Attention Multi-Scale Network for Automatic Layer Extraction of Ice Radar Topological Sequences

Analyzing the surface and bedrock locations in radar imagery enables the computation of ice sheet thickness, which is important for the study of ice sheets, their volume and how they may contribute to global climate change. However, the traditional handcrafted methods cannot quickly provide quantitative, objective and reliable extraction of information from radargrams. Most traditional handcrafted methods, designed to detect ice-surface and ice-bed layers from ice sheet radargrams, require complex human involvement and are difficult to apply to large datasets, while deep learning methods can obtain better results in a generalized way. In this study, an end-to-end multi-scale attention network (MsANet) is proposed to realize the estimation and reconstruction of layers in sequences of ice sheet radar tomographic images. First, we use an improved 3D convolutional network, C3D-M, whose first full connection layer is replaced by a convolution unit to better maintain the spatial relativity of ice layer features, as the backbone. Then, an adjustable multi-scale module uses different scale filters to learn scale information to enhance the feature extraction capabilities of the network. Finally, an attention module extended to 3D space removes a redundant bottleneck unit to better fuse and refine ice layer features. Radar sequential images collected by the Center of Remote Sensing of Ice Sheets in 2014 are used as training and testing data. Compared with state-of-the-art deep learning methods, the MsANet shows a 10% reduction (2.14 pixels) on the measurement of average mean absolute column-wise error for detecting the ice-surface and ice-bottom layers, runs faster and uses approximately 12 million fewer parameters.

Download Full-text

Antimicrobial peptide identification using multi-scale convolutional network

BMC Bioinformatics ◽

10.1186/s12859-019-3327-y ◽

2019 ◽

Vol 20 (1) ◽

Author(s):

Xin Su ◽

Jing Xu ◽

Yanbin Yin ◽

Xiongwen Quan ◽

Han Zhang

Keyword(s):

Deep Learning ◽

Antimicrobial Peptide ◽

State Of The Art ◽

Peptide Identification ◽

The State ◽

Fusion Model ◽

Learning Methods ◽

Convolutional Network ◽

Multi Scale ◽

Peptide Database

Abstract Background Antibiotic resistance has become an increasingly serious problem in the past decades. As an alternative choice, antimicrobial peptides (AMPs) have attracted lots of attention. To identify new AMPs, machine learning methods have been commonly used. More recently, some deep learning methods have also been applied to this problem. Results In this paper, we designed a deep learning model to identify AMP sequences. We employed the embedding layer and the multi-scale convolutional network in our model. The multi-scale convolutional network, which contains multiple convolutional layers of varying filter lengths, could utilize all latent features captured by the multiple convolutional layers. To further improve the performance, we also incorporated additional information into the designed model and proposed a fusion model. Results showed that our model outperforms the state-of-the-art models on two AMP datasets and the Antimicrobial Peptide Database (APD)3 benchmark dataset. The fusion model also outperforms the state-of-the-art model on an anti-inflammatory peptides (AIPs) dataset at the accuracy. Conclusions Multi-scale convolutional network is a novel addition to existing deep neural network (DNN) models. The proposed DNN model and the modified fusion model outperform the state-of-the-art models for new AMP discovery. The source code and data are available at https://github.com/zhanglabNKU/APIN.

Download Full-text

Impact of mid-glacial ice sheets on deep ocean circulation and global climate: Role of surface cooling on the AMOC

10.5194/cp-2020-75 ◽

2020 ◽

Author(s):

Sam Sherriff-Tadano ◽

Ayako Abe-Ouchi ◽

Akira Oka

Keyword(s):

