Iceberg Area Research Articles

Mapping the extent of giant Antarctic icebergs with deep learning

Abstract. Icebergs release cold, fresh meltwater and terrigenous nutrients as they drift and melt, influencing the local ocean properties, encouraging sea ice formation and biological production. To locate and quantify the fresh water flux from Antarctic icebergs, changes in their area and thickness have to be monitored along their trajectories. While the locations of large icebergs are operationally tracked by manual inspection, delineation of their extent is not. Here, we propose a U-net approach to automatically map the extent of giant icebergs in Sentinel-1 imagery. This greatly improves the efficiency compared to manual delineations, reducing the time for each outline from several minutes to less than 0.01 s. We evaluate the performance of our U-net and two state-of-the-art segmentation algorithms (Otsu and k-means) on 191 images. For icebergs larger than those covered by the training data, we find that U-net tends to miss parts. Otherwise, U-net is more robust in scenes with complex backgrounds – ignoring sea ice, smaller regions of nearby coast or other icebergs – and outperforms the other two techniques by achieving an F1 score of 0.84 and an absolute median deviation in iceberg area of 4.1 %.

Automated ArcticDEM iceberg detection tool: insights into area and volume distributions, and their potential application to satellite imagery and modelling of glacier–iceberg–ocean systems

Abstract. Iceberg calving accounts for up to half of mass loss from the Greenland Ice Sheet (GrIS), with their size distributions providing insights into glacier calving dynamics and impacting fjord environments through their melting and subsequent freshwater release. Iceberg area and volume data for the GrIS are currently limited to a handful of fjord locations, while existing approaches to iceberg detection are often time-consuming and are not always suited for long time series analysis over large spatial scales. This study presents a highly automated workflow that detects icebergs and appends their associated metadata within Google Earth Engine using high spatial resolution timestamped ArcticDEM (Arctic Digital Elevation Model) strip data. This is applied to three glaciers that exhibit a range of different iceberg concentrations and size distributions: Sermeq Kujalleq (Jakobshavn Isbræ), Umiammakku Isbræ and Kangiata Nunaata Sermia. A total of 39 ArcticDEM scenes are analysed, detecting a total of 163 738 icebergs with execution times of 6 min to 2 h for each glacier depending on the number of DEMs available and total area analysed, comparing well with the mapping of manually digitised outlines. Results reveal two distinct iceberg distributions at Sermeq Kujalleq and Kangiata Nunaata Sermia where iceberg density is high, and one distribution at Umiammakku Isbræ where iceberg density is low. Small icebergs (< 1000 m2) are found to account for over 80 % of each glacier's icebergs; however, they only contribute to 10 %–37 % of total iceberg volume suggesting that large icebergs are proportionally more important for glacier mass loss and as fjord freshwater reservoirs. The overall dataset is used to construct new area-to-volume conversions (with associated uncertainties) that can be applied elsewhere to two-dimensional iceberg outlines derived from optical or synthetic aperture radar imagery. When data are expressed in terms of total iceberg count and volume, insight is provided into iceberg distributions that have potential applicability to observations and modelling of iceberg calving behaviour and fjord freshwater fluxes. Due to the speed and automated nature of our approach, this workflow offers the potential to interrogate iceberg data on a pan-Arctic scale where ArcticDEM strip data coverage allows.

Significant contribution of small icebergs to the freshwater budget in Greenland fjords

Icebergs represent nearly half of the mass loss from the Greenland Ice Sheet and provide a distributed source of freshwater along fjords which can alter fjord circulation, nutrient levels, and ultimately the Meridional Overturning Circulation. Here we present analyses of high resolution optical satellite imagery using convolutional neural networks to accurately delineate iceberg edges in two East Greenland fjords. We find that a significant portion of icebergs in fjords are comprised of small icebergs that were not detected in previously-available coarser resolution satellite images. We show that the preponderance of small icebergs results in high freshwater delivery, as well as a short life span of icebergs in fjords. We conclude that an inability to identify small icebergs leads to inaccurate frequency-size distribution of icebergs in Greenland fjords, an underestimation of iceberg area (specifically for small icebergs), and an overestimation of iceberg life span.

