Predictability of Quasi-geostrophic Ocean Flow: Sensitivity to Varying Model Vertical Resolution

David Adamec Institute for Naval Oceanography, Stennis Space Center, Mississippi

Search for other papers by David Adamec in
Current site
Google Scholar
PubMed
Close
Restricted access

Abstract

Results from a set of 2- and 3-mode quasi-geostrophic simulations are used to estimate the predictability time scale of a Gulf Stream-like flow and investigate the sensitivity of the predictability time scale to changes in model vertical resolution. For this study, two simulations, which differ initially by a specified “small” amount diverge from each other with a doubling time of approximately 16 days for simulations that are resolved by 2 modes and 13 days for simulations that are resolved by 3 modes. On average, higher vertical modes have longer predictability time scales. A spectral analysis of the differences in streamfunctions between a control simulation and a simulation with perturbed initial conditions shows that higher modes have shorter horizontal scales of maximum growth, and the horizontal scale of maximum growth expands at later times for the baroclinic flow. An error energetics budget is calculated to show that transports of error kinetic and error potential energies are initially responsible for exponential error growth of the difference fields. After sufficient buildup in the error potential energy, the conversion between error potential and error kinetic energies becomes as important as the transport terms in the error energy budget.

Abstract

Results from a set of 2- and 3-mode quasi-geostrophic simulations are used to estimate the predictability time scale of a Gulf Stream-like flow and investigate the sensitivity of the predictability time scale to changes in model vertical resolution. For this study, two simulations, which differ initially by a specified “small” amount diverge from each other with a doubling time of approximately 16 days for simulations that are resolved by 2 modes and 13 days for simulations that are resolved by 3 modes. On average, higher vertical modes have longer predictability time scales. A spectral analysis of the differences in streamfunctions between a control simulation and a simulation with perturbed initial conditions shows that higher modes have shorter horizontal scales of maximum growth, and the horizontal scale of maximum growth expands at later times for the baroclinic flow. An error energetics budget is calculated to show that transports of error kinetic and error potential energies are initially responsible for exponential error growth of the difference fields. After sufficient buildup in the error potential energy, the conversion between error potential and error kinetic energies becomes as important as the transport terms in the error energy budget.

Save