North Atlantic ◽

General Circulation ◽

Global Climate ◽

Surface Wind ◽

Ice Sheets ◽

Ice Sheet ◽

Meridional Overturning Circulation ◽

Strengthening Effect ◽

Surface Cooling ◽

Glacial Ice

Abstract. This study explores the effect of southward expansion of mid-glacial ice sheets on the global climate and the Atlantic meridional overturning circulation (AMOC), as well as the processes by which the ice sheets modify the AMOC. For this purpose, simulations of Marine Isotope Stage (MIS) 3 and 5a are performed with an atmosphere-ocean general circulation model. In the MIS3 and MIS5a simulations, the global average temperature decreases by 5.0 °C and 2.2 °C, respectively, compared with the preindustrial climate simulation. The AMOC weakens by 3 % in MIS3, whereas it is enhanced by 16 % in MIS5a, both of which are consistent with a reconstruction. Sensitivity experiments extracting the effect of the expansion of glacial ice sheets from MIS5a to MIS3 show a global cooling of 1.1 °C, contributing to about 40 % of the total surface cooling from MIS5a to MIS3. These experiments also demonstrate that the ice sheet expansion leads to a surface cooling of 2 °C over the Southern Ocean as a result of colder North Atlantic deep water. We find that the southward expansion of the mid-glacial ice sheet exerts a small impact on the AMOC. Partially coupled experiments reveal that the global surface cooling by the glacial ice sheet tends to reduce the AMOC by increasing the sea ice at both poles, and hence compensates for the strengthening effect of the enhanced surface wind over the North Atlantic. Our results show that the total effect of glacial ice sheets on the AMOC is determined by the two competing effects, surface wind and surface cooling. The relative strength of surface wind and surface cooling depends on the ice sheet configuration, and the strength of the surface cooling can be comparable to that of surface wind when changes in the extent of ice sheet are prominent.

Download Full-text

Ice sheet velocity tracking by Sentinel-1 satellite images at Graham Coast Kyiv Peninsula

Ukrainian Antarctic Journal ◽

10.33275/1727-7485.1.2021.663 ◽

2021 ◽

pp. 24-31

Author(s):

S. Kadurin ◽

◽

K. Andrieieva ◽

Keyword(s):

Remote Sensing ◽

Global Climate ◽

Ice Sheets ◽

Ice Sheet ◽

Remote Sensing Data ◽

Ice Cover ◽

Satellite System ◽

Time Interval ◽

Radar Images ◽

Ocean Level

The study of Antarctic glaciers and ice sheets velocity is one of the most discussed topics. Such high interest in this topic is primarily because the ice from the Antarctic glaciers, which gets to the ocean, significantly affects the ocean level and the global climate. Development of modern satellite technologies for Earth remote sensing made it possible to elaborate a number of methods for ice sheets’ displacements estimation and calculation of such displacements velocities. This work uses remote sensing data from the satellite system Copernicus Sentinel-1 to estimate the ice cover velocities in the Kyiv Peninsula in the time interval from December 2020 to March 2021. To this end, 10 radar images of the study area from early December to the end of March were used with an interval of 12–14 days. All selected images were analyzed in pairs to establish changes on the surface for the selected time interval. GRD-format images from Copernicus Sentinel-1 satellite, corrected for Earth's ellipsoid shape, were used. Based on the offset tracking operation, we calculated the speeds of ice cover movements within the Kyiv Peninsula for each pair of images with approximately two weeks' time difference. As a result, the speed of ice movements varies considerably and at the glacier mouth can reach 3.5–4 meters per day. Also, the rate of ice displacement in the glacier body changed over time. Thus, the highest ice velocities were in the glacier's mouth. However, short-term time intervals of intensification were recorded for the rear and even the marginal parts of the glaciers in contact with the ice sheet. Thus, the lowest part of the glacier activating sequence leads to the upper part shifting. Notably, this increase in the displacement of ice cover was recorded in February, one of the warmest months in this part of Antarctica.

Download Full-text

Influence of albedo parameterization on surface mass balance in the perspective of Greenland ice sheet modelling in EC-Earth

10.5194/tc-2016-281 ◽

2016 ◽

Author(s):

Michiel Helsen ◽

Roderik Van de Wal ◽

Thomas Reerink ◽

Richard Bintanja ◽

Marianne Sloth Madsen ◽

...

