the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Jan 2025
The role of atmospheric large-scale patterns for recent warming periods in Greenland
Abstract. Atmospheric large-scale patterns strongly determine Greenland’s regional climate through air mass advection and local weather conditions, making them essential to understand atmospheric variability. This study analyses the occurrence of atmospheric large-scale patterns during two distinct warming periods of the recent past that we identify objectively in climatological data. The first warming period lasted from 1922 to 1932 and an average air temperature increase of 2.9 °C across all stations considered for this study. The second warming period lasted from 1993 to 2007 and had an average warming of 3.1 °C. We apply Self-Organizing Maps as a clustering technique based on the geopotential height of the 500 hPa pressure level using 20CRv3 reanalysis data to characterize prevalent atmospheric large-scale patterns and investigate their occurrence, persistence, and effects on air temperature anomalies at our study site (Qaamarujup Sermia) in West Greenland. Both warming periods show similar overall air temperature anomalies. However, the distribution of large-scale atmospheric patterns differs significantly, while the relationship between atmospheric large-scale patterns and local air temperature seems to be constant in time. This suggests that variations in Greenland’s warming are influenced by shifts in atmospheric circulation. This study emphasizes the critical role of changes in atmospheric large-scale patterns for understanding Greenland’s warming periods.
- Preprint
(1930 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2024-4060', Anonymous Referee #1, 27 Jan 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2024-4060/egusphere-2024-4060-RC1-supplement.pdf
-
AC1: 'Reply on RC1', Florina Roana Schalamon, 17 Mar 2025
Dear reviewer,
We are very grateful for your very constructive review and appreciate the valuable time put
into this. We have addressed your comments and suggestions in the attached file.Once again, thank you for your insightful input and support.
Best regards,
Florina Schalamon (on behalf of the author team)
-
AC1: 'Reply on RC1', Florina Roana Schalamon, 17 Mar 2025
-
RC2: 'Comment on egusphere-2024-4060', Anonymous Referee #2, 28 May 2025
This study examines how large-scale atmospheric circulation patterns affected surface air temperature during two historical warming periods in Greenland (1922–1932 and 1993–2007). Using Self-Organizing Maps (SOMs) performed on 500hPa heights from the NOAA 20th century reanalysis data, the authors identified key atmospheric patterns. Despite similar warming magnitudes, the patterns differed between the periods, suggesting that changes in atmospheric circulation patterns affect the pattern of warming.
Overall, the manuscript is well written and illustrated with appropriate references. The findings will be useful to scholars of polar climate.
Major questions:
- Large-scale, low-frequency patterns of atmospheric circulation will change over the seasons. Did the authors consider conducting the SOM analysis on a seasonal basis? While the seasonal analysis in Figure 5 is helpful, I question whether combining all the data to calculate the SOMs would provide the same patterns as SOMs conducted for each season.
- The authors should further explain the interest in the WEG_L station at Qaamarujup Sermia. It is unclear why they selected this station and how it was useful to the analysis.
- What is the value in looking at “warming periods”, periods of increasing temperature (i.e., first derivative of temperature), rather than warm periods (some threshold value above a long-term mean)? I would imagine the circulation patterns are more clearly associated with “warm periods” than “warming periods”.
- The authors mention an “additional approach” looking at the warmest and coldest days (l. 179-180). Should this be more prominent and mentioned earlier in the manuscript?
- The spatial pattern of the air temperature trend that reaches west Greenland appears to extend from Baffin Bay (Figs. 2b and c; also see final minor question). Have the authors considered looking at trends in SSTs or sea ice coverage?
- Previous work has used Empirical Orthogonal Functions (or Principal Components Analysis) to identify patterns in low-frequency modes of atmospheric circulation, including identifying NAO, using 700hPa or 500hPa heights. One of the key steps in such work is understanding the physical meaning of the patterns identified. What advantages are there to using the SOM analysis relative to EOFs? Does each pattern represent a physically meaningful mode of circulation?
Minor questions:
- 30: Reducing snow cover on the ice but also physically changing the structure of snow and firn, the key is reducing the albedo.
- 42: Cloud radiative processes dictated by cloud height, cloud optical thickness, and hydrometeor phase are also important in driving the radiative budget.
- 52: The NAO is a redistribution of atmospheric mass between the subpolar and subtropical regions of the North Atlantic that one can capture the NAO using surface pressure data, as mentioned here, but can also be identified using geopotential height data.
- 64: Thermal advections will not necessarily follow lines of constant geopotential height, but will follow lines of constant thickness. To a first order, this is a reasonable interpretation, but it is important to be careful when interpreting the height pattern as showing advection.
- 65: It appears there is a missing word or phrase after “west in”.
