patterns relate to the anomalous atmospheric circulation is evidence for a realistic flux-anomaly signal. The dominant atmospheric-circulation anomaly modes in the North Atlantic and North Pacific, represented as EOFs of the sea level pressure, produce systematic flux-anomaly patterns (DC2). The first SLP EOF in the North Pacific features a large pressure anomaly in the central North Pacific, and is associated with the well-known ''Pacific-North American" (PNA) pattern. Three correlation fields linking the PNA to surface variables are shown in Figure 1: the correlations between winter-month PNA time amplitudes and gridded fields of w (wind speed), Δq, and Ql+s anomalies. This set of maps, and others in DC2, show that the anomaly fields of w, Δq, and latent-plus-sensible flux are remarkably consistent with major SLP anomaly patterns over the North Pacific and North Atlantic. Negative SLP anomalies favor positive w anomalies to their south and negative w anomalies to their north, associated with shifts in storm tracks and changes in the mean wind field. Negative SLP anomalies favor positive Δq (and ΔT) to their west, and negative Δq to their east, probably because of meridional advection of air temperature and humidity. The flux anomalies are a hybrid of these patterns, with enhanced sea-to-air fluxes to the southwest and diminished fluxes to the east of negative SLP anomalies.
For the strong Aleutian low phase of the PNA, the anomalous wind-speed field is dominated by a single zonally oriented high-wind patch to the south of the Low in the central North Pacific, with out-of-phase tendencies to the north and the south. Δq and ΔT anomalies are arranged in pockets to the southwest (positive anomalies) and to the east (negative anomalies) of the low. Farther afield, a positive anomaly center appears in the Gulf of Mexico, reflecting the well-known downstream teleconnection of deep troughs and cold dry air outbreaks leading to positive ΔT and Δq anomalies over the southeast United States. The pattern has a broad center of positive anomalies to the southwest of the Aleutian Low, and a region of negative anomalies east of the Low along the West Coast. In the section of this paper that reports on the ocean model, it is shown that low-frequency changes in the PNA between the 1960s and 1980s were associated with marked variations in heat flux and wind stress, which produced striking variability in the upper-ocean thermal structure.
Does the heat flux drive the SST, or does the SST drive the heat flux? The relationship between SST and the climatological mean of total latent and sensible heat flux appears to support two different points of view. On one hand, the fluxes may be driven by the SST: High ocean-to-atmosphere fluxes are found in the tropics where there is warm water and high saturation-vapor pressure, and low fluxes occur
in high latitudes and along the eastern side of the ocean basins where surface temperatures are cool. Concerning the anomaly relationships, there is evidence for the SST-forcing-flux mechanism in the warm-season extratropics and in the tropics (Cayan, 1990; Liu and Gautier, 1990).
On the other hand, in the extratropics the fluxes appear