the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Benefits and challenges of dynamic sea-ice for weather forecasts
- 1European Centre for Medium Range Weather Forecasts, Shinfield Park, Reading, RG2 9AX, United Kingdom
- 2European Centre for Medium Range Weather Forecasts, Robert-Schuman-Platz 3, 53175 Bonn, Germany
- 1European Centre for Medium Range Weather Forecasts, Shinfield Park, Reading, RG2 9AX, United Kingdom
- 2European Centre for Medium Range Weather Forecasts, Robert-Schuman-Platz 3, 53175 Bonn, Germany
Abstract. The drive to develop environmental prediction systems that are seamless across both weather and climate timescales has culminated in the development and use of Earth system models, which include a coupled representation of the atmosphere, land, ocean and sea ice, for medium-range weather forecasts. One region where such a coupled Earth system approach has the potential to significantly influence the skill of weather forecasts is in the polar and sub-polar seas, where fluxes of heat, moisture and momentum are strongly influenced by the position of the sea ice edge. In this study we demonstrate that using a dynamically coupled ocean and sea ice model in ECMWF Integrated Forecasting System, results in improved sea ice edge position forecasts in the northern hemisphere in the medium-range. Further, this improves forecasts of boundary layer temperature and humidity downstream of the sea ice edge in some regions during periods of rapid change in the sea ice compared to forecasts in which the sea surface temperature anomalies and sea ice concentration do not evolve throughout the forecasts. Challenges and limitations, such as the quality of ocean and sea ice initial conditions or analyses, and the inability of the coupled system to capture the rate of sea ice concentration change during periods of ice advance and retreat will also be discussed.
- Preprint
(2815 KB) -
Supplement
(1192 KB) - BibTeX
- EndNote
Jonathan Day et al.
Status: closed
-
RC1: 'Comment on wcd-2022-5', Anonymous Referee #1, 25 Feb 2022
Review of “Benefits and challenges of dynamic sea-ice for weather forecasts” by Day et al.
Weather and Climate Dynamics
The ECMWF have recently introduced a dynamic sea-ice model (as part of their coupled atmosphere-ocean dynamical models) into their operational forecasting suites for their medium-range forecasts. This study evaluates one northern hemisphere winter of 10-day forecasts using three experimental configurations: (i) An atmospheric model forecast using persistent SST and sea ice surface condition; (ii) An atmospheric model forecast using ‘observed’ (i.e., daily updated) SST and sea ice surface condition; (iii) A coupled forecast. (i) is essentially the previous operational configuration, (ii) provides a sort of upper limit on forecast quality (assuming perfect observations) and (iii) is the new operational configuration so including the dynamic sea ice. The results are largely positive, but there are some limitations in the findings and in the ability to evaluate the results that are also discussed.
I thought this was a really nice study, one of the first to evaluate the benefits of dynamic sea-ice on weather forecasts for an extended period. The paper is concise, well written, generally placed into context well and illustrated with high quality and appropriate figures. I have a few specific concerns, primarily around caveating the largely positive benefits of this step forward in forecasting, and some minor comments.
General Comments
(1) The overall benefits of a dynamic sea ice are clearly evident and are nicely illustrated in Figures 1, 2 (which is a striking illustration) and 4. However, these general (and seasonally averaged) plots do show some caveats. Fig 4 illustrates that over the first day, the persistent forecast has lower IIEE than the coupled forecast for the northern hemisphere, and that this is always true in the Labrador Sea region, while in the Sea of Okhotsk it is true until around day 6. I suspect the reason persistence is better for these seas is that they are relatively small and enclosed regions, with sea-ice that advances/retreats when the winds are along the sea, thus pick out the advancing/retreating problems discussed later. These findings are noted briefly in section 3.1 (e.g., L90-95), but I think further discussion is really needed in section 3.1. This is explored a bit in section 3.2, where I think Fig 6 is used to explain that IIEE and changes in ice concentration are related (especially so for the more enclosed seas), but this is not very well linked back to the key figures of 2 and 4. I suggest the authors work on improving the links between Figs 2-4 and Fig 6-7 and explaining the different qualitative results of Fig 4.
