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

Polar Record ◽  
1995 ◽  
Vol 31 (177) ◽  
pp. 135-146 ◽  
Author(s):  
D.M. Smith ◽  
E.C. Barrett ◽  
J.C. Scott
Keyword(s):  
Sea Ice ◽  
Open Water ◽  
Speckle Noise ◽  
First Year ◽  
Grey Level ◽  
Texture Information ◽  
Texture Parameters ◽  
Sar Data ◽  
Speckle Noise Reduction

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.

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

2021 ◽  
Author(s):  
Anton Korosov ◽  
Hugo Boulze ◽  
Julien Brajard
Keyword(s):  
Random Forest ◽  
Sea Ice ◽  
Classification Accuracy ◽  
Thermal Noise ◽  
Noise Removal ◽  
The Arctic ◽  
First Year ◽  
Sar Data ◽  
The Impact

<p>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 a dataset from winter 2020 for retrieval of four classes: ice free, young ice, first-year ice and old ice. The accuracy of our classification is 91.6%. The error is a bit higher for young ice (76%) and first-year ice (84%). 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.</p><p> </p><p>Our study demonstrates that CNN can be successfully applied for classification of sea ice types in SAR data. The algorithm is applied in small sub-images extracted from a SAR image after preprocessing including thermal noise removal. Validation shows that the errors are mostly attributed to coarse resolution of ice charts or misclassification of training data by human experts.</p><p> </p><p>Several sensitivity experiments were conducted for testing the impact of CNN architecture, hyperparameters, training parameters and data preprocessing on accuracy. It was shown that a CNN with three convolutional layers, two max-pool layers and three hidden dense layers can be applied to a sub-image with size 50 x 50 pixels for achieving the best results. It was also shown that a CNN can be applied to SAR data without thermal noise removal on the preprocessing step. Understandably, the classification accuracy decreases to 89% but remains reasonable.</p><p> </p><p>The main advantages of the new algorithm are the ability to classify several ice types, higher classification accuracy for each ice type and higher speed of processing than in the previous studies. The relative simplicity of the algorithm (both texture analysis and classification are performed by CNN) is also a benefit. In addition to providing ice type labels, the algorithm also derives the probability of belonging to a class. Uncertainty of the method can be derived from these probabilities and used in the assimilation of ice type in numerical models. </p><p><br>Given the high accuracy and processing speed, the CNN-based algorithm is included in the Copernicus Marine Environment Monitoring Service (CMEMS) for operational sea ice type retrieval for generating ice charts in the Arctic Ocean. It is already released as an open source software and available on Github: https://github.com/nansencenter/s1_icetype_cnn.</p>

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

2020 ◽  
Vol 12 (13) ◽  
pp. 2165 ◽  
Author(s):  
Hugo Boulze ◽  
Anton Korosov ◽  
Julien Brajard
Keyword(s):  
Random Forest ◽  
Sea Ice ◽  
Learning Algorithm ◽  
Computing Time ◽  
Texture Features ◽  
Random Forest Classifier ◽  
Synthetic Aperture ◽  
First Year ◽  
Sar Data

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.

scholarly journals Albedo of Young and First-Year Antarctic Sea Ice

1990 ◽  
Vol 14 ◽  
pp. 331 ◽  
Author(s):  
Richard Brandt ◽  
Ian Allison ◽  
Stephen Warren
Keyword(s):  
Sea Ice ◽  
Radiation Flux ◽  
Open Water ◽  
First Year ◽  
Spectral Measurements ◽  
Antarctic Sea Ice ◽  
Antarctic Expedition ◽  
Downward Radiation ◽  
Radiation Measurements ◽  
Snow Thickness

Reflection of solar radiation was studied in the seasonal sea-ice zone off East Antarctica on a cruise of the Australian Antarctic Expedition, October-December 1988. Spectral and total albedos were measured for grease ice, nilas, young grey ice, grey-white ice, snow-covered ice, and open water. Spectral measurements covered the region 400–1000 nm wavelength. For ice too thin to support our weight, the radiometers were mounted at the end of a 1.5 m rod extended out the door of a helicopter or from a basket hung from the ship's crane, using a positioning and leveling rack. Corrections had to be applied to the downward radiation flux because the helicopter or the crane was in the field of view of the cosine-collector. The fractional coverage of each of the ice types (and open water) was estimated hourly for the region near the ship, as well as the thickness of each ice type, and the snow thickness. Observations were carried out continuously during the four weeks the ship was in the ice, supplemented by occasional helicopter surveys covering larger areas. These observations, together with the radiation measurements, make possible the computation of area-average albedo for the East Antarctic sea-ice zone in spring.

scholarly journals Albedo of the ice covered Weddell and Bellingshausen Seas

2012 ◽  
Vol 6 (2) ◽  
pp. 479-491 ◽  
Author(s):  
A. I. Weiss ◽  
J. C. King ◽  
T. A. Lachlan-Cope ◽  
R. S. Ladkin
Keyword(s):  
Sea Ice ◽  
Surface Albedo ◽  
Austral Summer ◽  
Open Water ◽  
Weddell Sea ◽  
First Year ◽  
Pack Ice ◽  
Bellingshausen Sea ◽  
The Mean ◽  
The Antarctic

