrieval Using the HaiYang-2B Scatterometer in the Arctic

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.

Arctic Sea Ice Type Classification by Combining CFOSCAT and AMSR‐2 Data

AbstractFirst‐year ice (FYI) and multi‐year ice (MYI) are the two most common ice types in the Arctic. In this article, the classification of FYI and MYI over the Arctic region in the winter of 2019/2020 and 2020/2021 is investigated by combining the data of the scatterometer on Chinese‐French Oceanography Satellite (CFOSCAT) and the Advanced Microwave Scanning Radiometer‐2 (AMSR‐2) based on the Tree Augmented Naive Bayes (TAN) classifier. The CFOSCAT/AMSR ice type classification results are validated by using Ocean and Sea Ice Satellite Application Facility (OSI SAF) sea ice type products, Canadian Ice Service ice charts, and synthetic‐aperture radar data. The results showed that the overall MYI extent change trend retrieved from CFOSCAT/AMSR was consistent with the OSI SAF product with a correlation coefficient of 0.89 for winter of 2019/2020 and 0.88 for 2020/2021. In addition, CFOSCAT/AMSR identified a slightly larger MYI extent than OSI SAF and the average deviation between them is 10.1% for 2019/2020 while 8.3% for 2020/2021. Besides, CFOSCAT/AMSR can identify more MYI pixels when the MYI concentration is relatively low in the Western Arctic region. We also used CFOSCAT data only to retrieve ice type and found that the active and passive microwave data fusion could capture more MYI pixels located near the boundary of MYI and FYI main body, and the introduction of AMSR‐2 data in ice type classification could reduce the error caused by the abnormal values of CFOSCAT parameters.

Classification of Sea Ice Types in the Arctic by Radar Echoes from SARAL/AltiKa

An important step in the sea ice freeboard to thickness conversion is the classification of sea ice types, since the ice type affects the snow depth and ice density. Studies using Ku-band CryoSat-2 have shown promise in distinguishing FYI and MYI based on the parametrisation of the radar echo. Here, we investigate applying the same classification algorithms that have shown success for Ku-band measurements to measurements acquired by SARAL/AltiKa at the Ka-band. Four different classifiers are investigated, i.e., the threshold-based, Bayesian, Random Forest (RF) and k-nearest neighbour (KNN), by using data from five 35 day cycles during Arctic mid-winter in 2014–2018. The overall classification performance shows the highest accuracy of 93% for FYI (Bayesian classifier) and 39% for MYI (threshold-based classifier). For all classification algorithms, more than half of the MYI cover falsely classifies as FYI, showing the difference in the surface characteristics attainable by Ka-band compared to Ku-band due to different scattering mechanisms. However, high overall classification performance (above 90%) is estimated for FYI for three supervised classifiers (KNN, RF and Bayesian). Furthermore, the leading-edge width parameter shows potential in discriminating open water (ocean) and sea ice when visually compared with reference data. Our results encourage the use of waveform parameters in the further validation of sea ice/open water edges and discrimination of sea ice types combining Ka- and Ku-band, especially with the planned launch of the dual-frequency altimeter mission Copernicus Polar Ice and Snow Topography Altimeter (CRISTAL) in 2027.

Sea Ice Classification via Deep Neural Network Semantic Segmentation

Sea ice monitoring plays a critical role in any icebreaker's journey, where standard procedures are in place to document and report sea ice types and concentration. In this paper, we propose semantic segmentation for automated detection and classification of sea ice types using camera feeds onboard an ice breaker. For this purpose, we evaluate the SegNet and PSPNet101 neural network architectures, which have proven success in navigation and mapping applications such as self-driving cars, remote sensing, and medical imagery. The networks are used to segment images based on two custom datasets, one with four classes: ice, ocean, vessel, and sky, i.e., sea ice detection dataset, and the second with eight classes: ocean, vessel, sky, lens artifacts, first-year ice, new ice, grey ice, and multiyear ice, i.e., sea ice classification dataset. The Nathaniel B. Palmer imagery, which captured 2-month footage of the icebreaker completing an Antarctic expedition was used in the creation of both datasets. A subset of the dataset was labeled to generate a 240-image training set for sea ice detection achieving an accuracy of 98% classification for the 26-image test set. The sea ice classification dataset consists of 1,090 labeled images achieving accuracies of 98.3% or greater for all ice types for the 104-image test set. These results validate the applicability of deep learning methods for sea ice detection and classification using images captured onboard an ice breaker, which can be further enhanced by incorporating additional ice types and operational data to support marine navigation and mapping applications.