(2) An interesting fact is noted with regard to Fig 4, that the ‘initialisation error (IIEE)’ is approximately half of the final IIEE error at day 10. This is rightly mentioned (L100) but this striking fact is not discussed further in Section 4 or the abstract. The authors note this is related to initialisation challenges and the use of only weakly coupled data assimilation. I know this is also a problem at other centres and is likely to be an issue for a number of years for coupled forecasts. I wonder if this finding should receive more prominence in the paper.
(3) The other related issue, which is briefly mentioned, is the veracity of the sea-ice analysis. The authors point out there are uncertainties in the sea-ice analyses and this will affect initialisation and the size of the errors (P10, L325) and that “guidance … from the remote sensing community” is needed. I agree here and I would perhaps suggest this limitation is added to the abstract. At present the last two lines of the abstract are a bit vague. It might be worth expanding these to state explicitly that the quality of satellite sea-ice products on daily to weekly timescales and on meso-scales (<500 km say) are not well characterised and this is a limitation for NWP.
Specific Comments
L42 – there is another recent idealised modelling study on the atmospheric response to sea-ice geometry and concentration that should be cited here:
Spensberger, C., & Spengler, T. (2021). Sensitivity of air-sea heat exchange in
cold-air outbreaks to model resolution and sea-ice distribution. Journal of
Geophysical Research: Atmospheres, 126, e2020JD033610. https://doi.
org/10.1029/2020JD033610
L49 – I am not an expert on the timelines here, but are you sure that ECMWF developed the first coupled global … system? Maybe for an ensemble? Not sure about NWP more generally. The Canadian ECCC have had a coupled forecast model for some time and this may pre-date the ECMWF development. You cite one paper for the Canadian system (Smith et al. 2018), but you should probably also cite earlier pioneering work that demonstrated the potential for improvement in atmospheric forecasts from such a coupled system withi NWP.
Pellerin P, Ritchie H, Saucier SJ, Roy F, Desjardins S, Valin M, Lee V. 2004.Impact of a two-way coupling between an atmospheric and an ocean – icemodel over the Gulf of St. Lawrence.Mon. Weather Rev.132: 1379 – 1398
Smith GC, Roy F, Brasnett B. 2013. Evaluation of an operational ice-ocean analysis and forecasting
system for the Gulf of St Lawrence. Q. J. R. Meteorol. Soc. 139: 419–433. DOI:10.1002/qj.1982
Smith, G.C., Roy, F., Reszka, M., Surcel Colan, D., He, Z., Deacu, D., Belanger, J.M., Skachko, S., Liu, Y., Dupont, F. and Lemieux, J.F., 2016. Sea ice forecast verification in the Canadian global ice ocean prediction system. Quarterly Journal of the Royal Meteorological Society, 142(695), pp.659-671.
L79-80 – I was slightly confused on reading the explanation for the persisted surface conditions for the first time, because ‘an anomaly is added each day’. On second reading I think this anomaly is only for the SST (not the sea ice)? Perhaps check for clarity here.
L128 – I think the Hersbach ERA5 reference is missing.
L142 – I’d replace “Atlantic coast’ with Labrador Sea coast, as it isn’t the main Atlantic basin.
L231 – ‘that region’ – it is unclear from this paragraph which region you are talking about. Maybe these lines should be merged into the previous paragraph?
L250-265 – this paragraph on internal boundary layer development at the ice edge is unreferenced – you could cite the Spensberger and Spengler 2021 paper here or the idealised 2D model of this internal BL development which also uses observations in
Renfrew, I.A. and King, J.C., 2000. A simple model of the convective internal boundary layer and its application to surface heat flux estimates within polynyas. Boundary-layer meteorology, 94(3), pp.335-356.
L285 – It was useful context to point out the differences in specific humidity (in g/kg and that this was 10% of the total value). You could also have expressed this as % of the standard deviation of this variable or something? And done similar for the difference in temperature. I think it useful to have an idea of the magnitude of these forecast differences in the context of day to day variability. If you can easily do such a metric? This is just a suggestion, not necessary.
L307 – The final section is more of a “Conclusions and Discussion” section.
L355 “weakly” not weekly.
Figures
Figure 1 – I would recommend changing the colour scale to one with white in the middle. At present the whole North Atlantic (which has no sea ice) is pink. It looks odd!
Fig 2 – the font size of the labels and legend is too small to read. Nice figure though!
Fig 3 & 2 – would it make sense to try and have the same colour for Ocean5 in these figures – this is red in 2 and 4 but green in fig 3.