Detailed Lacustrine Calving Iceberg Inventory from Very High Resolution Optical Imagery and Object-Based Image Analysis

In the field of iceberg and glacier calving studies, it is important to collect comprehensive datasets of populations of icebergs. Particularly, calving of lake-terminating glaciers has been understudied. The aim of this work is to present an object-based method of iceberg detection and to create an inventory of icebergs located in a proglacial lagoon of San Quintín glacier, Northern Patagonia Icefield, Chile. This dataset is created using high-resolution WorldView-2 imagery and a derived DEM. We use it to briefly discuss the iceberg size distribution and area–volume scaling. Segmentation of the multispectral imagery produced a map of objects, which were classified with use of Random Forest supervised classification algorithm. An intermediate classification product was corrected with a ruleset exploiting contextual information and a watershed algorithm that was used to divide multiple touching icebergs into separate objects. Common theoretical heavy-tail statistical distributions were tested as descriptors of iceberg area and volume distributions. Power law models were proposed for the area–volume relationship. The proposed method performed well over the open lake detecting correctly icebergs in all size bands except 5–15 m2 where many icebergs were missed. A section of the lagoon with ice melange was not reliably mapped due to uniformity of the area in the imagery and DEM. The precision of the DEM limited the scaling effort to icebergs taller than 1.7 m and larger than 99 m2, despite the inventory containing icebergs as small as 4 m2. The work demonstrates viability of object-based analysis for lacustrine iceberg detection and shows that the statistical properties of iceberg population at San Quintín glacier match those of populations produced by tidewater glaciers.

Mass Loss From Calving in Himalayan Proglacial Lakes

The formation and expansion of Himalayan glacial lakes has implications for glacier dynamics, mass balance and glacial lake outburst floods (GLOFs). Subaerial and subaqueous calving is an important component of glacier mass loss but they have been difficult to track due to spatiotemporal resolution limitations in remote sensing data and few field observations. In this study, we used near-daily 3 m resolution PlanetScope imagery in conjunction with an uncrewed aerial vehicle (UAV) survey to quantify calving events and derive an empirical area–volume relationship to estimate calved glacier volume from planimetric iceberg areas. A calving event at Thulagi Glacier in 2017 was observed by satellite from before and during the event to nearly complete melting of the icebergs, and was observed in situ midway through the melting period, thus giving insights into the melting processes. In situ measurements of Thulagi Lake’s surface and water column indicate that daytime sunlight absorption heats mainly just the top metre of water, but this heat is efficiently mixed downwards through the top tens of metres due to forced convection by wind-blown icebergs; this heat then is retained by the lake and is available to melt the icebergs. Using satellite data, we assess seasonal glacier velocities, lake thermal regime and glacier surface elevation change for Thulagi, Lower Barun and Lhotse Shar glaciers and their associated lakes. The data reveal widely varying trends, likely signifying divergent future evolution. Glacier velocities derived from 1960/70s declassified Corona satellite imagery revealed evidence of glacier deceleration for Thulagi and Lhotse Shar glaciers, but acceleration at Lower Barun Glacier following lake development. We used published modelled ice thickness data to show that upon reaching their maximum extents, Imja, Lower Barun and Thulagi lakes will contain, respectively, about 90 × 106 , 62 × 106 and 5 × 106 m3 of additional water compared to their 2018 volumes. Understanding lake–glacier interactions is essential to predict future glacier mass loss, lake formation and associated hazards.

Spatiotemporal distributions of icebergs in a temperate fjord: Columbia Fjord, Alaska

Abstract. Much of the world's ice enters the ocean via outlet glaciers terminating in fjords. Inside fjords, icebergs may affect glacier–ocean interactions by cooling incoming ocean waters, enhancing vertical mixing, or providing back stress on the terminus. However, relatively few studies have been performed on iceberg dynamics inside fjords, particularly outside of Greenland. We examine icebergs calved from Columbia Glacier, Alaska, over 8 months spanning late winter to mid-fall using 0.5 m resolution satellite imagery, identifying icebergs based on pixel brightness. Iceberg sizes fit a power-law distribution with an overall power-law exponent, m, of -1.26±0.05. Seasonal variations in the power-law exponent indicate that brittle fracture of icebergs is more prevalent in the summer months. Combining our results with those from previous studies of iceberg distributions, we find that iceberg calving rate, rather than water temperature, appears to be the major control on the exponent value. We also analyze icebergs' spatial distribution inside the fjord and find that large icebergs (10 000–100 000 m2 cross-sectional area) have low spatial correlation with icebergs of smaller sizes due to their tendency to ground on shallow regions. We estimate the surface area of icebergs in contact with incoming seawater to be 3.0±0.63×104 m2. Given the much larger surface area of the terminus, 9.7±3.7×105 m2, ocean interactions with the terminus may have a larger impact on ocean heat content than interactions with icebergs.

