the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Apr 2025
A 100-Year record of mineralogical variations in Northeastern Greenland ice-core dust: Insights from individual particle analysis
Abstract. Understanding the spatial and temporal variations in mineral dust sources in Greenland ice cores during the Holocene is challenging due to low dust concentration. Here, we present the first continuous records of the size and composition, as well as the temporal variations in potential sources, of mineral dust preserved in a northeastern Greenland ice core (EGRIP) covering the period from 1910 to 2013. Using a multi-proxy provenance approach based on individual particle analysis, we reconstruct variations in ice-core dust sources. We apply a recently developed provenance tracing technique, namely scanning electron microscopy (SEM)-cathodoluminescence (CL) analysis of single quartz particles, and SEM energy-dispersive X-ray spectroscopy (EDS) analysis to the Greenland ice core. The SEM-CL/EDS results reveal that the primary dust sources in the EGRIP ice core are Asian (Gobi Desert) and African (Sahara Desert) deserts. Their relative contributions have shifted since the 1970s–1980s: the contribution from the Gobi Desert has decreased whereas that from the Sahara Desert has increased. Our findings demonstrate that SEM-CL analysis is a valuable tool for identifying ice-core dust sources and reconstructing their variations during periods of low dust concentration. Additionally, we compare the EGRIP ice-core dust records with those of a northwestern Greenland ice core (SIGMA-D) to determine spatial variations in potential dust sources within the Greenland Ice Sheet over the past 100 years. The results reveal that the variability and mineral composition of the EGRIP ice-core dust differs significantly from those of the SIGMA-D ice-core dust, indicating that the dust in these two ice cores was likely transported from different geological sources. The SIGMA-D ice-core dust exhibits multidecadal variations, reflecting increased dust from the Greenland coastal region during warmer periods. Conversely, the EGRIP ice-core dust shows low temporal variation, suggesting a smaller contribution from local sources.
- Preprint
(5338 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2025-1522', Anonymous Referee #1, 08 May 2025
This paper attempts to characterise dust in the EGRIP ice core using mainly mineralogical methods. The data are new, covering around a century of dust input, and were probably the result of much painstaking work. I am reasonably happy with the dating as described. However, unfortunately, at high level my comments are the same as on previous versions of the paper. Even with the addition of CL data, the paper is unable to identify sources with any confidence. The statements in the abstract about changes in the relative proportions of Gobi and Sahara dust are simply not justified by the data in the text. The result is that the paper contains interesting data that is worth showing, but no gains in source apportionment that would justify publication in a climate journal such as CP. Additionally the paper is very long (with 14 figures) considering the low amount of new insight. I can only reiterate that a much shorter paper in a data journal might be more appropriate.
The paper hinges on three types of data. First is the mineralogical data, as shown first in Figure 6. We are invited to interpret these in the light of Table 1. However this table only illustrates why this cannot be done, with each mineral class having numerous sources, and not in any way differentiating even at the continent scale. The differences between samples from each decade at EGRIP are rather minor, and far from suggesting a change in source, to first order they reveal a constancy in mineral types (or at least the statistical analysis to establish otherwise is missing). For sure there are differences between EGRIP and SIGMA-D, but these are attributed mainly to the presence of more local material at the latter, an issue already discussed in a previous paper.
The second line of evidence is the CL data, shown in Fig 7 as ternary plots. I found the data treatment for this work surprising. Clusters derived from Asian dust are used as a basis, and then interpreted in terms of relative proportions of Asian and African desert dust. This seems inappropriate. It is never explained where the Sahara data come from, and I cannot see how clusters appropriate for Asia can be used. We simply have no information on other Asian and African sources (previous isotopic work, as I recall, does not suggest the Gobi as the most probably Asian source), and the differences between decades, with only 5 decades sampled, are not convincingly different (again a statistical treatment would be needed to establish that the raw CL data, before clustering, are different for the measured decades).
Finally the back trajectories in Fig 2 really show us nothing, except that air masses can come in from all directions to EGRIP.
Overall, I do not feel that the paper has constructed a story that warrants a long paper like this. I will go into some details in the following paragraphs, as well as giving some minor points and typos. I apologise that this comes across negatively – that the authors have tried to construct a long discussion where the data do not support it, and this leads to a lot of unsupported statements.
