Review of „Paleo±Dust: Quantifying uncertainty in paleo-dust deposition across archive types” Cosentino et al.

 Manuscript number: ESSD-2023-241

In recent years, there has been a renaissance in the field of spatio-temporal reconstruction of sediment mass accumulation rates (MAR), probably due to the fact that atmospheric dust does indeed play a significant role in shaping Earth's climate, in both direct and indirect ways. Yet there are considerable uncertainties about the actual role of dust that need to be clarified in order to better understand how the Earth system works under present-day or past (e.g. glacial) climate conditions. Research on MAR reconstructions is taking several directions: 1) synthesis of new data from one or more type(s) of paleo-dust archives, 2) methodological innovations in the calculation of MAR, 3) database construction, and/or 4) focusing on periods that have not been included in reconstructions so far (e.g. Marine Isotope Stage 4-5-6). As such, a new database called ChronoLoess (Bosq et al.) is currently appearing in the ESSD pages, which can be extended in a standardised format. In contrast to ChronoLoess, which focuses on only one archive, the authors of this manuscript have compiled data from a large number of paleodust archives of different types (marine/terrestrial) in the Paleo±Dust database, following a common methodology. I consider the most important advance of the present paper to be that they have discussed the uncertainties associated with the dust mass accumulation rate (DMAR) calculations of each archive type and attempted to estimate them quantitatively, which in many cases is not a simple task. In addition, paleo-dust flux estimates from peat bog cores are included in the paleodust compilation for the first time.

Having read the manuscript, I find it to be neatly written, logical and easy to follow, and I consider it to be a significant step forward in the reconstruction and comparison of LGM and Holocene dust fluxes. The authors have tried to organise the database in a way that it can be used for future Earth system modelling, which is also a positive aspect. I have, of course, some critical comments on the manuscript, which I write about below (and in the annotated pdf file attached), but these do not affect the conclusions to any significant extent and are rather just a further refinement of the methodology. On the whole, the manuscript, with some modifications, can be published in ESSD.

Line-by-line comments

Line 97, εDMAR: This is just a pedantic point, but the error of a quantity is usually denoted by the Greek sigma in statistics, I have never seen the use of epsilon for that purpose. What is the reason for using epsilon here? There is another reason of why I discourage the authors using epsilon: in dust provenance studies the Nd and Hf isotope compositions are often expressed in the epsilon notation (εNd, εHf), so this may potentially cause some misunderstandings.

Lines 98-100: The authors presented uncertainty estimations for MAR and MAR10, which disregard potential correlations between the variables in the MAR equations, so these are not full statistical treatments. I suggest the error propagations below: PLEASE SEE THE EQUATIONS IN THE ATTACHED REVIEW PDF FILE AS THEY COULD NOT BE INSERTED IN THE TEXT HERE!!!

Line 131: It is worthwile adding that this relation is true if the bottom/top ages of a segment is given as negative ages in the past calculating from e.g 1950 BP, or b2k. If the ages are positive as usually used (e. g. t_bottom=23500 years, t_top=21000), then this does not hold. This can be easily fixed by changing the direction of the inequality signs.

Lines 236-246: The authors use a dry bulk density (DBD) value of 1.48 g/cm³ for CLP loess, while 1.45 for others. For European loess, DBD is relatively well-defined (at least for East Central Europe, ECE): Újvári et al. (2010) [Quat.Sci.Rev. 29, 3157-3166] published a DRD value of 1.497+-0.079 g/cm³ derived from 6 test sample measurements (2 methods), which was later confirmed by e.g. Peric et al. (2020) [Boreas 49, 841-857]. So, this value should be used for the ECE loess sites, and perhaps for others as well considering how close this is to the CLP loess DBD value. I suggest mentioning these studies here, just to provide a less CLP-biased overview of the topic.

Line 381, SBMAR equation: Why is this Eq (1)? We have seen several equations before in this manuscript. Those must be numbered and indicated as well, in my view.

14/09/2023

Gabor Ujvari

Please find my further comments in the annotated pdf file.