Sea Ice Classification of SAR Imagery Based on Convolution Neural Networks

We explore new and existing convolutional neural network (CNN) architectures for sea ice classification using Sentinel-1 (S1) synthetic aperture radar (SAR) data by investigating two key challenges: binary sea ice versus open-water classification, and a multi-class sea ice type classification. The analysis of sea ice in SAR images is challenging because of the thermal noise effects and ambiguities in the radar backscatter for certain conditions that include the reflection of complex information from sea ice surfaces. We use manually annotated SAR images containing various sea ice types to construct a dataset for our Deep Learning (DL) analysis. To avoid contamination between classes we use a combination of near-simultaneous SAR images from S1 and fine resolution cloud-free optical data from Sentinel-2 (S2). For the classification, we use data augmentation to adjust for the imbalance of sea ice type classes in the training data. The SAR images are divided into small patches which are processed one at a time. We demonstrate that the combination of data augmentation and training of a proposed modified Visual Geometric Group 16-layer (VGG-16) network, trained from scratch, significantly improves the classification performance, compared to the original VGG-16 model and an ad hoc CNN model. The experimental results show both qualitatively and quantitatively that our models produce accurate classification results.

Incident Angle Dependence of Sentinel-1 Texture Features for Sea Ice Classification

Robust and reliable classification of sea ice types in synthetic aperture radar (SAR) images is needed for various operational and environmental applications. Previous studies have investigated the class-dependent decrease in SAR backscatter intensity with incident angle (IA); others have shown the potential of textural information to improve automated image classification. In this work, we investigate the inclusion of Sentinel-1 (S1) texture features into a Bayesian classifier that accounts for linear per-class variation of its features with IA. We use the S1 extra-wide swath (EW) product in ground-range detected format at medium resolution (GRDM), and we compute seven grey level co-occurrence matrix (GLCM) texture features from the HH and the HV backscatter intensity in the linear and logarithmic domain. While GLCM texture features obtained in the linear domain vary significantly with IA, the features computed from the logarithmic intensity do not depend on IA or reveal only a weak, approximately linear dependency. They can therefore be directly included in the IA-sensitive classifier that assumes a linear variation. The different number of looks in the first sub-swath (EW1) of the product causes a distinct offset in texture at the sub-swath boundary between EW1 and the second sub-swath (EW2). This offset must be considered when using texture in classification; we demonstrate a manual correction for the example of GLCM contrast. Based on the Jeffries–Matusita distance between class histograms, we perform a separability analysis for 57 different GLCM parameter settings. We select a suitable combination of features for the ice classes in our data set and classify several test images using a combination of intensity and texture features. We compare the results to a classifier using only intensity. Particular improvements are achieved for the generalized separation of ice and water, as well as the classification of young ice and multi-year ice.

Assessment of Arctic Sea Ice Classification Ability of Chinese HY-2B Dual-Band Radar Altimeter During Winter to Early Spring Conditions

Accurate identification of sea ice type from altimeter data is important for inversion of sea ice freeboard by radar altimeter, as well as for maritime navigation in polar regions. Previous sea ice classification studies have rarely involved multifrequency altimeter data. In this article, benefiting from the dual-frequency observation capability of the Chinese HY-2B altimeter, differences between waveform features of C- and Ku-band radar altimeter over different types of sea ice were investigated and analyzed for the first time. To validate and evaluate the sea ice classification capability of different feature combinations of the different bands, we extracted 14 features from the waveforms of C- and Ku-band HY-2B altimeter data acquired between December 2019 and April 2020. Our experiments demonstrated that the C- and Ku-band radar waveforms are sensitive to sea ice type, and that sea ice type classification could be improved by combining Ku- and C- band data. This work also proposed a sea ice classification methodology using C- and Ku-band altimeter waveforms. Comparison with results from previous studies indicated that the overall accuracy of the proposed method (up to 93.45%) is much higher than that of existing state-of-the-art methods.