Abstract “The SEM-CL/EDS results reveal that the primary dust sources in the EGRIP ice core are Asian (Gobi Desert) and African (Sahara Desert) deserts”. I’m afraid I don’t feel that the CL data showed this convincingly, and certainly the data are insufficient to justify the next sentence about their relative proportions.
Lines 44/45: surely GRIP is central Greenland and NEEM is north. The text says the opposite.
Line 50 “Greenland ice-core dust can also be used to estimate variations in the mass balance of the ice sheet”. This is incorrectly written. I think you mean that dust can influence the mass balance, not that the dust can tell us about the mass balance.
Line 107 “nearly 100 years (1910–2013)”. Nearly implies less than 100, but this is more than 100.
Fig 1. Hans Trausen should be Hans Tausen.
Line 205 and around. I do not follow the logic of using Nagashima’s clusters for Asia to differentiate between Asian and African sources. You need surely to use source materials to define an African cluster before attempting this analysis. I suspect it would also be important to look at the raw data (strength of each emission component) to ensure you are right that different Greenland samples are statistically significantly different before even attempting a cluster analysis.
Fig 4. What is R2 in these figures. Presumably not a correlation coefficient as these are always less than 1!
Line 300. Are the differences in mineralogy between decades statistically significant. My feeling is not but you need to address this.
Section 3.4 and Fig 7. To me these look like tiny changes in composition. I also wonder why we are only shown 5 samples.
Fig 6 and 8 (left) and 10 (EGRIP panel) seem to have the same information. How do they differ and why repeat the information?
Section 4.1 really doesn’t provide enough information to be definitive or convincing about sources or change of sources. I’d need to see a lot more source signature information, and an analysis of statistics of differences between decades.
Fig 9 – I am not really sure what I am meant to get from this figure, since most of the source signatures overlap close to the Greenland data.
Fig 10a GRIP panel – spelling of Svensson. Fig 10 needs a better caption, 10b has no information.
Line 388 (and Fig 11). “North America cannot be ruled out as an additional source”. But with the discussion as written nowhere can be ruled out.
Line 402 “with a decrease in contributions from Asia and an increase from Africa in the last 50 years”. I don’t feel you established this.
Line 412 “Therefore, the gradual increase in Fe/Al in the EGRIP”. Idon’t see the increase you are considering. It’s certainly not obvious in Fig 13, and Fig 12 is really unclear – what are the coloured shapes meant to show?
Fig 14 and the surrounding discussion was really unconvincing. It doesn’t seem to be related to sources in any case. As an example “a statistically significant negative correlation with the AMO index after 1950”: while I see the values in Table 4, in Fig 14 when I compare panels a and d, I see no correlation.
On the same discussion please clarify what average volume is. If it’s really “the volume concentration divided by the number concentration”, then it should be the volume of a dust particle, which would be of order 1 um^3, but in the figure it’s shown as cm^3. You need to explain what this is much better and also discuss why you think it is related to distance from the source, which is also not immediately obvious.
Fig 13 claims to include “annual average volume of dust particles” but it doesn’t.
Citation: https://doi.org/10.5194/egusphere-2025-1522-RC1 - AC1: 'Reply on RC1', Naoko Nagatsuka, 20 Jun 2025
-
RC2: 'Comment on egusphere-2025-1522', Anonymous Referee #2, 24 May 2025
The paper by Nagatsuka and colleagues provides analysis of the size and composition of mineral dust over the last 100 years based on particles collected from the EGRIP ice core in northern eastern Greenland. The work aims to present the obtained size and compositional datasets and provides analysis to determine the sources of dust in the ice core and their temporal variation. The results are discussed in comparison to data previously published and relying on another ice core in north western Greenland (SIGMA-D). The main result highlighted in the paper is the lower temporal variation of dust composition in EGRIP, suggesting more local dust sources compared to SIGMA-D.
The data from this work are very interesting and deserve to be published. On the other hand, the discussion and conclusions are not supported by the analysis, and no robust evidences are provided for the source identification. Detailed reasons for this are provided in the general comments.
