Antarctica, the coldest, windiest, and most remote continent on our planet, plays a crucial role in the global climate system. Its ice mass loss is a major driver of rising sea levels, with projections indicating a potential rise of several meters in the coming centuries. However, there remains considerable uncertainty about the future trajectory of Antarctic mass loss. A major area of uncertainty is the fate of ice shelves—floating extensions of land ice that surround much of Antarctica and act as barriers, slowing the flow of glaciers into the ocean. Ice shelves are affected by warm water from below, which thins them and increases their vulnerability to cracking, as well as by warm air from above, which melts the surface and forms ponds of meltwater.

This research focuses on surface melt, a phenomenon where meltwater forms and either refreezes or accumulates on the ice shelf surface. When the water accumulates, it can seep into cracks, causing them to deepen and widen, which can weaken the ice shelves. In today’s era of abundant satellite imagery and advanced deep learning techniques, we can efficiently process large volumes of data, enabling more comprehensive research on surface melt dynamics. The aim of this dissertation is to enhance the mapping and understanding of surface meltwater on Antarctic ice shelves using remote sensing and deep learning methods.

The introductory chapter provides an overview of the Antarctic Ice Sheet, emphasizing the continent's immense scale and importance. Written in an accessible style, it presents key concepts about Antarctica and explores how ice shelves and surface melt influence the continent. The chapter also describes the use of satellite data to map surface melt and discusses advancements in computational resources and deep learning, which have significantly improved our ability to analyze the expanding catalog of satellite data. It concludes with an overview of the research questions addressed in the thesis.

In the second chapter, various remote sensing datasets are compared to illustrate how and why satellite observations of surface melt differ. Using state-of-the-art melt detection algorithms, we analyze surface melt patterns and observe large differences, especially in icy areas, regions with subsurface melt, and during winter. These differences arise from factors such as satellite overpass times, spatial resolution, signal penetration, cloud cover, and detection methods. Despite these challenges, the variations create opportunities to combine data from multiple satellites, enhancing the overall accuracy of surface melt detection across Antarctica.

The third chapter builds on the previous findings and addresses the challenge of balancing spatial and temporal resolution in satellite observations. Surface melt in Antarctica is highly dynamic and varies regionally, making high-resolution mapping essential. To tackle this, we develop UMelt, a surface melt dataset for all Antarctic ice shelves with high spatial (500 m) and temporal (12 h) resolution, covering the period from 2016 to 2021. Our deep learning model integrates data from multiple satellites, allowing for detailed detection of surface melt while maintaining high temporal resolution. UMelt offers the potential for new insights into how ice shelves respond to changing atmospheric conditions.

In the fourth chapter, we shift from mapping the presence of surface melt to estimating its volume. Since surface melt is mainly driven by local processes, high-resolution regional climate models (RCMs) are necessary. However, current RCMs have a coarse resolution (25--30 km) that is insufficient for capturing small-scale melt processes. To address this, we introduce SUPREME, a deep learning method that downscales surface melt to 5.5 km resolution using a physically-informed super-resolution model. This model combines remote sensing data on albedo and elevation with a 27 km resolution Regional Atmospheric Climate Model (RACMO), accounting for the diverse drivers of surface melt across Antarctica. SUPREME demonstrates the potential of super-resolution techniques with physical constraints for high-resolution surface melt mapping, providing valuable insights into localized melting patterns.

The fifth chapter examines the hydrology of surface meltwater lakes on Antarctica, investigating whether they refreeze or drain into fractures at the end of the melt season, potentially destabilizing ice shelves. Monitoring these lakes with optical satellite imagery is often limited by cloud cover, complicating the tracking of their changes over time. To overcome this, we develop a spatiotemporal deep learning model using radar imagery from Sentinel-1, which allows us to classify the evolution of meltwater lakes regardless of cloud conditions. Our findings reveal no clear connections between lake evolution and ice shelf parameters, highlighting the need for further research and model refinement. The study is an initial step in using deep learning and Sentinel-1 data to monitor the evolution of supraglacial lakes on Antarctic ice shelves.

The sixth and final chapter reflects on the research and outlines future directions. It begins by summarizing the state of Antarctic surface melt research at the start of my PhD. The chapter then highlights the key contributions of this thesis and concludes with three proposed research ideas aimed at advancing our understanding of surface melt processes in Antarctica.

Damage features, such as rifts and crevasses, are the first signs of a weakened ice shelf and the precursor for retreat. Yet, damage changes are not widely quantified on Antarctic ice shelves, leaving future ice shelf weakening poorly understood. Here we use satellite imagery to detect both long-term (24-year) and short-term (annual, 2015–2021) Antarctic-wide damage changes, revealing a multiyear damage development cycle strongly correlated to ice shelf area changes, and a net decline in damaged area from 1997 to 2021. We establish a data-driven link between damage and ice flow characteristics, which shows that ice flow acceleration, strain rate increases and thinning lead to more damage development, in particular under high-emission climate scenarios. This sensitivity to warming suggests that without quantification of damage impacts by detailed physical models the (timing of) ice shelf retreat and Antarctic mass loss may currently be underestimated.

