WEATHER REGIMES IN THE PACIFIC FROM A GCM

Weather regimes have been sought by examining the 500-mb streamfunctio n of the UGAMP GCM run for 10 yr at T42 resolution with perpetual Janu ary forcing. Five-day low-pass EOFs provide a low-order phase space in which to study dynamical aspects of the variability. The PNA pattern shows up as the first EOF over the Northern Hemisphere representing 12 % of the variance, rising to 18.5% for Pacific-area-only EOFs. Within the phase space of three to five EOFs, two local minima of the area-av eraged psi tendency (based on rotational vorticity advection) are foun d. These two dow patterns both have a smaller implied tendency than th e climatolog and lie in the +/-PNA regions of the phase space. It is s uggested that these patterns may be acting as ''fixed points'' within the atmospheric attractor, encouraging persistent flows and the format ion of weather regimes. These dynamical attracting points are compared with a more conventional means of identifying weather regimes using a statistical maximum likelihood analysis of all model states during th e 10-yr GCM run. This analysis also indicates two preferred classes, s eparate from the climatology, in the +/-PNA regions of phase space. Th ese classes tend to be nearer the climatology than the dynamical state s but have similar appearance otherwise. Finally the role of low-frequ ency transients are examined to improve the dynamical interpretation o f the regime centers. The method is first demonstrated for the extende d Lorenz model of Molteni et al. The fixed points of the GCM attractor are assumed to be steady solutions to the 500-mb vorticity equation i n the absence of contributions from transient eddies. The eddy contrib utions to the climatological vorticity budget are first determined, an d then the deviations from the climatology that could provide similar contributions to the budget are found. Again two states in the +/-PNA regions of phase space are found to satisfy the above conditions. The authors speculate that the attractors themselves are determined by the large-scale steady effects of topography and land-sea contrasts.