Fig 5 – these figures illustrate the large variability between forecasts. Fig 5 is only very briefly mentioned in section 3.2 – I wonder if you should add a sentence or two emphasising the large variability.
-
RC2: 'Comment on wcd-2022-5', Anonymous Referee #2, 09 Mar 2022
I enjoyed reading this article very much. The results are relevant for the scientific community, numerical weather prediction centers, and forecast users. The forecast verification process is supported by a solid and sophisticated methodological base, and the forecast improvements and deficiencies are honestly highlighted without over or understating the findings. Furthermore, the manuscript is well written, and the figures illustrate the outcome of the study appropriately. I include below a few remarks and suggestions, which are mostly minor, and I hope that these will help the authors in the revision process. I recommend the publication of with manuscript once those (minor) points are addressed.
GENERAL COMMENTS
The Introduction and Method sections provide a very good overview of the system. However, I think some details on the probabilistic nature of the forecast are missing. Underlining the higher compatibility of a coupled model configuration with the ECMWF ensemble forecast system (i.e. the ice can evolve independently in each ensemble member, unlike in the persistence based strategy) would represent a nice addition to the study.
What about the melting season? I expect the impact of the sea ice on the ocean and land weather to be limited in summer because of the milder temperature gradients and winds. However, the demand for good ice forecasts might peak during this season. I would briefly mention whether the features of the dynamical system are appropriate also for the summer months. I also think a brief reference to what happens to the Southern Ocean sea ice might be appropriate.
The study focuses on a single winter season. Given the large number of forecasts analyzed, I expect the results to be solid. However, I think it might be good adding a characterization of the sea ice state during that winter in comparison to the climatological state, and discussing whether the results might be sensitive/influenced by potentially anomalous conditions (e.g. fast ice drift, abrupt melting events, etc.)
Figure 4 clearly shows that OCEAN5 reanalysis is biased, and you describe this well in the text. However, I think giving some more context on the origin of this bias would be helpful for the readers.
The fact that the thickness is not coupled implies that the thermodynamical transition at the ice edge is probably not well simulated also by the current dynamical system. Could you quantify the impact of this on the evolution of the internal boundary layer? Is the effect of a progressively reduced thickness towards the marginal ice zone negligible compared to the reduction in concentration? I expect this would also changes with the progressing of the freezing season. I think some more details on this in the discussion/conclusion section would be interesting for the reader.
I would like to point out that the using the AMSR2 derived sea ice concentration has also some drawbacks. It certainly comes with a desirable higher resolution because it uses higher frequencies. However, the effect of clouds on the microwave signal at higher frequencies is substantial and can penalize the quality of the retrieval, particularly across the marginal ice zone where clouds are not uncommon.
SPECIFIC COMMENTS
Line 71: I would not consider obs-SSTSIC a real forecast but rather an hindcast or an AMIP type simulation.
Line 75: I think it is worth mentioning here that the sea ice description of OSTIA comes from OSI-SAF. You mention this later in the result section, but I think stating this here would help the reader to understand the verification method.
Line 94: I suggest expressing the typical Arctic resolution of the ORCA025 grid also in km.
Line 96: Do you mean “…that is coupled to the atmosphere is…”?
FIGURES
Fig. 1: I suggest using a different colormap with the white color centered on zero. I don’t like seeing the rectangular domain of the polar stereographic grid in pink. Using red and green for the boxes might not be color friendly.
Fig. 2: Labels and titles are too small. I think it is ok to lose the outlier point in the OSI-SAF timeseries (plots a and c), probably caused by a partial observational coverage on that day.
Fig. 4 and 6: Labels are too small, and I suggest using the scientific notation to improve the readability of the plots.
-
AC1: 'Comment on wcd-2022-5', Jonathan Day, 29 Apr 2022
The comment was uploaded in the form of a supplement: https://wcd.copernicus.org/preprints/wcd-2022-5/wcd-2022-5-AC1-supplement.pdf
Status: closed
-
RC1: 'Comment on wcd-2022-5', Anonymous Referee #1, 25 Feb 2022
Review of “Benefits and challenges of dynamic sea-ice for weather forecasts” by Day et al.
