Final Discussion and Outlook

Twenty years from now you 'II be more disappointed by the things you didn't do, than by the ones you did do. So throw off the bowlines. Sail away from the safe harbor. Catch the trade winds in your sails. Explore. Dream. Discover.

Mark Twain

In this study climate variablity in the North Atlantic region is investigated with special emphasis on variability of the NAO and its interaction with the North Atlantic ocean as well as the Arctic sea ice export through Fram Strait.

The results of this study suggest that interdecadal NAO variability during the 20th century ( chapter 3) was associated with considerable changes in (i) the circulation of the North Atlantic ocean (chapter 4) and (ii) the link between the NAO and Arctic sea ice export through Fram Strait on interannual time scales ( chapter 6). From a statistical point of view NAO variability on intcrannual and quasi-decadal time scales is consistent with what is expected from realizations of "white noise" (chapter 3).

It is important, thus, to unravel the mcchanism(s) that gave rise to interdecadal NAO variability during the 20th century. The modelled response of the North Atlantic ocean to a surface flux forcing that was solely based on the observed NAO (chapter 4) pro-vides evidence that on interdecadal time scales oceanic changes lag behind the NAO by about 90° (Fig. 4.10). Since an atmospheric response to an oceanic forcing should be expected to occur instantaneously, the modelled phase-relationship between the NAO and the North Atlantic ocean on interdecadal time scales docs not support the presence of a two-way coupled mode of (interdecadal) variability, as found for instance, by Timmermann et al. (1998) in a century-scale integration of the ECHAM3/LSG model³⁵. Although the agreement between the observed and modelled evolution of interdecadal North Atlantic SST anomalies during the 20th century (Fig. 4.4) suggest that this simulation performed realistically, our confidence in this integration could be further enhanced by comparing modelled subsurface data with observations, at least where possible.

Since the results of this study provide little evidence for the presence of two-way air-sea interaction in the North Atlantic region on interdccadal time scales, other possible mechanisms have to be taken into account in order to explain interdccadal NAO variability during the 20th century. It cannot be ruled out that the recent positive trend of the NAO is forced by increased greenhouse gas concentrations. This notion is supported, for instance, by

• the strong projection of the NAO onto the recent Northern Hemisphere warming³⁶ (Hurrell, 1996),

35111 this model the NAO and the !\'orth ,\tlantic THC are approximately in phase on intcrdecadal time scales (Axel Tirurnerrnann, personal cornrnunication).

36The increasing frequency of El Nirio events in recent years also contributed to this \\'arming (flurrdl, 1996). :-.:otice, that Timmern1ann et al. (1999} found a v.·arming of tropical Pacific SST under increasing greenhouse ga<, concentrations iu the ECIL.\~I4/0P)'C3 model.

101

• increasing interdecadal NAO variability throughout the 20th century (Fig. 3.10) (see also Hurrell and van Loon, 1997),

• relatively weak interdecadal NAO variability before the mid-19th century (about 1650-1850) century as suggested by Greenland ice core data (Appenzeller et al., 1998b),

• decreasing SLP over the high-latitude Northern Hemisphere under increasing green-house gas concentrations in different CGCM experiments ( e.g., Ulbrich and Christoph, 1999; Shindel! et al., 1999; Fyfe et al., 1999; Paeth et al., 1999), although some un-certainties about the centers of the strongest secular changes remain.

• the similarity of the eastward shift of the NAO's centers of interannual variability as observed during the last two decades (Fig. 6.3d) and simulated under an enhanced greenhouse gas forcing (about 2020) by the ECHAM4/0PYC3 model (Fig. 6.12f) (see also Ulbrich and Christoph, 1999, their Fig. 6).

\Vhen discussing secular changes of the NAO under increasing greenhouse gas concentra-tions using state-of-the-art CGCMs, it has to be kept in mind, however, that the mecha-nisms giving rise to natural interdecadal climate variability in the North Atlantic region differ from model to model. These differences introduce considerable uncertainties in the context of modelled feedbacks.

