the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Sep 2022
Simulating marine neodymium isotope distributions using ND v1.0 coupled to the ocean component of the FAMOUS-MOSES1 climate model: sensitivities to reversible scavenging efficiency and benthic source distributions
Abstract. The neodymium (Nd) isotopic composition of seawater is a widely used ocean circulation tracer. However, uncertainty in quantifying the global ocean Nd budget, particularly constraining elusive non-conservative processes, remains a major challenge. A substantial increase in modern seawater Nd measurements from the GEOTRACES programme coupled with recent hypotheses that a seafloor-wide benthic Nd flux to the ocean may govern global Nd isotope distributions (εNd) presents an opportunity to develop a new scheme specifically designed to test these paradigms. Here, we present the implementation of Nd isotopes (143Nd and 144Nd) into the ocean component of the FAMOUS coupled atmosphere-ocean general circulation model (ND v1.0), a tool which can be widely used for simulating complex feedbacks between different Earth system processes on decadal to multi-millennial timescales.
Using an equilibrium pre-industrial simulation tuned to represent the largescale Atlantic Ocean circulation, we perform a series of sensitivity tests evaluating the new Nd isotope scheme. We investigate how Nd source/sink and cycling parameters govern global marine εNd distributions, and provide an updated compilation of 6,048 Nd concentration and 3,278 εNd measurements to assess model performance. Our findings support the notions that reversible scavenging is a key process for enhancing the Atlantic-Pacific basinal εNd gradient, and is capable of driving the observed increase in Nd concentration along the global circulation pathway. A benthic flux represents a major source of Nd to the deep ocean. However, model-data disparities in the North Pacific highlight that the source of εNd from seafloor sediment is too unradiogenic in our model with a constant benthic flux. Additionally, model-data mismatch in the northern North Atlantic suggests a missing source of Nd that is much more unradiogenic than the bulk sediment, alluding to the possibility of preferential contributions from ‘reactive’ detrital sediments under a benthic flux driven model of marine Nd cycling.
The new Nd isotope scheme forms an excellent tool for exploring global marine Nd cycling and the interplay between climatic and oceanographic conditions under both modern and palaeoceanographic contexts.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(9213 KB)
-
Supplement
(11064 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(9213 KB) - Metadata XML
-
Supplement
(11064 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-606', Benoit Pasquier, 24 Oct 2022
-
RC3: 'Reply on RC1', Benoit Pasquier, 01 Nov 2022
I just stumbled upon this GMD highlight paper on MAE vs RMSE that the authors may find useful:
Hodson, T. O.: Root-mean-square error (RMSE) or mean absolute error (MAE): when to use them or not, Geosci. Model Dev., 15, 5481–5487, https://doi.org/10.5194/gmd-15-5481-2022, 2022.My understanding from that paper is that MAE should be used for [Nd] (exponentially distributed) and RMSE should be used for εNd (normally distributed). I would still recommend reporting both MAE and RMSE however, to facilitate comparisons with past and future models, and also because the distribution assumptions are not exactly satisfied with the GEOTRACES IDP21 data:
Citation: https://doi.org/10.5194/egusphere-2022-606-RC3 - AC1: 'Reply on RC1', Suzanne Robinson, 10 Jan 2023
-
RC3: 'Reply on RC1', Benoit Pasquier, 01 Nov 2022
-
CC1: 'Comment on egusphere-2022-606', Tristan Vadsaria, 28 Oct 2022
The following comment of the preprint research article entitled “Simulating marine neodymium isotope distributions using ND v1.0 coupled to the ocean component of the FAMOUS-MOSES1 climate model: sensitivities to reversible scavenging efficiency and benthic source distributions” by Robinson et al. has been motivated by the next implementation of the Neodymium oceanic cycle in the iLOVECLIM model.
This article describes the modelling implementation of the Nd oceanic cycle in the fast global climate model (GCM) FAMOUS. After selecting the most appropriate reference simulation in terms of oceanic variable (temperature, salinity, AMOC pattern and strength), the authors scrutinized in detail the result of the implementation based on two decisive parameters. These parameters, being the scavenging coefficient and the Nd flux from the sediment, are known to be still poorly constrained by observational studies. The modelling work provided by Robinson et al., in line with the recent Nd modelling studies performed with GCM, helps to provide more insight on these processes with the “own” physics and parameterization of the FAMOUS model.
I found that the paper is overall very well written with good descriptions of the results which is very important to understand the relative importance of both the scavenging coefficient and the Nd flux from the sediment on the global oceanic distribution of Nd and εNd. I particularly like the effort put into reaching the most “appropriate” simulation regarding ocean physics in order to reduce the associated bias for the Nd and εNd interpretation. I also join the authors about the need of a shared protocol for an Nd modelling intercomparison project.
Below are some (minor) comments to 1) have some clarifications on some points of the manuscript, 2) suggest some modifications if it is feasible in time and in resources for the authors.
Because my field of expertise is mostly climate modelling, my comments will not focus on the geochemistry part of the paper.
Best regards,
Tristan Vadsaria
--------------------------------------------------------------------------------------------------------------------------------------
River forcings
For this point I am not sure at 100% that my comment is pertinent, sorry if I misunderstood the manuscript in this regard. As far as I understood from other Nd modelers, the river forcings in terms of (dissolved) Nd concentration and εNd come indeed from Goldstein and Jacobsen (1987) but is often updated with recent observations such as coming from Blanchet (2019). As I assumed that the authors are using the most updated data to force the model, would it not be better to indicate this in the manuscript (especially for the caption of figure 5)?