Weather and Climate Dynamics
The ECMWF have recently introduced a dynamic sea-ice model (as part of their coupled atmosphere-ocean dynamical models) into their operational forecasting suites for their medium-range forecasts. This study evaluates one northern hemisphere winter of 10-day forecasts using three experimental configurations: (i) An atmospheric model forecast using persistent SST and sea ice surface condition; (ii) An atmospheric model forecast using ‘observed’ (i.e., daily updated) SST and sea ice surface condition; (iii) A coupled forecast. (i) is essentially the previous operational configuration, (ii) provides a sort of upper limit on forecast quality (assuming perfect observations) and (iii) is the new operational configuration so including the dynamic sea ice. The results are largely positive, but there are some limitations in the findings and in the ability to evaluate the results that are also discussed.
I thought this was a really nice study, one of the first to evaluate the benefits of dynamic sea-ice on weather forecasts for an extended period. The paper is concise, well written, generally placed into context well and illustrated with high quality and appropriate figures. I have a few specific concerns, primarily around caveating the largely positive benefits of this step forward in forecasting, and some minor comments.
General Comments
(1) The overall benefits of a dynamic sea ice are clearly evident and are nicely illustrated in Figures 1, 2 (which is a striking illustration) and 4. However, these general (and seasonally averaged) plots do show some caveats. Fig 4 illustrates that over the first day, the persistent forecast has lower IIEE than the coupled forecast for the northern hemisphere, and that this is always true in the Labrador Sea region, while in the Sea of Okhotsk it is true until around day 6. I suspect the reason persistence is better for these seas is that they are relatively small and enclosed regions, with sea-ice that advances/retreats when the winds are along the sea, thus pick out the advancing/retreating problems discussed later. These findings are noted briefly in section 3.1 (e.g., L90-95), but I think further discussion is really needed in section 3.1. This is explored a bit in section 3.2, where I think Fig 6 is used to explain that IIEE and changes in ice concentration are related (especially so for the more enclosed seas), but this is not very well linked back to the key figures of 2 and 4. I suggest the authors work on improving the links between Figs 2-4 and Fig 6-7 and explaining the different qualitative results of Fig 4.
(2) An interesting fact is noted with regard to Fig 4, that the ‘initialisation error (IIEE)’ is approximately half of the final IIEE error at day 10. This is rightly mentioned (L100) but this striking fact is not discussed further in Section 4 or the abstract. The authors note this is related to initialisation challenges and the use of only weakly coupled data assimilation. I know this is also a problem at other centres and is likely to be an issue for a number of years for coupled forecasts. I wonder if this finding should receive more prominence in the paper.
(3) The other related issue, which is briefly mentioned, is the veracity of the sea-ice analysis. The authors point out there are uncertainties in the sea-ice analyses and this will affect initialisation and the size of the errors (P10, L325) and that “guidance … from the remote sensing community” is needed. I agree here and I would perhaps suggest this limitation is added to the abstract. At present the last two lines of the abstract are a bit vague. It might be worth expanding these to state explicitly that the quality of satellite sea-ice products on daily to weekly timescales and on meso-scales (<500 km say) are not well characterised and this is a limitation for NWP.
Specific Comments
L42 – there is another recent idealised modelling study on the atmospheric response to sea-ice geometry and concentration that should be cited here:
Spensberger, C., & Spengler, T. (2021). Sensitivity of air-sea heat exchange in
cold-air outbreaks to model resolution and sea-ice distribution. Journal of
Geophysical Research: Atmospheres, 126, e2020JD033610. https://doi.
org/10.1029/2020JD033610
L49 – I am not an expert on the timelines here, but are you sure that ECMWF developed the first coupled global … system? Maybe for an ensemble? Not sure about NWP more generally. The Canadian ECCC have had a coupled forecast model for some time and this may pre-date the ECMWF development. You cite one paper for the Canadian system (Smith et al. 2018), but you should probably also cite earlier pioneering work that demonstrated the potential for improvement in atmospheric forecasts from such a coupled system withi NWP.
