Klaus Fraedrich

This observational study investigates the impact of North Atlantic synoptic-scale wave breaking on the North Atlantic Oscillation (NAO) and its connection with the stratosphere in winter, as derived from the 40-yr ECMWF Re-Analysis (ERA-40). Anticyclonic (AB) and cyclonic wave breaking (CB) composites are compiled of the temporal and spatial components of the large-scale circulation using a method for the detection of AB and CB events from daily maps of potential vorticity on an isentropic surface. From this analysis a close link between wave breaking, the NAO, and the stratosphere is found: 1) a positive feedback between the occurrence of AB (CB) events and the positive (negative) phase of the NAO is suggested, whereas wave breaking in general without any reference to AB- or CB-like behavior does not affect the NAO, though it preferably emerges from its positive phase. 2) AB strengthens the North Atlantic eddy-driven jet and acts to separate it from the subtropical jet, while CB weakens the eddy-driven jet and tends to merge both jets. 3) AB (CB) events are associated with a stronger (weaker) lower-stratospheric polar vortex, characterized by the 50-hPa northern annular mode. During persistent weak vortex episodes, significantly more frequent CB than AB events are observed concurrently with a significant negative NAO response up to 55 days after the onset of the stratospheric perturbation. Finally, tropospheric wave breaking is related to nonannular stratospheric variability, suggesting an additional sensitivity of wave breaking and, thus, the NAO to specific distortions of the stratospheric polar vortex, rather than solely its strength.

A tangent linear adjoint for a low-resolution dynamical model of the atmosphere is used to derive the optimal forcing perturbations for all state variables such that after a specified lead time the model response has a given projection, in terms of an energy norm, on the pattern associated with the 51-yr trend in the Northern Hemisphere winter tropospheric circulation, 1948/49–1998/99. A feature of the derived forcing sensitivity is a Rossby wave–like feature that emanates from the western tropical Pacific and is associated with the deepening of the Aleutian low, whereas an annular pattern in the forcing sensitivity in the uppermost model level is shown to be associated with the pattern of the trend over the Euro-Atlantic/Asian sectors, including the upward trend in the North Atlantic Oscillation index. The authors argue that the Rossby wave–type feature is consistent with studies that have argued a role for the upward trend in tropical sea surface temperature during the 51-yr period. On the other hand, the authors interpret the annular pattern in the forcing sensitivity as being consistent with studies that have argued that the trend over the Euro-Atlantic sector was associated with influences from the stratosphere. In particular, a nonlinear model driven by the optimal forcing perturbation applied only to the top model level is successful at reproducing the trend pattern with the correct amplitude in the Euro-Atlantic sector, but implies a trend over the North Pacific toward a weaker Aleutian low, contrary to what was observed but similar to the spatial pattern associated with the northern annular mode. These results show that the adjoint approach can shed light on previous apparently different interpretations of the trend. The study also presents a successful application of a tangent linear adjoint model to a climate problem.

The low frequency variability of the near surface temperature in a climate simulation is compared with Greenland ice core δ¹⁸O time series during the holocene. The simulation is performed with the coupled CSIRO atmosphere‐ocean model under present‐day conditions. The variability, analyzed by the detrended fluctuation analysis, reveals power‐law scaling of the power‐spectrum for frequency f, S(f) ∼ f^−β, and long term memory (LTM) given by β > 0. The near surface temperature shows intense LTM in the North Atlantic south of Greenland, weak LTM in parts of the Antarctic ocean and the tropical Atlantic, and no LTM in the Pacific ocean. The power‐law exponent β ≈ 0.5 near Greenland agrees with ice core temperature proxies up to time scales of 1000 years. The LTM of the surface temperature is explained by the high low frequency variability of the zonally averaged stream‐function in the Atlantic with maxima in the Arctic ocean.

A Reply to the Comment by Armin Bunde et al.

Tropical cyclone tracks in the Australian basin are predicted by an analog ensemble forecast model. It is self-adapting in its search of optimal ensemble members from historic cyclone tracks by creating a metric that minimizes the error of the ensemble mean forecast. When compared with the climatology–persistence reference model, the adapted analog forecasts achieve great-circle errors that improve the reference model by 15%–20%. Ensemble mean forecast errors grow almost linearly with ensemble spread. Abstract Tropical cyclone tracks in the Australian basin are predicted by an analog ensemble forecast model. It is self-adapting in its search of optimal ensemble members from historic cyclone tracks by creating a metric that minimizes the error of the ensemble mean forecast. When compared with the climatology–persistence reference model, the adapted analog forecasts achieve great-circle errors that improve the reference model by 15%–20%. Ensemble mean forecast errors grow almost linearly with ensemble spread.

Idealised global circulation model simulations with circumpolar and localised (one and two) storm tracks are re-analysed to determine scaling, intermittency and phase-space structures. In a hundred year experiment with a circumpolar storm track, the spectrum S(f ) of the first principal component of the zonal wind fluctuations shows the following power law regimes: (a) a short-term memory between f^{- -4} and f  ^-2 up to 50 days and (b) a long-term memory f ^-1 from 50 to 400 days and f ^-0.24 beyond 400 days, similar to observed maritime single station near-surface air temperature data. In the presence of localised storm tracks, the wave number two dominates the dynamics and a long-term memory cannot be detected. The recurrence plot is introduced as a novel tool to comprehensively visualise the evolution of the dynamical system in terms of state separations (distances) in phase space. The patterns allow for a qualitative interpretation of the underlying local phenomena in phase space, such as waves, analogs, extremes, and global regimes. Attractor dimensions are, in general, larger than 10, but they appear to be lower in the wave-dominated regimes of the double storm track experiment.