scholarly journals A Novel Method for Automated Supraglacial Lake Mapping in Antarctica Using Sentinel-1 SAR Imagery and Deep Learning

2021 ◽  
Vol 13 (2) ◽  
pp. 197
Author(s):  
Mariel Dirscherl ◽  
Andreas J. Dietz ◽  
Christof Kneisel ◽  
Claudia Kuenzer
Keyword(s):  
Mass Loss ◽  
Ice Sheets ◽  
Greenland Ice Sheet ◽  
Detection Algorithm ◽  
Training Data ◽  
Ice Dynamics ◽  
Supraglacial Lakes ◽  
Spatio Temporal ◽  
Sar Imagery ◽  
Sentinel 2

Supraglacial meltwater accumulation on ice sheets can be a main driver for accelerated ice discharge, mass loss, and global sea-level-rise. With further increasing surface air temperatures, meltwater-induced hydrofracturing, basal sliding, or surface thinning will cumulate and most likely trigger unprecedented ice mass loss on the Greenland and Antarctic ice sheets. While the Greenland surface hydrological network as well as its impacts on ice dynamics and mass balance has been studied in much detail, Antarctic supraglacial lakes remain understudied with a circum-Antarctic record of their spatio-temporal development entirely lacking. This study provides the first automated supraglacial lake extent mapping method using Sentinel-1 synthetic aperture radar (SAR) imagery over Antarctica and complements the developed optical Sentinel-2 supraglacial lake detection algorithm presented in our companion paper. In detail, we propose the use of a modified U-Net for semantic segmentation of supraglacial lakes in single-polarized Sentinel-1 imagery. The convolutional neural network (CNN) is implemented with residual connections for optimized performance as well as an Atrous Spatial Pyramid Pooling (ASPP) module for multiscale feature extraction. The algorithm is trained on 21,200 Sentinel-1 image patches and evaluated in ten spatially or temporally independent test acquisitions. In addition, George VI Ice Shelf is analyzed for intra-annual lake dynamics throughout austral summer 2019/2020 and a decision-level fused Sentinel-1 and Sentinel-2 maximum lake extent mapping product is presented for January 2020 revealing a more complete supraglacial lake coverage (~770 km2) than the individual single-sensor products. Classification results confirm the reliability of the proposed workflow with an average Kappa coefficient of 0.925 and a F1-score of 93.0% for the supraglacial water class across all test regions. Furthermore, the algorithm is applied in an additional test region covering supraglacial lakes on the Greenland ice sheet which further highlights the potential for spatio-temporal transferability. Future work involves the integration of more training data as well as intra-annual analyses of supraglacial lake occurrence across the whole continent and with focus on supraglacial lake development throughout a summer melt season and into Antarctic winter.

A novel technique for automated mapping of Antarctic supraglacial lakes in Sentinel-1 SAR imagery using deep learning

2021 ◽  
Author(s):  
Mariel Dirscherl ◽  
Andreas Dietz ◽  
Celia Baumhoer ◽  
Christof Kneisel ◽  
Claudia Kuenzer
Keyword(s):  
Deep Learning ◽  
Sea Level Rise ◽  
Sea Level ◽  
Ice Sheets ◽  
Training Data ◽  
Supraglacial Lakes ◽  
Sar Imagery ◽  
Global Sea Level ◽  
The Antarctic ◽  
Sentinel 2

