A multi-dimensional spectral description of ocean variability with applications

Abstract

Efforts to monitor the ocean for signs of climate change are hampered by ever-present noise, in the form of stochastic ocean variability, and detailed knowledge of the character of this noise is necessary for estimating the significance of apparent trends. Typically, uncertainty estimates are made by a variety of ad hoc methods, often based on numerical model results or the variability of the data set being analyzed. We provide a systematic approach based on the four-dimensional frequency-wavenumber spectrum of low-frequency ocean variability. This thesis presents an empirical model of the spectrum of ocean variability for periods between about 20 days and 15 years and wavelengths of about 200-10,000 km, and describes applications to ocean circulation trend detection, observing system design, and satellite data processing. The horizontal wavenumber-frequency part of the model spectrum is based on satellite altimetry, current meter data, moored temperature records, and shipboard ADCP data. The spectrum is dominated by motions along a "nondispersive line". The observations considered are consistent with a universal [omega] -² power law at the high end of the frequency range, but inconsistent with a universal wavenumber power law. The model spectrum is globally varying and accounts for changes in dominant phase speed, period, and wavelength with location. The vertical structure of the model spectrum is based on numerical model results, current meter data, and theoretical considerations. We find that the vertical structure of kinetic energy is surface intensified relative to the simplest theoretical predictions. We present a theory for the interaction of linear Rossby waves with rough topography; rough topography can explain both the observed phase speeds and vertical structure of variability. The improved description of low-frequency ocean variability presented here will serve as a useful tool for future oceanographic studies.