What about the seafloor Nd concentration forcing?
This comment is not really a suggestion nor a critic since I think it’s beyond the scope of the study but rather an open question, I think it would need more discussion from and for the Nd modelling community.
For the Nd sediment source, I understand that this study is in line with the previous scheme initiated by Rempfer et al. (2011) that “[…]do not make any assumption regarding the nature of the Nd boundary source” and “[…]do not assume spatial variation...”, later followed by Gu et al. (2019) with the same depth limitation and also by Pöppelmeier et al. (2020) but without the depth limitation. This approach is indeed convenient for tuning “fsed” which is today still not very well known. However, how far is it reasonable to apply the same scheme while considering the whole seafloor, i.e., to not consider the spatial Nd concentration of the seafloor into the calculation of the sediment source (question also valid for Pöppelmeier et al., 2020)?
While the Nd sediment source was “confined” to the continental margin in Rempfer et al. (2011) it seemed more “reasonable” to make their assumptions especially regarding the horizontal resolution of the Bern3D model. However, now, without the depth limitation, I would guess that the spatial variations of the Nd concentration of the seafloor would have an impact on the deep and bottom Nd dissolved seawater distribution, don’t the authors think so?
In Arsouze et al. (2009), “Fsed is then determined for both 143Nd and 144Nd isotopes by multiplying this sediment flux to the concentration along the margin” and they fixed the sediment flux to only one value, without the ability to make a lot of simulation to tune this flux because of the resolution and the time consumption of their model. My question is: would it be possible to have an intermediate approach between Rempfer et al. (2011) and Arsouze et al. (2009), e. g., tuning the flux while applying widely the Nd seafloor concentration (obtained from the recent data of Robinson et al. 2021 for instance)?
Small edit: As I kept thinking about the previous point, I realized that Nd seafloor data was indeed scarcer than εNd and that extrapolating a wide Nd seafloor map would be less relevant especially in the Pacific Ocean (cf attached figure in supplement using Nd data provided by Robinson et al., 2021). Anyway, would it be possible to imagine a regional seafloor Nd (or basin-scale) signal such as used for the dust εNd?
Less sensitive response of Pacific εNd to reversible scavenging efficiency - deep seafloor wide sediment source
Concerning that point, the authors said that their results “[...] contrast with results from Rempfer et al. (2011) ...” (line 751) and “attribute this difference primarily to the spatial variation in the sediment Nd flux” (lines 752-753) due to the different modelling scheme used for the Nd sediment source.
Even though that explanation seems obvious, I would like to see a simulation output to confirm what the authors are suggesting (i.e., simulations governed by deep seafloor wide sediment source) with the “own” FAMOUS ocean dynamics: A set of simulations similar to the scavenging coefficient sensitivity simulations (first part of the result) but with the “initial” depth limitation of 3000km would be the best (but maybe too much for that purpose).
What about retaining the best simulation (“EXPT_RS4” as far as I understood) for the scavenging coefficient and run a parallel simulation with the depth limitation of 3000km for the sediment source? I think that putting only the result (a couple of 2d maps) of this new simulation in the supplementary compared with “EXPT_RS4”, while keeping the original text in the main paper would be enough. Taking this comment into consideration is obviously up to the authors regarding its feasibility.
Wrong residence time value?
As explained in the main manuscript, the residence time is equal to the Nd inventory divided by the total Nd flux (line 598). Following table 4 (and 5), it corresponds to (column 5*1000/column 4). If I apply this, I found the same results as the authors for all the experiments except for “EXPT_RS1” which should be (10.6*1000/5.27) = 2011 years, am I right? Anyway, it does not change anything to the results description and the conclusion since it is still the simulation with the highest residence time.
Very minor suggestion (1)
“The total global flux of river sourced dissolved Nd to seawater (friver) is 4.4 ×108g(Nd) yr-1“: this is the value that comes from the simulated runoff in FAMOUS combined with the prescribed Nd river concentration, isn’t it (as confirmed by looking at Table1)? In that case, I would suggest adding “in the model” to the sentence to enhance clarity.
Very minor suggestion (2)
In my opinion, the use of “pronounced” in line 642, 894 and 992 to describe the behavior of the vertical gradient of [Nd]d is not very descriptive, but I may be wrong since I am not a native English speaker.
Very minor suggestion (3)
Figures 11, 12, 15 and 16 are overall very nice but I would suggest, regarding the depth profiles, for more clarity and visibility, to not match the color of the observational data with the color bar of the central 2d map. Maybe rather use a unique color, also with the dots connected together, to reduce the confusion with the color of the simulated profiles (but at this stage it’s really a matter of taste).
Very minor suggestion (4)
Why not merge Table 2 and Table S3? I think that Table S3 would be very informative in the main text.
- AC2: 'Reply on CC1', Suzanne Robinson, 10 Jan 2023
-
RC2: 'Comment on egusphere-2022-606', Catherine Jeandel, 01 Nov 2022
The manuscript egusphere#2022-606 proposes the implementation of the oceanic cycles of the Nd isotopes (143 and 144) in the FAMOUS-MOSES1 climate model. There are interesting novelties in this work as 1) the use of the detailed epsNd map established by the author and comprising the bottom sediment signatures (Robinson et al, 2021); 2) exhaustive sensitivity test of two main parameters driving Nd and epsNd cycles: the reversible scavenging and the external flux, mostly the sediment one here. Actually, this represents a tremendous work; the manuscript is well written (although sometimes a bit wordy) and illustrated. It certainly deserves publication in egusphere. Nevertheless, I have some comments that I submit here to the authors.