Robustness of SAR Sea Ice Type Classification Across Incidence Angles and Seasons at L-Band

In recent years, space-borne synthetic aperture radar (SAR) polarimetry has become a valuable tool for sea ice type retrieval. L-band SAR has proven to be sensitive toward deformed sea ice and is complementary compared with operationally used C-band SAR for sea ice type classification during the early and advanced melt seasons. Here, we employ an artificial neural network (ANN)-based sea ice type classification algorithm on a comprehensive data set of ALOS-2 PALSAR-2 fully polarimetric images acquired with a range of incidence angles and during different environmental conditions. The variability within the data set means that it is ideal for making a novel assessment of the robustness of the sea ice classification, investigating the intraclass variability, the seasonal variations, and the incidence angle effect on the sea ice classification results. The images coincide with two different Arctic campaigns in 2015: the Norwegian Young Sea Ice Cruise 2015 (N-ICE2015) and the Polarstern’s (PS92) Transitions in the Arctic Seasonal Sea Ice Zone (TRANSSIZ). We find that it is essential to take into account seasonality and intraclass variability when establishing training data for machine learning-based algorithms though moderate differences in incidence angle are possible to accommodate by the classifier during the dry and cold winter season. We also conclude that the incidence angle dependence of backscatter for a given ice type is consistent for different Arctic regions.

Classification of sea ice types in Sentinel-1 synthetic aperture radar images

Abstract. A new Sentinel-1 image-based sea ice classification algorithm using a machine-learning-based model trained in a semi-automated manner is proposed to support daily ice charting. Previous studies mostly rely on manual work in selecting training and validation data. We show that the readily available ice charts from the operational ice services can reduce the amount of manual work in preparation of large amounts of training/testing data. Furthermore, they can feed highly reliable data to the trainer by indirectly exploiting the best ability of the sea ice experts working at the operational ice services. The proposed scheme has two phases: training and operational. Both phases start from the removal of thermal, scalloping, and textural noise from Sentinel-1 data and calculation of grey level co-occurrence matrix and Haralick texture features in a sliding window. In the training phase, the weekly ice charts are reprojected into the SAR image geometry. A random forest classifier is trained with the texture features on input and labels from the rasterized ice charts on output. Then, the trained classifier is directly applied to the texture features from Sentinel-1 images operationally. Test results from the two datasets spanning winter (January–March) and summer (June–August) seasons acquired over the Fram Strait and the Barents Sea showed that the classifier is capable of retrieving three generalized cover types (open water, mixed first-year ice, old ice) with overall accuracies of 87 % and 67 % in winter and summer seasons, respectively. For the summer season, the classifier failed in distinguishing mixed first-year ice from old ice with accuracy of only 12 %; however, it performed rather like an ice–water discriminator with high accuracy of 98 % as the misclassification between the mixed first-year ice and old ice was between them. The accuracy for five cover types (open water, new ice, young ice, first-year ice, old ice) in the winter season was 60 %. The errors are attributed both to incorrect manual classification on the ice charts and to the semi-automated algorithm. Finally, we demonstrate the potential for near-real-time service of the ice map using daily mosaicked Sentinel-1 images.

Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks

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 Remote Sensing Using GNSS-R: A Review

Knowledge of sea ice is critical for offshore oil and gas exploration, global shipping industries, and climate change studies. During recent decades, Global Navigation Satellite System-Reflectometry (GNSS-R) has evolved as an efficient tool for sea ice remote sensing. In particular, thanks to the availability of the TechDemoSat-1 (TDS-1) data over high-latitude regions, remote sensing of sea ice based on spaceborne GNSS-R has been rapidly growing. The goal of this paper is to provide a review of the state-of-the-art methods for sea ice remote sensing offered by the GNSS-R technique. In this review, the fundamentals of these applications are described, and their performances are evaluated. Specifically, recent progress in sea ice sensing using TDS-1 data is highlighted including sea ice detection, sea ice concentration estimation, sea ice type classification, sea ice thickness retrieval, and sea ice altimetry. In addition, studies of sea ice sensing using airborne and ground-based data are also noted. Lastly, applications based on various platforms along with remaining challenges are summarized and possible future trends are explored. In this review, concepts, research methods, and experimental techniques of GNSS-R-based sea ice sensing are delivered, and this can benefit the scientific community by providing insights into this topic to further advance this field or transfer the relevant knowledge and practice to other studies.

