THE ROLE OF THE NON-AXISYMMETRIC ANTARCTIC OROGRAPHY IN FORCING THE OBSERVED PATTERN OF VARIABILITY OF THE ANTARCTIC CLIMATE

Tom A. Lachlan-Cope, William M. Connolley, and John Turner.

British Antarctic Survey, Cambridge, UK. 26 April 2001

Abstract. The pattern of inter-annual variability in the atmospheric circulation around Antarctica has a maximum over the Amundsen-Bellinghausen Sea (ABS), which is particularly strong during the winter (June, July and August). By using an atmosphere-only general circulation model the causes of this maximum have been investigated. In particular we study the effect of the non-axisymmetric nature of the local surface forcing (sea surface temperatures, sea ice and orography) by imposing axi-symmetric forcing fields at high southern latitudes. The results of these experiments show that the the variability maximum in the ABS can be explained by the non-axisymmetric nature of the Antarctic orography is sufficient to explain the alone. variability maximum in the ABS .

1. Introduction

     The climate of Antarctica and its surrounding seas is very variable; temperature, and mean sea level pressure (MSLP) both show large inter-annual variability and any study of the Antarctic climate and how it changes must consider the role and cause of this variability. For example, the annual average surface air temperature of the western Antarctic Peninsula shows a large year-to-year variation that almost masks the systematic warming observed there over the last 50 years [King, 1994] and it is possible that the observed warming is a result of a change in the nature of the variability.

     The interannual variability of the MSLP and upper geopotential surfaces is not constant over the whole of the Antarctic region. Connolley [1997] showed that there is a maximum in the standard deviation of the annual average MSLP fields over the Amundsen-Bellingshausen Sea (ABS hereafter). The feature has been termed the ‘pole of variability’ and is very robust, being present in all available re-analysis data sets and most climate model runs at the surface and at 500hPa. The pole of variability is present to a certain extent in all seasons but is particularly strong during the southern winter (June, July and August). Figure 2i shows the standard deviation of 15 years of the winter season MSLP from the European Centre for Medium-range Weather Forecasts (ECMWF) reanalysis (ERA) data set. It might be thought that the variability seen in the ABS in the ERA data set is an artefact of a data-sparse region. However, a similar pattern is found in the Australian Bureau of Meteorology analysis (Connolley [1997]) and in the National Centers for Environmental Prediction (NCEP) reanalysis project data (not shown); furthermore the year-to-year variations are reproduced in the independent analyses.

     In this paper the cause of the pole of variability is investigated. In particular, we address the question; is its location determined by the non-axisymmetric nature of the Antarctic orography alone? The climate of the area surrounding Antarctica depends on the large-scale flow, which is controlled by planetary scale waves. These waves can be generated from many sources, such as sea surface temperature (SST) anomalies or the orography of large landmasses. There is some debate about the source of these waves found near Antarctica. James [1988] suggested the orography of the continent as the source for the planetary wave number one found around Antarctica. Also, Quintanar and Mechoso [1995 a and b] conducted a series of model experiments that looked at the effect of the Antarctic orography and zonal asymmetries in sea ice and SSTs. They concluded that the observed quasi-stationary wave one was predominately forced from lower latitudes. Both James [1988] and Quintanar and Mechoso [1995 a and b] looked at the planetary scale waves rather than the variability. This paper does not investigate the source of these waves, but rather looks at the variability of the climate around Antarctica that may be forced by these waves.

2. Method

     This paper examines the causes of the pole of variability by replacing the ‘real’ model orography of Antarctica with an idealised axi-symmetric orography, which has a surface slope close to that of the real orography and so still allows realistic gravity-forced outflow. The other main high latitude surface forcings, such as SST and sea ice extent, are also replaced with axi-symmetric fields around Antarctica.

     The model used for this work was the atmosphere-only version of the Hadley Centre General Circulation Model (GCM) (Cullen, 1993)  known as HadAM2. This model has 19 levels in the vertical and a 2.5º x 3.75º horizontal grid. The model is forced at the lower boundary by a set of ancillary fields giving the climatological characteristics of the surface, such as temperature and surface type, and these were modified to give axi-symmetric forcing in the Antarctic region in some of the runs reported here.

     The main modifications were made to the orography, sea ice and SST fields, although some other fields, such as the soil temperature fields also had to be altered for consistency.

     The axi-symmetric orography field was generated by zonally averaging the orography around the coast of East Antarctica. The radius and height of the Antarctic dome produced was then adjusted so that it had approximately the same height and volume as the highly smoothed original model orography. This axi-symmetric orography was used centred on the pole and also displaced from the pole by 10º of latitude to represent a highly idealised version of the asymmetric nature of the real orography. The three different Antarctic orography fields used in this study are shown in Figure 1.

