Surface Gravity Waves in Global Climate Models: Development, Evaluation and Optimization

• Main Author: Ikuyajolu, Olawale James
• Other Authors: Deng, Yi, Bracco, Annalisa, He, Jie, Luo, Jian, Montoya, Joseph, Roekel, Luke Van, Earth and Atmospheric Sciences
• Format: Doctoral or Postdoctoral Thesis
• Language: English
• Published: Georgia Institute of Technology 2023
• Subjects: E3SM; WW3; waves; GPU; GCM; HPC; Sea ice
• Online Access: https://hdl.handle.net/1853/72710

Surface gravity waves play a critical role in several processes at the air-sea interface, including mixing, coastal inundation, and surface fluxes. Yet wind–wave processes are usually excluded from Earth system models partly due to a lack of physical understanding and the high computational costs of spectral wave models. Most wave modeling studies utilize uncoupled short-term simulations and focus on the upper ocean. The impacts of wind-wave processes on coupled climate variability have yet to be thoroughly evaluated. This all underscores the need to advance surface gravity wave modeling frameworks within general circulation models (GCMs). Herein, the first half of this thesis partly addresses the high computational cost of running spectral wave models on a global grid. I identify the wave action source terms as the most computationally intensive part of the spectral wave model WAVEWATCH III (WW3), and then accelerate them on Graphics Processing Units (GPUs) using OpenACC. An average speedup of 1.4x was achieved, resulting in a reduction of 35-40% in runtime and resource usage. In the second half of this thesis, I incorporated a wave-state dependent bulk formula by fully coupling WW3 to the Energy Exascale Earth System Model (E3SM). Current state of the science GCM bulk parameterizations estimate the sea-state roughness as a function of surface wind speed, ignoring wave effects. The newly implemented parameterization includes two primary wave effects: first, a wave-state dependent surface roughness computed by WW3; second, the alteration of momentum flux from the atmosphere to the ocean due to wave growth and dissipation. I conducted numerical experiments with this new parameterization to investigate the sensitivity of the mean climate and Madden-Julian Oscillation (MJO) to different bulk flux parameterizations and the role of waves in air-sea coupling. My results highlight that discrepancies between bulk algorithms have nonnegligible impacts on mean climate such as ocean heat content, sea-ice concentration and ...