Keyword(s):

Mass Balance ◽

Climate Model ◽

Global Climate ◽

Global Climate Model ◽

Ice Sheets ◽

Ice Sheet ◽

Greenland Ice Sheet ◽

Surface Mass Balance ◽

Surface Mass

Abstract. The albedo of the surface of ice sheets changes as a function of time, due to the effects of deposition of new snow, ageing of dry snow, melting and runoff. Currently, the calculation of the albedo of ice sheets is highly parameterized within the Earth System Model EC-Earth, by taking a constant value for areas with thick perennial snow cover. This is one of the reasons that the surface mass balance (SMB) of the Greenland ice sheet (GrIS) is poorly resolved in the model. To improve this, eight snow albedo schemes are evaluated here. The resulting SMB is downscaled from the lower resolution global climate model topography to the higher resolution ice sheet topography of the GrIS, such that the influence of these different SMB climatologies on the long-term evolution of the GrIS is tested by ice sheet model simulations. This results in an optimised albedo parameterization that can be used in future EC-Earth simulations with an interactive ice sheet component.

Download Full-text

Asynchronous Antarctic and Greenland ice-volume contributions to the last interglacial sea-level highstand

10.5194/egusphere-egu2020-1513 ◽

2020 ◽

Author(s):

Eelco Rohling ◽

Fiona Hibbert

Keyword(s):

Sea Level Rise ◽

Sea Level ◽

Global Climate ◽

Ice Sheets ◽

Ice Sheet ◽

Greenland Ice Sheet ◽

Climate Forcing ◽

Potential Contribution ◽

Last Interglacial ◽

Ice Shelf

Sea-level rise is among the greatest risks that arise from anthropogenic global climate change. It is receiving a lot of attention, among others in the IPCC reports, but major questions remain as to the potential contribution from the great continental ice sheets. In recent years, some modelling work has suggested that the ice-component of sea-level rise may be much faster than previously thought, but the rapidity of rise seen in these results depends on inclusion of scientifically debated mechanisms of ice-shelf decay and associated ice-sheet instability. The processes have not been active during historical times, so data are needed from previous warm periods to evaluate whether the suggested rates of sea-level rise are supported by observations or not. Also, we then need to assess which of the ice sheets was most sensitive, and why. The last interglacial (LIG; ~130,000 to ~118,000 years ago, ka) was the last time global sea level rose well above its present level, reaching a highstand of +6 to +9 m or more. Because Greenland Ice Sheet (GrIS) contributions were smaller than that, this implies substantial Antarctic Ice Sheet (AIS) contributions. However, this still leaves the timings, magnitudes, and drivers of GrIS and AIS reductions open to debate. I will discuss recently published sea-level reconstructions for the LIG highstand, which reveal that AIS and GrIS contributions were distinctly asynchronous, and that rates of rise to values above 0 m (present-day sea level) reached up to 3.5 m per century. Such high pre-anthropogenic rates of sea-level rise lend credibility to high rates inferred by ice modelling under certain ice-shelf instability parameterisations, for both the past and future. Climate forcing was distinctly asynchronous between the southern and northern hemispheres as well during the LIG, explaining the asynchronous sea-level contributions from AIS and GrIS. Today, climate forcing is synchronous between the two hemispheres, and also faster and greater than during the LIG. Therefore, LIG rates of sea-level rise should likely be considered minimum estimates for the future.

Download Full-text

Noise Removal and Feature Extraction in Airborne Radar Sounding Data of Ice Sheets

Remote Sensing ◽

10.3390/rs14020399 ◽

2022 ◽

Vol 14 (2) ◽

pp. 399

Author(s):

Xueyuan Tang ◽

Sheng Dong ◽

Kun Luo ◽

Jingxue Guo ◽

Lin Li ◽

...

Keyword(s):

Machine Learning ◽

Feature Extraction ◽

Ice Sheets ◽

Ice Sheet ◽

Noise Removal ◽

Airborne Radar ◽

Learning Methods ◽

Ice Flow ◽

Machine Learning Methods ◽

Internal Structures

The airborne ice-penetrating radar (IPR) is an effective method used for ice sheet exploration and is widely applied for detecting the internal structures of ice sheets and for understanding the mechanism of ice flow and the characteristics of the bottom of ice sheets. However, because of the ambient influence and the limitations of the instruments, IPR data are frequently overlaid with noise and interference, which further impedes the extraction of layer features and the interpretation of the physical characteristics of the ice sheet. In this paper, we first applied conventional filtering methods to remove the feature noise and interference in IPR data. Furthermore, machine learning methods were introduced in IPR data processing for noise removal and feature extraction. Inspired by a comparison of the filtering methods and machine learning methods, we propose a fusion method combining both filtering methods and machine-learning-based methods to optimize the feature extraction in IPR data. Field data tests indicated that, under different conditions of IPR data, the application of different methods and strategies can improve the layer feature extraction.

