Dear Michael and colleagues:

This is a very cool manuscript, and I think the model issues are mostly addressed and fine. However, the true beauty and impact of this work will come if basic concepts and key points can be conveyed to potential readers who lie outside the GCM domain. This is because I strongly suspect that the easier you can convey concepts, points and problems to those who generate proxy records, the faster such records will arrive to constrain and to enlighten the models. My comments below are given in an effort to bridge this gap.

I am hoping that you will make several changes, as I think this will very much improve the importance of the manuscript.

Sincerely,

Gerald (Jerry) Dickens

Major Comments

(A) A proper “climo-stratigraphic” framework is mostly missing. Yes, the early Eocene (~56 to 49 Ma) was a time of general warmth relative to present-day (and indeed most of the Cenozoic). However, the early Eocene was also a very dynamic interval that includes the multi-million year interval of peak Cenozoic warmth (aka the EECO) and several hyperthermals (e.g., the PETM) -- all likely upon major long-term changes in pCO2 and temperature.

The tone of the manuscript, whether intended or not, is an awkward and unfortunately common one in most literature: comparing model simulations to proxy records over the Early Eocene, as if the ~7 Myr interval represents a single target for confluence and agreement. If one looks at detailed proxy records for the early Eocene, one sees amazing variance, probably with greater variations in earth surface temperature than all that has occurred over the last 7 Myr (Late Miocene-Holocene).

So, I ask a simple philosophical question: how would this manuscript be treated by members of our broad community if the goals were to compare model simulations and proxy targets over the last 7 Myr as if the Late Miocene-Holocene was considered all the same?

The conflation of time and the muting of climatic variance over such time remains a huge issue, and almost necessarily requires acknowledgement in the Introduction and briefly discussed to make this a great paper.

I think portions of the Abstract, Introduction and Proxy-model comparison almost have to be rewritten with this perspective in mind. I offer some basic and specific suggestions (the latter later).

Rewrite paragraph 2 (Lines 58-74) so it conforms to current knowledge of the Early Paleogene. This is one where Earth’s surface warmed, seemingly by about 4°C, from the late Paleocene (ca. 58 Ma) through the start of the EECO (ca. 53 Ma). Superimposed on this warming and within at least the beginning of the EECO were a series of hyperthermals, including the PETM ca 56 Ma, where Earth surface temperatures increased nominally 5°C above ambient background conditions. (In details, when one realizes that the PETM occurred well before the start of EECO, and the K/X event may have been the time of maximum warmth, and puts all together, Earth’s surface temperature probably varied by about 8°C in terms of magnitude over the latest Paleocene-early Eocene, and by at least 5°C over the early Eocene … and hence my comment about the last 7 Myr).

Add a new paragraph to the Introduction (would be number 3 or 4, depending on how you rewrite) that acknowledges the above issue, but briefly explains how and why this presents a truly great and wonderful problem. Basically, there is a 7 Myr window of time where physics and plate configurations did not change significantly, where highly elevated background pCO2 and temperature both occurred, but where BOTH changed significantly. And thus, there is a need for internally consistent GCMs that conform to modern conditions, but with a different past plate configuration, AND IMPORTANTLY a wide range of temperature and likely pCO2 forcing. (It is the latter that is problematic in the current manuscript, as the ranges are given, but presented almost as errors in past pCO2/Temp proxies rather than alternatively representing plausible ranges for pCO2/Temp in the time domain if we had perfect stratigraphy and interpretations of proxy records.)

You all are certainly welcome to explain to me where I am wrong on the above and several following comments. (Of course, my first response would be to blame this on sharing a hotel room for week with Matt Huber in Lower Hutt, New Zealand). I look at things as the modeling presents a range of possibilities with CO2-equivalent forcing, and that with a hypothetically perfect model such forcing is changing significantly through the early Eocene.

(B) The effect of seasonality in comparisons of model simulations and proxy information, especially at high latitudes and in warm worlds, is not really emphasized. It makes little sense, at least to me, to compare results of mean annual conditions in models to proxy formation in regions where mean annual conditions are heavily skewed toward a certain season.

Seriously, much of supposed early Eocene “model-data problems” for temperatures at high latitudes disappears if one assumes the proxies are not recording mean annual conditions but rather average spring-fall conditions when proxies are presumably being formed. I will speculate that the same is true for MAP in certain regions.




Specific comments

Lines 23-25: This is somewhat awkward writing in terms of stratigraphy, and compounds problematic issues that have arisen in numerous papers. (see above comment A)

Line 27: I do not follow the end clause beginning with “despite”. Rewrite.

Line 40: There is an alternative, namely that the proxies are not being interpreted correctly. Indeed, this is acknowledged in Lines 88-89.

Line 58: This is the correct approximate time for the Early Eocene. However, as noted above, surface temperatures were certainly not constant through the Early Eocene.

Lines 66-74: I do not believe that many of the temperature proxies at high-latitudes reflect MAT, because there is a huge potential bias for seasonality and skewness toward summer temperatures.

Line 81: As above, the word “maintained” is awkward.

Line 98: Without rewriting and a proper stratigraphic framework, the reference to the PETM will not make sense to some readers.

Line 153: Is this vegetation set at the start of the model as a boundary condition and then evolves with precipitation?

Line 154-159: Following from above, the inclusion of a vegetation model needs elaboration. Is the vegetation prescribed so as to affect precipitation or is it truly dynamic and coupled, so that it also responds to precipitation? For example, presumably in areas that become drier, or more seasonal in rainfall, the type and density of vegetation decreases.

Line 195: Svensen et al. should not be cited here, as it applies only to the PETM. Moreover, the volcanism invoked is intrusive.

Lines 577-579: Indeed, this may explain much of the data-model mismatch.

Figures general: It would be nice to have all figures internally consistent in terms of units. That is, have mm per day or cm/yr, 0-360 longitude or +/- 180 longitude, but do not alternate.

The colors are difficult to follow in some cases, as results for different models vary across different figures. Consistency would help.

Legends on the figures should state parameters explicitly. For example, Figure 7a should state “mean annual precipitation during extended summer (%)” not just “(%)”, Basically, try to make figures as stand alone as possible.

Figures 1 and 3: As there is room for an extra panel, it would be very helpful to have a generic panel on each that identifies key features such as ITCZ, SPCZ, etc.

Figure 1: Caption should state modern plate configuration.

Figure 2: It is not immediately obvious to realize the difference between the two different blue boxes representing two different models. Maybe have one symbol with a crossed-box?

Figure 3: Caption should state early Eocene plate configuration (nominally XX Ma).

Figures 4 and 9: Dashed lines need to be defined in legend. The x-axes need “latitude”

Figure S2: Needs information above scale bar.

This is a nice contribution. Thanks for amending.

Jerry