Monitoring the tabular icebergs C28A and C28B calved from the Mertz Ice Tongue using radar remote sensing data

We analyzed the evolution processes of two tabular icebergs, C28A and C28B, originating from iceberg C28, using continuous multisource radar remote sensing data. The monitoring began during the calving of iceberg C28 from the Mertz Ice Tongue in February 2010 and ended in April 2012. The evolution of the iceberg area was determined from ENVISAT ASAR images, and the freeboard change was derived from CryoSat-2 profiles. Two patterns of iceberg area decreases are observed during iceberg drifting. In austral summer (from October to next March), iceberg area shows a gradual decrease and it is interpreted to represent edge wasting, sidewall melting and wave erosion, the gradual decrease of the area of C28A occurred at rates between 8.1 km2/month and 16.5 km2/month. In austral winter (from April to September), iceberg area has several discrete decreases (sudden drop) while it remained constant for most of the time, and it is interpreted to be large rift calving primarily caused by ocean swell and basal melting. Calving area of the two studied icebergs is observed to be greater in the winter. The freeboard decreased at different rates at different locations on the same iceberg (between −0.02 ± 0.22 and −0.82 ± 0.28 m/month for C28A and C28B), suggesting a laterally inhomogeneous melting of icebergs. The average rate of change of the C28B freeboard during its 22-month grounding period was −0.03 ± 0.01 m/month, which is considerably smaller than the −0.76 ± 0.23 m/month observed when it was freely drifting in the ocean. The cold-water regime surrounding the iceberg can help the iceberg maintain a stable state. This study reveals the potential use of radar remote sensing when monitoring Antarctic icebergs.

Iceberg properties and distributions in three Greenlandic fjords using satellite imagery

ABSTRACT Icebergs calved from tidewater glaciers represent about one third to one half of the freshwater flux from the Greenland ice sheet to the surrounding ocean. Using multiple satellite datasets, we quantify the first fjord-wide distributions of iceberg sizes and characteristics for three fjords with distinct hydrography and geometry: Sermilik Fjord, Rink Isbræ Fjord and Kangerlussuup Sermia Fjord. We estimate average total iceberg volumes in summer in the three fjords to be 6.4 ± 1.5, 1.7 ± 0.40 and 0.16 ± 0.09 km3, respectively. Iceberg properties are influenced by glacier calving style and grounding line depth, with variations in size distribution represented by exponents of power law distributions that are −1.95 ± 0.06, −1.87 ± 0.05 and −1.62 ± 0.04, respectively. The underwater surface area of icebergs exceeds the subsurface area of glacial termini by at least one order of magnitude in all three fjords, underscoring the need to include iceberg melt in fjord freshwater budgets. Indeed, in Sermilik Fjord, we calculate summertime freshwater flux from iceberg melt of 620 m3 s−1 (±140 m3 s−1), similar in magnitude to subglacial discharge. The method developed here can be extended across Greenland to assess relationships between glacier calving, iceberg discharge and freshwater production.

Estimates of iceberg submarine melting from high-resolution digital elevation models: application to Sermilik Fjord, East Greenland

AbstractObserved increases in iceberg discharge from Greenland’s marine-terminating glaciers over the past two decades have altered the freshwater flux from glacial fjords into surrounding ocean basins. Although variations in freshwater flux due to ice-sheet discharge change have been investigated on a broad scale, the distribution of the freshwater flux due to melting of calved glacier ice (i.e. icebergs) has not been examined. Logistical challenges to collecting in situ data in glacial fjords have so far prevented a detailed examination of freshwater fluxes arising from melting beneath the waterline (i.e. submarine melting). Here we demonstrate that submarine melting of icebergs can be quantified using repeat digital elevation models derived from very high-resolution stereo satellite images. Analysis of volume changes for icebergs in Sermilik Fjord, East Greenland, yield area-averaged submarine melt rates of ~0.39 m d–1. These rates are in relatively good agreement with simulated winter melt rates along the submerged portion of the Helheim Glacier terminus, providing independent validation of the applied technique. Further, the volume flux of fresh water from iceberg melting scales with surface and submerged iceberg areas, which suggests that iceberg meltwater may be an important freshwater component in fjords with high iceberg concentrations and/or expansive ice melange.

A technique for the identification and analysis of icebergs in synthetic aperture radar images of Antarctica

This paper describes an image analysis technique developed to identify icebergs depicted in synthetic aperture radar images of Antarctica and to determine the outlines of these icebergs. The technique uses a pixel bonding process to delineate the edges of the icebergs. It then separates them from the background water and sea ice by an edge-guided image segmentation process. Characteristics such as centroid position and iceberg area were calculated for each iceberg segment and placed in a file for input to appropriate statistical data analysis software. The technique has been tested on three ERS-1 SAR sub-images in which it succeeded in identifying virtually all segments containing icebergs of size six pixels or larger. The images were first passed through an averaging filter to reduce speckle. This process produced a pixel size of 100m x 100m. As implemented, the technique overestimates iceberg areas by about 20% on average and the detection rate falls off rapidly for icebergs less than six pixels in size. Performance in these areas is expected to improve when additional stages, based on a more detailed analysis of pixel intensity, are implemented.