- Sediment flux vs Boundary Exchange (BE) processes
I think there is a misunderstanding or a confusion between these two terms that needs to be clarified.
At several places in the manuscript, it is written that sedimentary flux is encompassing Boundary Exchange (e.g lines 250-255, around 345 but also 998-1000 and at other places highlighted in the manuscript) while to me, it’s the opposite (ie BE is encompassing sedimentary flux, down to 3000 m -which is already deep!) in our preceding works).
Let’s consider lines 250 and after
“Seafloor sedimentary fluxes, an umbrella term that refers to a multitude of processes
encompassing boundary exchange (Lacan and Jeandel, 2005), submarine groundwater discharge (Johannesson and Burdige, 2007), and a benthic flux released from pore waters (Abbott et al., 2015a), are simulated via a combination of a sedimentary source applied across sediment-water interfaces together with a separate sink occurring via particle scavenging”
I suggest to write this paragraph differently (as well the other places where it’s a bit confusing, identified in my direct comments in the pdf). Indeed, when we proposed the “BE Concept” with F. Lacan (EPSL, 2005), we did not pretend to describe any specific processes that occur at the land-ocean interface and more specifically along the margins because we could not differentiate them. Later, I listed the potential processes that could explain the “BE (Jeandel, 2016). In other words, “BE” broadly comprises all the processes that could release Nd from the solid to the liquid but also those which would scavenge it, more or less at the same time and in the same area (note that this comprises reversible scavenging too!). More recently, one of the conclusions of the PAGES-GEOTRACES workshop (2018) pushed by Martin Franck was to "kill the BE", in other words to disentangle these processes, among them the seafloor sedimentary fluxes (either through early diagenesis or dissolution of resuspended sediments), low temperature hydrothermalism, SGD, benthic fluxes etc...
The point here is that the authors removed the depth limitation of 3 km which was forcing the model to consider sedimentary fluxes along the margins only. But the sedimentary flux they consider are occurring everywhere including along the slopes. This does not mean that the Sed Flux (a specific mechanism) encompasses the BE (a broader concept). This just means that this flux is extended to the whole ocean in the proposed work.
.
- “Top down” versus “Bottom up” processes (issue linked to what is discussed in 1)
I’d cautiously use this opposition which was never clear to me (and I had long debates with B. Haley on this issue). By the way, the earliest Nd budgets proposed 2 sources: dissolved rivers and hydrothermal (see the historical works of Goldstein, O’Nions etc…). To my knowledge, hydrothermal is not “top down”. Consider now the most recent budgets: most of them invoke “Boundary exchange” which includes processes that occur in the deep waters, down to 3000 m depth (see above), in other words they include “bottom processes”. Thus, although I agree with what is written line 103 and after (reported below), it seems to me that this was not “new” because the benthic flux is occurring at any place where there is a contact between sediment and water, in other words everywhere from the beach to the deepest parts of the ocean. Again, there is a confusion between the processes and the location. Thus, I strongly suggest to the authors to be cautious here. What was “new” is that it could concern sediments below 3 km depth.
“Recent pore fluid concentration profiles measured on the Oregon margin in the Pacific Ocean indicate that there may be a benthic flux of Nd from sedimentary pore fluids, presenting a new, potentially major seafloor-wide source of Nd to seawater (Abbott et al., 2015b, a).”
- The choice of the Ndp/Ndd ratio to conduct the Fsed sensitivity test.
I did not understand why the authors did not kept the value of 0.004 instead of that of 0.003 to do these sensitivity tests. Indeed, as underlined in Table 5, the value of 0.003 leas to residence times larger than 1000 y, leading to more moderate range of 40 years difference in the fsed simulations Nd
I did not see a clear justification of this choice in section 3.1 and would be keen to see the same sensitivity tests but with the Ndp/Ndd ratio of 0.004, which was the most consistent with the data.
- The discussion on the reasons leading to epsNd modelled profiles that do not fit the data is often too shy and not clear enough. Perhaps the sedimentary flux is too strong? Or the choice to attribute a constant flux for the deepest (bottom) and shallowest sediment (margin) is not appropriate? This is well exemplified by Figure 17 and the discussion lines 930-950. What would happen if the Fsed would allow differentiating the strength of the SedFlux deposited on the margin (fresh deposits from rivers, easy to remobilize) vs that of the bottom (too strong, “counterproductive” as it is written line 942)?
- Minor comments: they are highlighted in the attached pdf. I also identified some unit issues and rare typos but they are already listed by B. Pasquier ( who I thank for the exhaustive list!).
As a whole, I’d also suggest to the authors to shorten the manuscript by 10%-15% if possible.