I have thoroughly reviewed the manuscript titled "Paleo±Dust: Quantifying uncertainty in paleo-dust deposition across archive types" by Nicolás J. Cosentino and co-authors. The primary objective of this manuscript is to compile updated bulk and <10-μm paleo-dust deposition rates across various archive types between 2016 and 2022, referred to as the Paleo±Dust dataset. This dataset encompasses a total of 284 pre-industrial Holocene (pi-HOL) and 208 Last Glacial Maximum (LGM) dust flux records. Based on the Paleo±Dust dataset, the manuscript calculates that the global LGM:pi-HOL ratio of <10-μm dust deposition rate is 3.1 ± 0.8 (1σ), which is a critical parameter for future paleodust studies and simulations using Earth system models of high to intermediate complexity.

In my opinion, this dataset presented in the manuscript are novel and hold great potential being useful in the future. The materials and methods are described in detail y, and the dataset is complete and accessible through the provided identifier. The dataset is currently usable in its existing format and size. Additionally, the manuscript is well-structured, nicely illustrated, and well-written. However, I have a few suggestions that require further clarification:

Uncertainty in the Mass Fraction of Dust: It is acknowledged that most organic matter in loess is post-depositional in origin, it's important to note that carbonate in loess primarily originates from dust source areas and constitutes a significant portion (approximately 14wt%) of dust materials (Meng et al., 2019, Geophysical Research Letters 46, 4854-4862). After deposition, carbonate minerals in loess can undergo redistribution in the soil through processes such as dissolution, migration, and reprecipitation (Meng et al., 2015, Geophysical Research Letters 42, 10,391-310,398, and Meng et al., 2018, Earth and Planetary Science Letters 486, 61-69). Previous research on the Chinese Loess Plateau (Meng et al., 2015, Geophysical Research Letters 42, 10,391-310,398, and Meng et al., 2018, Earth and Planetary Science Letters 486, 61-69) and Central Asia (Zhang et al., 2023, Catena 232, 107420) has indicated that wet climates in interglacial periods (e.g., Holocene) tend to deplete carbonate in the soil, whereas carbonate loss during the last glacial period is limited. Therefore, it would be more reasonable to assume that 10±5wt% of carbonate was lost in the Holocene and none in the last glacial period. Similarly, for organic matter, it's essential to consider its varying content between the Holocene (high) and the last glacial period (low), as supported by some studies (e.g., Yang et al., 2015, PNAS, 112, 13178-13183). Based on previous works, it would be more reasonable to assume that TOC content was 1±0.5wt% for the Holocene and 0.2±0.2wt% for the last glacial period.

Minor Comments:

This manuscript is updated on the basis of Albani et al., (2015), if the updated sites are marked in Figure 5, it is easier to make the reader clear about the value of this dataset.

It should be noted that the last glacial-interglacial period corresponds to marine isotope stages (MIS) 2-5, and not to the Holocene (MIS 1) or the last glacial period (MIS 2-4). The present interglacial, which is the Holocene, is ongoing interglacial period, not last interglacial.

Lines 175-176：If dating uncertainty is not reported, 3.4%, 9.1% and 13.1% relative uncertainties are assumed for 14C-, OSL- and thermoluminescence-based age determinations. Please give the references for the uncertainties. In my understanding, the relative uncertainties associated with these dating methods are hardly accurate to one decimal point as the values mentioned (e.g., 9.1%). For instance, it is commonly accepted that the relative uncertainties in OSL dating method are often estimated to be around 10%.

Consider adding the dust data of Sihailongwan maar Lake in northeastern China (Zaarur, S., Stein, M., Adam, O., Mingram, J., Liu, J., Wu, J., Raveh-Rubin, S., Erel, Y., 2020. Synoptic stability and anomalies in NE China inferred from dust provenance of Sihailongwan maar sediments during the past ∼80 kyr. Quaternary Science Reviews 239, 106279). This could contribute valuable data to the manuscript.

Overall, the dataset shows promise and provides valuable contributions to the field of paleo-dust research, but addressing the above suggestions will further enhance its quality and impact.