Sedimentary noise and sea levels linked to land–ocean water exchange and obliquity forcing

In ancient hothouses lacking ice sheets, the origins of large, million-year (myr)-scale sea-level oscillations remain a mystery, challenging current models of sea-level change. To address this mystery, we develop a sedimentary noise model for sea-level changes that simultaneously estimates geologic time and sea level from astronomically forced marginal marine stratigraphy. The noise model involves two complementary approaches: dynamic noise after orbital tuning (DYNOT) and lag-1 autocorrelation coefficient (ρ1). Noise modeling of Lower Triassic marine slope stratigraphy in South China reveal evidence for global sea-level variations in the Early Triassic hothouse that are anti-phased with continental water storage variations in the Germanic Basin. This supports the hypothesis that long-period (1-2 myr) astronomically forced water mass exchange between land and ocean reservoirs is a missing link for reconciling geological records and models for sea-level change during non-glacial periods.

I recommend the following additional short discussion on the forcing of the Germanic Basin lakelevel: The authors should address the previously postulated connecting straits between the Germanic Basin and the Tethys ocean, like the Schlesian Strait. As it is shown that global oceanlevel rise was in anti-phase with the lake-level it follows, that marine ingressions through these straits cannot have been a significant contribution of water to the lake, as has been previously thought.
This paper stimulates exciting new research questions along the lines of the above thought. Obviously it should not lose conciseness by delving much into these aspects, yet I recommend placing a note on the need for research addressing these questions.
Using the attached annotated ms-file, please correct minor errors in writing and respond to the minor comments. Regarding data availability, it would be preferable if the data of this study could be stored in an online database.

Jens Wendler
Reviewer #3 (Remarks to the Author): Li et al provide a novel method for estimating medium term (1 to 2 Myr) variations in sea level using analysis of time series that are proxies for sediment composition. As explained below I would support publication after minor revision.
Firstly they developed a new method for analysis demonstrated for a Pleistocene to Recent section from ODP Leg 181 off New Zealand and then applied the method to two Chinese natural gamma ray logs from the Triassic. The method simply involves removing the quasi-regular orbitally-driven components of the time series and then find the time-varying ratio of the "non-orbital" variance to the total (untreated) time series variance. As a check they have used the time-varying lag-1 autocorrelation (which is largely independent of the regular components). Their method is novel and, accepting their interpretation that the results indicate 1-2 Myr variations in sea level, potentially of great value.
It should be pointed out that they have only used one Pleistocene case of just 1.4 Myr long to test their data. Much longer sea-level related records from passive continental margins could be used for testing (e.g. the New Jersey margin investigated by the IODP Leg 313 following up earlier ODP studies). On the other hand, this idea deserves investigation so I'm happy to go along with their argument that they have recovered a long-term sea level change off New Zealand in the Pleistocene and that they have correctly interpreted the meaning of the Chinese results too.
Secondly having judged their Chinese results to represent global sea level variations they explain the variations themselves in terms of 1-2 Myr cycles in palaeoclimate as driven by long terms of the orbital obliquity.
Overall the treatment is detailed and convincing, but I did not immediately understand the key points listed above because there are some aspects of the paper that could do with clarifying. The suggestions made here are provided to help improve clarity. They do NOT represent a criticism of the arguments or of the conclusions.
Suggestions for improvement: 1) The title could be much clearer. The author describe the "non-orbital" variance extracted with their main method as "Dynamic sedimentary noise". This is a strange use of the word dynamic and as it is not standard I think most readers will not have a clue what the paper refers to from the title. Similarly the second part of the title refers to "land-ocean water balance dynamics". This also fails to connect the reader with the idea that 1-2 Myr sea level variation in the Triassic was driven by long-term obliquity variations. Why not change the title to something like: "Sedimentary noise and Triassic sea levels linked to long-term obliquity forcing"? 2) The method is said to remove the orbitally forced components of the compositional variability from the time series (built into the term DYNOT). However, strictly it only removes the orbital components between 405 kyr and 18 kyr. The reasoning that Triassic (hothouse) sea level variations over 1-2 Myr are related to long-term oblquity forcing of climate means the authors think there are long-term orbital components in the DYNOT estimates. Therefore it needs to be made much clearer that the DYNOT estimates only eliminate the "short term" orbital components.
To clarify this the authors should change the abstract on line 1 from "high-frequency sea-level" to "high-frequency (1/1-2 Myr) sea-level" [though I personally don't think of 1/1-2 Myr as high frequency]. Similarly on line 32 of the abstract change "astronomically forced water mass exchange" to "long-period (1-2 Myr) astronomically forced water mass exchange".
Other points:

