Arctic Oscillation regime behavior in an idealized atmospheric circulation model as a result of almost-intransitivity

• Main Authors: Sempf, M., Dethloff, Klaus, Handorf, Dörthe, Kurgansky, Michael
• Format: Conference Object
• Language: unknown
• Published: 2005
• Subjects: Arctic
• Online Access: https://epic.awi.de/id/eprint/14037/; https://hdl.handle.net/10013/epic.24386

A quasi-geostrophic, hemispheric three-level atmospheric modelwith horizontal T21 resolution is driven by Northern Hemisphere's T21topography, whereas its thermal and surface forcingare determined by an automated, iterative tuning procedure.The zonal parts of the forcings are tuned toproduce a realistic zonal wind profile for northern winter, while non-zonalthermal forcings are adjusted to obtain time-averaged non-zonal diabaticheating fields equal to wintertime observations. The perpetual wintermodel simulation reproduces observed wintertime climatology and patterns oflow-frequency variability with accuracy. The model exhibits two significantcirculation regimes which correspond to the positive and negative phase ofthe Arctic Oscillation (AO), respectively.Steady solutions of the model equations have been determined, but they do notcoincide with the regime centroids. They are even far outside of the system'sattractor and therefore do not directly influence the model's dynamics. Butan explanationof the dynamical structures underlying the model's regime behaviour issuggested by the results of a series of model experiments, in which thetuning procedure is repeated using lower values of surface friction.The weaker surface friction is, the moredistant and pronounced the two AO regimes become, indicated by increasinggeopotential standard deviation at polar latitudes and alsoby the AO index distribution, the bimodality of which is becoming more andmore extreme. The regime persistence, but also themodel's sensitivity with respect to forcing changes dramatically increase.Due to this sensitivity, the tuning procedure fails to reproduce the observedzonal climate if the strength of the surface friction is below some criticalvalue. Rather, the model's climate flips between the two extreme AO phasesfrom one tuning iteration to another, but still allows for rare jumpsto the other regime, respectively.Based on these results, the two regimes observed in the control simulationare interpreted as a feature of the attractor's ...