Cryosphere Lab @ CUHK

Abstract Rock glaciers are distinctive debris landforms found worldwide in cold mountainous regions. They express the long-term movement of perennially frozen ground. Rock Glacier Velocity (RGV), defined as the time series of the annualized surface velocity of a rock glacier unit or a part of it, has been accepted as an Essential Climate Variable Permafrost Quantity in 2022. This review aims to highlight the relationship between rock glacier velocity and climatic factors, emphasizing the scientific relevance of interannual rock glacier velocity in generating RGV products within the context of observed rock glacier kinematics. Under global warming, rock glacier velocity exhibits widespread (multi-)decennial acceleration. This acceleration varies regionally in onset timing (from the 1950s to the 2010s) and magnitude (up to a factor of 10), and has been observed in regions such as the European Alps, High Mountain Asia, and the Andes. Despite different local conditions, a synchronous interannual velocity pattern prevails in the European Alps since the 2000s, highlighting the primary influence of climate. A common pattern is the seasonal velocity rhythm, which peaks in late summer to autumn and declines in spring. RGV assesses permafrost evolution via (multi-)decennial and interannual changes in rock glacier velocity, influenced by air temperature shifts with varying time lags and snow cover effects. Although not integrated into the RGV products, seasonal variations should be examined. This rhythmic behavior is attributed to alterations in pore water pressure influenced by air temperature, snow cover, and ground water conditions.

Rock glaciers – periglacial landforms commonly found in high-mountain systems – are of significant scientific value for inferring the presence of permafrost, understanding mountain hydrology, and assessing climate impacts on high-mountain environments. However, inventories remain patchy in many alpine regions, and as a result they are poorly understood for some areas of High Mountain Asia such as the Tibetan Plateau. To address this gap, we compiled a comprehensive inventory of rock glaciers for the Tibetan Plateau, i.e., TPRoGI (v1.0), developed using an innovative deep learning method. This inventory consists of a total of 44 273 rock glaciers, covering approximately 6000 km2, with a mean area of 0.14 km2. They are predominantly situated at elevations ranging from 4000 to 5500 m a.s.l., with a mean of 4729 m a.s.l. They tend to occur on slopes with gradients between 10 and 25\,^∘, with a mean of 17.7\,^∘. Across the plateau, rock glaciers are widespread in the northwestern and southeastern areas, with dense concentrations in the Western Pamir and Nyainqêntanglha, while they are sparsely distributed in the inner part. Our inventory serves as a benchmark dataset, which will be further maintained and updated in the future. This dataset constitutes a significant contribution towards understanding, future monitoring, and assessment of permafrost on the Tibetan Plateau in the context of climate change.

The Greenland ice sheet (GrIS) is at present the largest single contributor to global-mass-induced sea-level rise, primarily because of Arctic amplification on an increasingly warmer Earth. However, the processes of englacial water accumulation, storage and ultimate release remain poorly constrained. Here we show that a noticeable amount of the summertime meltwater mass is temporally buffered along the entire GrIS periphery, peaking in July and gradually reducing thereafter. Our results arise from quantifying the spatiotemporal behaviour of the total mass of water leaving the GrIS by analysing bedrock elastic deformation measured by Global Navigation Satellite System (GNSS) stations. The buffered meltwater causes a subsidence of the bedrock close to GNSS stations of at most approximately 5 mm during the melt season. Regionally, the duration of meltwater storage ranges from 4.5 weeks in the southeast to 9 weeks elsewhere. We also show that the meltwater runoff modelled from regional climate models may contain systematic errors, requiring further scaling of up to about 20% for the warmest years. These results reveal a high potential for GNSS data to constrain poorly known hydrological processes in Greenland, forming the basis for improved projections of future GrIS melt behaviour and the associated sea-level rise.

Retrogressive thaw slumps (RTSs), formed by abrupt degradation of ice-rich permafrost, are widely distributed on the Qinghai-Tibet Plateau, causing infrastructure damage and enhancing soil carbon emissions. We compiled annual RTS inventories across the plateau from 2016 to 2022 using a deep-learning-aided method to quantify the spatial-temporal variations. We found that RTS-affected locations increased from 1,592 to 3,805 in 2016–2022, which increased affected areas by 2.8 times from 1,714 to 6,507 ha. The most active initiation and expansion periods were in 2016–2017 and 2018–2019. RTSs tend to be clustered, showing local heterogeneity among clusters characterized by various responses toward high temperatures and precipitation and tendencies to be on different topography and vegetation types. This research reveals the rapid development, wide distribution and regional heterogeneity of RTS activities, serving as a crucial step toward understanding how RTSs respond to climate change and regional environmental varieties.