<p>Supraglacial meltwater accumulation on ice sheets and ice shelves can have considerable impact on ice discharge, mass balance and global sea-level-rise. With further increasing surface air temperatures, surface melting and resulting processes including hydrofracturing, meltwater penetration to the glacier bed as well as surface runoff will cumulate and most likely trigger unprecedented ice mass loss from the Greenland and Antarctic ice sheets. To date, the Antarctic surface hydrological network remains understudied calling for increased monitoring efforts and circum-Antarctic mapping strategies. This is particularly important given that Antarctica’s future contribution to global sea-level-rise is the largest uncertainty in current projections.</p><p>In this study, we present a novel methodology for Antarctic supraglacial lake extent mapping in Sentinel-1 Synthetic Aperture Radar imagery using state-of-the-art deep learning techniques. The method was implemented with the aim of complementing a previously developed supraglacial lake detection algorithm applying Machine Learning on optical Sentinel-2 data in order to deliver a more complete picture of Antarctic meltwater ponding compared to single-sensor mapping products. The deep learning model was trained on 21,200 Sentinel-1 image patches using a modified ResUNet for semantic segmentation of supraglacial lakes and evaluated by means of ten spatially or temporally independent Sentinel-1 test acquisitions distributed across the Antarctic continent. Besides, George VI Ice Shelf is analyzed for intra-annual lake dynamics throughout austral summer 2019/2020 and decision-level fused Sentinel-1 and Sentinel-2 maximum lake extent mapping products are presented for selected time periods. Future work involves the integration of more training data as well as the generation of circum-Antarctic mapping products using both, Sentinel-2 and Sentinel-1 derived lake extent mappings. These will be crucial for intra-annual analyses on supraglacial lake occurrence across the whole continent and associated drivers and impacts.</p>

scholarly journals Ice sheet mass loss caused by dust and black carbon accumulation

2015 ◽  
Vol 9 (2) ◽  
pp. 2563-2596
Author(s):  
T. Goelles ◽  
C. E. Bøggild ◽  
R. Greve
Keyword(s):  
Mass Loss ◽  
Black Carbon ◽  
Ice Sheet ◽  
Greenland Ice Sheet ◽  
Carbon Accumulation ◽  
Model Framework ◽  
Ice Dynamics ◽  
Ablation Area ◽  
Ice Surface ◽  
Surface Melt

Abstract. Albedo is the dominating factor governing surface melt variability in the ablation area of ice sheets and glaciers. Aerosols such as mineral dust and black carbon (soot) accumulate on the ice surface and cause a darker surface and therefore a lower albedo. The dominant source of these aerosols in the ablation area is melt-out of englacial material which has been transported via ice flow. The darkening effect on the ice surface is currently not included in sea level projections, and the effect is unknown. We present a model framework which includes ice dynamics, aerosol transport, aerosol accumulation and the darkening effect on ice albedo and its consequences for surface melt. The model is applied to a simplified geometry resembling the conditions of the Greenland ice sheet, and it is forced by several temperature scenarios to quantify the darkening effect of aerosols on future mass loss. The effect of aerosols depends non-linearly on the temperature rise due to the feedback between aerosol accumulation and surface melt. The effect of aerosols in the year 3000 is up to 12% of additional ice sheet volume loss in the warmest scenario.

scholarly journals Dark Glacier Surface of Greenland’s Largest Floating Tongue Governed by High Local Deposition of Dust

2020 ◽  
Vol 12 (22) ◽  
pp. 3793
Author(s):  
Angelika Humbert ◽  
Ludwig Schröder ◽  
Timm Schultz ◽  
Ralf Müller ◽  
Niklas Neckel ◽  
...  
Keyword(s):  
Black Carbon ◽  
Laser Scanner ◽  
Greenland Ice Sheet ◽  
Remote Sensing Data ◽  
Mineralogical Composition ◽  
Glacier Surface ◽  
Supraglacial Lakes ◽  
Governing Factor ◽  
Local Deposition ◽  
Sentinel 2

Surface melt, driven by atmospheric temperatures and albedo, is a strong contribution of mass loss of the Greenland Ice Sheet. In the past, black carbon, algae and other light-absorbing impurities were suggested to govern albedo in Greenland’s ablation zone. Here we combine optical (MODIS/Sentinel-2) and radar (Sentinel-1) remote sensing data with airborne radar and laser scanner data, and engage firn modelling to identify the governing factors leading to dark glacier surfaces in Northeast Greenland. After the drainage of supraglacial lakes, the former lake ground is a clean surface represented by a high reflectance in Sentinel-2 data and aerial photography. These bright spots move with the ice flow and darken by more than 20% over only two years. In contrast, sites further inland do not exhibit this effect. This finding suggests that local deposition of dust, rather than black carbon or cryoconite formation, is the governing factor of albedo of fast-moving outlet glaciers. This is in agreement with a previous field study in the area which finds the mineralogical composition and grain size of the dust comparable with that of the surrounding soils.