Throughout this study interannual to interdecadal NAO variability was studied using winter-averaged data. As demonstrated by Madden and Jones (1999) the analysis of winter-averaged data is associated with considerable aliasing: "In the case of one sam-ple per year (e.g., time series of Januaries or Winter Seasons), the aliased variance at resolved frequencies is an increadible 92% and 75% for 30 and 90 days averages". Thus, care has to be taken when interpreting the results, since a considerable amount of "in-terannual" variability may result from intra.seasonal sampling variability. The aliasing problem may be expressed in more physical terms: Atmospheric climate variability can be understood in terms of changes in the occupation statistics of intraseasonal large-scale cir-culation patterns (e.g., Palmer, 1993, 1999; Corti et al., 1999). Hence, in order to improve our understanding about interannual to intcrdecadal NAO variability, it is important to understand the nature of intrascasonal variability (so-called weather regimes).

There is a growing body of evidence that low-frequency intra.seasonal variability of the extratropical atmosphere is governed by nonlinear dynamics (e.g., Kimoto and Ghil, 1993a,b; Corti et al., 1999; Smyth et al., 1999). If the extratropical atmosphere is treated as a nonlinear dynamical system, then the interpretation of a possible external forcing of the atmosphere is influenced (Palmer, 1993, 1999): An external forcing may primarily appear as a change in the frequency of occurrence of fixed natural modes of atmospheric variability, rather than as a change in their location. The observed eastward shift of the NAO's centers of interannual variability around the late 1970s may be seen as evidence against Palmer's nonlinear paradigm. On the other hand, it cannot be ruled-out that this shift resulted from changes in the occupation statistic of more than one natural mode of variability. Thus, it is of importance to study this shift from the point of view of intra.seasonal variability. Since a similar shift is observed in ECHAM4/0PYC3 model under increased greenhouse gas concentrations, this model integration provides a valueable data source to further study the nature of this shift.

)

Figure 7.1: Daily SLP at Iceland (solid in hPa) during the (a) high NAO winter 1988/1989 and the (b) low NAO winter 1995/1996. Winter averaged SLP (December through March) is given as dashed line.

It is worth mentioning that this shift may well be in agreement with Palmer's non-linear paradigm taking into account the possibility for bifurcations of the extratropical atmosphere under an external forcing. Modelling evidence has been presented by Molteni and Corti (1998) that the extratropical atmosphere over the North Pacific resides in the neighbourhood of a bifurcation point³⁷. The results by Molteni and Corti (1998) suggest that in this model changes in the tropical forcing can bring the atmosphere beyond a bifurcation point. This bifurcation becomes evident from differences in the intraseasonal variance between winters with strong and weak projections onto the Pacific/North Amer-ican pattern (PNA pattern). Similar differences are associated with high and low NAO winters in the North Atlantic region (Fig. 3.5a and Fig. 5.lOb). To further illustrate this difference, Fig. 7.1 shows daily SLP time series from Iceland during the high and low NAO winters 1988/1989 and 1995/1996, respectively. (Compare Fig. 6.1 for the choice of these years.) Obviously, SLP at Iceland was relatively persistent during the high NAO winter 1988/1989, whereas considerable variability at the ]ow-frequency end of the intra.seasonal SLP spectrum took place at Iceland during the low NAO winter 1995/1996. These differ-ences may be of importance when considering the physics, predictability, and impacts of intraseasonal atmospheric variability.

From the results of this study and their dicussion in the context of other recent publi-cations (see above) - in my view - the following items deserve special focus in future studies in order to improve our understanding about natural variability and possible an-thropogenically induced changes of the North Atlantic climate system:

• Further assessment how the North Atlantic ocean responds to a forcing by the NAO, especially on interdecadal time scales. A thorough understand-ing of this response is a prerequisite to realistically model low-frequency changes of the coupled North Atlantic climate system.

• Future studies on extratropical climate variability should be based on a

37Note, that an eastward shift of the dominant mode of atrno!-pheric variability around the late 1970s is also evident in the :North Pacific basin (Fig. 6.13).

103 broader range of frequencies taking into account atmospheric variability from days to decades (compare Fig. 1.1).

104 A SHIFT OF NAO AND IMPACT ON NORTH ATLANTIC CLIMATE