the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Forcing and impact of the Northern Hemisphere continental snow cover in 1979–2014
Guillaume Gastineau
Claude Frankignoul
Yongqi Gao
Yu-Chiao Liang
Young-Oh Kwon
Annalisa Cherchi
Rohit Ghosh
Eliza Manzini
Daniela Matei
Jennifer Mecking
Lingling Suo
Tian Tian
Shuting Yang
Ying Zhang
Abstract. The role of surface ocean anomalies for the continental Northern Hemisphere snow cover is investigated, together with the interactions between snow cover and atmosphere. Four observational datasets and two large multi-model ensembles of atmosphere-only simulations are used, with prescribed sea surface temperature (SST) and sea ice concentration (SIC). A first ensemble uses observed interannually varying SST and SIC conditions for 1979–2014, while a second ensemble is identical except for SIC where a repeated climatological cycle is used.
SST and external forcing typically explain 10 to 25 % of the snow cover variance in model simulations, with a dominant forcing from the tropical and North Pacific SST, while no robust influence of the SIC is found. In observations, the Ural blocking is the main driver of the November and April snow cover over Eastern Eurasia, while the North Atlantic Oscillation (NAO) dominates the snow cover forcing in January. In November and more robustly in January, dipolar anomalies of snow cover over Eurasia, with positive anomalies over Europe and negative anomalies over Southern Siberia, also precede the Arctic Oscillation (AO) by one month. In models, snow cover over western Eurasia in January also precedes by one or two months a negative AO phase. The detailed outputs from one of the models suggest that both the western Eurasia snow cover and polar vortex are generated by Ural blocking, and that both snow cover and polar vortex anomalies act to generate the AO one or two months later.
-
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
(18344 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(18344 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Guillaume Gastineau et al.
Interactive discussion
Status: closed
-
CC1: 'Comment on egusphere-2022-939', Joaquin Munoz-Sabater, 23 Nov 2022
Please, note that the ERA5-Land citation is wrong. In the preprints it is cited as:
- Copernicus Climate Change Service (C3S): C3S ERA5-Land reanalysis. Copernicus Climate Change Service, April 9th 2020. https://cds.climate.copernicus.eu/cdsapp#!/home, 2019.
The correct citations are provided at the ERA5-Land online documentation: https://confluence.ecmwf.int/display/CKB/ERA5-Land%3A+data+documentation#ERA5Land:datadocumentation-HowtocitetheERA5-Landdataset
If using only monthly means then you need to use:
-Muñoz Sabater, J., (2019): ERA5-Land hourly data from 1981 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS). (Accessed on < DD-MMM-YYYY >), 10.24381/cds.e2161bac
and the corresponding scientific publication is:
-J. Muñoz-Sabater, Dutra, E., Agustí-Panareda, A., Albergel, C., Arduini, G., Balsamo, G., Boussetta, S., Choulga, M., Harrigan, S., Hersbach, H., Martens, B., Miralles, D. G., Piles, M., Rodríguez-Fernández, N. J., Zsoter, E., Buontempo, C., and Thépaut, J.-N.: ERA5-Land: A state-of-the-art global reanalysis dataset for land applications, Earth Syst. Sci. Data,13, 4349–4383, 2021. https://doi.org/10.5194/essd-13-4349-2021.
Please, kindly correct the citation.
Regards
Citation: https://doi.org/10.5194/egusphere-2022-939-CC1 -
CC2: 'Reply on CC1', Guillaume Gastineau, 05 Jan 2023
Dear Joaquin,
This will be corrected in the revised manuscript. I apologize for the inappropriate citation.
Thank you!
Citation: https://doi.org/10.5194/egusphere-2022-939-CC2
-
CC2: 'Reply on CC1', Guillaume Gastineau, 05 Jan 2023
-
RC1: 'Comment on egusphere-2022-939', Anonymous Referee #1, 14 Dec 2022
The paper uses an ensemble of AMIP simulations, from 8 different atmospheric models, to explore drivers of Northern Hemisphere (NH) snow cover trend/interannual variability, and its feedback on the winter atmospheric circulation, over 1979-2014. Two types of experiments are analyzed. ALL includes prescription of observed external forcings, sea surface temperature (SST) and sea ice concentration (SIC), while no-SIC is similar but with climatological SIC, such that the difference between ALL and SIC highlights the influence of observed SIC variability. Daily SST and SIC are prescribed to the atmospheric models.
The paper is well-written and clear. One of the main result is that SIC has little impact on both the trend and interannual variability in NH snow. This is convincingly demonstrated and described in the paper/abstract. Although internal variability dominates, the authors find that SST/external forcings have a greater influence than SIC in driving NH snow variability. The influence of snow on the atmospheric circulation and NH climate is then explored. No robust influence of November and April snow is found, but the authors identify a significant polar vortex anomaly (and associated surface climate response) lagging the 1st mode of variability of January snow cover (snow cover EOF1int) in the models. This connection is furthe explored singling out the LMDZOR6 model, for data availability consideration.