To investigate the drivers of lake drainages in Antarctica, we analyzed optical remote sensing data from the Shackleton Ice Shelf in East Antarctica over seven melt seasons, 2016 to 2023. Our study identified seven drainage event in 2016-2017, one in 2018-2019, fifteen in 2019-2020, and two in 2020-2021. All identified drainages occurred in regions with relatively medium to high levels of satellite-derived ice shelf damage and, except one, all with active damage development. Additionally, 17 out of 25 drainages coincided with increases in tidal heights. These findings provide insights into the factors influencing current lake drainages in Antarctica in both timing and distribution.

Assessing the Surface Mass Balance (SMB) of the Antarctic Ice Sheet is crucial for understanding its response to climate change. Synthetic Aperture Radar observations from Sentinel-1 provide the potential to monitor the variability of SMB processes through changes in the scattering response of near-surface and internal snow layers. However, the interplay between several factors, such as accumulation, wind erosion, deposition, and melt, complicates the interpretation of scattering changes of the microwave signal. Additionally, lack of reliable ground truth measurements of the snow surface limits our capability to associate the SMB processes with dominant scattering mechanism. In this study, we aim to quantify the dominant scattering in Sentinel-1 signal and evaluate the scattering changes in drifting snow-dominated regions of East Antarctica. We introduce a scattering indicator, alpha -{text{scat},varepsilon }, derived from scattering-type and entropy descriptors, providing a measure between volume and pure scattering. By relating the field measurements to alpha -{text{scat},varepsilon }, we establish that the evolution of dominant scattering in the presence of snowdrift is complex. First, alpha -{text{scat},varepsilon } strongly correlates with surface roughness (R^{2}=0.92, RMSE =2^circ). Spatially variable erosion patterns significantly increase the roughness and result in a strong affinity towards pure scattering despite net accumulation. Second, high surface densities also tend to influence pure scattering; however, the effect is dependent on the accumulation rate. With more accumulation, we observe an increasing dominance of volume scattering from internal snow layers. Long-term trends in alpha -{text{scat},varepsilon } (2017/2023) further suggest that it is challenging to address the causes behind the scattering source based on a single snow surface process. We thus demonstrate the potential and limitations of alpha -{text{scat},varepsilon } to infer the variability in dominant scattering from changes in surface processes.

Correction to: Nature Reviews Earth & Environment https://doi.org/10.1038/s43017-023-00507-9, published online 23 January 2024.

In the version of the article initially published, in Fig. 5, under “Radar altimeter”, “O(16–160 m)” previously read “O(16–160 km)”. This has now been corrected in the HTML and PDF versions of the article.

Most of the Greenland and Antarctic ice sheets are covered with firn — the transitional material between snow and glacial ice. Firn is vital for understanding ice-sheet mass balance and hydrology, and palaeoclimate. In this Review, we synthesize knowledge of firn, including its formation, observation, modelling and relevance to ice sheets. The refreezing of meltwater in the pore space of firn currently prevents 50% of meltwater in Greenland from running off into the ocean and protects Antarctic ice shelves from catastrophic collapse. Continued atmospheric warming could inhibit future protection against mass loss. For example, warming in Greenland has already contributed to a 5% reduction in firn pore space since 1980. All projections of future firn change suggest that surface meltwater will have an increasing impact on firn, with melt occurring tens to hundreds of kilometres further inland in Greenland, and more extensively on Antarctic ice shelves. Although progress in observation and modelling techniques has led to a well-established understanding of firn, the large uncertainties associated with meltwater percolation processes (refreezing, ice-layer formation and storage) must be reduced further. A tighter integration of modelling components (firn, atmosphere and ice-sheet models) will also be needed to better simulate ice-sheet responses to anthropogenic warming and to quantify future sea-level rise.

Surface melting occurs across many of Antarctica’s ice shelves, mainly during the austral summer. The onset, duration, area and fate of surface melting varies spatially and temporally, and the resultant surface meltwater is stored as ponded water (lakes) or as slush (saturated firn or snow), with implications for ice-shelf hydrofracture, firn air content reduction, surface energy balance and thermal evolution. This study applies a machine-learning method to the entire Landsat 8 image catalogue to derive monthly records of slush and ponded water area across 57 ice shelves between 2013 and 2021. We find that slush and ponded water occupy roughly equal areas of Antarctica’s ice shelves in January, with inter-regional variations in partitioning. This suggests that studies that neglect slush may substantially underestimate the area of ice shelves covered by surface meltwater. Furthermore, we found that adjusting the surface albedo in a regional climate model to account for the lower albedo of surface meltwater resulted in 2.8 times greater snowmelt across five representative ice shelves. This extra melt is currently unaccounted for in regional climate models, which may lead to underestimates in projections of ice-sheet melting and ice-shelf stability.

