|Compared to the previous version, this version has much improved. For instance, in my view, the comparison of the CCLM results with ECHAM5, including the application of the cost function, has strengthened the manuscript considerably. However, there are still several points that must be addressed before this manuscript can be considered for publication. |
Concerning my first main comment in the previous review, I find that the manuscript is still rather poorly written. It still includes many language mistakes, so I am quite sure that the text has not been checked by a native speaker as I suggested. Such a check would significantly improve the readability of the manuscript, so I stress this issue once more here.
Furthermore, regarding my 5th previous comment, the authors have not really provided a more in-depth discussion of the uncertainty of the summer temperature reconstructions for Southern Europe. In addition, their reply to my comment is not clear to me on this point. The authors note that "the previous analyses of mid-to-late Holocene temperature evolution were misleading. In fact, simply considering regional means, they did not allow to have a proper overview of the trends at different locations, possibly resulting in a mismatch in the comparison against other proxies. The new maps presented in Fig. 4 show now a more heterogeneous behaviour, and are in better agreement with other indipendent reconstructions". From the reply, it is not clear who "they" are, or what "previous analyses" the authors are referring to. Furthermore, Figure 4 presents maps of summer temperatures for the period 1991-2000, so not for the mid-to-late Holocene. A major outcome of the present paper is the mismatch between the CCLM model results in Southern Europe and the pollen-based reconstructions of Mauri et al. (2014), showing an opposite trends in summer. In Mauri et al. (2014), the pollen-based reconstruction is compared against independent reconstructions (their Figure 8), but this was not done for summer temperatures in Southern Europe, except for the Alps. In explaining the noted mismatch, Russo and Cubasch only consider a model bias in surface heat fluxes, but do not reflect on the possibility that the pollen-based reconstruction does not capture the right temperature trend in the Mediterranean area during the Holocene. The latter possibility has been discussed in the literature before as I mentioned in my review of the CPD manuscript, and I urge the authors again to make the readers aware of this discussion. As mentioned in Section 3.3.1., all PMIP3 models suggest warmer conditions in Southern Europe during the mid-Holocene (see also Table S2), so if the reconstruction of Mauri et al. (2014) is correctly indicating cooler conditions here, this implies that all these models are not capable of capturing the right climate response in Southern Europe under influence of astronomical forcing. An alternative to consider is that the reconstructions are incorrect. This important issue is still under debate, and in my opinion it is crucial to provide the arguments of both sides of this debate.
Line 157: It is mentioned here that effects of greenhouse gas forcing is not considered in the analyses. Does this mean that the trace gas concentrations were kept fixed at their preindustrial values in all experiments? Or are the concentrations adjusted in the experiments based on ice-core measurements, but is the effect of this forcing not separated from the impact of insolation changes? Please clarify.
Line 195: Please explain what you mean by 'satisfactory results'.
Lines 213 and 219: Why do you use "instead" in these sentences?
Line 219 to 228: I suggest to make this section more quantitative, as it is very descriptive.
Section 3.3.1: This new section on the results of PMIP3 experiments is a useful addition to the manuscript. However, it is very descriptive and it could be made more informative by making it more quantitative.
Line 270: I suggest to use "proxy-based reconstructions", as the proxies themselves are not reconstructed.
Line 281: "The coastline is also better reproduced in this case, resulting in more suitable informations for possible comparison against proxy-data". Please clarify how a more detailed coastline would improve the model-data comparison. See also line 448 in the Conclusions.
Line 283: "CCLM shows better defined patterns as a consequence of higher resolution". What kind of patterns, please clarify. See also line 449 in the conclusions.
Line 287: Zhang et al. (2010) is missing in the reference list.
Line 306: "It is important to mention that the scale considered in our analysis is closer to the resolution of ECHAM5 than the one of the CCLM". What scale are you referring to here? The scale of the reconstructions?
Line 347: In Fig 9 the winter map for CCLM is gray (i.e. the trend is not significant) in Scandinavia, where the reconstructions show a strong, significant cooling trend. So here the model results do not match the proxy data, and I suggest to mention this here.
Line 387: "the first CCA pair (Fig. 11 a,b)". Is this correct? The MSLP pattern explaining most of the variance is shown in Fig. 11c, so should Fig 11c,d not be the first CCA pair?
Line 409: "to capture this trend". What trend? In soil moisture during Summer? Or in winter or spring precipitation. Please clarify.
Figure 1: In the caption, a reference for the presented data should be added.
Caption Figure 5: Typo, "refernce period"
Figure 6: I suggest to use the same projection in all six panels to facilitate the inter-model comparison.
Figure 9: Please clarify what the unit of the slopes. Is it in °C per 6kyr? Or °C per kyr?
Figure 12: In my view, this figure does not add much to the manuscript, so it could be described only, or it could be moved to the supplementary information.