scholarly journals Greenland high-elevation mass balance: inference and implication of reference period (1961–90) imbalance

2015 ◽  
Vol 56 (70) ◽  
pp. 105-117 ◽  
Author(s):  
William Colgan ◽  
Jason E. Box ◽  
Morten L. Andersen ◽  
Xavier Fettweis ◽  
Beáta Csathó ◽  
...  
Keyword(s):  
Mass Loss ◽  
Mass Balance ◽  
Greenland Ice Sheet ◽  
Mass Gain ◽  
High Elevation ◽  
Reference Period ◽  
Input Output ◽  
Specific Mass ◽  
Ice Dynamics ◽  
Dynamic Mass

AbstractWe revisit the input–output mass budget of the high-elevation region of the Greenland ice sheet evaluated by the Program for Arctic Regional Climate Assessment (PARCA). Our revised reference period (1961–90) mass balance of 54±48 Gt a–1 is substantially greater than the 0±21 Gt a–1 assessed by PARCA, but consistent with a recent, fully independent, input–output estimate of high-elevation mass balance (41±61 Gt a–1). Together these estimates infer a reference period high-elevation specific mass balance of 4.8±5.4 cm w.e. a–1. The probability density function (PDF) associated with this combined input–output estimate infers an 81% likelihood of high-elevation specific mass balance being positive (>0 cm w.e. a–1) during the reference period, and a 70% likelihood that specific balance was >2 cm w.e. a–1. Given that reference period accumulation is characteristic of centurial and millennial means, and that in situ mass-balance observations exhibit a dependence on surface slope rather than surface mass balance, we suggest that millennial-scale ice dynamics are the primary driver of subtle reference period high-elevation mass gain. Failure to acknowledge subtle reference period dynamic mass gain can result in underestimating recent dynamic mass loss by ~17%, and recent total Greenland mass loss by ~7%.

Trends and projections in ice sheet mass balance

2020 ◽  
Author(s):  
Andrew Shepherd ◽  
Keyword(s):  
Mass Balance ◽  
Sea Level Rise ◽  
Sea Level ◽  
Ice Sheets ◽  
Ice Sheet ◽  
Greenland Ice Sheet ◽  
Ice Dynamics ◽  
Surface Mass ◽  
Sea Levels ◽  
Global Sea Level

<p>In recent decades, the Antarctic and Greenland Ice Sheets have been major contributors to global sea-level rise and are expected to be so in the future. Although increases in glacier flow and surface melting have been driven by oceanic and atmospheric warming, the degree and trajectory of today’s imbalance remain uncertain. Here we compare and combine 26 individual satellite records of changes in polar ice sheet volume, flow and gravitational potential to produce a reconciled estimate of their mass balance. <strong>Since the early 1990’s, ice losses from Antarctica and Greenland have caused global sea-levels to rise by 18.4 millimetres, on average, and there has been a sixfold increase in the volume of ice loss over time. Of this total, 41 % (7.6 millimetres) originates from Antarctica and 59 % (10.8 millimetres) is from Greenland. In this presentation, we compare our reconciled estimates of Antarctic and Greenland ice sheet mass change to IPCC projection of sea level rise to assess the model skill in predicting changes in ice dynamics and surface mass balance.  </strong>Cumulative ice losses from both ice sheets have been close to the IPCC’s predicted rates for their high-end climate warming scenario, which forecast an additional 170 millimetres of global sea-level rise by 2100 when compared to their central estimate.</p>

Automated mapping of Antarctic supraglacial lakes and streams using machine learning

2020 ◽  
Author(s):  
Mariel Dirscherl ◽  
Andreas Dietz ◽  
Celia Baumhoer ◽  
Christof Kneisel ◽  
Claudia Kuenzer
Keyword(s):  
Machine Learning ◽  
Mass Balance ◽  
Antarctic Peninsula ◽  
Ice Sheet ◽  
Ice Shelf ◽  
Ice Dynamics ◽  
Antarctic Ice Sheet ◽  
Supraglacial Lakes ◽  
The Antarctic ◽  
Sentinel 2