I think the paper is interesting and overall in good shape, but I have a few comments that may help improving the study. Please find them hereafter.
1) My first comment is about the observed trend in Eurasia/North America snow cover in November. The CanSISE observations you use stop in 2010, but it is notorious that snow cover in fall has had a tendency to increase since 2010, roughly. See the timeserie of snow cover from the Rutgers University Global Snow Lab for November, that shows the Eurasia and North America snow cover extent for the 1966-2022 period. There is a clear positive trend in both domains. This is for 1966-2022, but even over the 1979-2010 period that you use in your study, I do not see a decreasing trend as shown in your Fig. 5. Could you elaborate on this discrepancy? Is this due to uncertainties in observations, periods, both?
Link to the Rutgers Univ. snow extent timeseries:
https://climate.rutgers.edu/snowcover/chart_anom.php?ui_set=1&ui_region=nhland&ui_month=112)2) In link to my previous comment, section 3.4 shows there are large differences between the observed datasets in November. This is something that could be discussed further. Which dataset is more reliable in fall? NOAA-CDR, because it consists of direct satellite measurement of snow cover?
3) The most interesting result of the study, other than highlighting the limited impact of SIC on snow cover variability, is the potential feedback of January snow cover anomalies on polar vortex warming events (section 4.3). You find that snow cover EOF1int is preceded then followed by a significant weakening of the polar vortex, and you hypothesize that snow may act as a feedback in increasing the persistence, possibly amplitude, of the anomaly in the stratosphere. This is interesting but only speculation since no analyses are shown to demonstrate it. I think this is where the paper can be improved, and I have a few suggestions. In the analyses of LMDZOR6, you could select winters that exhibit persistent polar vortex weakening (similar to Fig. 15), and differentiate these events between those that also exhibit high snow cover EOF1int anomalies, and those that do not (composite analyses). This would be a way to verify whether snow cover EOF1int anomalies are indeed necessary to enhance the persistence and amplitude of the polar vortex warming. If possible, it would also be nice to see how snow cover EOF1int affects the stationary wave structure over Eurasia, and wave activity, that could potentially cause a higher persistence of polar warming anomalies. This section needs improvement to be more convincing.
Citation: https://doi.org/10.5194/egusphere-2022-939-RC1 -
AC1: 'Reply on RC1', Guillaume Gastineau, 23 Jan 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-939/egusphere-2022-939-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Guillaume Gastineau, 23 Jan 2023
-
RC2: 'Comment on egusphere-2022-939', Anonymous Referee #2, 06 Jan 2023
Review of “Forcing and impact of the Northern Hemisphere continental snow cover in 1979-2014” by Gastineau et al.
In this manuscript the authors analyze snow variability (primilarily snow cover variability) from two ensembles of land-atmosphere models, one forced by boundary conditions (radiative forcings, SSTs, and sea ice) that include sea ice variability and one which does not. The authors provide two key pieces of analysis. First they analyze which boundary conditions are able to influence snow cover variability including a demonstration of the limited extent to which sea ice variability can influence snow cover variability (an important null result) and a demonstration of how ocean SSTs are correlated with the dominant patterns of snow cover variability over the 1981-2014 period. Secondly, the authors provide an analysis of how snow cover variability is related to internal atmospheric variability. Comparisons to observations are provided where appropriate.
Overall the manuscript is clearly written with logically argued results. The use of multiple realizations (10-30 members) from a suite of 8 different land-atmosphere models helps to improve the robustness of results which have previously relied on ensembles of more limited size or a more limited range of models. While this study’s use of atmosphere-only simulations omits the influence of ocean and ice feedbacks, the analysis is an excellent baseline for future experiments using coupled models. My comments below are generally minor in nature.
Specific Comments:
Abstract: Please review the entirety of the abstract to make sure it's consistent with the results in the paper and as stated in the conclusions. E.g. It’s stated here that snow cover/PV anomalies act to generate the AO, which isn't consistent with the paper results or interpretations discussed within. The statement in your conclusions, that Ural blocking, which itself projects onto the AO, may drive important elements of both snow cover and PV variability, is consistent with my interpretation of your results.
L23: You analyze more than just the role of SST anomalies.
L30: I agree that topical/North Pacific SSTs are a dominant influence over the 1981-2014 period, but given the strong change in IPO signals over the period the statement may not apply more generally. Please restate to include the period of analysis.
L37: rephrase to be more specific. E.g. “snow cover variability across western Eurasia and an important contribution to polar vortex variability are both generated by Ural blocking”
L154: A snow density of 330 kg/m^3 is unreasonably high for a spatial and seasonal mean. A seasonal mean of 220-240 kg/m^3 would be more appropriate and would correct some of the over-estimated snow mass for the LMDZ6 and CMCC models shown in your Figure 1 and mentioned at line 249. Sturm et al (https://doi.org/10.1175/2010JHM1202.1) is the classic reference and the "test" data described there (Fig 1-2, Table 3) is consistent with a more recent analysis (https://doi.org/10.5194/tc-2022-227; in fact, Fig 3 of this recent analysis plots the spatially averaged snow density as a function of day of the year).