My judgment is that in the present form the paper would require major revisions to narrow the scope and/or better support conclusions. My suggestion would be to present the data (Section 3) and provide a concise and objective analysis of elements resuming the relatively long discussion now in Section 4. A “data paper” of a “measurement report” type could be more suitable for this. I also note that in several points of the paper, including the abstract and the conclusions, the authors mention that “the findings demonstrate that the SEM-CL analysis is a valuable tool for identifying ice-core dust sources and reconstructing their variation during periods of low dust concentration”. This kind of sentence mostly belongs to methodological papers dedicated to test and evaluate techniques, instead of this kind of papers that should focus on scientific interpretation of results based on well-established analysis tools. This kind of statement also question me about the most appropriate journal and targeted scope of the work.
General comments
Section 4.1 lines 333-340 and Fig. 7: the analysis referring to Fig. 7 is not very robust. Only 5 points are considered from each decade and compared to unpublished data. Apart from the fact that unpublished datasets are considered (data which however should be made available to the scientific community prior to be used as interpretation of other data) the five points from the EGRIP core are at the edges between the West Africa and East Asia data, without – to my reading of the plot - an emerging clear pattern.
Fig. 8 and more in general the interpretation of the clustering analysis (Sect. 4.1 and 4.2): the categorization into Types A-E described in Sect. 2.4 refers to mineral types and associate them to potential sources, but those are very broadly defined, which makes difficult to understand their usefulness for the data interpretation.
Section 4.1 lines 356-367, Fig. 10: the comparison against other datasets and the similarities/differences discussed, ultimately in relation to dust origin, is difficult to evaluate due to the methodological differences mentioned, potentially affecting the reasoning and the implications of identified mineralogical patterns
Back trajectory analysis: I am unsure how to read and interpret the analysis of back trajectories. These are performed considering 20-days, a choice that is questionable. Indeed, this is very long, also compared to an aerosol lifetime of few days in the troposphere. Apart from the trajectory duration, I am not sure to understand how to use the information of Fig. 11, also based on the results of Fig. 2 showing that there is not a preferential direction of back trajectory air masses – which somehow can be expected as all data for more than 50 years are put together. Also attention as there is an inconsistency: in Sect. 2.6 the 20-days trajectories are mentioned, while in Fig. 11 is 25 days. Instead, Fig. 2 shows 7-days trajectories.
Section 4.2 lines 40-414, Figures 12 and 13: to my reading of Fig. 12 I do not see a clear pattern in the data points, which instead seems quite dispersed. The authors should provide some statistical analysis to support their interpretation. I also do not identify an increase in Fe/Al in Fig. 13 as stated, and more statistical analysis should be provided to support this statement (line 413).
Detailed comments
Section 2.7, lines 223-226: this kind of sentences is not necessary / relevant
Section 2.7, lines 230-235: this part can stay in supplementary information
Section 2.7, lines 236-237: this is relevant to the analysis of the present paper and should be better detailed
Section 3.2: the diameter is a projected-area diameter – please provide more details
Figure 4: y-axis should read “particle number concentration, x-axis should refer to “diameter” or “projected-area diameter” (if this is the case)
Citation: https://doi.org/10.5194/egusphere-2025-1522-RC2 - AC2: 'Reply on RC2', Naoko Nagatsuka, 20 Jun 2025
Status: closed
-
RC1: 'Comment on egusphere-2025-1522', Anonymous Referee #1, 08 May 2025
This paper attempts to characterise dust in the EGRIP ice core using mainly mineralogical methods. The data are new, covering around a century of dust input, and were probably the result of much painstaking work. I am reasonably happy with the dating as described. However, unfortunately, at high level my comments are the same as on previous versions of the paper. Even with the addition of CL data, the paper is unable to identify sources with any confidence. The statements in the abstract about changes in the relative proportions of Gobi and Sahara dust are simply not justified by the data in the text. The result is that the paper contains interesting data that is worth showing, but no gains in source apportionment that would justify publication in a climate journal such as CP. Additionally the paper is very long (with 14 figures) considering the low amount of new insight. I can only reiterate that a much shorter paper in a data journal might be more appropriate.
The paper hinges on three types of data. First is the mineralogical data, as shown first in Figure 6. We are invited to interpret these in the light of Table 1. However this table only illustrates why this cannot be done, with each mineral class having numerous sources, and not in any way differentiating even at the continent scale. The differences between samples from each decade at EGRIP are rather minor, and far from suggesting a change in source, to first order they reveal a constancy in mineral types (or at least the statistical analysis to establish otherwise is missing). For sure there are differences between EGRIP and SIGMA-D, but these are attributed mainly to the presence of more local material at the latter, an issue already discussed in a previous paper.