- AC3: 'Reply on RC2', Suzanne Robinson, 10 Jan 2023
-
RC4: 'Comment on egusphere-2022-606', Ed Hathorne, 02 Nov 2022
This paper describes the results of a model of the marine Nd cycle implemented in the ocean part of the fast climate model derived from the Hadley centre GCM. The use of such a model for simulating the Nd isotopes of seawater is a useful development as the fast run times allow more experiments to be conducted, although care must be taken that the ocean circulation is resolved correctly as this is what we hope to trace with Nd isotopes. From the standpoint of a geochemical oceanographer this paper is very interesting because it directly tests the hypothesis that the distribution of Nd isotopes in seawater is mostly controlled by a flux from marine sediments, sometimes known as the “bottom flux hypothesis”. Along with other models of the marine Nd isotope cycle published very recently, this work affords many insights into the processes that are likely, and unlikely, to control the distribution of Nd isotopes in seawater. The discussions paper is rather long but with some editing, clarification in places and discussion of the other very recently published works, I would gladly recommend this for publication.
This could be an important contribution as the authors have the most up to date data compilation available, but this should be utilised throughout. For example, in Figure 9 the global marine Nd inventory of 4.2 x10^12 g from Tachikawa et al. (2003) is used to asses which reversable scavenging scenarios are realistic. Although this ground breaking study is clearly still relevant, many samples have been taken and measured in the intervening decades as shown in Figure 8. Would it not make sense to estimate the marine Nd inventory with all the available data? Perhaps it will make little difference but in 2003 there were very few data available for the entire Southern Ocean and North Pacific (Table 1 in Tachikawa et al., 2003). Using these realistic reversable scavenging values (can it please be clarified if this is also 100% released like in Tachikawa et al., 2003?) a very simple universal sediment flux is tested. Although it is very interesting that this fails to simulate the tails of the observed data, both radiogenic in the Pacific and unradiogenic in the N Atlantic, this is not proof that the bottom flux hypothesis is wrong. Assuming a constant flux over the entire ocean bottom is clearly unrealistic and this point should be clearly stated. With rare earth element concentrations >2 times that of shale, the red clay sediments covering large parts of the abyssal Pacific (e.g. Kato et al., 2011, Nature Geoscience 4) are most likely a sink for Nd. Here and also in areas influenced by hydrothermal particles (German et al., 1990, Nature 345, 516-518) the bottom flux is likely to be negative. The fact that Pasquier et al. (2022) use a parameterisation which increases the sediment flux at both radiogenic and unradiogenic extremes of sediment composition should be mentioned in the context of a constant bottom flux not simulating the highest and lowest seawater values.
Detailed comments and suggestions are provided in an annotated PDF. I still hope publishers will provide a tool for extracting comments from PDFs.
- AC4: 'Reply on RC4', Suzanne Robinson, 10 Jan 2023
-
RC5: 'Comment on egusphere-2022-606', Torben Stichel, 04 Nov 2022
This discussion paper by Robinson et al. discusses the incorporation of Nd (isotopes and elemental concentration) into FAMOUS GCM’s ocean component to better understand the GLOBAL marine Nd cycle. I really like the paper and it addresses the current debate on the direction of control in Nd (and REE) distribution. Acknowledging the paper has set its focus on the sensitivity of the scavenging efficiency and benthic fluxes, it leans towards comparing itself with previous modelling studies (Rempfer et al., Siddall et al., Pöppelmeier et al). The overall reality test is done by comparing their results with a global data base. They conclude that reversible scavenging is important for the Atlantic-Pacific gradient in eNd, but again the modelled Pacific Ocean does not match the observed data there. They admit that a global constant sediment flux in the model runs could be the issue here as the Pacific Ocean supposedly provides more reactive material (young, mafic rocks) than the Atlantic, which would support different sediment fluxes within these basins.
As a non-modeller, I would like to avoid evaluating the technical parts of this paper, but I would like to highlight some important aspects on the biogeochemical cycles of Nd. Overall, I found the paper very well written, a bit wordy though. I do admit, I am bit surprised that recent particle studies (e.g. Lagarde et al. 2020, Paffrath et al., Stichel et al. 2020) were very marginally used in this paper. From those studies, we now have information on eNd in particles, different mineral fractions, different kDs etc. from the same locations as the dissolved fraction. We also know that pNd/dNd unfortunately is not uniform in the ocean. Also, in the last paragraph of the discussion (lines 927 ff.), where the authors compared their model outcome with observations in the North Atlantic – the area where the aforementioned papers have their study area – the composition of particles would help to assess the NADW composition (e.g. fig.6 in Stichel et al. 2020). In the search for end member composition, the authors might want to consider the very dynamic particle composition in that area (pointing towards different sources) and not necessarily from bulk sediments. If I am not mistaken, in those papers the observational pNd/dNd are often one order of magnitude higher in the North Atlantic, compared to the global average assumed in the modelling studies (0.001 to 0.006). Or do I miss something here? Of course, it is reasonable to assume that Pacific pNd are very much lower than in the North Atlantic and therefore the pNd/dNd is very much skewed towards lower values. This discrepancy should at least be mentioned and justified.
I would like to point out another rather minor issue, which is the database used. I acknowledge that with the now very impressive global Nd data sets available, it is very convenient to cite the GEOTRACES IDP. However, the authors want to double-check whether data actually IS in the data product. For instance, large parts of the eNd from GA03 (or US-GEOTRACES North Atlantic Zonal Section) are not included in the IDP 2021 but was used a citation here. For those data sets you can find the correct citations here:
http://data.bco-dmo.org/jg/info/BCO/GEOTRACES/NorthAtlanticTransect/Nd_GT10%7Bdir=data.bco-dmo.org/jg/dir/BCO/GEOTRACES/NorthAtlanticTransect/,data=data.bco-dmo.org:80/jg/serv/BCO/GEOTRACES/NorthAtlanticTransect/Nd_GT10_v8_joined.html0%7D?. I apologise for this rather shameless self-advertisement…
I fully support the publication of this paper eventually. It is an important work and will be key for a better understanding of the Nd cycle as it is one of (if not) the most complete modelling papers for marine Nd isotopes and concentrations. The supplement’s profound eNd/YREE data set is also great! Thanks for providing this with your publication.