- 65-66: As large accumulation and ablation events are often driven by cyclonic events, it would be worth saying more about how low-frequency circulation drives the storm track and the subsequent impact on mass balance.
- 114: Why not use a standard 30-year reference period, such as 1981-2010?
- 153: Is there any issue including a domain that reaches the pole?
- 180: Are the warmest 15% “abnormal”?
- 220-224: It’s not just the direction of flow that’s important, but the geopotential height is a function of the thickness of the lower troposphere, which is a function of the mean (virtual) temperature of that layer. If, like pattern 5, you are sitting under a trough, you would expect colder than normal weather. The patterns may also represent potential storm tracks, such as LSP6, which I expect would be more conducive to storms traveling up Baffin Bay and affecting the west coast of Greenland.
- 265: I really like this figure, especially the heat map and persistence plot! Very helpful to understand the patterns.
- 319-324: I had been wondering why 20CR3 was used instead of the ERA-20C product when the 20CR3 product was first mentioned. This should be discussed earlier in the manuscript.
- 349-350: It might be helpful to produce composite AT and SST anomalies (like Fig 2b,c) for each of the 8 patterns (overall and during each of the two periods studies).
Citation: https://doi.org/10.5194/egusphere-2024-4060-RC2 -
RC3: 'Comment on egusphere-2024-4060', Anonymous Referee #3, 12 Jun 2025
The manuscript addresses an important topic and is interesting and clearly written. However, it includes methodological issues related to the application of Self-Organizing Maps (SOM). These issues generate uncertainty in the attribution of the observed temperature changes in Greenland. Hence, I recommend major revisions to be made.
Major comments
1. The number of SOM nodes (cluster centres) is only eight. The authors give an argument that they preferred to avoid the need the carry out two SOM analyses, first using a larger number of nodes and then regrouping to smaller number. However, the regrouping is not necessarily needed, and the number of nodes should be large enough to allow robust conclusions. In the results obtained, some SOM nodes (particularly 1, 3 and 8) may result in either warm or cold conditions at the study site. This seems to be attributed to the role of local processes. I agree that local processes often have a strong impact on near-surface air temperatures, particularly in regions of complex orography. However, the non-systematic sign of the temperature anomaly for a single SOM node may also result from the fact the smaller is the number of nodes, the larger is the variability of circulation patters within the a single node. This makes it impossible to understand (based on the analyses made here) what is the relative importance of small-scale processes and the variations between large-scale patterns within a SOM node. I suggest performing a new SOM analysis with a larger number of nodes (e.g., 20 or even more).
2. The SOMs are grouped in a one-dimensional array. Although the plots in Figure 3 are not shown in a row, it is evident that the analysis is made for an 1D array. This is demonstrated by the fact that neighboring nodes (e.g., 1 and 4, 2 and 5, 3 and 6, 5 and 8) differ a lot from each other, which is not the case for a 2D SOM array. Using a 1D array is very unusual in the field of meteorology. In general, 2D SOM is preferred for most applications, in particular when it is relevant to visualize clusters or relationships, and the data does not have a natural linear ordering. A 1D SOM may be relevant for simple, linear datasets or specific ranking tasks, but it is less interpretable for general clustering or visualization. I suggest that the new SOM analysis will be made using a 2D array (e.g., 5 x 4 nodes).
3. The SOM analysis is made for a rather small area around Greenland (Figure 3). Even if certain circulation patters appear rather similar in the study area, fitting into a same SOM node, they may include considerable differences outside the boundaries of the study area. This may result in large differences in the transports of heat and moisture to Greenland. I cannot be sure if the effects on the results and conclusions of this study are large, but it is possible. Hence, I recommend that in the new SOM analyses the study area is enlarged.
4. The analysis addresses near-surface air temperatures with focus on coastal site at an elevation of 940 m. There the typical atmospheric pressure may be roughly 910–920 hPa. However, the analysis is made on the basis of 500 hPa geopotential height fields. These indeed well characterize the large-scale circulation patterns, but in a baroclinic atmosphere the wind vector at the 500 hPa may deviate a lot from that at the 910.920 hPa level. This should be discussed when making conclusions on the role heat advection associated with various SOM nodes. Over a flat surface, the effect of large-scale circulation on 2-m air temperature might be best analysed on the basis of 850-hPa fields but in Greenland this is naturally liable to errors. However, the authors could consider, if the SOM analysis would be better to do for 700 instead of 500 hPa fields.
5. I recommend carrying out seasonal analyses, as a certain geopotential height pattern may have very different effects on near-surface air temperatures in winter and summer.
6. I am not convinced if it is an optimal approach to focus on a single study site (WEG_L).
Minor comments
Lines 65-66: “cyclones following the North Atlantic Oscillation” is unclear expression.
