Indices of strength and location for the North Pacific Subtropical and Subpolar Gyres

The adjustment of the North Pacific Subtropical and Subpolar Gyres towards changes in wind stress leads to different time-scale variabilities, which plays a significant role in climate changes. Based on the Simple Ocean Data Assimilation (SODA) and Global Ocean Data Assimilation System (GODAS) datasets, the variations of the Subtropical and Subpolar Gyres are diagnosed using "three-dimension Ocean Circulation Diagnostic Method", and established three types of index series describe the strength, meridional and depth center of the Subtropical and Subpolar Gyres. The above indices present the seasonal, interannual and interdecadal variabilities of the Subtropical and Subpolar Gyres, which proves well. Both the Gyres are the strongest in winter, but the Subtropical Gyre is the weakest in summer and the Subpolar Gyre is the weakest in autumn. The Subtropical Gyre moves northward from February to March, southward in October, and to the southernmost in around January, while the Subpolar Gyre moves northward in spring, southward in summer, northward again in autumn and reaching the extreme point in winter to the south. The common feature of the interannual and interdecadal variabilities is that the two gyres were weaker and to the north before 1976-1977, while they were stronger and to the south after 1976-1977. The Subpolar Gyre has made a paramount contribution to the variability on interdecadal scales. As is indicated with the Subpolar Gyre strength indices, there was an important shift from weak to strong around 1976-1977, and the correlation coefficient with the North Pacific Decadal Oscillation (PDO) indices was 0.45, which was far better than that between the Subtropical Gyre strength indices and the PDO. Tests show that influenced by small and mesoscale eddies, the magnitude of large-scale gyres strength is strongly dependent on data resolution. But seasonal interannual and interdecadal large-scale variabilities of the two gyres presented with indices is less affected by model resolution.