Antarctic Peninsula mass trends from 2003 - 2016 using a Bayesian hierarchical model approach

Long-term continuous monitoring of Antarctic Ice Sheet mass balance is imperative to better understand its multi-decadal response to changes in climate and ocean forcing. Additionally, more accurate knowledge of contemporaneous mass balance is key for improved parameterisations in ice sheet models. The Antarctic Peninsula has undergone rapid changes in mass balance and ice dynamics over the last two decades, with satellite observations showing the presence of grounding line retreat and increases in ice sheet velocity. This is particularly the case after the collapse of the Larsen A and B ice shelves in 1995 and 2002, and more recently the glaciers draining the southern Antarctic Peninsula. As a result, this region provides analogues for future ice sheet response to ice shelf collapse in other regions of Antarctica. 

Despite the region’s importance to understanding ice sheet dynamics, it is challenging to accurately assess mass balance due its geometry and mountainous topography. Conventional pulse-limited altimetry suffers from poor coverage and data loss over steep mountainous terrain, particularly before the launch of CryoSat-2 in 2010. In the case of gravimetry, the geometry of the region means the coarse spatial resolution of the GRACE mission (~300 km) cannot resolve small spatial scale glacier changes (particularly over northern Antarctic Peninsula) and suffers from signal leakage into the ocean. For the mass budget approach, the challenge of accurately modelling surface mass balance over the region’s mountainous topography coupled with the sparsity of ice thickness observations at the grounding line for many sectors can result in large uncertainties. As a result, it can be difficult to reconcile the results from different conventional approaches in this region. 

To resolve this, we have developed and optimised the BHM framework used previously over the Antarctic Ice Sheet to specifically investigate the Antarctic Peninsula. This enables each latent process driving ice sheet mass change to be resolved at a higher spatial resolution compared to previous implementations across Antarctica as a whole. The new regional solution also incorporates more recent and higher resolution observations including: CryoSat-2 swath altimetry, stereo-image DEM differencing and NASA Operation Ice Bridge laser altimetry elevation rates. This is the first time such a range of observations of varying spatio-temporal resolutions will be combined into one assessment for the region. We will present results from the regionally optimised model from 2003 until present, including basin-scale mass trends and changes in spatial latent processes at an annual resolution. Additionally, we will discuss future opportunities, such as extending the record from this approach into the next decade and further understanding of the GIA response in this region.