The dominant EOF (Fig. 11) shows the salinity trend, accounting for 36% of the variance, though a significant fraction of the variation of this EOF is associated with short term events (perhaps due to eddies). Salinity and dissolved oxygen decrease in concert, and potential vorticity increases. This is also consistent with advection by a perturbation velocity from the south or southeast. In particular, a large pool of low potential vorticity waters exists to the east and northeast of the Hawaiian Islands, so it is clear that westward or southwestward advection of the mean potential vorticity gradient is not consistent with this result. The maximum freshening occurs at 25 σθ while the maximum decrease of oxygen is near 26σθ. The density distribution of the potential vorticity signal is somewhat noisy, but the maximum signal is near 24.2 σθ. These are roughly coincident with the level of maximum north-south gradient in the respective variables. Because the climatology is heavily smoothed, and because it has been mapped onto mean isopycnal surfaces, a more quantitative analysis does not make much sense. New climatologies are being constructed with higher resolution and in a more natural vertical framework, and these will be used to estimate the velocity structure needed to account for the observed variations at ALOHA.
Figure 11. First EOF of salinity, oxygen and potential vorticity anomalies (top) versus potential density, and the associated time series (bottom). The smooth line in the lower panel is a spline fit to the individual cruise values.
CONCLUSIONS
Between 1991 and 1997, a pronounced freshning trend (〜0.15 psu, 0.5℃) was observed in the upper pycnocline at the HOT site. The salinity signals appear later on deeper isopycnals. This trend was associated with decreasing dissolved oxygen and increasing potential vorticity in the upper pycnocline, although these changes are not as apparent in the observations as for the salinity signal. The freshening was reversed in the 24-25 σθ layer during the first half of 1998, probably associated with the 1997-98 ENSO drought in the region. The freshening trend apparently was towards the long-term climatology for this site, which experienced saltier conditions than average during the 10 years of HOT observations. The freshening trend is consistent with both a northward displacement of the North Pacific subtropical gyre in this region, and also consistent with subduction carrying the freshening of the surface layer in the North Pacific into the pycnocline and advection carrying the signal to Station ALOHA. The relative contributions of subduction and gyre wobble requires quantification. Appropriate models forced with estimated winds, heat fluxes and freshwater fluxes are needed to deconvolve the causes of observed variations. The models must be combined with sufficiently dense observations to quantify surface heat and freshwater fluxes; Ekman pumping and subduction; and circulation pathways. Additional time series would be very helpful in distinguishing between forcing errors and model physics deficiencies.