L141: This is a small point but changing the name of the second ensemble to ‘NoSICvar’ in the text and plots would read more accurately to me.
L155: The monthly binary fields resulting from this procedure would produce reasonable time series, but I don't know that the EOF patterns from these 4 models would have the same spatial variability as those based on monthly mean snow cover fraction (which would average over sub-monthly changes in the snow line). It might not matter in this analysis if the important part of the signal is just snow vs no-snow, but please check that the EOF_BC and EOF_SIC patterns (used in Fig 8) for models 1-4 (which use binary fields calculated from SWE output) are the same as those from models 5-8 (which use model-derived SCF).
L198: Do you use a convention to assign positive/negative values to the EOF patterns? For the patterns with multiple centers of action it’s not always clear how the EOFs relate to one another among the different plots. E.g. Fig 7d and Fig 10e.
L260: “simulate more snow cover… and too little snow cover over…”
L268-269: I think you meant to write that ECHAM6 and IAP4.1 both under-estimate snow water equivalent over Eurasia rather than overestimate.
L290: Please confirm you are using snow cover output from ERA5-Land, not ERA5 and correct in the text and figure captions.
Figure 5: The figure caption which specifies snow cover vs snow mass does not reflect the figure labels. Please confirm the labels are placed on the correct plot and correct the description if necessary. Also please change the plot so that the small grey circles in Jan and Apr are plotted on top of the other symbols and can be seen. The one in Apr is hard to see and I don't see it at all in Jan assuming it is underneath one of the other symbols. Same for the dark blue cross in NDJFM for plot c.
Figure 6: This is the first time they appear and it’s not clear what distinguishes the models marked with the * symbol (specified on Figure 11).
Figure 10: Please adjust the lower latitude limit of Figures 10b,c,f,g,j,k to match those in Figures 7,8,12,13 etc.
L 459: The similarity of the January snow cover variability patterns over Eurasia between Figs 7 and 10 suggests the NAO is the dominant source of variability over Eurasia in January rather than external forcings (as opposed to Nov and Apr where snow loss trends from external forcings are an important source of variability and alter the observed EOF1 patterns). This might be worth pointing out and commenting on here and in the discussion since it’s consistent with the strong influence of the NAO analyzed in the model simulations during January.
L480: I suggest being more nuanced about the maximum loading locations: in Nov it is in western Russia, in January it shifts towards eastern Europe, in Apr it shifts back eastwards to central Siberia. The dipoles of the EOF2 patterns seem to be positioned on the northwestern and southeastern ends of these EOF1 patterns.
L500: A little more summary/guidance for the reader would be helpful here. Maybe something like “The comparison between Fig. 10 for observations and Figs. 11-12 for models suggests that the models reproduce fairly well the main mode of variability found in observations. The NAO is the dominant mode of variability during January in both models and observations. During November and April, the dominant mode of variability found in the observations is a blocking pattern with a trough over the Ural region. This pattern also occurs in the model simulations but with less associated variance (it is reproduced in the EOF2 patterns rather than the EOF1 patterns). However the analysis of observations is based on…”
L500: I know the paper already includes a lot of analysis, but I presume the projections of SLP and temperature onto the observed PC2 time series either isn’t very interesting or doesn’t relate to the other modes of variability seen in the models?
L509: This claim that the SLP pattern in Fig 10c resembles the AO is not convincing. Nor do the models suggest anything of this sort in November in Fig 13a. I would accept that as for January the Nov-lagged pattern shares some similarities with the original Nov pattern and hence has somewhat persisted into the following month.
L539: Should this read “The comparison of Fig 10 with Figures 12 and 13….”?
L572: Fig 16b?
L578: Fig 16a,b?
L591: In the abstract, consider adding your conclusion that in uncoupled models sea ice loss drives a detectable but insignificant fraction of snow cover anomalies. I know it’s effectively a null result, but I think it’s still important to highlight.
L592: In the results you also state sea ice variability drives a small and insignificant fraction of snow mass anomalies, but don’t explicitly show the results.
L620: Please remove this claim unless further justified (see comment at line 509).
Technical Comments:
L199: “The first EOF analysis performed is based on the MMM calculated from the ALL experiment. The EOF pattern is denoted as EOF_BC, where….”
L208: “to highlight the effect of the SIC variability.”
L296: For clarity I suggest: “The overall impact of the sea-ice variations on the snow cover area and snow mass is limited, as shown by the differences between the MMM of ALL and NoSIC (Fig. 4e-h). Figures 4f-h have no clear trend and are not significantly related to observations...”