Despite in-situ observations of perennial firn aquifers (PFAs) at specific locations of the Antarctic ice sheet, a comprehensive continent-wide mapping of PFA distribution is currently lacking. We present an estimate of their distribution across Antarctica in the form of a probability assessment using a Monte Carlo technique. Our approach involves a novel methodology that combines observations from Sentinel-1 and Advanced SCATterometer (ASCAT) with output from a regional climate model. To evaluate our method, we conduct an extensive comparison with Operation Ice Bridge observations from the Greenland Ice Sheet. Application to Antarctica reveals high PFA probabilities in the Antarctic Peninsula (AP), particularly along its northern, northwestern, and western coastlines, as well as on the Wilkins, Müller, and George VI ice shelves. Outside the AP, PFA probability is low, except for some locations with marginally higher probabilities, such as on the Abbot, Totten, and Shackleton ice shelves.

While the influence of surface melt on Antarctic ice shelf stability can be large, the duration and affected area of melt events are often small. Therefore, melt events are difficult to capture with remote sensing, as satellite sensors always face the trade-off between spatial and temporal resolution. To overcome this limitation, we developed UMelt: a surface melt record for all Antarctic ice shelves with a high spatial (500 m) and high temporal (12 h) resolution for the period 2016–2021. Our approach is based on a deep learning model, specifically a U-Net, which was developed in Google Earth Engine. The U-Net combines microwave remote sensing observations from three sources: Sentinel-1, Special Sensor Microwave Imager/Sounder (SSMIS), and Advanced Scatterometer (ASCAT). The U-Net was trained on the Shackleton Ice Shelf for melt seasons 2017–2021, using the fine-scale melt patterns of Sentinel-1 as reference data and SSMIS, ASCAT, a digital elevation model, and multi-year Sentinel-1 melt fraction as predictors. The trained U-Net performed well on the Shackelton Ice Shelf for test melt season 2016–2017 (accuracy: 91.3%; F1-score: 86.9%), and the Larsen C Ice Shelf, which was not considered during training (accuracy: 91.0%; F1-score: 89.3%). Using the trained U-Net model, we have successfully developed the UMelt record. UMelt allows Antarctic-wide surface melt to be detected at a small scale while preserving a high temporal resolution, which could lead to new insights into the response of ice shelves to a changing atmospheric forcing.

Surface melt is an important driver of ice shelf disintegration and its consequent mass loss over the Antarctic Ice Sheet. Monitoring surface melt using satellite remote sensing can enhance our understanding of ice shelf stability. However, the sensors do not measure the actual physical process of surface melt, but rather observe the presence of liquid water. Moreover, the sensor observations are influenced by the sensor characteristics and surface properties. Therefore, large inconsistencies can exist in the derived melt estimates from different sensors. In this study, we apply state-of-the-art melt detection algorithms to four frequently used remote sensing sensors, i.e., two active microwave sensors, which are Advanced Scatterometer (ASCAT) and Sentinel-1, a passive microwave sensor, i.e., Special Sensor Microwave Imager/Sounder (SSMIS), and an optical sensor, i.e., Moderate Resolution Imaging Spectroradiometer (MODIS). We intercompare the melt detection results over the entire Antarctic Ice Sheet and four selected study regions for the melt seasons 2015-2020. Our results show large spatiotemporal differences in detected melt between the sensors, with particular disagreement in blue ice areas, in aquifer regions, and during wintertime surface melt. We discuss that discrepancies between sensors are mainly due to cloud obstruction and polar darkness, frequency-dependent penetration of satellite signals, temporal resolution, and spatial resolution, as well as the applied melt detection methods. Nevertheless, we argue that different sensors can complement each other, enabling improved detection of surface melt over the Antarctic Ice Sheet.

River ice is a major contributor to flood risk in cold regions due to the physical impediment of flow caused by ice jamming. Although a variety of classifiers have been developed to distinguish ice types using HH or VV intensity of SAR data, mostly based on data from RADARSAT-1 and -2, these classifiers still experience problems with breakup classification, because meltwater development causes overlap in co-polarization backscatter intensities of open water and sheet ice pixels. In this study, we develop a Random Forest classifier based on multiple features of Sentinel-1 data for three main classes generally present during breakup: rubble ice, sheet ice and open water, in a case study over the Athabasca River in Canada. For each ice stage, intensity of the VV and VH backscatter, pseudo-polarimetric decomposition parameters and Grey Level Co-occurrence Matrix texture features were computed for 70 verified sample areas. Several classifiers were developed, based on i) solely intensity features or on ii) a combination of intensity, pseudo-polarimetric and texture features and each classifier was evaluated based on Recursive Feature Elimination with Cross-Validation and pair-wise correlation of the studied features. Results show improved classifier performance when including GLCM mean of VV intensity, and VH intensity features instead of the conventional classifier based solely on intensity. This highlights the complementary nature of texture and intensity for the classification of breaking river ice. GLCM mean incorporates spatial patterns of the co-polarized intensity and sensitivity to context, while VH intensity introduces cross-polarized surface and volume scattering signals and is less sensitive to wind than the commonly used co-polarized intensity. We conclude that the proposed method based on the combination of texture and intensity features is suitable for and performs well in physically complex situations such as breakup, which are hard to classify otherwise. This method has a high potential for classifying river ice operationally, also for data from other SAR missions. Since it is a generic approach, it also has potential to classify river ice along other rivers globally.