Detail of the circa 9 year peak in N.Atlantic N.Pacific cross-correlation ( best estimation 9.051 year ) is documented here:
This appears to be a single peak rather than two close peaks at 8.85259 and 18.5996/2. In comparison to the peaks at 5.15 and 5.5, with approximately twice the frequency spacing which were clearly resolved as separate peaks, there would be clear evidence of some separation here.
Superposition of purely sinusoidal 8.85259 and 18.5996/2 year cycles is mathematically identical to an 9.0696 cycle, amplitude modulated by a 366.45 year cycle. Normally Fourier type analysis will resolve a modulated signal in to the two constant amplitude additive components. However, either the data quality or real physical mixing can disrupt the phase relationship allowing the resolution of two separate peaks. In which case the spectral analysis will return an average amplitude of the hybrid frequency.
It is noted that 366.256 is the number of earth rotations per orbit of the sun (sidereal year) and probably relates to the celestial geometry causing the lunar cycles.
The former is very close to the central value of the peak detected here. The latter being beyond the scope and resolution of this data.
Since 18.5996/2 is the periodic change in the magnitude of the lunar declination angle ( ie its displacement from the equator ) and 8.85259 represents the perigee cycle governing the magnitude of the variation in the lunar tidal force, such an interpretation is physically meaningful as a periodic influence of the moon on the horizontal displacement of water (and hence thermal energy) by lunar tidal forces.