No abstract

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 5 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 10 frequently used remote sensing sensors: two active microwave sensors, ASCAT (Advanced Scatterometer) and Sentinel-1, a passive microwave sensor SSMIS (Special Sensor Microwave Imager/Sounder), and an optical sensor MODIS (Moderate Resolution Imaging Spectroradiometer). We intercompare the 15 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. 20We discuss that discrepancies between sensors are mainly due to (1) cloud obstruction and polar darkness, (2) frequencydependent penetration of satellite signals, (3) temporal resolution, and (4) spatial resolution, as well as (5) the applied melt detection methods. Nevertheless, we argue that different sensors can 25 complement each other, enabling improved detection of surface melt over the Antarctic Ice Sheet.

Regional climate models (RCMs) have a high computational cost due to their higher spatial resolution compared to global climate models (GCMs). Therefore, various downscaling approaches have been developed as a surrogate for the dynamical downscaling of GCMs. This study assesses the potential of using a cost-efficient machine learning alternative to dynamical downscaling by using the example case study of emulating surface mass balance (SMB) over the Antarctic Peninsula. More specifically, we determine the impact of the training framework by comparing two training scenarios: (1) a perfect and (2) an imperfect model framework. In the perfect model framework, the RCM-emulator learns only the downscaling function, similar to classical super-resolution approaches; therefore, it was trained with upscaled RCM features at GCM resolution. This emulator accurately reproduced SMB when evaluated on upscaled RCM features (mean RMSE of 0.27 mm w.e./day), but its predictions on GCM data conserved RCM-GCM inconsistencies and led to underestimation. In the imperfect model framework, the RCM-emulator was trained with GCM features and downscaled the GCM while exposed to RCM-GCM inconsistencies. This emulator predicted SMB close to the truth, showing it learned the underlying inconsistencies and dynamics. Our results suggest that a deep learning RCM-emulator can learn the proper GCM to RCM downscaling function while working directly with GCM data. Furthermore, the RCM-emulator presents a significant computational gain compared to an RCM simulation. We conclude that machine learning emulators can be applied to produce fast and fine-scaled predictions of RCM simulations from GCM data.

Regional climate models (RCMs) have a high computational cost due to their higher spatial resolution compared to global climate models (GCMs). Therefore, various downscaling approaches have been developed as a surrogate for the dynamical downscaling of GCMs. This study assesses the potential of using a cost‐efficient machine learning alternative to dynamical downscaling by using the example case study of emulating surface mass balance (SMB) over the Antarctic Peninsula. More specifically, we determine the impact of the training framework by comparing two training scenarios: (a) a perfect and (b) an imperfect model framework. In the perfect model framework, the RCM‐emulator learns only the downscaling function; therefore, it was trained with upscaled RCM (UPRCM) features at GCM resolution. This emulator accurately reproduced SMB when evaluated on UPRCM, but its predictions on GCM data conserved RCM‐GCM inconsistencies and led to underestimation. In the imperfect model framework, the RCM‐emulator was trained with GCM features and downscaled the GCM while exposed to RCM‐GCM inconsistencies. This emulator predicted SMB close to the truth, showing it learned the underlying inconsistencies and dynamics. Our results suggest that a deep learning RCM‐emulator can learn the proper GCM to RCM downscaling function while working directly with GCM data. Furthermore, the RCM‐emulator presents a significant computational gain compared to an RCM simulation. We conclude that machine learning emulators can be applied to produce fast and fine‐scaled predictions of RCM simulations from GCM data.