Line 108: 6.5 K/1000 m
Line 217: Add southwest?
Line 381: Should the phase of Arctic Oscillation be somehow seen in changes in the occurrence of various SOM nodes?
Lines 398-400: Also the increase in CO2 concentration matters.
Line 403: globally?
In summary, based on my evaluation, the results obtained using the current methodology are not robust. However, I believe the study has significant potential to yield important results if more appropriate SOM analyses are conducted.
Citation: https://doi.org/10.5194/egusphere-2024-4060-RC3
Status: closed
-
RC1: 'Comment on egusphere-2024-4060', Anonymous Referee #1, 27 Jan 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2024-4060/egusphere-2024-4060-RC1-supplement.pdf
-
AC1: 'Reply on RC1', Florina Roana Schalamon, 17 Mar 2025
Dear reviewer,
We are very grateful for your very constructive review and appreciate the valuable time put
into this. We have addressed your comments and suggestions in the attached file.Once again, thank you for your insightful input and support.
Best regards,
Florina Schalamon (on behalf of the author team)
-
AC1: 'Reply on RC1', Florina Roana Schalamon, 17 Mar 2025
-
RC2: 'Comment on egusphere-2024-4060', Anonymous Referee #2, 28 May 2025
This study examines how large-scale atmospheric circulation patterns affected surface air temperature during two historical warming periods in Greenland (1922–1932 and 1993–2007). Using Self-Organizing Maps (SOMs) performed on 500hPa heights from the NOAA 20th century reanalysis data, the authors identified key atmospheric patterns. Despite similar warming magnitudes, the patterns differed between the periods, suggesting that changes in atmospheric circulation patterns affect the pattern of warming.
Overall, the manuscript is well written and illustrated with appropriate references. The findings will be useful to scholars of polar climate.
Major questions:
- Large-scale, low-frequency patterns of atmospheric circulation will change over the seasons. Did the authors consider conducting the SOM analysis on a seasonal basis? While the seasonal analysis in Figure 5 is helpful, I question whether combining all the data to calculate the SOMs would provide the same patterns as SOMs conducted for each season.
- The authors should further explain the interest in the WEG_L station at Qaamarujup Sermia. It is unclear why they selected this station and how it was useful to the analysis.
- What is the value in looking at “warming periods”, periods of increasing temperature (i.e., first derivative of temperature), rather than warm periods (some threshold value above a long-term mean)? I would imagine the circulation patterns are more clearly associated with “warm periods” than “warming periods”.
- The authors mention an “additional approach” looking at the warmest and coldest days (l. 179-180). Should this be more prominent and mentioned earlier in the manuscript?
- The spatial pattern of the air temperature trend that reaches west Greenland appears to extend from Baffin Bay (Figs. 2b and c; also see final minor question). Have the authors considered looking at trends in SSTs or sea ice coverage?
- Previous work has used Empirical Orthogonal Functions (or Principal Components Analysis) to identify patterns in low-frequency modes of atmospheric circulation, including identifying NAO, using 700hPa or 500hPa heights. One of the key steps in such work is understanding the physical meaning of the patterns identified. What advantages are there to using the SOM analysis relative to EOFs? Does each pattern represent a physically meaningful mode of circulation?
Minor questions:
- 30: Reducing snow cover on the ice but also physically changing the structure of snow and firn, the key is reducing the albedo.
- 42: Cloud radiative processes dictated by cloud height, cloud optical thickness, and hydrometeor phase are also important in driving the radiative budget.
- 52: The NAO is a redistribution of atmospheric mass between the subpolar and subtropical regions of the North Atlantic that one can capture the NAO using surface pressure data, as mentioned here, but can also be identified using geopotential height data.
- 64: Thermal advections will not necessarily follow lines of constant geopotential height, but will follow lines of constant thickness. To a first order, this is a reasonable interpretation, but it is important to be careful when interpreting the height pattern as showing advection.
- 65: It appears there is a missing word or phrase after “west in”.
- 65-66: As large accumulation and ablation events are often driven by cyclonic events, it would be worth saying more about how low-frequency circulation drives the storm track and the subsequent impact on mass balance.
- 114: Why not use a standard 30-year reference period, such as 1981-2010?
- 153: Is there any issue including a domain that reaches the pole?
- 180: Are the warmest 15% “abnormal”?
- 220-224: It’s not just the direction of flow that’s important, but the geopotential height is a function of the thickness of the lower troposphere, which is a function of the mean (virtual) temperature of that layer. If, like pattern 5, you are sitting under a trough, you would expect colder than normal weather. The patterns may also represent potential storm tracks, such as LSP6, which I expect would be more conducive to storms traveling up Baffin Bay and affecting the west coast of Greenland.