Response to referees (manuscript NCOMMS-17-21491)
All line numbers are original manuscript line numbers.
Response to reviewer #1: Dr. Rob Duller Comment: I like the paper for its rigour and its theme. The major claim of the paper is that, during hothouse conditions, water storage on the continents and water storage in ocean basins is antiphased. This correlation is used to suggest that continental water storage-and-release is responsible for high amplitude, obliquity-driven sea level changes during the Triassic. To get to this point the authors use a 'dynamic sedimentary noise model', which appears to be a novel methodological contribution in its own right. However I must say that I am in no position to evaluate the robustness of the methodology and techniques, but I can say that the analysis appears to be very detailed.
The data support the idea that continental water storage could have a major influence on global sea level change. This appears to be new but the idea is not a completely new one. No doubt these results are of significant interest to many branches of Earth, atmospheric and hydrological sciences, but the broad relevance of the work is not highlighted.

Reply:
The original manuscript had a paragraph on long-term future projections of sea level. We have now moved that paragraph to the end of the paper. The paragraph has also been slightly extended to highlight the broad relevance of the work.
The conclusions are poorly written and fail to focus on the real outcome of the work. This may be because there is a dual focus to the paper: (1) Methodological, and (2) The water storage and release story. Overall the paper sways towards the methodological aspect, which dilutes the water storage aspect (the intended focus of the paper). I realise that you MUST demonstrate the robustness of your model to unequivocally demonstrate the 'antiphase', so it is a difficult one.

Reply:
We needed to demonstrate the robustness of the model before applying it to the Early Triassic. As explained in the introduction, two leading methods (i.e., oxygen isotope and sequence stratigraphy) for sea-level reconstruction in deep time have significant problems. Therefore, this "sedimentary noise" model was developed. Description, discussion and verification of this new model were necessary to lay down a solid foundation for the reconstruction of global sea-level during the Early Triassic.
The paper seems to skip key explanations of what data was collected and how, e.g. the Germanic fluvial and lacustrine dataset. Clearly this dataset is key but I was left wondering "what was actually measured to allow the authors to reconstruct water storage over time?". Yet the authors went into a lot of poorly-explained detail about how their worked'. Im not necessariliy saying that this material should be removed, but rather to focus on the broad scientific issue of 'water storage' and refer the authors to details of the model in the SI. This would free up some space for you to 'beef-up' your results section (evidence) and discussion of the origin continental water storage versus water storage in ocean basins is anti-phased. I would like more clarity of the methods and results of the continental Germanic data collection. The work is 80% convincing, which could be increased to 95% if my comments are answered. The authors do a rigorous job in their data analysis on the marine sections, but to demonstrate the point of the paper (i.e. sea-saw water storage fluctuations) more clarification is need on what data and analysis was done on the Germanic terrestrial/lacustrine section. This was unclear to me.
Reply: In this revision, the hypothesis of aquifer eustasy is now presented in detail. Problems with the aquifer eustasy hypothesis are also discussed. We have rewritten the section about linkages among sequences, lake level, groundwater table, and continental water storage in the Germanic Basin. We have now also moved one key figure on the Germanic sequences from the SI of the original submission to the main paper (new Fig. 5). Previous studies have demonstrated that sequences in a closed terrestrial basin indicate relative changes in lake level and groundwater table; therefore sequences can be used as an proxy for ancient continental aquifer changes. Consequently, our work draws on the widely accepted sequence stratigraphy of the Germanic Basin.
Furthermore, I was not satisfied with the mechanism the authors put forward for continental storage and release over the timescales assumed. What are the mechanics of this process other than 'obliquity forcing'. Can O-isotopes of the terrestrial successions offer a future dataset that can link terrestrial to marine? How long does it take to fill and release water to and from a continental reservoir? Could this timescale be sufficient to explain any anti-phased relationship or amplification? How are these reservoirs filled and emptied?
The paper will certainly influence thinking in the field but requires more careful crafting of arguments and a clearer explanation of: 1) data collection and analysis at Germanic site; 2) the linkage between the Germanic site and marine sections; and 3) the mechanism (and timescales) responsible for storage and release in continental reservoirs.

Reply:
The mechanics of the astronomically forced recharge and discharge of the continental aquifer has been rewritten in this revision. We now present more details for the evidence of obliquity forcing in the South China marine sections, precipitation in the continental Germanic basin, the timescale of the aquifer-eustasy process, and the filling and discharge dynamics that drive continental aquifer variations. As to O-isotopes and groundwater variations in terrestrial basins, Holocene-Quaternary speleothems (e.g., Hulu Cave) have been useful in this regard. Alternatively, "future climate modeling and documentation of groundwater variations in other Early Triassic climate zones will further clarify this hypothesis of anti-phasing between global sea level and continental water reservoir changes".