Abstract
Wavelet spectrum analysis was applied to sea surface temperature images
derived from NOAA/AVHRR for the area around Japan. A two-dimensional
wavelet spectrum for each scale (wave number) was computed at each position
using the Haar wavelet basis, and then a specially-averaged wavelet spectrum was
calculated for each scale. The wavelet spectral slope derived from the averaged
wavelet spectra was approximately -2.0 in the wavelength range between 4 and 280
km in the Kuroshio area and the East China Sea, while in the Japan Sea and the
Perturbed area (between the Kuroshio and the Oyashio), the slope was approximate
ly -5/3. However, the calculation of the slope at every position showed the values
of the slope were randomly distributed, and even in the Japan Sea and the Perturbed
area, the number of positions where the slope was -5/3 was very small (about 10%).
The results gave insight into the structure of horizontal turbulence, which would be
impossible using a traditional Fourier transform approach.
