Centennial-scale shifts in the position of the Southern Hemisphere westerly wind belt over the past millennium

Introduction Conclusions References Tables Figures

AC: We have added a new section focused specifically on this topic (4.7.2) and have included the TSI reconstruction of Steinhilber et al (2009) in our figure of paleoclimate records (now Figure 7).
(3) As the authors mention it (P.3131, lines 14-16), it is still impossible to differentiate between shifts and changes in the intensity of the SWW in the past.As the proxy used in this study (CPP) cannot differentiate between these two mechanisms as well, the authors should always mention both possibilities, i.e. either "weakening and/or equatorward shift of the SWW", or "strengthening and/or poleward shift of the SWW", what is not always the case throughout the manuscript, as well as on Figure 9.The literature suggest that in the present day the SH storm track activity related to the SWW is persistent over the year in both intensity and location.Strongest activity is found in austral autumn and activity is extending over broader latitudes in austral winter, but it is always close to 50_S (Trenberth, 1991, Journal of the Atmospheric Sciences 48, 2159-2178).Garreaud et al. (2009, Palaeogeography, Palaeoclimatology, Palaeoecology 281, 180-195) wrote: "In particular, over the southern tip of South America and the adjacent south Pacific, the westerlies are strongest during austral summer, peaking between 45_ and 55_S.During the austral winter, the jet stream moves into subtropical latitudes (its axis is at about 30_S) and the low-level westerlies expand equatorward but weaken, particularly at â Lij50_S."So my impression is that the SWW are shifting poleward AND strengthen, while they are expanding equatorward AND weaken (see also Varma et al., 2012, GRL 39, L20704), but this may be valid only for the present day southeast Pacific region.The authors could mention this general pattern of the SWW in e.g.Section 3.