Figure 6: separatly -> separately
L619: In observations…
Citation: https://doi.org/10.5194/egusphere-2022-939-RC2 -
AC2: 'Reply on RC2', Guillaume Gastineau, 23 Jan 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-939/egusphere-2022-939-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Guillaume Gastineau, 23 Jan 2023
-
RC3: 'Comment on egusphere-2022-939', Anonymous Referee #3, 10 Jan 2023
In this article, the variability of the Eurasian snow cover and the North American snow cover (in terms of snow cover extent and snow water equivalent) are examined using several observational or observation-driven datasets and multi-model ensembles of atmosphere-only simulations, forced by either prescribed daily-varying sea ice and SST else daily-varying SST and climatological sea ice. The period investigated is 1979-2014. The interdependency of the continental snow cover and the atmospheric circulation is examined, including the feedback of the snow onto atmospheric circulation and the snow cover.
A detailed statistical analysis of the main modes of snow variability, as well as of trends, are carried out. The role of Ural blocking and of the NAO in driving the snow cover variability are examined, as well as the snow feedback on the circulation during specific months.
The paper is mostly well-written, although it is a bit long with a large number of figures. I find the paper acceptable for publication in The Cryosphere, providing that the comments below are addressed.
Major Comments
- That the Eurasian snow cover in autumn leads a negative phase of the (N)AO in winter was originally been proposed by Cohen and co-authors (cited). The idea that the snow cover exerts a (weak) feedback that reinforces a pre-existing negative (N)AO phase during winter, was proposed in Orsolini et al (2015), based on a case study contrasting a pair of coupled forecast ensembles where the snow-atmosphere feedback could be switched off (or at least scrambled at the initial time). This idea was further explored by Garfinkel et al (2020), using a suite of coupled (S2S) forecast models, who showed some transient feedback from snow cover onto the atmospheric circulation in the models with a better stratosphere.
It seems to me that one of the main findings of the current study is along the same lines, albeit using a different set of atmosphere-only ensemble of simulations. Namely, that a negative AO is not forced by snow as it arises from internal variability but, yet, is re-inforced or prolonged by the snow feedback (Fig 13), especially in January: is this correct interpretation? This could be stressed more clearly in the Abstract, and the appropriate references included.
- In their observational case study of the 2018 SSW, Lü Z. et al highlighted the potential role of the Siberian snow cover fluctuations in forcing planetary waves into the stratosphere, with the pulses of upward wave propagation preceded by snow increases in January-February by about a week (see their Figs 10-12), which modulate land-sea longitudinal temperature contrast over the Eurasian continent. Although such a lag is not a proof of causality, I wonder if this is consistent with the lagged effect on surface temperature/SLP highlighted here (Figs 13,15). A map of geopotential height in the stratosphere might be useful to complement Fig 15.
- Earlier studies of the snow-NAO linkage argued that the observed snow cover variability in the fall is underestimated by climate models. Here, model ensemble means are used, which damp the variability, but it would be of interest to document of actual range of snow variability in each model using all members, across the snow season.
- The role of the sea ice change on the continental-scale snow cover trend indeed appears small. Yet, it is interesting that there appears to be a regional effect in Western Russia during November (Fig 5) where there is some decrease downstream and south of the Barents-Kara seas. Could the authors comment on that?
- I believe that the CanCISE snow product is a multi-instrument/model product which comes with a measure of uncertainty. Would it be of interest to incorporate that “observational” uncertainty in some of the Figures (e.g. Figs 4)?
- The role of spring snow cover over the Tibetan Plateau Mongolian Plateau and its impact on the monsoons is alluded to on several occasions, with a reference to Barnett et al (1989). There has been a large body of literature on this topic since 1989, which is not mentioned. Since the paper focuses on continental Eurasia and North American snow cover in autumn and winter, and this precipitation and snow biases in models and re-analyses over this Tibet region are well documented elsewhere, the authors could skip this issue and keep the paper more focused.
- Garfinkel C.I, C. Schwartz, I. White and J. Rao (2020), Predictability of the early winter Arctic Oscillation from autumn Eurasian snowcover in subseasonal forecast models, Clim. Dyn., 5:961-974
- Orsolini, Y.J., Senan, R., Vitart, F., Weisheimer, A., Balsamo, G., Doblas-Reyes F., Influence of the Eurasian snow on the negative North Atlantic Oscillation in subseasonal forecasts of the cold winter 2009/10, Clim. Dyn., DOI: 10.1007/s00382-015-2903-8 (2015)
- Lü, Z., Li, F., Orsolini, Y. J., Gao, Y., & He, S. (2020). Understanding of European Cold Extremes, Sudden Stratospheric Warming, and Siberian Snow Accumulation in the Winter of 2017/18, Journal of Climate, 33(2), 527-545.