Citation: https://doi.org/10.5194/egusphere-2022-606-RC5 - AC5: 'Reply on RC5', Suzanne Robinson, 10 Jan 2023
- AC6: 'Comment on egusphere-2022-606', Suzanne Robinson, 10 Jan 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-606', Benoit Pasquier, 24 Oct 2022
-
RC3: 'Reply on RC1', Benoit Pasquier, 01 Nov 2022
I just stumbled upon this GMD highlight paper on MAE vs RMSE that the authors may find useful:
Hodson, T. O.: Root-mean-square error (RMSE) or mean absolute error (MAE): when to use them or not, Geosci. Model Dev., 15, 5481–5487, https://doi.org/10.5194/gmd-15-5481-2022, 2022.My understanding from that paper is that MAE should be used for [Nd] (exponentially distributed) and RMSE should be used for εNd (normally distributed). I would still recommend reporting both MAE and RMSE however, to facilitate comparisons with past and future models, and also because the distribution assumptions are not exactly satisfied with the GEOTRACES IDP21 data:Citation: https://doi.org/10.5194/egusphere-2022-606-RC3 - AC1: 'Reply on RC1', Suzanne Robinson, 10 Jan 2023
-
RC3: 'Reply on RC1', Benoit Pasquier, 01 Nov 2022
-
CC1: 'Comment on egusphere-2022-606', Tristan Vadsaria, 28 Oct 2022
The following comment of the preprint research article entitled “Simulating marine neodymium isotope distributions using ND v1.0 coupled to the ocean component of the FAMOUS-MOSES1 climate model: sensitivities to reversible scavenging efficiency and benthic source distributions” by Robinson et al. has been motivated by the next implementation of the Neodymium oceanic cycle in the iLOVECLIM model.
This article describes the modelling implementation of the Nd oceanic cycle in the fast global climate model (GCM) FAMOUS. After selecting the most appropriate reference simulation in terms of oceanic variable (temperature, salinity, AMOC pattern and strength), the authors scrutinized in detail the result of the implementation based on two decisive parameters. These parameters, being the scavenging coefficient and the Nd flux from the sediment, are known to be still poorly constrained by observational studies. The modelling work provided by Robinson et al., in line with the recent Nd modelling studies performed with GCM, helps to provide more insight on these processes with the “own” physics and parameterization of the FAMOUS model.
I found that the paper is overall very well written with good descriptions of the results which is very important to understand the relative importance of both the scavenging coefficient and the Nd flux from the sediment on the global oceanic distribution of Nd and εNd. I particularly like the effort put into reaching the most “appropriate” simulation regarding ocean physics in order to reduce the associated bias for the Nd and εNd interpretation. I also join the authors about the need of a shared protocol for an Nd modelling intercomparison project.
Below are some (minor) comments to 1) have some clarifications on some points of the manuscript, 2) suggest some modifications if it is feasible in time and in resources for the authors.
Because my field of expertise is mostly climate modelling, my comments will not focus on the geochemistry part of the paper.
Best regards,
Tristan Vadsaria
--------------------------------------------------------------------------------------------------------------------------------------
River forcings
For this point I am not sure at 100% that my comment is pertinent, sorry if I misunderstood the manuscript in this regard. As far as I understood from other Nd modelers, the river forcings in terms of (dissolved) Nd concentration and εNd come indeed from Goldstein and Jacobsen (1987) but is often updated with recent observations such as coming from Blanchet (2019). As I assumed that the authors are using the most updated data to force the model, would it not be better to indicate this in the manuscript (especially for the caption of figure 5)?
What about the seafloor Nd concentration forcing?
This comment is not really a suggestion nor a critic since I think it’s beyond the scope of the study but rather an open question, I think it would need more discussion from and for the Nd modelling community.
For the Nd sediment source, I understand that this study is in line with the previous scheme initiated by Rempfer et al. (2011) that “[…]do not make any assumption regarding the nature of the Nd boundary source” and “[…]do not assume spatial variation...”, later followed by Gu et al. (2019) with the same depth limitation and also by Pöppelmeier et al. (2020) but without the depth limitation. This approach is indeed convenient for tuning “fsed” which is today still not very well known. However, how far is it reasonable to apply the same scheme while considering the whole seafloor, i.e., to not consider the spatial Nd concentration of the seafloor into the calculation of the sediment source (question also valid for Pöppelmeier et al., 2020)?
While the Nd sediment source was “confined” to the continental margin in Rempfer et al. (2011) it seemed more “reasonable” to make their assumptions especially regarding the horizontal resolution of the Bern3D model. However, now, without the depth limitation, I would guess that the spatial variations of the Nd concentration of the seafloor would have an impact on the deep and bottom Nd dissolved seawater distribution, don’t the authors think so?
