63 
this situation corresponds to the high energy density in the wavelet modulus and to the up and 
downs in the time series. When the Kuroshio touches the cape, the dynamic pressure rises and the 
cyclonic eddy starts to be formed. When the eddy is shed from the cape, the Kuroshio detaches 
being pushed offshore to make the K-U index fall. 
2.3 Inverse cascade and interlocking processes inside the Shikoku-Basin 
Eddies shed from Giant-Cusps feed the boundary layer to make it grow. The process lasted 
until 20 February 1993, when the boundary layer detached the Cape-Shionomisaki. Then, vortex 
generators are made inactive, and cyclonic eddies are transported downstream by the advection. 
Figure 5 shows the situation, 2 April 1993. In the wake of Cape-Shionomisaki, we find a cyclone 
of 200km in diameter growing, inside which smaller scale eddies advected from upstream are 
amalgamating to form a larger scale cyclone. This is an inverse cascade process of turbulence. 
Another inverse cascade process followed as shown in Figure 1; i.e. a dipole of 500km in 
size was formed through the pairing of cyclone and offshore anticyclone. Here, we should remark 
that the dipole is interlocked tight inside the Shikoku-Basin. Figure 1(b) shows the anticyclone just 
centered overhead Koshu-Seamount, suggesting the vertical structure reaching 2,000m in depth. 
In this case, however, the meandering path of Kuroshio didn't last long. Figure 3(a) shows 
an eddy shedding from Cape-Shionomisaki, 06 June 1993. Owing to the entrainment of Kuroshio 
watermass, the coastwise cold watermass decayed to be followed by the nonmeandering path of the 
Kuroshio. In these ways, the eddy sheddings from Giant-Cusps lead to growing or decaying 
process of coastal cold watermass, and cause the meander or nonmeander of the Kuroshio path. 
Figure 4. Eddy sheddings from Giant-Cusps in the Pacific-Coast boundary layer, 4 February 1993. 
Figure 5. Inverse cascade process in the wake of Cape-Shionomisaki, 2 April 1993.