Download Full-text

Modelling dynamic ice-sheet boundaries and grounding line migration using the level set method

Journal of Glaciology ◽

10.1017/jog.2020.45 ◽

2020 ◽

Vol 66 (259) ◽

pp. 766-776

Author(s):

M. Alamgir Hossain ◽

Sam Pimentel ◽

John M. Stockie

Keyword(s):

Level Set Method ◽

Level Set ◽

Ice Sheets ◽

Ice Sheet ◽

Numerical Approach ◽

Ice Surface ◽

Grounding Line ◽

High Degree ◽

Ice Water ◽

Benchmark Solutions

AbstractComputing predictions of future sea level that include well-defined uncertainty bounds requires models that are capable of robustly simulating the evolution of ice sheets and glaciers. Ice flow behaviour is known to be sensitive to the location and geometry of dynamic ice boundaries such as the grounding line (GRL), terminus position and ice surface elevation, so that any such model should track these interfaces with a high degree of accuracy. To address this challenge, we implement a numerical approach that uses the level-set method (LSM) that accurately models the evolution of the ice–air and ice–water interface as well as capturing topological changes in ice-sheet geometry. This approach is evaluated by comparing simulations of grounded and marine-terminating ice sheets to various analytical and numerical benchmark solutions. A particular advantage of the LSM is its ability to explicitly track the moving margin and GRL while using a fixed grid finite-difference scheme. Our results demonstrate that the LSM is an accurate and robust approach for tracking the ice surface interface and terminus for advancing and retreating ice sheets, including the transient marine ice-sheet interface and GRL positions.

Download Full-text

Impact of mid-glacial ice sheets on deep ocean circulation and global climate

10.5194/egusphere-egu21-13538 ◽

2021 ◽

Author(s):

Sam Sherriff-Tadano ◽

Ayako Abe-Ouchi ◽

Akira Oka

Keyword(s):

North Atlantic ◽

Global Climate ◽

Surface Wind ◽

Ice Sheets ◽

Ice Sheet ◽

Meridional Overturning Circulation ◽

Strengthening Effect ◽

Surface Cooling ◽

Glacial Ice ◽

Cooling Effects

This study explores the effect of southward expansion of Northern Hemisphere (American) mid-glacial ice sheets on the global climate and the Atlantic Meridional Overturning Circulation (AMOC), as well as the processes by which the ice sheets modify the AMOC. For this purpose, simulations of Marine Isotope Stage (MIS) 3 (36ka) and 5a (80ka) are performed with an atmosphere-ocean general circulation model. In the MIS3 and MIS5a simulations, the global average temperature decreases by 5.0 &#176;C and 2.2 &#176;C, respectively, compared with the preindustrial climate simulation. The AMOC weakens by 3% in MIS3, whereas it strengthens by 16% in MIS5a, both of which are consistent with an estimate based on 231Pa/230Th. Sensitivity experiments extracting the effect of the southward expansion of glacial ice sheets from MIS5a to MIS3 show a global cooling of 1.1 &#176;C, contributing to about 40% of the total surface cooling from MIS5a to MIS3. These experiments also demonstrate that the ice sheet expansion leads to a surface cooling of 2 &#176;C over the Southern Ocean as a result of colder North Atlantic deep water. We find that the southward expansion of the mid-glacial ice sheet exerts a small impact on the AMOC. Partially coupled experiments reveal that the global surface cooling by the glacial ice sheet tends to reduce the AMOC by increasing the sea ice at both poles, and hence compensates for the strengthening effect of the enhanced surface wind over the North Atlantic. Our results show that the total effect of glacial ice sheets on the AMOC is determined by the two competing effects, surface wind and surface cooling. The relative strength of surface wind and surface cooling effects depends on the ice sheet configuration, and the strength of the surface cooling can be comparable to that of surface wind when changes in the extent of ice sheet are prominent.