Abstract. This study investigates the surface albedo of the sea ice areas adjacent to the Antarctic Peninsula during the austral summer. Aircraft measurements of the surface albedo, which were conducted in the sea ice areas of the Weddell and Bellingshausen Seas show significant differences between these two regions. The averaged surface albedo varied between 0.13 and 0.81. The ice cover of the Bellingshausen Sea consisted mainly of first year ice and the sea surface showed an averaged sea ice albedo of αi = 0.64 ± 0.2 (± standard deviation). The mean sea ice albedo of the pack ice area in the western Weddell Sea was αi = 0.75 ± 0.05. In the southern Weddell Sea, where new, young sea ice prevailed, a mean albedo value of αi = 0.38 ± 0.08 was observed. Relatively warm open water and thin, newly formed ice had the lowest albedo values, whereas relatively cold and snow covered pack ice had the highest albedo values. All sea ice areas consisted of a mixture of a large range of different sea ice types. An investigation of commonly used parameterizations of albedo as a function of surface temperature in the Weddell and Bellingshausen Sea ice areas showed that the albedo parameterizations do not work well for areas with new, young ice.

scholarly journals Predicting Melt Pond Fraction on Landfast Snow Covered First Year Sea Ice from Winter C-Band SAR Backscatter Utilizing Linear, Polarimetric and Texture Parameters

2018 ◽  
Vol 10 (10) ◽  
pp. 1603 ◽  
Author(s):  
Saroat Ramjan ◽  
Torsten Geldsetzer ◽  
Randall Scharien ◽  
John Yackel
Keyword(s):  
Sea Ice ◽  
Regression Models ◽  
Incidence Angle ◽  
Early Summer ◽  
Aerial Photographs ◽  
Late Winter ◽  
First Year ◽  
Melt Pond ◽  
Texture Parameters ◽  
Snow Thickness

Early-summer melt pond fraction is predicted using late-winter C-band backscatter of snow-covered first-year sea ice. Aerial photographs were acquired during an early-summer 2012 field campaign in Resolute Passage, Nunavut, Canada, on smooth first-year sea ice to estimate the melt pond fraction. RADARSAT-2 Synthetic Aperture Radar (SAR) data were acquired over the study area in late winter prior to melt onset. Correlations between the melt pond fractions and late-winter linear and polarimetric SAR parameters and texture measures derived from the SAR parameters are utilized to develop multivariate regression models that predict melt pond fractions. The results demonstrate substantial capability of the regression models to predict melt pond fractions for all SAR incidence angle ranges. The combination of the most significant linear, polarimetric and texture parameters provide the best model at far-range incidence angles, with an R 2 of 0.62 and a pond fraction RMSE of 0.09. Near- and mid- range incidence angle models provide R 2 values of 0.57 and 0.61, respectively, with an RMSE of 0.11. The strength of the regression models improves when SAR parameters are combined with texture parameters. These predictions also serve as a proxy to estimate snow thickness distributions during late winter as higher pond fractions evolve from thinner snow cover.

scholarly journals Comparison of Sea-ice Type Identification Between Airborne Dual Frequency Passive Microwave Radiometry and Standard Laser/Infrared Techniques

1975 ◽  
Vol 15 (73) ◽  
pp. 225-239
Author(s):  
S. G. Tooma ◽  
R. A. Mennella ◽  
J. P. Hollinger ◽  
R. D. Ketchum
Keyword(s):  
Sea Ice ◽  
Ground Truth ◽  
Open Water ◽  
Weather Conditions ◽  
Passive Microwave ◽  
Single Frequency ◽  
Naval Research ◽  
First Year ◽  
Dual Frequency ◽  
Brightness Temperatures

AbstractDuring December 1973, the Naval Oceanographie Offirc (NAVOCKANO) and the Naval Research Laboratory (NRL) conducted a joint remote-sensing experiment over the sea-ice fields off Scoresby Sound on the east coast of Greenland using NAVOCEANO’s RP3-A Birdseye aircraft, laser profiler, and infrared scanner, and NRL’s 19.34 and 31.0 GHz nadir-looking radiometers. The objectives of this mission were: (1) to develop skills for interpreting sea-ice passive microwave data. (2) to expand, if possible, the two-category capability (multi-year ice and first-year ice) of passive microwave sensors over sea ice, (3) to compare two frequencies (19 and 31 GHz) to determine which may be more useful in a scanning radiometer now under development at NRL, and (4) to determine the value of multi-frequency as compared to single-frequency study of sea ice.Since, because of darkness and remoteness, no photography or in situ ground truth were possible for this mission, it was necessary to rely on the interpretations of the laser and infrared (IR) data to evaluate the performance of the microwave radiometers. Fortunately, excellent laser and IR data were collected, and a confident description of the ice overflown was possible.Five ice conditions: (1) open water/new ice, (2) smooth first-year ice, (3) ridged first-year ice, (4) multi-year ice, and (5) a higher brightness temperature form of multi-year ice interpreted as second-year ice were identifiable, regardless of weather conditions, by comparing the average of the two microwave brightness temperatures at the two frequencies with their difference.

Contextual Classification of Sea-Ice Types Using Compact Polarimetric SAR Data

2019 ◽  
Vol 57 (10) ◽  
pp. 7476-7491 ◽  
Author(s):  
Mohsen Ghanbari ◽  
David A. Clausi ◽  
Linlin Xu ◽  
Mingzhe Jiang
Keyword(s):  
Sea Ice ◽  
Polarimetric Sar ◽  
Sar Data
Sign in / Sign up

Export Citation Format

Share Document