An Arctic sea ice multi-step classification based on GNSS-R data from the TDS-1 mission

This study examines the potential of using bistatic radar reflections from the Global Navigation Satellite System (GNSS) to classify sea ice types in the Arctic Ocean during the sea ice formation period. For this study, we used data obtained from the United Kingdom (UK) TechDemoSat-1 (TDS-1) satellite mission. The main objective of the TDS-1 mission is to provide L-band radar reflections of the ocean surface in order to infer wind speed through an estimation of the ocean surface roughness. Given the orbit inclination of TDS-1, polar coverage is obtained providing datasets over the Arctic sea ice cover. Recent studies demonstrated the use of TDS-1 data for accurately deriving ocean-ice detection using the shape of the bistatic radar waveforms derived from delay-Doppler maps. We aim to advance this previous research by demonstrating the sensitivity of GNSS bistatic radar signals to sea ice types. The originality of the study presented in this manuscript is the classification of sea ice types, never implemented before with any GNSS-R mission. For this study, we examined the fall period of October 2015 in the Beaufort and Chukchi seas region, when considerable expanses of young ice, first-year ice (FYI), and multi-year ice (MYI) were present. We developed a sea ice multi-step classification approach based on bistatic radar reflections to generate not only an ocean-ice classification but to classify between the three dominant ice types present. The classification was done in multiple steps based on GNSS bistatic radar observations only and includes checks for spatio-temporal consistency with each dominant class. Validation results are based on the comparison of the GNSS classification against SAR-derived sea ice type maps produced at the US National Ice Center. First, we derived classifications of sea ice–open water samples with a success rate of 97%, comparable to previous studies. Next, the sea ice type classification results identified FYI, MYI and young ice with success rates of 70%, 82% and 81%, respectively, indicating a strong sensitivity of the L-band GNSS bistatic radar signals to the different surface scattering properties of these primary ice types.With this manuscript we demonstrate the potential of the GNSS bistatic radar signals to classify ice types during the sea ice formation period. GNSS bistatic radar signals provide unique forward scattered measurements at L-band, potentially on a daily temporal revisit, that can be used to produce enhanced information on sea ice characteristics, such as those currently generated by other sensors. In order to determine the use of GNSS-R measurements on an operational basis, further studies are needed over a year-long period of sea ice growth and melt to determine the full benefit of GNSS for improving and providing complementary information to that generated by both microwave and optical sensors.

Sea Ice Type Classification with Optical Remote Sensing Data

Sea Ice Type Classification with Optical Remote Sensing Data

Classification of Sea Ice Types in ENVISAT Synthetic Aperture Radar Images

In this paper, sea ice in the Central Arctic has been classified in synthetic aperture radar (SAR) images from ENVISAT using a neural network (NN)-based algorithm and a Bayesian algorithm. Since different sea ice types can have similar backscattering coefficients at C-band HH polarization, it is necessary to use textural features in addition to the backscattering coefficients. The analysis revealed that the most informative texture features for the classification of multiyear ice (MYI), deformed first-year ice (FYI) (DFYI), and level FYI (LFYI) and open water/nilas are correlation, inertia, cluster prominence, energy, homogeneity, and entropy, as well as third and fourth central statistical moments of image brightness. The optimal topology of the NN, trained for ENVISAT wide-swath SAR sea ice classification, consists of nine neurons in input layer, six neurons in hidden layer, and three neurons in output layer. The classification results for a series of 20 SAR images, acquired in the central part of the Arctic Ocean during winter months, were compared to expert analysis of the images and ice charts. The results of the NN classification show that the average correspondences with the expert analysis amount to 85 $\%$ , 83 $\%$ , and 80 $\%$ for LFYI, DFYI, and MYI, respectively. The Bayesian pixel-based method can provide a higher resolution in the classified image and, therefore, better capability to identify leads compared to the NN method. Both methods may be effectively used in the Central Arctic where MYI is predominant.