     Axi-symmetric sea ice and SST fields were produced for the area around Antarctica by zonally averaging these fields over the Southern Ocean. The zonally averaged SST fields were merged with climatological yearly repeating sea ice and SST data between 50ºS and 40ºS. The runs have been carried out using the various forcing fields shown in Table 1.

     The variability studied in this paper is represented by fields of the standard deviation of winter (June, July and August) means of the surface pressure. The standard deviations of winter means of other fields, such as the 500hPa height, give similar results and are not shown. The yearly means also show a similar, but somewhat reduced, maximum in variability in the ABS. The F-statistic has been used to identify areas where the variance is statistically different, at the 95% level, between the various runs. On the plots presented in this paper runs A and C (Figures 2ii and 2iv) are compared to the axi-symmetric case, run B, and areas where the variance is significantly different, at the 95% level, are shaded.

     To study the propagation of planetary scale stationary waves the horizontal generalised Eliasen-Palm flux (F_h) is calculated using the expression derived by Plumb [1985].

3. Results

     The standard deviation of the winter averaged surface pressure for the control run A is shown in Figure 2ii. This run shows a clear maximum in the variability over the ABS. This pattern is very similar, although smaller in extent and displaced slightly to the west, to that seen in the ERA data (Figure 2i) and also to the results reported by Connolley (1997).  Also, the area of lower variability between 50-60º S near the Greenwich meridian can be seen and this is also present in the ERA data. The pattern of variability is seen to be different between figures 2ii and 2iii, with the variability maximum in the ABS missing from 2iii. The difference in the ABS is confirmed as significant by the F-statistic shown in figure 2ii. Also, the largest differences, globally, between Figures 2ii and 2iii are found in the ABS. There are of course other areas of difference because of the large change we have made to the orography. However, in this study we are primarily interested in the forcing of the variability in the ABS.  In Figure 2ii the maximum in the ABS is statistically significant (at the 95% level) when compared to the axi-symmetric case.

     In the run made with pole-centred axi-symmetric orography (run B), shown in Figure 2iii, the ABS maximum is not present, although other features seen in the control run (Figure 2ii), such as the lower variability near the Greenwich meridian and the increased variability around 50º S and 230º W, are present.

     The final run (C) was made with the axi-symmetric orography displaced to 90ºE, 80ºS: this is intended to simulate the gross effect of the displaced centre of mass of the real Antarctic orography. Figure 2iv shows the results for this model run. This time a clear maximum of variability can be seen in the ABS and this maximum is statistically significant (at the 95% level), when compared to the axi-symmetric run. However, in this case the minimum near the Greenwich meridian and the secondary maximum at 230ºW are not observed. In fact a statistically significant maximum is seen near the Greenwich meridian n.and this maximum is not present in reality. However the displaced axi-symmetric orography is only intended to crudely simulate the main feature of the Antarctic continent and should not be expected to reproduce all the details of the observed variability field. In particular, the axi-symmetric orography does not include any representation of the Antarctic Peninsula  and we believe that the maximum near the Greenwich meridian may be caused by more synoptic scale weather systems passing into the South Atlantic from the ABS unimpeded by the Peninsula.

     The final Figures 3a and b show the horizontal generalised Elliasen-Palm (E-P) flux calculated on the 500hPa surface for runs A and B. The horizontal E-P flux is plotted over the zonal anomaly of the 500hPa height and this shows clearly a wave 1 pattern with higher pressure off the coast of East Antarctica. It is the variation of this high-pressure cell that is responsible for the high variability in this area. The E-P flux in Figure 3a (run A) indicates that large-scale stationary planetary wave activitys, generated close to the Ross Sea area, are being transported from East Antarctica towards the ABS. This E-P flux into the ABS is missing from the run (B) using axi-symmetric Antarctic orography shown in Figure 3b. The general pattern of the E-P flux for the rest of the area shown is very similar between the two cases. In particular the pattern over the Weddell Sea and the pattern near 60ºS 150ºW are very similar and in these areas there is very little difference in the variability between the two runs. There are some large differences in the E-P flux over the continent and this reflects the lack of steep topography in the axi-symmetric case to generate large scale planetary waves.

4. Conclusions

     The results show clearly that the asymmetric nature of the orography of Antarctica, and in particular the high plateau of East Antarctica, are necessary for the presence of the maximum of variability seen in the Amundsen-Bellinghausen Sea.

