What factors affect the mass balance of Antarctic glaciers? Can we use observations, satellite data, and a regional climate model to predict climate change trends in Antarctica and the Arctic? What can be done to prevent climate change on Earth?

Traditionally, understanding small-scale (like individual glaciers) mass changes in Antarctica has primarily relied on the flux gate method based on satellite imagery. This study advances our understanding of mass balance processes by combining two independent satellite observations measuring gravity and surface elevation changes to assess Antarctic glacier-scale ice mass changes from 2003 to 2020. We examined the contributing factors, namely the mass gains/losses from snowfall and ice discharge. Our results indicate increased ice discharge as the main factor in long-term ice mass change, albeit with significant regional variations. Comparisons of our estimates to those from satellite imagery reveal agreement in West Antarctica but notable regional differences, underscoring the importance of collective efforts to improve data coverage and model accuracy.We examine Antarctic glacier mass changes from 2003 to 2020 and divide them into contributions from surface mass balance (SMB) and ice discharge, using high-resolution ice mass change estimates derived from a combination of two different types of satellite observations (gravimetry and altimetry) and outputs from a regional climate model. Our analysis shows that changes in ice discharge have played a dominant role in current ice mass trends and their acceleration, particularly in glaciers near the Amundsen and Bellingshausen Seas in West Antarctica. In particular, mass losses at Thwaites and Pine Island glaciers have been largely (>90%) controlled by ice discharge, while the contribution from SMB has been relatively minor. In East Antarctica, SMB accounts for a significant portion (>50%) of the glacier ice mass imbalance in, for example, Drenning Mud Land and Wilkes Land. Ice discharge has also played a significant role in the overall mass gain in the region. While our estimates of ice discharge rates are in good agreement with previous estimates from satellite imagery in West Antarctica, significant differences are found in the glaciers of East Antarctica and the Antarctic Peninsula. This highlights the need for more observations and improved numerical models to refine these estimates.he Antarctic Ice Sheet (AIS) has been losing ice in recent decades, contributing significantly to global sea level rise. From 1992 to 2020, the AIS lost an average of about 92 gigatonnes per year (Gt/yr) of ice, contributing to a 7.4 mm rise in global sea level (1). As global warming continues, Antarctic ice mass loss is expected to continue and even accelerate, potentially contributing to a sea level rise of up to 0.34 m by the end of this century (2). However, predicting these changes is challenging due to the complexity of atmosphere-ocean-ice sheet interactions and the limitations of observational data (3). Therefore, improving observability and deepening our understanding of the factors influencing ice mass changes are essential for more accurate predictions of future sea level rise. Two main processes are responsible for changes in ice mass: changes in surface mass balance (SMB) and/or in solid ice discharge. A third process, base melting, contributes little to the overall mass balance of floating ice. SMB represents the accumulation and erosion of ice mass due to atmospheric factors, while ice discharge is influenced by changes in ice flow velocity and ice thickness at the landfall line. The main factor affecting SMB in the AIS is changes in snowfall (4), which occur on subannual, interannual, or longer (decadal) time scales (5). Antarctic ice discharge largely influences mass change on interannual and longer time scales and is determined by factors such as glacier geometry, ice physical properties, and interactions between ice and bedrock or ocean (6–8). For any given glacial basin, an imbalance between SMB and ice discharge results in ice mass loss or gain. For example, if warm ocean water thins or breaks an ice shelf, this could reduce the upstream glacier’s resistance to flow, potentially leading to increased ice discharge (9). This process has recently been observed in West Antarctic and Antarctic Peninsula glaciers (10–12). In addition, increased/decreased snowfall can cause glacier mass to increase/decrease, as has recently been observed in Dronning Maud Land in East Antarctica (13) and on the West Antarctic Coast (14). These non-uniform ice mass changes make future predictions more difficult and highlight the importance of separately determining the contribution of SMB and ice discharge (1). AIS mass changes have been observed primarily using three remote sensing techniques: satellite gravity measurements (GRACE), satellite altimetry, and a partial approach based on satellite imagery and SMB estimates. Despite differences in their sensitivity to various sources of uncertainty, these methods are employed independently by different systems.Research groups have been working to obtain the most accurate determination of AIS ice mass changes. The modern flux gate method, also known as the mass budget method, the input-output method (1), or the component method (15), estimates ice mass change using ice discharge observations and SMB modeling. In this method, ice discharge is calculated by multiplying ice thickness by ice velocity along a flux gate, which is usually located at or slightly above the grounding line. Thanks to the availability of high-resolution glacier flow velocity data, this method has been widely used to study ice mass balance at the glacier scale (15). However, the flux gate method has its challenges. For example, in regions with complex bedrock topography, such as the East Coast of Antarctica and the Antarctic Peninsula, ice thickness measurements at flux gates are often uncertain, as are SMB estimates from coarse-grained atmospheric models (15, 16). In such cases, scaling of ice discharge with limited observations over the entire basin is required, potentially introducing significant errors. In addition, uncertainties in flow velocity data pose another problem for this method. While these data are generally reliable for