Minor comment:
- I find it confusing that, in Fig 4, the anomaly (ALL minus SIC, hence a small quantity), representing the potential role of the sea ice, is correlated with the full-field snow from ERA5-land. Wouldn’t it be clearer to show the relation to the snow from ERA5-land for each simulation ensemble separately, next to one another?
- The inset in Fig 4 should specifically mention ERA5-land, not be confused with ERA5 re-analyses, which assimilate snow observations.
Wording:
L23: The first sentence of the Abstract is a bit unclear.
L42: which aspect of “ecosystems”: management? Understanding the inner working of ecosystems?
Citation: https://doi.org/10.5194/egusphere-2022-939-RC3 -
AC3: 'Reply on RC3', Guillaume Gastineau, 23 Jan 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-939/egusphere-2022-939-AC3-supplement.pdf
Interactive discussion
Status: closed
-
CC1: 'Comment on egusphere-2022-939', Joaquin Munoz-Sabater, 23 Nov 2022
Please, note that the ERA5-Land citation is wrong. In the preprints it is cited as:
- Copernicus Climate Change Service (C3S): C3S ERA5-Land reanalysis. Copernicus Climate Change Service, April 9th 2020. https://cds.climate.copernicus.eu/cdsapp#!/home, 2019.
The correct citations are provided at the ERA5-Land online documentation: https://confluence.ecmwf.int/display/CKB/ERA5-Land%3A+data+documentation#ERA5Land:datadocumentation-HowtocitetheERA5-Landdataset
If using only monthly means then you need to use:
-Muñoz Sabater, J., (2019): ERA5-Land hourly data from 1981 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS). (Accessed on < DD-MMM-YYYY >), 10.24381/cds.e2161bac
and the corresponding scientific publication is:
-J. Muñoz-Sabater, Dutra, E., Agustí-Panareda, A., Albergel, C., Arduini, G., Balsamo, G., Boussetta, S., Choulga, M., Harrigan, S., Hersbach, H., Martens, B., Miralles, D. G., Piles, M., Rodríguez-Fernández, N. J., Zsoter, E., Buontempo, C., and Thépaut, J.-N.: ERA5-Land: A state-of-the-art global reanalysis dataset for land applications, Earth Syst. Sci. Data,13, 4349–4383, 2021. https://doi.org/10.5194/essd-13-4349-2021.
Please, kindly correct the citation.
Regards
Citation: https://doi.org/10.5194/egusphere-2022-939-CC1 -
CC2: 'Reply on CC1', Guillaume Gastineau, 05 Jan 2023
Dear Joaquin,
This will be corrected in the revised manuscript. I apologize for the inappropriate citation.
Thank you!
Citation: https://doi.org/10.5194/egusphere-2022-939-CC2
-
CC2: 'Reply on CC1', Guillaume Gastineau, 05 Jan 2023
-
RC1: 'Comment on egusphere-2022-939', Anonymous Referee #1, 14 Dec 2022
The paper uses an ensemble of AMIP simulations, from 8 different atmospheric models, to explore drivers of Northern Hemisphere (NH) snow cover trend/interannual variability, and its feedback on the winter atmospheric circulation, over 1979-2014. Two types of experiments are analyzed. ALL includes prescription of observed external forcings, sea surface temperature (SST) and sea ice concentration (SIC), while no-SIC is similar but with climatological SIC, such that the difference between ALL and SIC highlights the influence of observed SIC variability. Daily SST and SIC are prescribed to the atmospheric models.
The paper is well-written and clear. One of the main result is that SIC has little impact on both the trend and interannual variability in NH snow. This is convincingly demonstrated and described in the paper/abstract. Although internal variability dominates, the authors find that SST/external forcings have a greater influence than SIC in driving NH snow variability. The influence of snow on the atmospheric circulation and NH climate is then explored. No robust influence of November and April snow is found, but the authors identify a significant polar vortex anomaly (and associated surface climate response) lagging the 1st mode of variability of January snow cover (snow cover EOF1int) in the models. This connection is furthe explored singling out the LMDZOR6 model, for data availability consideration.
I think the paper is interesting and overall in good shape, but I have a few comments that may help improving the study. Please find them hereafter.
1) My first comment is about the observed trend in Eurasia/North America snow cover in November. The CanSISE observations you use stop in 2010, but it is notorious that snow cover in fall has had a tendency to increase since 2010, roughly. See the timeserie of snow cover from the Rutgers University Global Snow Lab for November, that shows the Eurasia and North America snow cover extent for the 1966-2022 period. There is a clear positive trend in both domains. This is for 1966-2022, but even over the 1979-2010 period that you use in your study, I do not see a decreasing trend as shown in your Fig. 5. Could you elaborate on this discrepancy? Is this due to uncertainties in observations, periods, both?
Link to the Rutgers Univ. snow extent timeseries:
https://climate.rutgers.edu/snowcover/chart_anom.php?ui_set=1&ui_region=nhland&ui_month=112)2) In link to my previous comment, section 3.4 shows there are large differences between the observed datasets in November. This is something that could be discussed further. Which dataset is more reliable in fall? NOAA-CDR, because it consists of direct satellite measurement of snow cover?