AC:
The reviewer makes a valid point, in that a single record cannot distinguish be-C3461 tween a latitudinal shift and a change in strength of the SWW.And we agree that modern observations indicate that SWW seasonal variability includes both strength and position changes.However, as we discuss in Section 4.6.4,by using records spanning the latitudinal range of the westerlies' influence, we believe we can differentiate between changes in strength and changes in position on decadal to centennial timescales.For example, where the GeoB-3313 Fe record (Lamy et al., 2001) indicates a southward shift and/or weakening of the SWW at the same time as our CPP record indicates a southward shift and/or strengthening, we can infer that the core of the westerlies shifted southward, thereby diminishing precipitation in the Chilean mid-latitudes while enhancing cyclone activity and the deposition of coarse particles at the WAIS Divide site.Thus, we prefer to keep our language consistent with this interpretation.
(4) While the CPP data do show an increase during the MCA, suggesting a poleward shift and/or a strengthening of the SWW, there is no real signal during the LIA (one would expect a decrease in the CPP) in comparison to the period before the MCA (I mean here 1950-950 yrs BP).The CPP data alone do not give any clue about changes in the SWW during the LIA.Is the proxy mainly sensitive to changes in the southern side/boarder of the SWW during warm periods, but not during cold periods, when the poleward side of the SWW weakens and the equatorward one broadens?Is there a threshold mechanism, which explains that a CPP decrease related to a wind weakening cannot be recorded ?Or were the climatic conditions during the LIA similar to the ones before the MCA ?Could you comment on this issue?AC: Yes, we believe the CPP record indicates times of increased SWW strength and/or southerly position, and that it is not as sensitive to northward shifts.It seems plausible that there is a threshold mechanism, as the reviewer suggests.We have added several sentences addressing this question in Section 4.6.and 10 ?).It would be helpful for the reader to broaden/stretch the graphic and to increase (double) the increment of the X axis.AC: See above.We prefer to keep these datasets plotted linearly.We have broadened the graphic and increased the x-increment (x4), as suggested.(2008), and add a 10Be record, the _18O record from the WDC ice core and e.g. the Y content or the _18O stalagmite records from Schimpf et al. (2011).AC: We have followed the reviewer's recommendations.Koffman and co-authors present a dust particle dataset from the West Antarctic Ice Sheet Divide ice core over the past 2400 yrs.Measured dust size changes are compared with zonal wind strength between 1979 and 2002 in an attempt to calibrate the dust record in terms of wind strength.The dust size is interpreted as reflecting Southern Hemisphere westerly wind strength and is subsequently compared with published datasets from the mid-latitude and tropical Pacific.The paper reads well and the datasets appear to have been obtained in a rigorous manner.However, I have several major comments below which should be addressed before the manuscript is published in its final form.In general I feel that more effort should be made to understand what the measured dust parameters actually represent before fitting the data with existing hypotheses about SWW shifts.
Major points: The authors argue convincingly that the relatively coarse size of the dust particles at WAIS suggests that dust sources are local, and are thought to be in Marie Byrd Land; P3140).Yet, in terms of the link between wind strength and CPP, there is little correlation between dust size and wind strength in the Marie Byrd Land region (Fig. 6).The authors should explain why this is the case.There is an arrow in their schematic (Fig. 9) suggesting the authors must have reason to think that winds were stronger in this region during periods of coarser CPP.One reason for that lack of correlation may be that the main winds in this region (and the arrow on the schematic) are onshore i.e. they are meridional as well as zonal.Therefore, the authors could C3469 also analyse the correlation with meridional winds in this region.Until a mechanism linking CPP and SWW in the Marie Byrd Land region is shown, the interpretation of CPP remains speculative.
AC: As the reviewer points out and as we show in Section 3.2, winds in the central West Antarctic region are predominantly onshore (i.e., from the NE).Although we had not made the link between zonal and meridional circulation explicit in the text, we do observe that the CPP correlates with zonal winds in the middle latitudes and with meridional (onshore) winds in Marie Byrd Land.This is consistent with our interpretation that changes in the mid-latitude westerlies are translated to the WAIS Divide site via cyclonic systems, as summarized in Figure 9 -a dynamic that can also been seen on a synoptic scale in satellite imagery.We have added a figure that shows the spatial correlations between meridional wind speed and WAIS Divide CPP (Figure 6b), and have added a paragraph in Section 3.3 to describe the circulation patterns more thoroughly.
The timing of the increase in CPP seems not to fit so well with the northern hemisphere climate anomalies, the MCA and LIA.The maximum in CPP is at 1300 and seems to span between about 1200 and 1450, hence covering the boundary between both MCA and LIA periods.The authors should still compare their record to the northern hemisphere climate anomalies such as the MCA and LIA, but I do not think that the CPP peak at about 1300 CE should necessarily be attributed to the MCA as is stated in the abstract and conclusions.Similarly, the timing appears to be later than climate changes in other records that are shown in Fig. 8.If the control on all of the other records is simple shifts of the SWW as the authors argue in section 4.4, the mismatches in timing should be better justified than the explanation that is presently given (P3149, L8-9).
AC: The reviewer makes a good point.Upon further consideration, we have adjusted the language throughout the paper to de-emphasize links with Northern Hemisphere climate anomalies.The addition of several new datasets to Figure 7 (previously Fig. 8) provides evidence for a close link with the tropical Pacific, and we have shifted the focus C3470 of the discussion toward discussing this relationship.In addition, we explicitly address (Section 4.6.4) the apparent temporal offsets among records, which we expect are a function of their respective timescales.We have plotted the age control points, with uncertainty, for the Lamy et al. (2001) and Schimpf et al. (2011) records, to help make the point visually that the three records' timescales are not equally resolved.
The link between SWW speed and dust emission/transport to the WAIS Divide site is an important aspect of the story but it is only briefly mentioned in the discussion (P3144 L20-26).The mechanisms by which increased cyclogenesis causes an increased CPP should be explained more clearly here, in terms of this region and the WAIS dust record and should be included in the abstract.flux would be masked by larger-scale temperature variability.
In the introduction there is a lengthy description of the SWW and their importance for global climate.However, considering that the manuscript aims to reconstruct changes in dust sources and dust transport pathways, very little information is presented on known Antarctic dust sources and on the mechanisms by which this dust is transported towards the ice core site.The introduction should be modified to focus more on Antarctic dust sources and transport pathways and less on the significance of the SWW.
AC: We see the reviewer's point, and recognize the challenge of describing and interpreting a 2400-year dust flux and size distribution record, while also comparing our record to a latitudinal range of paleoclimate reconstructions from the tropics and middle latitudes, and doing justice to all these topics.Given the primary focus of the paper on reconstructing past atmospheric circulation changes, we believe the introductory material on the westerlies is important.We chose to discuss dust sources and transport in Section 4.1, as this seemed the most relevant place to go into detail on these topics.
In the discussion there is a great deal of comparison with other records from South America.However, there is little comparison with downcore changes in dust from other high-resolution ice core records.I suggest to reduce the discussion of SWW records from South America but to add some plots and comparison with published ice core records (for e.g.those that are mentioned: Ruth et al, 2004;McConnell et al, 2007;Mosley-Thompson et al, 1990 or the cores using nssCa).This might help to assess the regional nature of dust deposition.

