Two Probabilistic Methods for Attributing the Sources of Sea-Level Rise from Sparse Tide Gauge Records

Abstract:

Recent estimates of 20^th century global mean sea-level (GMSL) rise are in the range 1.6-1.9 mm/yr. However, these estimates use a temporally and spatially sparse network of tide gauge observations that may result in a biased estimate due to the incomplete sampling of a global field. Furthermore, this sparse sampling makes resolving the GMSL change into individual sources (e.g., mass loss from individual ice sheets and mountain glaciers, ocean thermal expansion, etc.) challenging. We have employed two different statistical techniques, a multi-model Kalman smoother (KS) and Gaussian process regression (GPR), to address the above challenges. Both techniques naturally accommodate spatio-temporal changes in the availability of tide gauge observations and use models of glacial isostatic adjustment, ocean dynamics, and the sea-level fingerprints of rapid land ice melt to exploit both the spatial and temporal information within observations of the sparsely-sampled global field. We present reconstructions of the global sea-level field estimated with both methods, as well as the associated constraints on the underlying sources of global mean sea-level change. Our results include estimates of the individual contributions of the world’s ice sheets and mountain glaciers to 20^th century sea-level change and provide new estimates of the spatial and temporal variability in sea level since 1900.