Download Full-text

Individual and combined effects of ice sheets and precession on MIS-13 climate

Climate of the Past ◽

10.5194/cp-5-229-2009 ◽

2009 ◽

Vol 5 (2) ◽

pp. 229-243 ◽

Cited By ~ 41

Author(s):

Q. Z. Yin ◽

A. Berger ◽

M. Crucifix

Keyword(s):

Global Climate ◽

Wave Train ◽

Ice Sheets ◽

Ice Sheet ◽

Summer Precipitation ◽

East China ◽

Thermal Contrast ◽

The Last Glacial Maximum ◽

The Last Glacial

Abstract. An Earth System Model of Intermediate Complexity is used to investigate the role of insolation and of the size of ice sheets on the regional and global climate of marine isotope stage (MIS) 13. The astronomical forcing is selected at two dates with opposite precession, one when northern hemisphere (NH) summer occurs at perihelion (at 506 ka (1 ka=1000 years) BP), and the other when it occurs at aphelion (at 495 ka BP). Five different volumes of the Eurasian ice sheet (EA) and North American ice sheet (NA), ranging from 0 to the Last Glacial Maximum (LGM) one, are used. The global cooling due to the ice sheets is mainly related to their area, little to their height. The regional cooling and warming anomalies caused by the ice sheets intensify with increasing size. Precipitation over different monsoon regions responds differently to the size of the ice sheets. Over North Africa and India, precipitation decreases with increasing ice sheet size due to the southward shift of the Intertropical Convergence Zone (ITCZ), whatever the astronomical configuration is. However, the situation is more complicated over East Asia. The ice sheets play a role through both reducing the land/ocean thermal contrast and generating a wave train which is topographically induced by the EA ice sheet. This wave train contributes to amplify the Asian land/ocean pressure gradient in summer and finally reinforces the precipitation. The presence of this wave train depends on the combined effect of the ice sheet size and insolation. When NH summer occurs at perihelion, the EA is able to induce this wave train whatever its size is, and this wave train plays a more important role than the reduction of the land/ocean thermal contrast. Therefore, the ice sheets reinforce the summer precipitation over East China whatever their sizes are. However, when NH summer occurs at aphelion, there is a threshold in the ice volume beyond which the wave train is not induced anymore. Therefore, below this threshold, the wave train effect is dominant and the ice sheets reinforce precipitation over East China. Beyond this threshold, the ice sheets reduce the precipitation mainly through reducing the land/ocean thermal contrast.

Download Full-text

Mapping and Discriminating Rural Settlements Using Gaofen-2 Images and a Fully Convolutional Network

Sensors ◽

10.3390/s20216062 ◽

2020 ◽

Vol 20 (21) ◽

pp. 6062

Author(s):

Ziran Ye ◽

Bo Si ◽

Yue Lin ◽

Qiming Zheng ◽

Ran Zhou ◽

...

Keyword(s):

Deep Learning ◽

Rural Areas ◽

Ground Truth ◽

Representation Learning ◽

Rural Settlement ◽

Spatial Characteristic ◽

Rural Settlements ◽

Convolutional Network ◽

Essential Information ◽

Multi Scale

New ongoing rural construction has resulted in an extensive mixture of new settlements with old ones in the rural areas of China. Understanding the spatial characteristic of these rural settlements is of crucial importance as it provides essential information for land management and decision-making. Despite a great advance in High Spatial Resolution (HSR) satellite images and deep learning techniques, it remains a challenging task for mapping rural settlements accurately because of their irregular morphology and distribution pattern. In this study, we proposed a novel framework to map rural settlements by leveraging the merits of Gaofen-2 HSR images and representation learning of deep learning. We combined a dilated residual convolutional network (Dilated-ResNet) and a multi-scale context subnetwork into an end-to-end architecture in order to learn high resolution feature representations from HSR images and to aggregate and refine the multi-scale features extracted by the aforementioned network. Our experiment in Tongxiang city showed that the proposed framework effectively mapped and discriminated rural settlements with an overall accuracy of 98% and Kappa coefficient of 85%, achieving comparable and improved performance compared to other existing methods. Our results bring tangible benefits to support other convolutional neural network (CNN)-based methods in accurate and timely rural settlement mapping, particularly when up-to-date ground truth is absent. The proposed method does not only offer an effective way to extract rural settlement from HSR images but open a new opportunity to obtain spatial-explicit understanding of rural settlements.

Download Full-text