Dr. Samar Khatiwala
Earth Institute Contact: Dr. Samar Khatiwala
The slow dynamical adjustment of the deep ocean creates a computational bottleneck for models of the ocean circulation. In particular, integrating ocean GCMs to equilibrium remains prohibitively expensive, limiting our ability, for example, to simulate climate under different (paleo) boundary conditions, systematically explore sensitivity to parameters and representations of sub-grid scale physics, or to study the ventilation of the ocean using tracers such as natural radiocarbon. The goal of the proposed study accelerating the dynamical spin up of ocean models with a recently developed method for efficient tracer simulation in ocean models. This scheme, known as the "matrix method", makes it feasible to directly obtain equilibrium solutions to the tracer equations without explicit time integration. The matrix method, which is linear and exact for passive tracers, will be adapted to the nonlinear spin up problem by embedding it into an iterative algorithm to "shoot" toward an equilibrium circulation. The proposed study will make extensive use of recent developments in scientific computing and numerical analysis, in particular, the "Jacobian-Free Newton-Krylov" class of methods. The proposed study is exploratory and high-risk because the linear tracer model will be extended to the non-linear spin-up problem. The accuracy and numerical efficiency of the method is hard to predict until after it is developed. Broader impacts: An important outcome of this research will be the development of accurate and efficient methods for tracer simulation and GCM spin up. These tools will be especially useful for paleoclimate studies. The proposed algorithms will be implemented in the widely used MOM ocean GCM to ensure its accesibility to a broad group of researchers. However, the techniques can be applied to any existing GCM, and numerical code developed as part of this research will be made freely available to the research community.
Cross Cutting Themes:
National Science Foundation