fast-moving glaciers, such as Thwaites Glacier in West Antarctica, the uncertainties for slower-moving glaciers (e.g., many in East Antarctica) (17) are proportionally higher, potentially affecting the accuracy of the results. There have also been attempts to investigate mass balance at the glacier scale using GRACE (18, 19), which observes gravitational potential changes inferred from perturbation of the range rate between two satellites. However, these estimates suffer from low spatial resolution (about 300 km) due to signal leakage from land to ocean or from one basin to another. Uncertainties in the correction for solid-earth isostatic adjustment also introduce significant uncertainties in gravity-based estimates of ice mass change (1, 20). Satellite altimetry, which measures changes in ice surface elevation, has difficulty detecting glacier-scale ice mass changes in areas with complex ice topography, particularly on the Antarctic Peninsula. Converting observed changes in surface elevation to ice mass changes also introduces significant uncertainties, largely due to variability in ice density (21).To combine the strengths and reduce the limitations of GRACE and altimetry, a recent study (22) has developed a new method that combines both methods to improve the spatial resolution of GRACE data and reduce the uncertainty of altimetry data. This results in high-spatial resolution ice mass changes that are comparable to ice-corrected altimetry data (~27 km) and have an accuracy similar to GRACE. The estimates with higher spatial resolution and accuracy combined with SMB data from a regional climate model allow us to estimate ice mass changes at the glacier scale and investigate their main source. In this study, we estimate AIS mass change over the years 2003–2020 using the method developed in the aforementioned study, with the inclusion of a new satellite mission (Materials and Methods). We assess the relative contributions of two main processes, SMB and ice discharge, in determining ice mass trends and accelerations. Various sources of uncertainty affecting the linear trend and acceleration terms are considered to ensure the robustness of our results. Finally, we compare our ice discharge estimates with values ​​obtained from satellite imagery and ice thickness.Fig. 1.   (A) ΔM  , ΔSMB  , and −ΔD  over AIS. Note that ΔSMB  and −ΔD  were resampled to the time intervals of ΔM  (GRACE observation). The ΔD  is presented with its negative sign ( −ΔD  ) for a more intuitive interpretation. (B and C) Maps of linear trends (B) and accelerations (C) in ΔM  , accompanied by pie charts showing the contribution of ΔSMB  (light colors) and −ΔD  (dark colors). Circle sizes indicate the combined absolute values of SMB and ice discharge, represented by blue (positive values) and red (negative values) colors, respectively. Only statistically significant values of ΔM  (Table 1) are displayed as pie charts. The black lines represent the basin boundaries from ICESat altimetry observation (23). The boundaries of the Pine Island (purple), Thwaites (green), and Totten (cyan) Glaciers were obtained from satellite radar imagery (24, 25). The background shading represents the optical image from the MODIS Mosaic of Antarctica 2013–2014 (MOA2014) project (26).Conclusion We used high-resolution ice mass change estimates based on combined satellite gravimetry and altimetry and a numerical SMB model to investigate the continental- to regional-scale ice mass balance in Antarctica over the past 20 years. The results show that the spatially nonuniform ice mass changes were due to different contributions from SMB and ice discharge. Increased ice discharge was found to be the primary cause of the significant ice mass loss and acceleration of, notably, the glaciers flowing into the Amundsen and Bellingshausen Seas, as well as the glaciers in the Antarctic Peninsula. In contrast, SMB variability mostly explains the ice mass gain in Dronning Maud Land, mass loss in Wilkes Land, and the accelerations in the eastern part of East Antarctica (basins 12 to 17). However, the spatially varying linear trends in SMB canceled out across the entire continent, making ice discharge the dominant contributor (~100%) to the ice mass loss. However, SMB did explain 50% of the total acceleration in mass loss. It should be noted that our results are limited to the past two decades (2003–2020) and thus are susceptible to interannual to decadal climate variability. For example, high interannual variability in mass changes, modulated by natural phenomena like El Ninõ Southern Oscillation (ENSO) and the Southern Annular Mode (SAM), are dominant signals in both regional and continental-scale SMB anomalies ( 5 , 50 , 51 ). Extreme snowfall events, partly caused by the occurrence of atmospheric rivers, would also significantly affect the contribution of SMB to ice mass changes ( 52 ). While some relatively gradual changes in ice discharge are obviously caused by large-scale (oceanic) processes, such as in the Amundsen and Bellingshausen Sea coast of West Antarctica, some individual glaciers exhibit strong interannual changes, probably as a result of local oceanographic variability ( 5 , 15 , 37 ). Consequently, accurate observation of these short-term dynamics is crucial for understanding of regional contribution to future ice mass balance. Our results not only provide insight into the intricate imbalance of ice discharge and SMB but also underscore the need for integrating multiple observational datasets and associated numerical modeling to refine our understanding of the AIS ice mass balance. Precise delineation of the rheological properties of the Antarctic upper mantle could refine viscoelastic response to historical and contemporary ice mass unloading, thereby improving ice mass observations from satellite gravimetry and altimetry. Additionally, refining bed topography is crucial for the flux gate method to reduce potential systematic errors, and enhancing the SMB models is key for accurately identifying the drivers of short-term ice mass/ volume changes across various observational methods and reducing associated uncertainties ( 53 ). With the ongoing impact of Antarctic ice mass loss on global sea levels, such efforts will become crucial for accurately projecting future sea level rise.