3) The most interesting result of the study, other than highlighting the limited impact of SIC on snow cover variability, is the potential feedback of January snow cover anomalies on polar vortex warming events (section 4.3). You find that snow cover EOF1int is preceded then followed by a significant weakening of the polar vortex, and you hypothesize that snow may act as a feedback in increasing the persistence, possibly amplitude, of the anomaly in the stratosphere. This is interesting but only speculation since no analyses are shown to demonstrate it. I think this is where the paper can be improved, and I have a few suggestions. In the analyses of LMDZOR6, you could select winters that exhibit persistent polar vortex weakening (similar to Fig. 15), and differentiate these events between those that also exhibit high snow cover EOF1int anomalies, and those that do not (composite analyses). This would be a way to verify whether snow cover EOF1int anomalies are indeed necessary to enhance the persistence and amplitude of the polar vortex warming. If possible, it would also be nice to see how snow cover EOF1int affects the stationary wave structure over Eurasia, and wave activity, that could potentially cause a higher persistence of polar warming anomalies. This section needs improvement to be more convincing.
Citation: https://doi.org/10.5194/egusphere-2022-939-RC1 -
AC1: 'Reply on RC1', Guillaume Gastineau, 23 Jan 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-939/egusphere-2022-939-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Guillaume Gastineau, 23 Jan 2023
-
RC2: 'Comment on egusphere-2022-939', Anonymous Referee #2, 06 Jan 2023
Review of “Forcing and impact of the Northern Hemisphere continental snow cover in 1979-2014” by Gastineau et al.
In this manuscript the authors analyze snow variability (primilarily snow cover variability) from two ensembles of land-atmosphere models, one forced by boundary conditions (radiative forcings, SSTs, and sea ice) that include sea ice variability and one which does not. The authors provide two key pieces of analysis. First they analyze which boundary conditions are able to influence snow cover variability including a demonstration of the limited extent to which sea ice variability can influence snow cover variability (an important null result) and a demonstration of how ocean SSTs are correlated with the dominant patterns of snow cover variability over the 1981-2014 period. Secondly, the authors provide an analysis of how snow cover variability is related to internal atmospheric variability. Comparisons to observations are provided where appropriate.
Overall the manuscript is clearly written with logically argued results. The use of multiple realizations (10-30 members) from a suite of 8 different land-atmosphere models helps to improve the robustness of results which have previously relied on ensembles of more limited size or a more limited range of models. While this study’s use of atmosphere-only simulations omits the influence of ocean and ice feedbacks, the analysis is an excellent baseline for future experiments using coupled models. My comments below are generally minor in nature.
Specific Comments:
Abstract: Please review the entirety of the abstract to make sure it's consistent with the results in the paper and as stated in the conclusions. E.g. It’s stated here that snow cover/PV anomalies act to generate the AO, which isn't consistent with the paper results or interpretations discussed within. The statement in your conclusions, that Ural blocking, which itself projects onto the AO, may drive important elements of both snow cover and PV variability, is consistent with my interpretation of your results.
L23: You analyze more than just the role of SST anomalies.
L30: I agree that topical/North Pacific SSTs are a dominant influence over the 1981-2014 period, but given the strong change in IPO signals over the period the statement may not apply more generally. Please restate to include the period of analysis.
L37: rephrase to be more specific. E.g. “snow cover variability across western Eurasia and an important contribution to polar vortex variability are both generated by Ural blocking”
L154: A snow density of 330 kg/m^3 is unreasonably high for a spatial and seasonal mean. A seasonal mean of 220-240 kg/m^3 would be more appropriate and would correct some of the over-estimated snow mass for the LMDZ6 and CMCC models shown in your Figure 1 and mentioned at line 249. Sturm et al (https://doi.org/10.1175/2010JHM1202.1) is the classic reference and the "test" data described there (Fig 1-2, Table 3) is consistent with a more recent analysis (https://doi.org/10.5194/tc-2022-227; in fact, Fig 3 of this recent analysis plots the spatially averaged snow density as a function of day of the year).
L141: This is a small point but changing the name of the second ensemble to ‘NoSICvar’ in the text and plots would read more accurately to me.
L155: The monthly binary fields resulting from this procedure would produce reasonable time series, but I don't know that the EOF patterns from these 4 models would have the same spatial variability as those based on monthly mean snow cover fraction (which would average over sub-monthly changes in the snow line). It might not matter in this analysis if the important part of the signal is just snow vs no-snow, but please check that the EOF_BC and EOF_SIC patterns (used in Fig 8) for models 1-4 (which use binary fields calculated from SWE output) are the same as those from models 5-8 (which use model-derived SCF).