Bistatic Scattering of GPS Signals Off Arctic Sea Ice

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> This paper evaluates the potential of a global positioning system (GPS) bistatic radar for the retrieval of information concerning the presence and condition of sea ice cover. For this purpose, the <emphasis emphasistype="boldital">permittivity</emphasis> and <emphasis emphasistype="boldital">roughness</emphasis> of a ground scattering target at L-band are extracted from reflected GPS waveforms collected from an airborne platform using the Kirchhoff approximation for the surface cross section. The retrieved GPS estimates are then evaluated against collocated measurements of surface roughness obtained from a lidar profiler and a reference classification of sea ice types inferred from a multisensor data set that includes polarimetric microwave emissions, RADARSAT backscatter, and Moderate Resolution Imaging Spectroradiometer infrared/visible imagery. </para>

Classification of sea ice types in the Ross Sea, Antarctica from SAR and AVHRR imagery

ERS-2 synthetic aperture radar (SAR) and Advanced Very High Resolution Radiometer (AVHRR) imagery are used to examine spectral characteristics of late winter/early spring ice in the Ross Sea, Antarctica. The combined spectral signatures are used to distinguish six ice types: fast ice, new ice, smooth first year ice, rough first year ice, thin new ice/wind roughened open water and glacial ice. The procedure firstly involves 'picking' class boundaries from SAR imagery based on the morphology of a speckle reduced backscatter spectrum. These class boundaries are then used as input to an iterative segmentation procedure that involves the repeated application of a speckle reduction filter to the image. For an image from late September 1996 the segmentation procedure enabled separation of five general ice categories each with a characteristic backscatter range. However because of the combined contributions of ice thickness, surface roughness, salinity and water content to the SAR backscatter, further decision criteria are required to separate some physical ice types unable to be resolved individually using this method. Coincident and co-registered infrared data from the AVHRR sensor are used to extract spectral characteristics for the final ice classes. Using this procedure we were able to distinguish floating glacier ice from thin new ice/wind roughened open water and new ice from nearshore fast ice. These ice types were unable to be separated using SAR backscatter intensity alone. In addition image subtraction was also able to clearly delineate areas of shore fast ice.

Sea-ice type classification from ERS-1 SAR data based on grey level and texture information

AbstractThis paper describes the development of a practical algorithm for the classification of sea-ice types from ERS-1 synthetic aperture radar (SAR) data. The algorithm was based on a combination of grey level and texture information in order to overcome ambiguous grey level values of different ice types. The problem of calculating texture parameters for windows containing more than one ice type was overcome by first segmenting the image so that only pixels from the same segment were included in the calculation of the texture measure. The segmentation procedure was based on the iterative application of a speckle noise reduction filter, and was thus crucially dependent on the ability of such a filter to smooth out noise without destroying edges and fine features. In order to achieve this, a modification to the sigma filter of Lee (1983b) was developed; it out-performed the sigma filter for a model problem. Two ERS-1 SAR scenes of the marginal ice zone east of Spitsbergen in March 1992 were analysed by calculating values of grey level and range for different ice types contained within raw data extracts. Although the grey levels of some of the ice types overlapped, most of the ambiguity was removed through the additional use of range. It was also necessary to test for the wave-like appearance of open water. The classification scheme was demonstrated to identify correctly most of the grease/new ice, first-year ice, multiyear ice, rough ice, pancake ice, and open water in the two SAR scenes, although there was some misclassification of open water as first-year ice.

Classification of sea ice types with single‐band (33.6 GHz) airborne passive microwave imagery

During March 1983 extensive high‐quality airborne passive Ka band (33.6 GHz) microwave imagery and coincident high‐resolution aerial photography were obtained of ice along a 378‐km flight line in the Beaufort Sea. Analysis of these data suggests that four classes of winter surfaces can be distinguished solely on the basis of 33.6‐GHz brightness temperature: open water, frazil, old ice, and young/first‐year ice. New ice (excluding frazil) and nilas display brightness temperatures that overlap the range of temperatures characteristic of old ice and, to a lesser extent, young/first‐year ice. Scenes in which a new ice or nilas are present in appreciable amounts are subject to substantial errors in classification if static measures of Ka band radiometric brightness temperature alone are considered. Textural characteristics of nilas and new ice, however, differ significantly from textural features characteristic of other ice types and probably can be used with brightness temperature data to classify ice type in high‐resolution single‐band microwave images. In any case, open water is radiometrically the coldest surface observed in any scene. Lack of overlap between brightness temperatures characteristic of other surfaces indicates that estimates of the areal extent of open water based on only 33.6‐GHz brightness temperatures are accurate.