Quasi-geostrophic Ocean Response to Real Wind Forcing: The Effects of Temporal Smoothing

View More View Less
  • 1 National Center for Atmospheric Research, Boulder, Colorado
  • | 2 University of South Carolina Columbia, South Carolina
© Get Permissions Rent on DeepDyve
Restricted access

Abstract

A one-degree, flat bottom, eight-layer quasi-geostrophic model of the North Pacific Ocean is forced by six different wind stress curl datasets, all derived from seven years of twice daily analyses at the European Centre for Medium Range Weather Forecasts, 1980–86. The six datasets, with nominal averaging times of 1, 2, 3, 7, 14, and 30 days, are obtained by carefully filtering in the frequency domain. This filtering greatly reduces the variance, typically by 90% for 30-day averaging, because the wind stress curl spectrum is nearly white in frequency. It also smoothes spatially, reducing high wavenumber variance to a greater degree than variance near wavenumber zero. The climatology of the ECMWF wind stress curl does not show any unexpected differences from climatologies based on historical marine wind observations. The wind stress curl is neither temporally nor spatially stationary with the high frequency variance being much larger during the winter season and over the northern half of the North Pacific. Its spectrum does not appear to be isotropic in wavenumber.

From a common initial state, the baroclinic fields in the model ocean runs evolve nearly identically regardless of the forcing bemuse their frequencies are all lower than the Nyquist frequency of even 30-day sampling. The higher frequency forcing generates Rossby waves that dominate the instantaneous barotropic stream function throughout the basin. These barotropic waves are not found at frequencies above the Nyquist frequency of the forcing. There is negligible rectification into basin scale, six year mean flows. There are only small scale differences in mean monthly barotropic streamfunction fields. Thus, the barotropic ocean response diminishes as the nominal averaging time increases. Furthermore, these Rossby waves appear to be natural modes of the model basin, and they could be artificially forced unless the wind data processing carefully avoids aliasing unresolved frequencies. Overall the spectrum of ocean response is found to be more red in frequency than the nearly white wind curl spectrum and even more red in wavenumber.

High frequency forcing produces higher kinetic energies at all depths with an annual cycle that is related to the annual cycle of the high frequency variance in the wind stress curl. The deep kinetic energy and the intra-annual streamfunction variance are used to quantify the relative importance of the high frequency barotropic Rossby waves. There is considerable advantage in using 3-day average (over 7- and 14-day average) wind forcing, however, there is little more to be gained with 2-day averaging and nothing further added by 1-day averaging. When forced with 30-day averaged wind curls, the intra-annual stream function variance is typically only 30% of its value when forced with 3-day or shorter averages.

Abstract

A one-degree, flat bottom, eight-layer quasi-geostrophic model of the North Pacific Ocean is forced by six different wind stress curl datasets, all derived from seven years of twice daily analyses at the European Centre for Medium Range Weather Forecasts, 1980–86. The six datasets, with nominal averaging times of 1, 2, 3, 7, 14, and 30 days, are obtained by carefully filtering in the frequency domain. This filtering greatly reduces the variance, typically by 90% for 30-day averaging, because the wind stress curl spectrum is nearly white in frequency. It also smoothes spatially, reducing high wavenumber variance to a greater degree than variance near wavenumber zero. The climatology of the ECMWF wind stress curl does not show any unexpected differences from climatologies based on historical marine wind observations. The wind stress curl is neither temporally nor spatially stationary with the high frequency variance being much larger during the winter season and over the northern half of the North Pacific. Its spectrum does not appear to be isotropic in wavenumber.

From a common initial state, the baroclinic fields in the model ocean runs evolve nearly identically regardless of the forcing bemuse their frequencies are all lower than the Nyquist frequency of even 30-day sampling. The higher frequency forcing generates Rossby waves that dominate the instantaneous barotropic stream function throughout the basin. These barotropic waves are not found at frequencies above the Nyquist frequency of the forcing. There is negligible rectification into basin scale, six year mean flows. There are only small scale differences in mean monthly barotropic streamfunction fields. Thus, the barotropic ocean response diminishes as the nominal averaging time increases. Furthermore, these Rossby waves appear to be natural modes of the model basin, and they could be artificially forced unless the wind data processing carefully avoids aliasing unresolved frequencies. Overall the spectrum of ocean response is found to be more red in frequency than the nearly white wind curl spectrum and even more red in wavenumber.

High frequency forcing produces higher kinetic energies at all depths with an annual cycle that is related to the annual cycle of the high frequency variance in the wind stress curl. The deep kinetic energy and the intra-annual streamfunction variance are used to quantify the relative importance of the high frequency barotropic Rossby waves. There is considerable advantage in using 3-day average (over 7- and 14-day average) wind forcing, however, there is little more to be gained with 2-day averaging and nothing further added by 1-day averaging. When forced with 30-day averaged wind curls, the intra-annual stream function variance is typically only 30% of its value when forced with 3-day or shorter averages.

Save