Master Thesis Defense by Chenan Di

Title: Initialization of the Greenland Ice Sheet Using the Parallel Ice Sheet Model

Abstract: 
The Greenland Ice Sheet is a major contributor to global sea level rise and plays a crucial role in the Earth’s climate system. Accurate modeling of this ice sheet is essential to understand its dynamics and assess its past and future impacts on climate. A critical step in ice sheet modeling is the initialization process, which ensures that model simulations closely match the observations. In this thesis, the Parallel Ice Sheet Model (PISM) is used to provide such an initialization by exploring its parameter space and minimizing the misfit between the simulated and observed Greenland ice sheet. Forced by a historical (1850-2014) simulated surface mass balance (SMB) data set from the Earth system model EC-Earth3-ESM, a linear gradient method is applied to account for altitude differences between the simulated ice sheet in PISM and the reference topography in EC-Earth3-ESM.

We focus on the selection and calibration of key model settings, including the parameters of ice deformation, basal sliding, and MB adjustment. Since the SMB of EC-Earth3-ESM can differ from those produced by regional climate models, these parameters may be different from the settings in the existing literature. Because such exploration is computationally intensive and involves many parameters simultaneously, we introduce the Gaussian process (GP) as an emulator to approximate the PISM’s output. It successfully locates parameter combinations that closely match the observed ice surface elevation and ice surface velocity magnitudes in different trade-offs. We then introduce the cumulative ice mass change as an additional constraint to further investigate new optimal parameter combinations. In addition, a paleo spin-up over the last glacial cycle is included to explore its impact during the initialization process.

Our results provide a strategy guide for parameter selection during the ice sheet model initialization and highlight the importance of accounting for the paleo-glacial history of the ice sheet during the spin-up process. We show that the resulting optimal parameter combinations using this method enable PISM to simulate a Greenland ice sheet that reasonably resembles its present-day state under the EC-Earth3-ESM climate, capturing the present-day ice mass loss. As such, our results offer an efficient calibration method for existing and future coupling efforts involving PISM and EC-Earth.

Supervisor: Christine Hvidberg, Shuting Yang (external), and Jorge Bernales (external)
Censor: Jens Olaf Pepke Pedersen