L198: Do you use a convention to assign positive/negative values to the EOF patterns? For the patterns with multiple centers of action it’s not always clear how the EOFs relate to one another among the different plots. E.g. Fig 7d and Fig 10e.
L260: “simulate more snow cover… and too little snow cover over…”
L268-269: I think you meant to write that ECHAM6 and IAP4.1 both under-estimate snow water equivalent over Eurasia rather than overestimate.
L290: Please confirm you are using snow cover output from ERA5-Land, not ERA5 and correct in the text and figure captions.
Figure 5: The figure caption which specifies snow cover vs snow mass does not reflect the figure labels. Please confirm the labels are placed on the correct plot and correct the description if necessary. Also please change the plot so that the small grey circles in Jan and Apr are plotted on top of the other symbols and can be seen. The one in Apr is hard to see and I don't see it at all in Jan assuming it is underneath one of the other symbols. Same for the dark blue cross in NDJFM for plot c.
Figure 6: This is the first time they appear and it’s not clear what distinguishes the models marked with the * symbol (specified on Figure 11).
Figure 10: Please adjust the lower latitude limit of Figures 10b,c,f,g,j,k to match those in Figures 7,8,12,13 etc.
L 459: The similarity of the January snow cover variability patterns over Eurasia between Figs 7 and 10 suggests the NAO is the dominant source of variability over Eurasia in January rather than external forcings (as opposed to Nov and Apr where snow loss trends from external forcings are an important source of variability and alter the observed EOF1 patterns). This might be worth pointing out and commenting on here and in the discussion since it’s consistent with the strong influence of the NAO analyzed in the model simulations during January.
L480: I suggest being more nuanced about the maximum loading locations: in Nov it is in western Russia, in January it shifts towards eastern Europe, in Apr it shifts back eastwards to central Siberia. The dipoles of the EOF2 patterns seem to be positioned on the northwestern and southeastern ends of these EOF1 patterns.
L500: A little more summary/guidance for the reader would be helpful here. Maybe something like “The comparison between Fig. 10 for observations and Figs. 11-12 for models suggests that the models reproduce fairly well the main mode of variability found in observations. The NAO is the dominant mode of variability during January in both models and observations. During November and April, the dominant mode of variability found in the observations is a blocking pattern with a trough over the Ural region. This pattern also occurs in the model simulations but with less associated variance (it is reproduced in the EOF2 patterns rather than the EOF1 patterns). However the analysis of observations is based on…”
L500: I know the paper already includes a lot of analysis, but I presume the projections of SLP and temperature onto the observed PC2 time series either isn’t very interesting or doesn’t relate to the other modes of variability seen in the models?
L509: This claim that the SLP pattern in Fig 10c resembles the AO is not convincing. Nor do the models suggest anything of this sort in November in Fig 13a. I would accept that as for January the Nov-lagged pattern shares some similarities with the original Nov pattern and hence has somewhat persisted into the following month.
L539: Should this read “The comparison of Fig 10 with Figures 12 and 13….”?
L572: Fig 16b?
L578: Fig 16a,b?
L591: In the abstract, consider adding your conclusion that in uncoupled models sea ice loss drives a detectable but insignificant fraction of snow cover anomalies. I know it’s effectively a null result, but I think it’s still important to highlight.
L592: In the results you also state sea ice variability drives a small and insignificant fraction of snow mass anomalies, but don’t explicitly show the results.
L620: Please remove this claim unless further justified (see comment at line 509).
Technical Comments:
L199: “The first EOF analysis performed is based on the MMM calculated from the ALL experiment. The EOF pattern is denoted as EOF_BC, where….”
L208: “to highlight the effect of the SIC variability.”
L296: For clarity I suggest: “The overall impact of the sea-ice variations on the snow cover area and snow mass is limited, as shown by the differences between the MMM of ALL and NoSIC (Fig. 4e-h). Figures 4f-h have no clear trend and are not significantly related to observations...”
Figure 6: separatly -> separately
L619: In observations…
Citation: https://doi.org/10.5194/egusphere-2022-939-RC2 -
AC2: 'Reply on RC2', Guillaume Gastineau, 23 Jan 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-939/egusphere-2022-939-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Guillaume Gastineau, 23 Jan 2023
-
RC3: 'Comment on egusphere-2022-939', Anonymous Referee #3, 10 Jan 2023
In this article, the variability of the Eurasian snow cover and the North American snow cover (in terms of snow cover extent and snow water equivalent) are examined using several observational or observation-driven datasets and multi-model ensembles of atmosphere-only simulations, forced by either prescribed daily-varying sea ice and SST else daily-varying SST and climatological sea ice. The period investigated is 1979-2014. The interdependency of the continental snow cover and the atmospheric circulation is examined, including the feedback of the snow onto atmospheric circulation and the snow cover.
A detailed statistical analysis of the main modes of snow variability, as well as of trends, are carried out. The role of Ural blocking and of the NAO in driving the snow cover variability are examined, as well as the snow feedback on the circulation during specific months.
