Premium
Comment on ‘Improved global maps and 54‐year history of wind‐work on ocean inertial motions’ by M. H. Alford
Author(s) -
Feser F.,
von Storch H.,
Weisse R.,
Zorita E.
Publication year - 2003
Publication title -
geophysical research letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.007
H-Index - 273
eISSN - 1944-8007
pISSN - 0094-8276
DOI - 10.1029/2003gl018026
Subject(s) - spectral line , meteorology , environmental science , buoy , geology , wind speed , atmospheric sciences , climatology , physics , oceanography , astronomy
In his paper 'Improved global maps and 54-year history of wind-work on ocean inertial motions', the author is using near surface (10 meter) winds from a number of different sources as input for subsequent analyses. One of the sources is the gridded multi-decadal reconstruction prepared by Feser et al. (2001) with the help of the regional atmospheric model REMO (Jacob et al., 1995). Another source used are the NCEP reanalyses (Kalnay et al., 1996). Analyzing both types of data Alford (2003) comes to the conclusion that "The NCEP and the REMO winds are highly coherent at all frequencies over the entire domain. However, the spectra of the REMO winds at ƒ are lower than the NCEP winds by a constant factor. To account for this attenuation, the REMO winds are multiplied by 1.32." This conclusion is based on the fact that a comparison of NCEP spectra with buoy data in the Pacific yields good agreement and that a visual comparison of spectra obtained from REMO and NCEP wind fields showed a clear underestimation at high frequencies for the REMO winds (Alford 2003, pers. comm.). We tried to reconstruct the spectra that led Alford (2003) to the conclusion that REMO winds have too little variance at high frequencies. In Figure 1 spectra of the zonal wind component at 10 m height are shown for hourly sampled REMO and 6-hourly sampled NCEP winds. In addition a spectrum of 6-hourly sampled REMO winds is shown. While all three spectra agree well at low frequencies there is indeed discrepancy between the 6-hourly sampled NCEP (black curve) and the hourly sampled REMO winds (blue curve) at the high-frequency end of the spectrum. If the hourly REMO time-series is sampled at 6-hour intervals (red line), the resemblance with NCEP in the high-frequency range is obvious. This suggests that the hourly sampled REMO winds do not underestimate spectral densities but that the 6-hourly sampled data suffer from an aliasing effect In such cases, variations at time scales shorter than the Nyquist folding frequency (2∆)-1 (where ∆ represents the sampling interval), are folded onto the resolved frequencies. A poor choice of sampling intervals can obviously lead to non-fitting spectra and misleading interpretation. In fact, multiplication of the hourly REMO winds by a factor of 1.32 artificially raises the spectral levels at high frequencies close to those obtained from 6-hourly sampled NCEP (and REMO) data but in the same …