The second line of evidence is the CL data, shown in Fig 7 as ternary plots. I found the data treatment for this work surprising. Clusters derived from Asian dust are used as a basis, and then interpreted in terms of relative proportions of Asian and African desert dust. This seems inappropriate. It is never explained where the Sahara data come from, and I cannot see how clusters appropriate for Asia can be used. We simply have no information on other Asian and African sources (previous isotopic work, as I recall, does not suggest the Gobi as the most probably Asian source), and the differences between decades, with only 5 decades sampled, are not convincingly different (again a statistical treatment would be needed to establish that the raw CL data, before clustering, are different for the measured decades).
Finally the back trajectories in Fig 2 really show us nothing, except that air masses can come in from all directions to EGRIP.
Overall, I do not feel that the paper has constructed a story that warrants a long paper like this. I will go into some details in the following paragraphs, as well as giving some minor points and typos. I apologise that this comes across negatively – that the authors have tried to construct a long discussion where the data do not support it, and this leads to a lot of unsupported statements.
Abstract “The SEM-CL/EDS results reveal that the primary dust sources in the EGRIP ice core are Asian (Gobi Desert) and African (Sahara Desert) deserts”. I’m afraid I don’t feel that the CL data showed this convincingly, and certainly the data are insufficient to justify the next sentence about their relative proportions.
Lines 44/45: surely GRIP is central Greenland and NEEM is north. The text says the opposite.
Line 50 “Greenland ice-core dust can also be used to estimate variations in the mass balance of the ice sheet”. This is incorrectly written. I think you mean that dust can influence the mass balance, not that the dust can tell us about the mass balance.
Line 107 “nearly 100 years (1910–2013)”. Nearly implies less than 100, but this is more than 100.
Fig 1. Hans Trausen should be Hans Tausen.
Line 205 and around. I do not follow the logic of using Nagashima’s clusters for Asia to differentiate between Asian and African sources. You need surely to use source materials to define an African cluster before attempting this analysis. I suspect it would also be important to look at the raw data (strength of each emission component) to ensure you are right that different Greenland samples are statistically significantly different before even attempting a cluster analysis.
Fig 4. What is R2 in these figures. Presumably not a correlation coefficient as these are always less than 1!
Line 300. Are the differences in mineralogy between decades statistically significant. My feeling is not but you need to address this.
Section 3.4 and Fig 7. To me these look like tiny changes in composition. I also wonder why we are only shown 5 samples.
Fig 6 and 8 (left) and 10 (EGRIP panel) seem to have the same information. How do they differ and why repeat the information?
Section 4.1 really doesn’t provide enough information to be definitive or convincing about sources or change of sources. I’d need to see a lot more source signature information, and an analysis of statistics of differences between decades.
Fig 9 – I am not really sure what I am meant to get from this figure, since most of the source signatures overlap close to the Greenland data.
Fig 10a GRIP panel – spelling of Svensson. Fig 10 needs a better caption, 10b has no information.
Line 388 (and Fig 11). “North America cannot be ruled out as an additional source”. But with the discussion as written nowhere can be ruled out.
Line 402 “with a decrease in contributions from Asia and an increase from Africa in the last 50 years”. I don’t feel you established this.
Line 412 “Therefore, the gradual increase in Fe/Al in the EGRIP”. Idon’t see the increase you are considering. It’s certainly not obvious in Fig 13, and Fig 12 is really unclear – what are the coloured shapes meant to show?
Fig 14 and the surrounding discussion was really unconvincing. It doesn’t seem to be related to sources in any case. As an example “a statistically significant negative correlation with the AMO index after 1950”: while I see the values in Table 4, in Fig 14 when I compare panels a and d, I see no correlation.
On the same discussion please clarify what average volume is. If it’s really “the volume concentration divided by the number concentration”, then it should be the volume of a dust particle, which would be of order 1 um^3, but in the figure it’s shown as cm^3. You need to explain what this is much better and also discuss why you think it is related to distance from the source, which is also not immediately obvious.
Fig 13 claims to include “annual average volume of dust particles” but it doesn’t.