In Arsouze et al. (2009), “Fsed is then determined for both 143Nd and 144Nd isotopes by multiplying this sediment flux to the concentration along the margin” and they fixed the sediment flux to only one value, without the ability to make a lot of simulation to tune this flux because of the resolution and the time consumption of their model. My question is: would it be possible to have an intermediate approach between Rempfer et al. (2011) and Arsouze et al. (2009), e. g., tuning the flux while applying widely the Nd seafloor concentration (obtained from the recent data of Robinson et al. 2021 for instance)?
Small edit: As I kept thinking about the previous point, I realized that Nd seafloor data was indeed scarcer than εNd and that extrapolating a wide Nd seafloor map would be less relevant especially in the Pacific Ocean (cf attached figure in supplement using Nd data provided by Robinson et al., 2021). Anyway, would it be possible to imagine a regional seafloor Nd (or basin-scale) signal such as used for the dust εNd?
Less sensitive response of Pacific εNd to reversible scavenging efficiency - deep seafloor wide sediment source
Concerning that point, the authors said that their results “[...] contrast with results from Rempfer et al. (2011) ...” (line 751) and “attribute this difference primarily to the spatial variation in the sediment Nd flux” (lines 752-753) due to the different modelling scheme used for the Nd sediment source.
Even though that explanation seems obvious, I would like to see a simulation output to confirm what the authors are suggesting (i.e., simulations governed by deep seafloor wide sediment source) with the “own” FAMOUS ocean dynamics: A set of simulations similar to the scavenging coefficient sensitivity simulations (first part of the result) but with the “initial” depth limitation of 3000km would be the best (but maybe too much for that purpose).
What about retaining the best simulation (“EXPT_RS4” as far as I understood) for the scavenging coefficient and run a parallel simulation with the depth limitation of 3000km for the sediment source? I think that putting only the result (a couple of 2d maps) of this new simulation in the supplementary compared with “EXPT_RS4”, while keeping the original text in the main paper would be enough. Taking this comment into consideration is obviously up to the authors regarding its feasibility.
Wrong residence time value?
As explained in the main manuscript, the residence time is equal to the Nd inventory divided by the total Nd flux (line 598). Following table 4 (and 5), it corresponds to (column 5*1000/column 4). If I apply this, I found the same results as the authors for all the experiments except for “EXPT_RS1” which should be (10.6*1000/5.27) = 2011 years, am I right? Anyway, it does not change anything to the results description and the conclusion since it is still the simulation with the highest residence time.
Very minor suggestion (1)
“The total global flux of river sourced dissolved Nd to seawater (friver) is 4.4 ×108g(Nd) yr-1“: this is the value that comes from the simulated runoff in FAMOUS combined with the prescribed Nd river concentration, isn’t it (as confirmed by looking at Table1)? In that case, I would suggest adding “in the model” to the sentence to enhance clarity.
Very minor suggestion (2)
In my opinion, the use of “pronounced” in line 642, 894 and 992 to describe the behavior of the vertical gradient of [Nd]d is not very descriptive, but I may be wrong since I am not a native English speaker.
Very minor suggestion (3)
Figures 11, 12, 15 and 16 are overall very nice but I would suggest, regarding the depth profiles, for more clarity and visibility, to not match the color of the observational data with the color bar of the central 2d map. Maybe rather use a unique color, also with the dots connected together, to reduce the confusion with the color of the simulated profiles (but at this stage it’s really a matter of taste).
Very minor suggestion (4)
Why not merge Table 2 and Table S3? I think that Table S3 would be very informative in the main text.
- AC2: 'Reply on CC1', Suzanne Robinson, 10 Jan 2023
-
RC2: 'Comment on egusphere-2022-606', Catherine Jeandel, 01 Nov 2022
The manuscript egusphere#2022-606 proposes the implementation of the oceanic cycles of the Nd isotopes (143 and 144) in the FAMOUS-MOSES1 climate model. There are interesting novelties in this work as 1) the use of the detailed epsNd map established by the author and comprising the bottom sediment signatures (Robinson et al, 2021); 2) exhaustive sensitivity test of two main parameters driving Nd and epsNd cycles: the reversible scavenging and the external flux, mostly the sediment one here. Actually, this represents a tremendous work; the manuscript is well written (although sometimes a bit wordy) and illustrated. It certainly deserves publication in egusphere. Nevertheless, I have some comments that I submit here to the authors.
- Sediment flux vs Boundary Exchange (BE) processes
I think there is a misunderstanding or a confusion between these two terms that needs to be clarified.
At several places in the manuscript, it is written that sedimentary flux is encompassing Boundary Exchange (e.g lines 250-255, around 345 but also 998-1000 and at other places highlighted in the manuscript) while to me, it’s the opposite (ie BE is encompassing sedimentary flux, down to 3000 m -which is already deep!) in our preceding works).
Let’s consider lines 250 and after
“Seafloor sedimentary fluxes, an umbrella term that refers to a multitude of processes
encompassing boundary exchange (Lacan and Jeandel, 2005), submarine groundwater discharge (Johannesson and Burdige, 2007), and a benthic flux released from pore waters (Abbott et al., 2015a), are simulated via a combination of a sedimentary source applied across sediment-water interfaces together with a separate sink occurring via particle scavenging”
I suggest to write this paragraph differently (as well the other places where it’s a bit confusing, identified in my direct comments in the pdf). Indeed, when we proposed the “BE Concept” with F. Lacan (EPSL, 2005), we did not pretend to describe any specific processes that occur at the land-ocean interface and more specifically along the margins because we could not differentiate them. Later, I listed the potential processes that could explain the “BE (Jeandel, 2016). In other words, “BE” broadly comprises all the processes that could release Nd from the solid to the liquid but also those which would scavenge it, more or less at the same time and in the same area (note that this comprises reversible scavenging too!). More recently, one of the conclusions of the PAGES-GEOTRACES workshop (2018) pushed by Martin Franck was to "kill the BE", in other words to disentangle these processes, among them the seafloor sedimentary fluxes (either through early diagenesis or dissolution of resuspended sediments), low temperature hydrothermalism, SGD, benthic fluxes etc...