     With the displaced axi-symmetric orography, using axi-symmetric surface forcing at high southern latitudes and climatological yearly repeating SSTs elsewhere (run C), the observed pattern of year-to-year variability in the ABS was reproduced with a larger than observed magnitude. However the run with the pole-centred axi-symmetric orography (B) does not have the observed maximum in the variability over the ABS. This suggests that it is the centre of mass of Antarctica displaced from the pole that is responsible for the observed pattern of variability. The mechanism for the process involved in the formation of this pattern is likely to be planetary scale Rossby waves generated from the displaced orography. The pattern of the Elliasen-Palm flux shown in Figure 36 for the run with yearly repeating SSTs and the original ‘real’ model orography does indeed show a flux of planetary scale waves being produced by the high plateau of East Antarctica. Although we have not considered the effects of tropical SST anomalies, such as those associated with the El Niño/Southern Oscillation, it is likely that they will play a role in controlling the variability of this region. Future work will address the role played by Rossby waves generated in the tropical Pacific. However, as this paper demonstrates, it is not necessary to have real year-to-year variations of tropical SSTs to generate the area of maximum variability.

References

Connolley, W.M., Variability in annual mean circulation in southern high latitudes, Climate Dynamics,  1997.

Cullen, M.J., The unified forecast/climate model, Met. Mag., 122, 81-94, 1993.

James, I.N., On the forcing of planetary scale Rossby waves by Antarctica, Quart. J. Roy. Met. Soc., 114, 619-637, 1988.

King, J.C., Recent climate variability in the vicinity of the Antarctic Peninsula, Int. J. Climatol., 14, 357-369, 1994.

Plumb, R.A., On the three dimensional propagation of stationary waves, J. Atmos. Sci., 8, 217-229, 1985.

Quintanar, A.I. and C.R. Mechoso, Quasi-stationary waves in the Southern Hemisphere. Part I: Observational data., J. Clim., 8, 2659-2672, 1995a.

Quintanar, A.I. and C.R. Mechoso, Quasi-stationary waves in the Southern Hemisphere. Part II: Generation mechanisms., J. Clim., 8, 2673-2690, 1995b.

(Received May 17, 2001; revised July 12, 2001; accepted August 9 2001.)

Figure 1.The orographies used in the model runs; a) the standard models orography used in run A. b) the axi-symmetric orography used in run B with the position of the displaced orography used in run C shown as a dashed line.

Figure 1.The orographies used in the model runs; a) the standard models orography used in run A. b) the axi-symmetric orography used in run B with the position of the displaced orography used in run C shown as a dashed line.

Figure 2. i) Standard deviation of the winter (June, July and August)  average of the MSLP from the ECMWF reanalysis data set. ii) The standard deviation of the winter average of the surface pressure for model run A, using standard orography and a yearly repeating climatology of sea surface temperatures and sea ice. iii) The standard deviation of the winter average of the surface pressure for model run B, using axi-symmetric forcing for orography, sea ice and high latitude sea surface temperatures. iv) The standard deviation of the winter average of the surface pressure for model run C, using the displaced orography and axi symmetric forcing for sea ice and high latitude sea surface temperatures. The significant difference between runs AC and CD and the run with axi-symmetric Antarctic orography (AB) at the 95% level are shown shaded. The standard deviations show here are in hPa.

Figure 2. i) Standard deviation of the winter (June, July and August)  average of the MSLP from the ECMWF reanalysis data set. ii) The standard deviation of the winter average of the surface pressure for model run A, using standard orography and a yearly repeating climatology of sea surface temperatures and sea ice. iii) The standard deviation of the winter average of the surface pressure for model run B, using axi-symmetric forcing for orography, sea ice and high latitude sea surface temperatures. iv) The standard deviation of the winter average of the surface pressure for model run C, using the displaced orography and axi symmetric forcing for sea ice and high latitude sea surface temperatures. The significant difference between runs AC and CD and the run with axi-symmetric Antarctic orography (AB) at the 95% level are shown shaded. The standard deviations show here are in hPa.

Figure3. a) The zonal mean anomaly of 500 hPa height from the long term zonal mean for run A with the horizontal component of the Elliasen-Palm flux superimposed. b) The zonal mean anomaly of 500 hPa height from the long term zonal mean for run B with the horizontal component of the Elliasen-Palm flux superimposed.

Figure3. a) The zonal mean anomaly of 500 hPa height from the long term zonal mean for run A with the horizontal component of the Elliasen-Palm flux superimposed. b) The zonal mean anomaly of 500 hPa height from the long term zonal mean for run B with the horizontal component of the Elliasen-Palm flux superimposed.

Table 1 The forcing fields used and length of the three model runs.