Spatial distribution of amplitudes of SST's annual cycle and their pentadal anomalies are shown in Figure 2. With some exceptions during 1951-1970 anomalies of amplitudes of annual cycle generally tend to be attributed to anomalies of annual means of the same sign. Particularly, it is clearly seen in plots for pentads 1961-1965 and 1966-1970.
Beginning from pentad 1971-1975 relationship between anomalies of annual cycle and ones of annual means is reversed, i. e. positive anomalies of amplitudes are attributed to negative ones of annual means and vice versa. We note here that this change in relationships between anomalies has occurred before North Pacific climate shift.
In an attempt to find the reasons for such a change we have considered space-time structure of pentadal anomalies of seasonal (winter and summer) means. Climates and pentadal anomalies of winter (December-January-February) and summer (June-July-August) mean SST's are presented in Figure 3 and Figure 4 respectively. Not only winter, as it was noted earlier, but summer anomalies reveal North Pacific climate shift of mid 70-ties pretty well. Before 1971 both positive and negative summer anomalies of SST's were generally stronger than winter ones. For example, it is well seen in 1951-1955 and 1961-1965. Thus, during 1951-1970 summer anomalies played a leading role in formation of anomalies of annual cycle. And it explains why during this period anomalies of amplitudes of annual cycle were attributed to anomalies of annual means of the same sign. Since pentad 1971-1975 summer anomalies have passed their goveming role in formation of annual cycle of SST's to winter anomalies. Winter anomalies became stronger, and relationship between anomalies of annual cycle (Fig. 2) and anomalies of annual means (Fig. 1) is reversed, i. e. anomalies of amplitudes of annual cycle are attributed to anomalies of annual means of opposite sign. It looks however that beginning from pentad 1986-1990 governing role in formation of annual cycle of SST's comes back to summer anomalies again.
To detect more correctly when winter anomalies have became stronger than summer ones we considered seasonal anomalies year by year. It is appeared that this change has occurred in 1973, i. e. three years earlier than North Pacific climate shift.
Climate and pentadal anomalies of amplitudes of semiannual cycle of SST are shown in Figure 5. While values of amplitudes are in general by a factor of five lower than ones of annual cycle, magnitudes of anomalies are only two to three times lower than ones of annual cycle are (Fig. 2). Analysis of long-term evolution of anomalies of semiannual cycle has not revealed association neither with changes in annual cycle nor with those of annual and seasonal means. It appears that positive anomalies of semiannual cycle can be attributed both to positive and to negative anomalies of amplitudes of annual cycle. The same is true about negative anomalies of semiannual cycle.
It would be very tempting to find anomalies of phases of annual and semiannual cycles of SST's related to above described changes in amplitudes' and annual and seasonal means but our analysis has not revealed such an anomalies of phases. It appears that over the most of the North Pacific during considered period (1951-1995) anomalies of phases are equal to zero. There are some small and statistically insignificant anomalies only in the near equatorial region.