<p>Antarctica stores ~91 % of the global ice mass making it the biggest potential contributor to global sea-level-rise. With increased surface air temperatures during austral summer as well as in consequence of global climate change, the ice sheet is subject to surface melting resulting in the formation of supraglacial lakes in local surface depressions. Supraglacial meltwater features may impact Antarctic ice dynamics and mass balance through three main processes. First of all, it may cause enhanced ice thinning thus a potentially negative Antarctic Surface Mass Balance (SMB). Second, the temporary injection of meltwater to the glacier bed may cause transient ice speed accelerations and increased ice discharge. The last mechanism involves a process called hydrofracturing i.e. meltwater-induced ice shelf collapse caused by the downward propagation of surface meltwater into crevasses or fractures, as observed along large coastal sections of the northern Antarctic Peninsula. Despite the known impact of supraglacial meltwater features on ice dynamics and mass balance, the Antarctic surface hydrological network remains largely understudied with an automated method for supraglacial lake and stream detection still missing. Spaceborne remote sensing and data of the Sentinel missions in particular provide an excellent basis for the monitoring of the Antarctic surface hydrological network at unprecedented spatial and temporal coverage.</p><p>In this study, we employ state-of-the-art machine learning for automated supraglacial lake and stream mapping on basis of optical Sentinel-2 satellite data. With more detail, we use a total of 72 Sentinel-2 acquisitions distributed across the Antarctic Ice Sheet together with topographic information to train and test the selected machine learning algorithm. In general, our machine learning workflow is designed to discriminate between surface water, ice/snow, rock and shadow being further supported by several automated post-processing steps. In order to ensure the algorithm’s transferability in space and time, the acquisitions used for training the machine learning model are chosen to cover the full circle of the 2019 melt season and the data selected for testing the algorithm span the 2017 and 2018 melt seasons. Supraglacial lake predictions are presented for several regions of interest on the East and West Antarctic Ice Sheet as well as along the Antarctic Peninsula and are validated against randomly sampled points in the underlying Sentinel-2 RGB images. To highlight the performance of our model, we specifically focus on the example of the Amery Ice Shelf in East Antarctica, where we applied our algorithm on Sentinel-2 data in order to present the temporal evolution of maximum lake extent during three consecutive melt seasons (2017, 2018 and 2019).</p>

scholarly journals Modeling precipitation over ice sheets: an assessment using Greenland

2002 ◽  
Vol 48 (160) ◽  
pp. 70-80 ◽  
Author(s):  
Gerard H. Roe
Keyword(s):  
Accumulation Rate ◽  
Ice Sheets ◽  
Ice Sheet ◽  
Greenland Ice Sheet ◽  
Climate Forcing ◽  
Steady State Response ◽  
Ice Dynamics ◽  
Accumulation Pattern ◽  
State Response ◽  
Long Time

AbstractThe interaction between ice sheets and the rest of the climate system at long time-scales is not well understood, and studies of the ice ages typically employ simplified parameterizations of the climate forcing on an ice sheet. It is important therefore to understand how an ice sheet responds to climate forcing, and whether the reduced approaches used in modeling studies are capable of providing robust and realistic answers. This work focuses on the accumulation distribution, and in particular considers what features of the accumulation pattern are necessary to model the steady-state response of an ice sheet. We examine the response of a model of the Greenland ice sheet to a variety of accumulation distributions, both observational datasets and simplified parameterizations. The predicted shape of the ice sheet is found to be quite insensitive to changes in the accumulation. The model only differs significantly from the observed ice sheet for a spatially uniform accumulation rate, and the most important factor for the successful simulation of the ice sheet’s shape is that the accumulation decreases with height according to the ability of the atmosphere to hold moisture. However, the internal ice dynamics strongly reflects the influence of the atmospheric circulation on the accumulation distribution.

scholarly journals Modelling channelized surface drainage of supraglacial lakes

2015 ◽  
Vol 61 (225) ◽  
pp. 185-199 ◽  
Author(s):  
J. Kingslake ◽  
F. Ng ◽  
A. Sole
Keyword(s):  
Surface Water ◽  
Ice Sheets ◽  
Frictional Heat ◽  
Lake Area ◽  
Ice Dynamics ◽  
Surface Drainage ◽  
Supraglacial Lakes ◽  
Lake Model ◽  
Lake Drainage ◽  
Area Expansion

