Seminar Abstract

DATE:  Friday, March 01, 2019
TIME:  2:30 pm (refreshments at 2:15 pm)
PLACE: ENR building, room 223
       14 College Farm Road, New Brunswick, NJ

Daniel Gilford
Department of Earth and Planetary Sciences, Rutgers University

Investigating 21st Century Antarctic Contributions to Sea Level Rise with a Supervised Machine Learning Technique

Antarctic ice loss has accelerated in recent decades, and its contributions to global sea level rise could be as high as ~1 meter by the end of the century. Modeled contributions to future global mean sea level from the Antarctic ice-sheet are deeply uncertain, however, in part because ice-sheet model parameters are poorly constrained. We use Gaussian Process regression to develop a statistical "emulator" designed to mimic the behavior of an ice-sheet simulator. Gaussian process modeling is a non-parametric supervised machine learning technique which maps inputs (e.g. model parameters) to target outputs (e.g. sea-level contributions from Antarctica) with explicit quantification of model uncertainties. Emulation is applied over the last interglacial (a past warm period ~125,000 years ago) and a future high greenhouse gas emissions scenario, and trained on ice-sheet model ensembles constructed by varying the parameters of maximum rate of ice-cliff loss and the coefficient of ice-shelf hydrofracturing. The emulator fill gaps between discrete ice-sheet model outcomes, and results illustrate how assumptions about the quality/interpretation of paleoclimate data influence projections of future sea level rise. We invert emulated high and low probability sea level contributions in 2100 to explore 21st century evolution pathways; results show there remains deep uncertainty in ice-sheet model physics, limiting determination of what ice-loss pathway we are traveling on until at least 2060.