The point here is that the authors removed the depth limitation of 3 km which was forcing the model to consider sedimentary fluxes along the margins only. But the sedimentary flux they consider are occurring everywhere including along the slopes. This does not mean that the Sed Flux (a specific mechanism) encompasses the BE (a broader concept). This just means that this flux is extended to the whole ocean in the proposed work.
.
- “Top down” versus “Bottom up” processes (issue linked to what is discussed in 1)
I’d cautiously use this opposition which was never clear to me (and I had long debates with B. Haley on this issue). By the way, the earliest Nd budgets proposed 2 sources: dissolved rivers and hydrothermal (see the historical works of Goldstein, O’Nions etc…). To my knowledge, hydrothermal is not “top down”. Consider now the most recent budgets: most of them invoke “Boundary exchange” which includes processes that occur in the deep waters, down to 3000 m depth (see above), in other words they include “bottom processes”. Thus, although I agree with what is written line 103 and after (reported below), it seems to me that this was not “new” because the benthic flux is occurring at any place where there is a contact between sediment and water, in other words everywhere from the beach to the deepest parts of the ocean. Again, there is a confusion between the processes and the location. Thus, I strongly suggest to the authors to be cautious here. What was “new” is that it could concern sediments below 3 km depth.
“Recent pore fluid concentration profiles measured on the Oregon margin in the Pacific Ocean indicate that there may be a benthic flux of Nd from sedimentary pore fluids, presenting a new, potentially major seafloor-wide source of Nd to seawater (Abbott et al., 2015b, a).”
- The choice of the Ndp/Ndd ratio to conduct the Fsed sensitivity test.
I did not understand why the authors did not kept the value of 0.004 instead of that of 0.003 to do these sensitivity tests. Indeed, as underlined in Table 5, the value of 0.003 leas to residence times larger than 1000 y, leading to more moderate range of 40 years difference in the fsed simulations Nd
I did not see a clear justification of this choice in section 3.1 and would be keen to see the same sensitivity tests but with the Ndp/Ndd ratio of 0.004, which was the most consistent with the data.
- The discussion on the reasons leading to epsNd modelled profiles that do not fit the data is often too shy and not clear enough. Perhaps the sedimentary flux is too strong? Or the choice to attribute a constant flux for the deepest (bottom) and shallowest sediment (margin) is not appropriate? This is well exemplified by Figure 17 and the discussion lines 930-950. What would happen if the Fsed would allow differentiating the strength of the SedFlux deposited on the margin (fresh deposits from rivers, easy to remobilize) vs that of the bottom (too strong, “counterproductive” as it is written line 942)?
- Minor comments: they are highlighted in the attached pdf. I also identified some unit issues and rare typos but they are already listed by B. Pasquier ( who I thank for the exhaustive list!).
As a whole, I’d also suggest to the authors to shorten the manuscript by 10%-15% if possible.
- AC3: 'Reply on RC2', Suzanne Robinson, 10 Jan 2023
-
RC4: 'Comment on egusphere-2022-606', Ed Hathorne, 02 Nov 2022
This paper describes the results of a model of the marine Nd cycle implemented in the ocean part of the fast climate model derived from the Hadley centre GCM. The use of such a model for simulating the Nd isotopes of seawater is a useful development as the fast run times allow more experiments to be conducted, although care must be taken that the ocean circulation is resolved correctly as this is what we hope to trace with Nd isotopes. From the standpoint of a geochemical oceanographer this paper is very interesting because it directly tests the hypothesis that the distribution of Nd isotopes in seawater is mostly controlled by a flux from marine sediments, sometimes known as the “bottom flux hypothesis”. Along with other models of the marine Nd isotope cycle published very recently, this work affords many insights into the processes that are likely, and unlikely, to control the distribution of Nd isotopes in seawater. The discussions paper is rather long but with some editing, clarification in places and discussion of the other very recently published works, I would gladly recommend this for publication.
This could be an important contribution as the authors have the most up to date data compilation available, but this should be utilised throughout. For example, in Figure 9 the global marine Nd inventory of 4.2 x10^12 g from Tachikawa et al. (2003) is used to asses which reversable scavenging scenarios are realistic. Although this ground breaking study is clearly still relevant, many samples have been taken and measured in the intervening decades as shown in Figure 8. Would it not make sense to estimate the marine Nd inventory with all the available data? Perhaps it will make little difference but in 2003 there were very few data available for the entire Southern Ocean and North Pacific (Table 1 in Tachikawa et al., 2003). Using these realistic reversable scavenging values (can it please be clarified if this is also 100% released like in Tachikawa et al., 2003?) a very simple universal sediment flux is tested. Although it is very interesting that this fails to simulate the tails of the observed data, both radiogenic in the Pacific and unradiogenic in the N Atlantic, this is not proof that the bottom flux hypothesis is wrong. Assuming a constant flux over the entire ocean bottom is clearly unrealistic and this point should be clearly stated. With rare earth element concentrations >2 times that of shale, the red clay sediments covering large parts of the abyssal Pacific (e.g. Kato et al., 2011, Nature Geoscience 4) are most likely a sink for Nd. Here and also in areas influenced by hydrothermal particles (German et al., 1990, Nature 345, 516-518) the bottom flux is likely to be negative. The fact that Pasquier et al. (2022) use a parameterisation which increases the sediment flux at both radiogenic and unradiogenic extremes of sediment composition should be mentioned in the context of a constant bottom flux not simulating the highest and lowest seawater values.