The paper is mostly well-written, although it is a bit long with a large number of figures. I find the paper acceptable for publication in The Cryosphere, providing that the comments below are addressed.
Major Comments
- That the Eurasian snow cover in autumn leads a negative phase of the (N)AO in winter was originally been proposed by Cohen and co-authors (cited). The idea that the snow cover exerts a (weak) feedback that reinforces a pre-existing negative (N)AO phase during winter, was proposed in Orsolini et al (2015), based on a case study contrasting a pair of coupled forecast ensembles where the snow-atmosphere feedback could be switched off (or at least scrambled at the initial time). This idea was further explored by Garfinkel et al (2020), using a suite of coupled (S2S) forecast models, who showed some transient feedback from snow cover onto the atmospheric circulation in the models with a better stratosphere.
It seems to me that one of the main findings of the current study is along the same lines, albeit using a different set of atmosphere-only ensemble of simulations. Namely, that a negative AO is not forced by snow as it arises from internal variability but, yet, is re-inforced or prolonged by the snow feedback (Fig 13), especially in January: is this correct interpretation? This could be stressed more clearly in the Abstract, and the appropriate references included.
- In their observational case study of the 2018 SSW, Lü Z. et al highlighted the potential role of the Siberian snow cover fluctuations in forcing planetary waves into the stratosphere, with the pulses of upward wave propagation preceded by snow increases in January-February by about a week (see their Figs 10-12), which modulate land-sea longitudinal temperature contrast over the Eurasian continent. Although such a lag is not a proof of causality, I wonder if this is consistent with the lagged effect on surface temperature/SLP highlighted here (Figs 13,15). A map of geopotential height in the stratosphere might be useful to complement Fig 15.
- Earlier studies of the snow-NAO linkage argued that the observed snow cover variability in the fall is underestimated by climate models. Here, model ensemble means are used, which damp the variability, but it would be of interest to document of actual range of snow variability in each model using all members, across the snow season.
- The role of the sea ice change on the continental-scale snow cover trend indeed appears small. Yet, it is interesting that there appears to be a regional effect in Western Russia during November (Fig 5) where there is some decrease downstream and south of the Barents-Kara seas. Could the authors comment on that?
- I believe that the CanCISE snow product is a multi-instrument/model product which comes with a measure of uncertainty. Would it be of interest to incorporate that “observational” uncertainty in some of the Figures (e.g. Figs 4)?
- The role of spring snow cover over the Tibetan Plateau Mongolian Plateau and its impact on the monsoons is alluded to on several occasions, with a reference to Barnett et al (1989). There has been a large body of literature on this topic since 1989, which is not mentioned. Since the paper focuses on continental Eurasia and North American snow cover in autumn and winter, and this precipitation and snow biases in models and re-analyses over this Tibet region are well documented elsewhere, the authors could skip this issue and keep the paper more focused.
- Garfinkel C.I, C. Schwartz, I. White and J. Rao (2020), Predictability of the early winter Arctic Oscillation from autumn Eurasian snowcover in subseasonal forecast models, Clim. Dyn., 5:961-974
- Orsolini, Y.J., Senan, R., Vitart, F., Weisheimer, A., Balsamo, G., Doblas-Reyes F., Influence of the Eurasian snow on the negative North Atlantic Oscillation in subseasonal forecasts of the cold winter 2009/10, Clim. Dyn., DOI: 10.1007/s00382-015-2903-8 (2015)
- Lü, Z., Li, F., Orsolini, Y. J., Gao, Y., & He, S. (2020). Understanding of European Cold Extremes, Sudden Stratospheric Warming, and Siberian Snow Accumulation in the Winter of 2017/18, Journal of Climate, 33(2), 527-545.
Minor comment:
- I find it confusing that, in Fig 4, the anomaly (ALL minus SIC, hence a small quantity), representing the potential role of the sea ice, is correlated with the full-field snow from ERA5-land. Wouldn’t it be clearer to show the relation to the snow from ERA5-land for each simulation ensemble separately, next to one another?
- The inset in Fig 4 should specifically mention ERA5-land, not be confused with ERA5 re-analyses, which assimilate snow observations.
Wording:
L23: The first sentence of the Abstract is a bit unclear.
L42: which aspect of “ecosystems”: management? Understanding the inner working of ecosystems?
Citation: https://doi.org/10.5194/egusphere-2022-939-RC3 -
AC3: 'Reply on RC3', Guillaume Gastineau, 23 Jan 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-939/egusphere-2022-939-AC3-supplement.pdf
Peer review completion
Journal article(s) based on this preprint
Guillaume Gastineau et al.
Guillaume Gastineau et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 391 | 149 | 25 | 565 | 8 | 6 |
- HTML: 391
- PDF: 149
- XML: 25
- Total: 565
- BibTeX: 8
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(18344 KB) - Metadata XML