Figures: Fig. 1 :
Figures: Fig.1: Latitude and longitude coordinates are missing on both maps.Please indicate where the Amundsen-Bellingshausen Sea region is located.AC: We have added lat/long coordinates and place names to both maps.Fig.2: I suggest plotting the data on log scales (as well for the CPP records on Figures8 and 10 ?).It would be helpful for the reader to broaden/stretch the graphic and to increase (double) the increment of the X axis.

Figs. 6
Figs. 6 & 7: These two figures could be merged into one figure only.AC: Done.We combined figures 6 and 7 and added a spatial correlation map for the CPP and meridional wind speed, as requested by the second reviewer.

Fig. 8 :
Fig. 8: I recommend removing the records from Hall et al. (2010) and Moy et al.(2008), and add a 10Be record, the _18O record from the WDC ice core and e.g. the Y content or the _18O stalagmite records fromSchimpf et al. (2011).

Fig. 9 :
Fig. 9: Repeat in the figure capture that PSD = particle size distribution.AC: Done.

Fig. 10 :
Fig. 10: As well as for Figure 8, you may add here the 10Be record, the _18O record from the WDC ice core and e.g. the Y content or the _18O stalagmite records from Schimpf et al. (2011), and also the TEX86 record from Shevenell et al. (2011).Why C3468

Fig. 1 .
Fig. 1.Dust time series plots as in original Figure 2, except plotted on log scales.
Lubin et al. (2008)eimportant, the authors could test this by analysis of temperature indicators as suggested byLubin et al 2008.AC: One of the challenges to a more complete discussion of dust emission/transport to the WAIS Divide site is the fact that at present, no dust provenance data exist for any site in West Antarctica.Unlike the Dome C, Talos Dome, Berkner Island, and Vostok sites in East Antarctica, we do not yet have Pb, Sr, Nd or major/minor element geochemical data that can be used to constrain dust sources.For this reason, our discussion of dust sources and transport relies on grain size data coupled with known atmospheric circulation patterns.We suspect that the McMurdo Dry Valleys supply dust to WAIS Divide via cyclonic systems (as discussed in Section 4.1), but we cannot yet test this hypothesis.One of us (B.G.K.) is working currently with a group at Lamont-Doherty Earth Observatory to develop provenance datasets from West Antarctica that can be used to resolve some of these questions.Regarding the observation ofLubin et al. (2008)that mesoscale cyclone activity is associated with positive (primarily) winter temperature anomalies, we do not have adequately resolved long-term data to address this question.Lubin et al. (2008)were able to observe this relationship using daily-average temperature data, computed into monthly anomalies.Because the WAIS Divide oxygen isotope record is interpreted as an annual-mean surface air temperature record (WAIS Divide Project Members, 2013), synoptic-scale changes in sensible heat C3471