Assessment of High‐Resolution Dynamical and Machine Learning Models for Prediction of Sea Ice Concentration in a Regional Application

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

Revealing Unknown Controlled Substances and New Psychoactive Substances Using High-Resolution LC–MS-MS Machine Learning Models and the Hybrid Similarity Search Algorithm

Journal of Analytical Toxicology ◽

10.1093/jat/bkab098 ◽

2021 ◽

Author(s):

So Yeon Lee ◽

Sang Tak Lee ◽

Sungill Suh ◽

Bum Jun Ko ◽

Han Bin Oh

Keyword(s):

Machine Learning ◽

High Resolution ◽

Carbonyl Compound ◽

Similarity Search ◽

New Psychoactive Substances ◽

Support Vector ◽

Learning Models ◽

Psychoactive Substances ◽

Controlled Substances ◽

Machine Learning Models

Abstract High-resolution liquid chromatography (LC)–tandem mass spectrometry (MS-MS)-based machine learning models are constructed to address the analytical challenge of identifying unknown controlled substances and new psychoactive substances (NPSs). Using a training set composed of 770 LC–MS-MS barcode spectra (with binary entries 0 or 1) obtained generally by high-resolution mass spectrometers, three classification machine learning models were generated and evaluated. The three models are artificial neural network (ANN), support vector machine (SVM) and k-nearest neighbor (k-NN) models. In these models, controlled substances and NPSs were classified into 13 subgroups (benzylpiperazine, opiate, benzodiazepine, amphetamine, cocaine, methcathinone, classical cannabinoid, fentanyl, 2C series, indazole carbonyl compound, indole carbonyl compound, phencyclidine and others). Using 193 LC–MS-MS barcode spectra as an external test set, accuracy of the ANN, SVM and k-NN models were evaluated as 72.5%, 90.0% and 94.3%, respectively. Also, the hybrid similarity search (HSS) algorithm was evaluated to examine whether this algorithm can successfully identify unknown controlled substances and NPSs whose data are unavailable in the database. When only 24 representative LC–MS-MS spectra of controlled substances and NPSs were selectively included in the database, it was found that HSS can successfully identify compounds with high reliability. The machine learning models and HSS algorithms are incorporated into our home-coded artificial intelligence screener for narcotic drugs and psychotropic substances standalone software that is equipped with a graphic user interface. The use of this software allows unknown controlled substances and NPSs to be identified in a convenient manner.

Download Full-text

A sea ice concentration estimation algorithm utilizing radiometer and SAR data

The Cryosphere ◽

10.5194/tc-8-1639-2014 ◽

2014 ◽

Vol 8 (5) ◽

pp. 1639-1650 ◽

Cited By ~ 19

Author(s):

J. Karvonen

Keyword(s):

High Resolution ◽

Sea Ice ◽

Estimation Algorithm ◽

Sea Ice Concentration ◽

Finnish Meteorological Institute ◽

Brightness Temperatures ◽

Concentration Estimation ◽

Ice Concentration ◽

Bootstrap Algorithm ◽

The University

Abstract. We have studied the possibility of combining the high-resolution synthetic aperture radar (SAR) segmentation and ice concentration estimated by radiometer brightness temperatures. Here we present an algorithm for mapping a radiometer-based concentration value for each SAR segment. The concentrations are estimated by a multi-layer perceptron (MLP) neural network which has the AMSR-2 (Advanced Microwave Scanning Radiometer 2) polarization ratios and gradient ratios of four radiometer channels as its inputs. The results have been compared numerically to the gridded Finnish Meteorological Institute (FMI) ice chart concentrations and high-resolution AMSR-2 ASI (ARTIST Sea Ice) algorithm concentrations provided by the University of Hamburg and also visually to the AMSR-2 bootstrap algorithm concentrations, which are given in much coarser resolution. The differences when compared to FMI daily ice charts were on average small. When compared to ASI ice concentrations, the differences were a bit larger, but still small on average. According to our comparisons, the largest differences typically occur near the ice edge and sea–land boundary. The main advantage of combining radiometer-based ice concentration estimation and SAR segmentation seems to be a more precise estimation of the boundaries of different ice concentration zones.

Download Full-text

HIGH-RESOLUTION NATIONAL-SCALE HYDROGEOLOGIC DATASETS IN SUPPORT OF MACHINE LEARNING MODELS

10.1130/abs/2020am-357568 ◽

2020 ◽

Author(s):

Kenneth Belitz ◽

Keyword(s):

Machine Learning ◽

High Resolution ◽

Learning Models ◽

National Scale ◽

Machine Learning Models

Download Full-text

Exploring the link between microseism and sea ice in Antarctica by using machine learning

Scientific Reports ◽

10.1038/s41598-019-49586-z ◽

2019 ◽

Vol 9 (1) ◽

Cited By ~ 4

Author(s):

Andrea Cannata ◽

Flavio Cannavò ◽

Salvatore Moschella ◽

Stefano Gresta ◽

Laura Spina

Keyword(s):

Machine Learning ◽

Sea Ice ◽

Seismic Energy ◽

Middle Latitude ◽

Energy Coupling ◽

Seismic Signal ◽

Machine Learning Techniques ◽

Sea Ice Concentration ◽

Ocean Wave Energy ◽

Ice Concentration

Abstract The most continuous and ubiquitous seismic signal on Earth is the microseism, closely related to ocean wave energy coupling with the solid Earth. A peculiar feature of microseism recorded in Antarctica is the link with the sea ice, making the temporal pattern of microseism amplitudes different with respect to the microseism recorded in low-middle latitude regions. Indeed, during austral winters, in Antarctica the oceanic waves cannot efficiently excite seismic energy because of the sea ice in the Southern Ocean. Here, we quantitatively investigate the relationship between microseism, recorded along the Antarctic coasts, and sea ice concentration. In particular, we show a decrease in sea ice sensitivity of microseism, due to the increasing distance from the station recording the seismic signal. The influence seems to strongly reduce for distances above 1,000 km. Finally, we present an algorithm, based on machine learning techniques, allowing to spatially and temporally reconstruct the sea ice distribution around Antarctica based on the microseism amplitudes. This technique will allow reconstructing the sea ice concentration in both Arctic and Antarctica in periods when the satellite images, routinely used for sea ice monitoring, are not available, with wide applications in many fields, first of all climate studies.

Download Full-text

High-resolution maps of the sea-ice concentration from MODIS satellite data

Geophysical Research Letters ◽

10.1029/2004gl020808 ◽

2004 ◽

Vol 31 (20) ◽

Cited By ~ 25

Author(s):

Clemens Drüe

Keyword(s):

High Resolution ◽

Sea Ice ◽

Satellite Data ◽

Sea Ice Concentration ◽

Ice Concentration

Download Full-text

Supplementary material to "High-resolution mapping of regional traffic emissions by using land-use machine learning models"

10.5194/acp-2021-281-supplement ◽

2021 ◽

Author(s):

Xiaomeng Wu ◽

Daoyuan Yang ◽

Jiajun Gu ◽

Yifan Wen ◽

Shaojun Zhang ◽

...

Keyword(s):

Machine Learning ◽

Land Use ◽

High Resolution ◽

Traffic Emissions ◽

Learning Models ◽

High Resolution Mapping ◽

Resolution Mapping ◽

Supplementary Material ◽

Machine Learning Models

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