AbstractSupraglacial lakes can drain to the bed of ice sheets, affecting ice dynamics, or over their surface, relocating surface water. Focusing on surface drainage, we first discuss observations of lake drainage. In particular, for the first time, lakes are observed to drain >70 km across the Nivlisen ice shelf, East Antarctica. Inspired by these observations, we develop a model of lake drainage through a channel that incises into an ice-sheet surface by frictional heat dissipated in the flow. Modelled lake drainage can be stable or unstable. During stable drainage, the rate of lake-level drawdown exceeds the rate of channel incision, so discharge from the lake decreases with time; this can prevent the lake from emptying completely. During unstable drainage, discharge grows unstably with time and always empties the lake. Model lakes are more prone to drain unstably when the initial lake area, the lake input and the channel slope are larger. These parameters will vary during atmospheric-warming-induced ablation-area expansion, hence the mechanisms revealed by our analysis can influence the dynamic response of ice sheets to warming through their impact on surface-water routing and storage.

scholarly journals ON THE CLASSIFIER PERFORMANCE FOR SIMULATION BASED DEBRIS DETECTION IN SAR IMAGERY

2021 ◽  
Vol XLIII-B1-2021 ◽  
pp. 45-50
Author(s):  
S. Kuny ◽  
H. Hammer ◽  
K. Schulz
Keyword(s):  
Random Forest ◽  
Damage Detection ◽  
Urban Areas ◽  
Detection Algorithm ◽  
Performance Comparison ◽  
Random Forest Classifier ◽  
Training Data ◽  
Support Vector ◽  
Effective Performance ◽  
Sar Imagery

Abstract. Urban areas struck by disasters such as earthquakes are in need of a fast damage detection assessment. A post-event SAR image often is the first available image, most likely with no matching pre-event image to perform change detection. In previous work we have introduced a debris detection algorithm for this scenario that is trained exclusively with synthetically generated training data. A classification step is employed to separate debris from similar textures such as vegetation. In order to verify the use of a random forest classifier for this context, we conduct a performance comparison with two alternative popular classifiers, a support vector machine and a convolutional neural network. With the direct comparison revealing the random forest classifier to be best suited, the effective performance on the prospect of debris detection is investigated for the post-earthquake Christchurch scene. Results show a good separation of debris from vegetation and gravel, thus reducing the false alarm rate in the damage detection operation considerably.

scholarly journals Wintertime storage of water in buried supraglacial lakes across the Greenland Ice Sheet

2014 ◽  
Vol 8 (4) ◽  
pp. 3999-4031 ◽  
Author(s):  
L. S. Koenig ◽  
D. J. Lampkin ◽  
L. N. Montgomery ◽  
S. L. Hamilton ◽  
J. B. Turrin ◽  
...  
Keyword(s):  
Mass Loss ◽  
Total Mass ◽  
Ice Sheet ◽  
Greenland Ice Sheet ◽  
Subsurface Water ◽  
Blue Color ◽  
Total Mass Loss ◽  
Supraglacial Lakes ◽  
Hydrologic System ◽  
Surface Melt

Abstract. Surface melt over the Greenland Ice Sheet (GrIS) is increasing and estimated to account for half or more of the total mass loss. Little, however, is known about the hydrologic pathways that route surface melt within the ice sheet. In this study, we present over-winter storage of water in buried supraglacial lakes as one hydrologic pathway for surface melt, referred to as buried lakes. Airborne radar echograms are used to detect the buried lakes that are distributed extensively around the margin of the GrIS. The subsurface water can persist through multiple winters and is, on average, ~4.2 + 0.4 m below the surface. The few buried lakes that are visible at the surface of the GrIS have a~unique visible signature associated with a darker blue color where subsurface water is located. The volume of retained water in the buried lakes is likely insignificant compared to the total mass loss from the GrIS but the water will have important implications locally for the development of the englacial hydrologic network, ice temperature profiles and glacial dynamics. The buried lakes represent a small but year-round source of meltwater in the GrIS hydrologic system.

Sign in / Sign up

Export Citation Format

Share Document