Citation: https://doi.org/10.5194/egusphere-2025-1522-RC1 - AC1: 'Reply on RC1', Naoko Nagatsuka, 20 Jun 2025
-
RC2: 'Comment on egusphere-2025-1522', Anonymous Referee #2, 24 May 2025
The paper by Nagatsuka and colleagues provides analysis of the size and composition of mineral dust over the last 100 years based on particles collected from the EGRIP ice core in northern eastern Greenland. The work aims to present the obtained size and compositional datasets and provides analysis to determine the sources of dust in the ice core and their temporal variation. The results are discussed in comparison to data previously published and relying on another ice core in north western Greenland (SIGMA-D). The main result highlighted in the paper is the lower temporal variation of dust composition in EGRIP, suggesting more local dust sources compared to SIGMA-D.
The data from this work are very interesting and deserve to be published. On the other hand, the discussion and conclusions are not supported by the analysis, and no robust evidences are provided for the source identification. Detailed reasons for this are provided in the general comments.
My judgment is that in the present form the paper would require major revisions to narrow the scope and/or better support conclusions. My suggestion would be to present the data (Section 3) and provide a concise and objective analysis of elements resuming the relatively long discussion now in Section 4. A “data paper” of a “measurement report” type could be more suitable for this. I also note that in several points of the paper, including the abstract and the conclusions, the authors mention that “the findings demonstrate that the SEM-CL analysis is a valuable tool for identifying ice-core dust sources and reconstructing their variation during periods of low dust concentration”. This kind of sentence mostly belongs to methodological papers dedicated to test and evaluate techniques, instead of this kind of papers that should focus on scientific interpretation of results based on well-established analysis tools. This kind of statement also question me about the most appropriate journal and targeted scope of the work.
General comments
Section 4.1 lines 333-340 and Fig. 7: the analysis referring to Fig. 7 is not very robust. Only 5 points are considered from each decade and compared to unpublished data. Apart from the fact that unpublished datasets are considered (data which however should be made available to the scientific community prior to be used as interpretation of other data) the five points from the EGRIP core are at the edges between the West Africa and East Asia data, without – to my reading of the plot - an emerging clear pattern.
Fig. 8 and more in general the interpretation of the clustering analysis (Sect. 4.1 and 4.2): the categorization into Types A-E described in Sect. 2.4 refers to mineral types and associate them to potential sources, but those are very broadly defined, which makes difficult to understand their usefulness for the data interpretation.
Section 4.1 lines 356-367, Fig. 10: the comparison against other datasets and the similarities/differences discussed, ultimately in relation to dust origin, is difficult to evaluate due to the methodological differences mentioned, potentially affecting the reasoning and the implications of identified mineralogical patterns
Back trajectory analysis: I am unsure how to read and interpret the analysis of back trajectories. These are performed considering 20-days, a choice that is questionable. Indeed, this is very long, also compared to an aerosol lifetime of few days in the troposphere. Apart from the trajectory duration, I am not sure to understand how to use the information of Fig. 11, also based on the results of Fig. 2 showing that there is not a preferential direction of back trajectory air masses – which somehow can be expected as all data for more than 50 years are put together. Also attention as there is an inconsistency: in Sect. 2.6 the 20-days trajectories are mentioned, while in Fig. 11 is 25 days. Instead, Fig. 2 shows 7-days trajectories.
Section 4.2 lines 40-414, Figures 12 and 13: to my reading of Fig. 12 I do not see a clear pattern in the data points, which instead seems quite dispersed. The authors should provide some statistical analysis to support their interpretation. I also do not identify an increase in Fe/Al in Fig. 13 as stated, and more statistical analysis should be provided to support this statement (line 413).
Detailed comments
Section 2.7, lines 223-226: this kind of sentences is not necessary / relevant
Section 2.7, lines 230-235: this part can stay in supplementary information
Section 2.7, lines 236-237: this is relevant to the analysis of the present paper and should be better detailed
Section 3.2: the diameter is a projected-area diameter – please provide more details
Figure 4: y-axis should read “particle number concentration, x-axis should refer to “diameter” or “projected-area diameter” (if this is the case)
Citation: https://doi.org/10.5194/egusphere-2025-1522-RC2 - AC2: 'Reply on RC2', Naoko Nagatsuka, 20 Jun 2025
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 396 | 113 | 20 | 529 | 18 | 36 |
- HTML: 396
- PDF: 113
- XML: 20
- Total: 529
- BibTeX: 18
- EndNote: 36
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1