Detailed comments and suggestions are provided in an annotated PDF. I still hope publishers will provide a tool for extracting comments from PDFs.
- AC4: 'Reply on RC4', Suzanne Robinson, 10 Jan 2023
-
RC5: 'Comment on egusphere-2022-606', Torben Stichel, 04 Nov 2022
This discussion paper by Robinson et al. discusses the incorporation of Nd (isotopes and elemental concentration) into FAMOUS GCM’s ocean component to better understand the GLOBAL marine Nd cycle. I really like the paper and it addresses the current debate on the direction of control in Nd (and REE) distribution. Acknowledging the paper has set its focus on the sensitivity of the scavenging efficiency and benthic fluxes, it leans towards comparing itself with previous modelling studies (Rempfer et al., Siddall et al., Pöppelmeier et al). The overall reality test is done by comparing their results with a global data base. They conclude that reversible scavenging is important for the Atlantic-Pacific gradient in eNd, but again the modelled Pacific Ocean does not match the observed data there. They admit that a global constant sediment flux in the model runs could be the issue here as the Pacific Ocean supposedly provides more reactive material (young, mafic rocks) than the Atlantic, which would support different sediment fluxes within these basins.
As a non-modeller, I would like to avoid evaluating the technical parts of this paper, but I would like to highlight some important aspects on the biogeochemical cycles of Nd. Overall, I found the paper very well written, a bit wordy though. I do admit, I am bit surprised that recent particle studies (e.g. Lagarde et al. 2020, Paffrath et al., Stichel et al. 2020) were very marginally used in this paper. From those studies, we now have information on eNd in particles, different mineral fractions, different kDs etc. from the same locations as the dissolved fraction. We also know that pNd/dNd unfortunately is not uniform in the ocean. Also, in the last paragraph of the discussion (lines 927 ff.), where the authors compared their model outcome with observations in the North Atlantic – the area where the aforementioned papers have their study area – the composition of particles would help to assess the NADW composition (e.g. fig.6 in Stichel et al. 2020). In the search for end member composition, the authors might want to consider the very dynamic particle composition in that area (pointing towards different sources) and not necessarily from bulk sediments. If I am not mistaken, in those papers the observational pNd/dNd are often one order of magnitude higher in the North Atlantic, compared to the global average assumed in the modelling studies (0.001 to 0.006). Or do I miss something here? Of course, it is reasonable to assume that Pacific pNd are very much lower than in the North Atlantic and therefore the pNd/dNd is very much skewed towards lower values. This discrepancy should at least be mentioned and justified.
I would like to point out another rather minor issue, which is the database used. I acknowledge that with the now very impressive global Nd data sets available, it is very convenient to cite the GEOTRACES IDP. However, the authors want to double-check whether data actually IS in the data product. For instance, large parts of the eNd from GA03 (or US-GEOTRACES North Atlantic Zonal Section) are not included in the IDP 2021 but was used a citation here. For those data sets you can find the correct citations here:
http://data.bco-dmo.org/jg/info/BCO/GEOTRACES/NorthAtlanticTransect/Nd_GT10%7Bdir=data.bco-dmo.org/jg/dir/BCO/GEOTRACES/NorthAtlanticTransect/,data=data.bco-dmo.org:80/jg/serv/BCO/GEOTRACES/NorthAtlanticTransect/Nd_GT10_v8_joined.html0%7D?. I apologise for this rather shameless self-advertisement…
I fully support the publication of this paper eventually. It is an important work and will be key for a better understanding of the Nd cycle as it is one of (if not) the most complete modelling papers for marine Nd isotopes and concentrations. The supplement’s profound eNd/YREE data set is also great! Thanks for providing this with your publication.
Citation: https://doi.org/10.5194/egusphere-2022-606-RC5 - AC5: 'Reply on RC5', Suzanne Robinson, 10 Jan 2023
- AC6: 'Comment on egusphere-2022-606', Suzanne Robinson, 10 Jan 2023
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 551 | 136 | 31 | 718 | 45 | 5 | 5 |
- HTML: 551
- PDF: 136
- XML: 31
- Total: 718
- Supplement: 45
- BibTeX: 5
- EndNote: 5
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 182 | 24 |
| China | 2 | 152 | 20 |
| United Kingdom | 3 | 69 | 9 |
| Switzerland | 4 | 54 | 7 |
| Germany | 5 | 53 | 7 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 182
Cited
Ruza Ivanovic
Lauren Gregoire
Julia Tindall
Tina van de Flierdt
Yves Plancherel
Frerk Pöppelmeier
Kazuyo Tachikawa
Paul Valdes
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(9213 KB) - Metadata XML
-
Supplement
(11064 KB) - BibTeX
- EndNote
- Final revised paper