the evolution of variability features such as the 1976-1977 shift (Miller et al., 1994a,b). In the Coastal region, heat-flux input tends to be about four times larger than either horizontal advection or entrainment effects. In the mid-Pacific region, in contrast, which is nearer the more variable winds of the storm track, heat-flux input is only twice as large as horizontal advection, and typically the same size as entrainment effects. Diffusion tends to be slightly weaker than horizontal advection, but inspection of the time series shows that diffusion simply acts in opposition to the cumulative effects of the other heating term.
During the six-month period preceding the 1976-1977 climate shift, a long period of warming via heat-flux input occurred in the Coastal region (not shown). Although this period was not particularly strong, it was persistent compared to most of the rest of the time intervals (Miller et al., 1994a). What makes this period particularly striking, however, is that the strongest occurrence (in the 1970-1988 time interval) of warming by horizontal advection took place during fall 1976 and the subsequent winter. Together, these two effects resulted in a mixed layer 10 to 15 m shallower and more than half a degree of warming in the surface temperature. In spite of a strong loss of heat during the late winter of 1977, the system managed to remain in a shallow mixed-layer, warm-SST state through the winter of 1988. Throughout 1977, several months of strong heat-flux cooling began to cool the upper ocean toward the mean, but strong heating in summer 1979 thinned the mixed layer and warmed the SST to such a degree that the mixed-layer warmth persisted into winter 1980.
In the Mid-Pacific region, the 1976-1977 climate shift is likewise instigated by the effects of an anomalously long and strong period of cooling by horizontal advection combined with sizable cooling by heat flux (Figure 11). During the six-year period preceding the shift, anomalously weak entrainment effects (partially due to lower TKE input) serve to maintain the system in a shallow mixed-layer, warm-SST state. In the years following the shift, the effects of stronger entrainment are confined to the summer months.
It therefore appears that the 1976-1977 shift in both the Coastal and Mid-Pacific regions was caused by an unusual atmospheric state that took hold several months before the 1976-1977 winter, and by conditions in the winter atmospheric circulation that persisted for several winters thereafter. This atmospheric circulation produced large-scale shifts in ocean-current advection that acted in concert with large-scale heat-transfer processes to significantly alter the upper-ocean thermal structure, and thus its stratification. The Mid-Pacific region subsequently remained in that perturbed state through the maintenance effects of reduced TKE input.
Two experimental approaches indicate that bulk parameterizations yield fairly realistic estimates of anomalous sea-
air fluxes on seasonal-to-interannual scales over the tropical and northern extratropical oceans. Monthly parameterizations of wind stress and heat flux derived from COADS historical summaries provide reasonable coverage over much of the Pacific basin back to the 1960s. The validity of these derived variables was demonstrated in two ways: (1) by the consistency of the spatial and temporal variability of heat-flux anomalies with independently observed data on the atmospheric circulation and SST tendency; and (2) by the success in simulating two decades of Pacific Ocean SST anomalies with a model driven by the parameterized wind and heat-flux forcings. The bulk parameterizations appear to be adequate for useful diagnoses of several unobserved thermal and dynamic processes (advection, vertical mixing, etc.) that operate in the upper ocean on seasonal-to-decadal time scales.
The model's air-sea heat fluxes were shown to be consistent with two sets of independently observed data. First, the latent and sensible flux anomalies are spatially organized by the atmospheric circulation into systematic patterns about the major cyclonic and anticyclonic features. This organization occurs because the fluxes are determined by wind speed and by air-mass temperature and humidity, which have characteristic structures relating to the general circulation. The flux anomaly/circulation patterns that appear in the North Pacific are confirmed by very similar patterns found