- 265: I really like this figure, especially the heat map and persistence plot! Very helpful to understand the patterns.
- 319-324: I had been wondering why 20CR3 was used instead of the ERA-20C product when the 20CR3 product was first mentioned. This should be discussed earlier in the manuscript.
- 349-350: It might be helpful to produce composite AT and SST anomalies (like Fig 2b,c) for each of the 8 patterns (overall and during each of the two periods studies).
Citation: https://doi.org/10.5194/egusphere-2024-4060-RC2 -
RC3: 'Comment on egusphere-2024-4060', Anonymous Referee #3, 12 Jun 2025
The manuscript addresses an important topic and is interesting and clearly written. However, it includes methodological issues related to the application of Self-Organizing Maps (SOM). These issues generate uncertainty in the attribution of the observed temperature changes in Greenland. Hence, I recommend major revisions to be made.
Major comments
1. The number of SOM nodes (cluster centres) is only eight. The authors give an argument that they preferred to avoid the need the carry out two SOM analyses, first using a larger number of nodes and then regrouping to smaller number. However, the regrouping is not necessarily needed, and the number of nodes should be large enough to allow robust conclusions. In the results obtained, some SOM nodes (particularly 1, 3 and 8) may result in either warm or cold conditions at the study site. This seems to be attributed to the role of local processes. I agree that local processes often have a strong impact on near-surface air temperatures, particularly in regions of complex orography. However, the non-systematic sign of the temperature anomaly for a single SOM node may also result from the fact the smaller is the number of nodes, the larger is the variability of circulation patters within the a single node. This makes it impossible to understand (based on the analyses made here) what is the relative importance of small-scale processes and the variations between large-scale patterns within a SOM node. I suggest performing a new SOM analysis with a larger number of nodes (e.g., 20 or even more).
2. The SOMs are grouped in a one-dimensional array. Although the plots in Figure 3 are not shown in a row, it is evident that the analysis is made for an 1D array. This is demonstrated by the fact that neighboring nodes (e.g., 1 and 4, 2 and 5, 3 and 6, 5 and 8) differ a lot from each other, which is not the case for a 2D SOM array. Using a 1D array is very unusual in the field of meteorology. In general, 2D SOM is preferred for most applications, in particular when it is relevant to visualize clusters or relationships, and the data does not have a natural linear ordering. A 1D SOM may be relevant for simple, linear datasets or specific ranking tasks, but it is less interpretable for general clustering or visualization. I suggest that the new SOM analysis will be made using a 2D array (e.g., 5 x 4 nodes).
3. The SOM analysis is made for a rather small area around Greenland (Figure 3). Even if certain circulation patters appear rather similar in the study area, fitting into a same SOM node, they may include considerable differences outside the boundaries of the study area. This may result in large differences in the transports of heat and moisture to Greenland. I cannot be sure if the effects on the results and conclusions of this study are large, but it is possible. Hence, I recommend that in the new SOM analyses the study area is enlarged.
4. The analysis addresses near-surface air temperatures with focus on coastal site at an elevation of 940 m. There the typical atmospheric pressure may be roughly 910–920 hPa. However, the analysis is made on the basis of 500 hPa geopotential height fields. These indeed well characterize the large-scale circulation patterns, but in a baroclinic atmosphere the wind vector at the 500 hPa may deviate a lot from that at the 910.920 hPa level. This should be discussed when making conclusions on the role heat advection associated with various SOM nodes. Over a flat surface, the effect of large-scale circulation on 2-m air temperature might be best analysed on the basis of 850-hPa fields but in Greenland this is naturally liable to errors. However, the authors could consider, if the SOM analysis would be better to do for 700 instead of 500 hPa fields.
5. I recommend carrying out seasonal analyses, as a certain geopotential height pattern may have very different effects on near-surface air temperatures in winter and summer.
6. I am not convinced if it is an optimal approach to focus on a single study site (WEG_L).
Minor comments
Lines 65-66: “cyclones following the North Atlantic Oscillation” is unclear expression.
Line 108: 6.5 K/1000 m
Line 217: Add southwest?
Line 381: Should the phase of Arctic Oscillation be somehow seen in changes in the occurrence of various SOM nodes?
Lines 398-400: Also the increase in CO2 concentration matters.
Line 403: globally?
In summary, based on my evaluation, the results obtained using the current methodology are not robust. However, I believe the study has significant potential to yield important results if more appropriate SOM analyses are conducted.
Citation: https://doi.org/10.5194/egusphere-2024-4060-RC3
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 407 | 88 | 20 | 515 | 21 | 38 |
- HTML: 407
- PDF: 88
- XML: 20
- Total: 515
- BibTeX: 21
- EndNote: 38
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1