Improving sea ice projections with the modern-era satellite altimetry record of freeboard and thickness

Future projections of Arctic sea ice remain poorly constrained, due in large part to the significant inter-model spread across the CMIP6 archive. Various recent studies have explored novel calibration methods to better constrain future sea ice conditions, especially the timing of an ice-free Arctic, but can come to different conclusions based on the calibration approach taken. Some approaches exclude outlier models, while others seek to exploit sensitivities of sea ice area/extent to global warming (e.g., Northern Hemisphere temperatures) to better constrain the multi-model ensemble. As long-term and reliable observations of sea ice thickness are lacking, calibration efforts have primarily relied on sea ice concentration/area data from the passive microwave satellite record. 

We are leading a new NASA-funded effort to utilize the multi-year record of observed sea ice freeboard and derived sea ice thickness from ICESat-2, together with similar data from the original ICESat and ESA’s CryoSat-2 missions, to reduce inter-model uncertainty in future sea ice projections. Although the sea ice altimetry data record is more limited in space and time than the longer-term passive microwave concentration record, it can offer significant advantages; for example, freeboard is measured very accurately by modern laser altimetry satellites, providing more information within the consolidated ice pack, and is now output by some of the CMIP6 models, enabling direct comparisons between the model and the satellite measurements. Only a small fraction of CMIP6 models provide the direct output of derived freeboard so assumptions (mainly related to the bulk ice and snow density) need to be made when estimating freeboard with the core model output. Initial results suggest this can have an important impact on the comparisons. Converting the observations of freeboard to sea ice thickness introduces significantly more uncertainty to the observed data but can radically simplify the model comparison effort. 

In this presentation we will showcase our initial efforts to better utilize the satellite altimetry record for calibrating CMIP6 simulations of future sea ice conditions across both poles, but with a primary focus on projections of the timing of an ice-free Arctic. We discuss some of the nuances of using freeboard as a more direct observational constraint compared to thickness, providing motivation for more modeling groups to provide the direct ice density and freeboard outputs in the lead-up to CMIP7. 

Finally, our analysis has all been carried out within the NASA-supported CryoCloud compute environment using the cloud-based (AWS) CMIP6 data archive, so we include additional insight into the benefits of this analysis approach and the small but important impact from differences in model output availability (compared to the recent IPCC analysis).