What is our Focus Group About? 

Focus Group Statement

In Greenland (GrIS), roughly half of the mass loss is driven by decreased ice sheet surface mass balance (SMB), i.e., the difference between snowfall accumulation, and ablation from meltwater runoff and sublimation. The remainder stems from increased solid ice discharge (D), i.e., the mass flux across grounding lines. In Antarctica (AIS), mass loss is driven by increased D, while its temporal variability is governed by fluctuations in SMB, notably snowfall accumulation. To project ice sheets’ future climate and SMB, earth system models (ESMs) at low spatial resolution, i.e., typically 100 km for CMIP-style efforts, are commonly used. However, ESMs cannot resolve e.g., high melt (resp. snowfall) rates across low-lying (resp. elevated) GrIS outlet glaciers or AIS ice shelves (resp. mountain divides). Our focus group aims at projecting reliable, high-resolution ice sheets’ SMB by 2100 and beyond, that is suitable on the one hand to capture ESMs sensitivity to prescribed emissions, and on the other hand to force high-resolution ISMIP7 ice sheet models (Tasks E4 and E5).

ISMIP6 used different protocols to yield GrIS and AIS SMB forcing for ice sheet models (Nowicki et al., 2020). For the GrIS, ESMs climate and SMB projections were first dynamically downscaled using regional climate models (RCMs), i.e., typically 10 km, and in some cases statistically downscaled to a higher-resolution grid (e.g., 1 km). For the AIS, native ESMs climate and SMB were used as ice sheet model forcing. Forcing for both ice sheets were provided either as yearly SMB gradient anomalies for direct mass flux forcing, or yearly precipitation and temperature anomalies for models that used a positive degree day (PDD) method. This approach has garnered some scrutiny because of its under-sampling of SMB-related uncertainty. In addition, the use of a regional climate model for generating SMB requires an additional non-trivial and computational intensive step for data preparation, which in turn limits the number of future climate scenarios that one can ask the ice sheet models to simulate. It also limits the number of climate models that can be used as climate drivers for the ice sheet simulations, because not all climate models have output at the required temporal resolutions.

Alternative methods for ice-atmosphere interactions exist and this task is for the surface mass balance focus group to assess via a proposed SMBMIP exercise for both ice sheets. Techniques that will be assessed include: dynamic downscaling using multiple RCMs (see e.g., Fettweis et al., 2020 and Mottram et al., 2021), statistical downscaling (e.g., Noël et al., 2022 and Noël et al., 2023), positive degree day downscaling (e.g.,  Wilton et al., 2017)), machine learning statistical downscaling (Harilal et al., 2021), surface energy balance downscaling (e.g., Krebs-Kanzow et al., 2021), surface energy and hydrology model (e.g., Cullather et al., 2014), surface energy and snow model (Born et al., 2019) and a novel spatial SMB-temperature regression. Applying the above techniques, we aim at yielding high-resolution (GrIS at 1-10 km; AIS at 2-12 km) SMB and components including, e.g., runoff, precipitation, and (near-)surface temperature. Vertical gradients will be estimated based on the ice sheet topography prescribed in each model. The SMBMIP exercise will inform us on the robustness, efficiency, and accuracy of each method, enabling us to rank the techniques, propose methods to be used for ISMIP7 and prepare the relevant datasets for ice sheet model simulations.