ABSTRACT Quantifying Climate Change: Bayesian Model for Antarctic Surface Mass Balance

• Main Author: Philip Andrew White
• Other Authors: The Pennsylvania State University CiteSeerX Archives
• Format: Text
• Language: English
• Published: 2014
• Subjects: Bayesian statistics; climate modeling; Gaussian process; global warming; Metropolis- Hastings; spatial statistics; surface mass balance; Antarctic; Hastings; The Antarctic; Antarc*; Ice Sheet
• Online Access: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.636.4675; http://www.physics.byu.edu/docs/thesis/397/

Because of the importance of Antarctic surface mass balance (SMB) in predicting sea level change, models are created to predict SMB on the Antarctic ice sheet. Using Favier et al.’s quality-controlled aggregate data set N = 3529, a fully Bayesian spatial model has been utilized to predict Antarctic SMB (Favier et al. 2013). Utilizing Markov random fields constructed through Gaussian process models, SMB is predicted over the entire Antarctic ice sheet. An SMB surface over the Antarctic ice sheet is computed by this model and compared with previous maps. An SMB predic-tion error surface is created to identify regions of high prediction uncertainty. This model estimates total SMB to be 1.75±0.335 ·1012 m3 ·w.e ·yr−1 and mean SMB as 124.80±23.85 mm ·w.e. ·yr−1. These results suggest lower Antarctic water accumulation than previously purported. The calcu-lated SMB surface showed more negative SMB regions and higher spatial variation than is likely plausible. Lastly, Antarctic boundary regions and areas with little data show high prediction un-certainty by the generated SMB prediction uncertainty surface.