Permafrost on the Qinghai-Tibet Plateau (QTP) undergoes significant thawing and degradation, which affects the hydrological processes, ecosystems and infrastructure stability. The ground deformation, a key indicator of permafrost degradation, can be quantified via geodetic observations, especially using multi-temporal InSAR techniques. The previous InSAR studies, however, either rely on data-driven models or Stefan-equation-based models, which are both lacking of consideration of the spatial-temporal variations of freeze-thaw processes. Furthermore, the magnitudes and patterns of the permafrost-related ground deformation over large scales (e.g., 1×105km2 or larger) is still insufficiently quantified or poorly understood. In this study, to account for the spatial heterogeneity of freeze-thaw processes, we develop a permafrost-tailored InSAR approach by incorporating a MODIS-land-surface-temperature-integrated ground deformation model to reconstruct the seasonal and long-term deformation. Utilizing the approach to Sentinel-1 SAR images on the vast regions of about 140,000km2 of the central QTP during 2014–2019, we observe widespread seasonal deformation up to about 80mm with a mean value of about 10mm and linear subsidence up to 20mm/year. We apply the geographical detector to determine the controlling factors on the permafrost-related deformation. We find that the slope angle is the primary controller on the seasonal deformation: strong magnitudes and variations of seasonal deformation are most pronounced in flat or gentle-slope regions. The aspect angle, vegetation and soil bulk density exhibit a certain correlation with seasonal deformation as well. Meanwhile, we find that a linear subsidence is higher in the regions with high ground ice content and warm permafrost. It indicates that warm and ice-rich permafrost regions are more vulnerable to extensive long-term subsidence. We also observe that the cold permafrost regions experience lower linear subsidence even with high ground ice content, which indicate ice loss is limited. Thus, we infer that under continuously warming, the transition from cold permafrost to warm permafrost may lead to more extensive ground ice melting. Moreover, the strong subsidence/uplift signals surrounding some lakes suggesting that the change of local hydrological conditions may induce localized permafrost degradation/aggradation. Our study demonstrates the capability of the permafrost-tailored InSAR approach to quantify the permafrost freeze-thaw dynamics as well as their spatial-temporal patterns over large scales in vast permafrost areas.

In the past two decades, the data volume of remote sensing imagery in the polar regions has increased dramatically. The calving fronts of many Greenland glaciers have been undergoing substantial variations, and a comprehensive front dataset is necessary for better understanding such frontal dynamics. Therefore, there is a need for an automated approach to identifying glaciological features such as calving fronts. In 2019, three deep-learning-based methods were applied to calving front delineation, but were restricted to a specific area or dataset. Here, we develop a more generalized method that can be applied to a major outlet glacier or remote sensing datasets that are not included in the training. We integrate seven remote sensing datasets into a single deep learning network. The core datasets include optical (Landsat-8 and Sentinel-2) and synthetic aperture radar images (Envisat, ALOS-1 TerraSAR-X, Sentinel-1, and ALOS-2) taken over Jakobshavn Isbræ, Kangerlussuaq, and Helheim, spanning from 2002 to 2019. We evaluate four neural network architectures (e.g., U-Net, DeepLabv3+ with ResNet, DRN, and MobileNet as the backbones) and three histogram modification strategies (e.g., histogram normalization, linear stretching, and no histogram modification). We find that the combination of histogram normalization and DRN-DeepLabv3+ has the lowest test error, at 86 m. These promising results show that our method has a high generalization ability on various glaciers and data types.

The Greenland Ice Sheet is the largest land ice contributor to sea level rise. This will continue in the future but at an uncertain rate and observational estimates are limited to the last few decades. Understanding the long-term glacier response to external forcing is key to improving projections. Here we use historical photographs to calculate ice loss from 1880–2012 for Jakobshavn, Helheim, and Kangerlussuaq glacier. We estimate ice loss corresponding to a sea level rise of 8.1 \pm1.1 millimetres from these three glaciers. Projections of mass loss for these glaciers, using the worst-case scenario, Representative Concentration Pathways 8.5, suggest a sea level contribution of 9.1–14.9 mm by 2100. RCP8.5 implies an additional global temperature increase of 3.7 \,^∘C by 2100, approximately four times larger than that which has taken place since 1880. We infer that projections forced by RCP8.5 underestimate glacier mass loss which could exceed this worst-case scenario.

Conventional benchmark-based survey and Global Positioning System (GPS) have been used to measure surface elevation changes over permafrost areas, usually once or a few times a year. Here we use reflected GPS signals to measure temporal changes of ground surface elevation due to dynamics of the active layer and near-surface permafrost. Applying the GPS interferometric reflectometry technique to the multipath signal-to-noise ratio data collected by a continuously operating GPS receiver mounted deep in permafrost in Barrow, Alaska, we can retrieve the vertical distance between the antenna and reflecting surface. Using this unique kind of observables, we obtain daily changes of surface elevation during July and August from 2004 to 2015. Our results show distinct temporal variations at three timescales: regular thaw settlement within each summer, strong interannual variability that is characterized by a sub-decadal subsidence trend followed by a brief uplift trend, and a secular subsidence trend of 0.26 \pm 0.02 cm year−1 during 2004 and 2015. This method provides a new way to fully utilize data from continuously operating GPS sites in cold regions for studying dynamics of the frozen ground consistently and sustainably over a long time.