Pellerin P, Ritchie H, Saucier SJ, Roy F, Desjardins S, Valin M, Lee V. 2004.Impact of a two-way coupling between an atmospheric and an ocean – icemodel over the Gulf of St. Lawrence.Mon. Weather Rev.132: 1379 – 1398
Smith GC, Roy F, Brasnett B. 2013. Evaluation of an operational ice-ocean analysis and forecasting
system for the Gulf of St Lawrence. Q. J. R. Meteorol. Soc. 139: 419–433. DOI:10.1002/qj.1982
Smith, G.C., Roy, F., Reszka, M., Surcel Colan, D., He, Z., Deacu, D., Belanger, J.M., Skachko, S., Liu, Y., Dupont, F. and Lemieux, J.F., 2016. Sea ice forecast verification in the Canadian global ice ocean prediction system. Quarterly Journal of the Royal Meteorological Society, 142(695), pp.659-671.
L79-80 – I was slightly confused on reading the explanation for the persisted surface conditions for the first time, because ‘an anomaly is added each day’. On second reading I think this anomaly is only for the SST (not the sea ice)? Perhaps check for clarity here.
L128 – I think the Hersbach ERA5 reference is missing.
L142 – I’d replace “Atlantic coast’ with Labrador Sea coast, as it isn’t the main Atlantic basin.
L231 – ‘that region’ – it is unclear from this paragraph which region you are talking about. Maybe these lines should be merged into the previous paragraph?
L250-265 – this paragraph on internal boundary layer development at the ice edge is unreferenced – you could cite the Spensberger and Spengler 2021 paper here or the idealised 2D model of this internal BL development which also uses observations in
Renfrew, I.A. and King, J.C., 2000. A simple model of the convective internal boundary layer and its application to surface heat flux estimates within polynyas. Boundary-layer meteorology, 94(3), pp.335-356.
L285 – It was useful context to point out the differences in specific humidity (in g/kg and that this was 10% of the total value). You could also have expressed this as % of the standard deviation of this variable or something? And done similar for the difference in temperature. I think it useful to have an idea of the magnitude of these forecast differences in the context of day to day variability. If you can easily do such a metric? This is just a suggestion, not necessary.
L307 – The final section is more of a “Conclusions and Discussion” section.
L355 “weakly” not weekly.
Figures
Figure 1 – I would recommend changing the colour scale to one with white in the middle. At present the whole North Atlantic (which has no sea ice) is pink. It looks odd!
Fig 2 – the font size of the labels and legend is too small to read. Nice figure though!
Fig 3 & 2 – would it make sense to try and have the same colour for Ocean5 in these figures – this is red in 2 and 4 but green in fig 3.
Fig 5 – these figures illustrate the large variability between forecasts. Fig 5 is only very briefly mentioned in section 3.2 – I wonder if you should add a sentence or two emphasising the large variability.
-
RC2: 'Comment on wcd-2022-5', Anonymous Referee #2, 09 Mar 2022
I enjoyed reading this article very much. The results are relevant for the scientific community, numerical weather prediction centers, and forecast users. The forecast verification process is supported by a solid and sophisticated methodological base, and the forecast improvements and deficiencies are honestly highlighted without over or understating the findings. Furthermore, the manuscript is well written, and the figures illustrate the outcome of the study appropriately. I include below a few remarks and suggestions, which are mostly minor, and I hope that these will help the authors in the revision process. I recommend the publication of with manuscript once those (minor) points are addressed.
GENERAL COMMENTS
The Introduction and Method sections provide a very good overview of the system. However, I think some details on the probabilistic nature of the forecast are missing. Underlining the higher compatibility of a coupled model configuration with the ECMWF ensemble forecast system (i.e. the ice can evolve independently in each ensemble member, unlike in the persistence based strategy) would represent a nice addition to the study.
What about the melting season? I expect the impact of the sea ice on the ocean and land weather to be limited in summer because of the milder temperature gradients and winds. However, the demand for good ice forecasts might peak during this season. I would briefly mention whether the features of the dynamical system are appropriate also for the summer months. I also think a brief reference to what happens to the Southern Ocean sea ice might be appropriate.
The study focuses on a single winter season. Given the large number of forecasts analyzed, I expect the results to be solid. However, I think it might be good adding a characterization of the sea ice state during that winter in comparison to the climatological state, and discussing whether the results might be sensitive/influenced by potentially anomalous conditions (e.g. fast ice drift, abrupt melting events, etc.)
Figure 4 clearly shows that OCEAN5 reanalysis is biased, and you describe this well in the text. However, I think giving some more context on the origin of this bias would be helpful for the readers.
The fact that the thickness is not coupled implies that the thermodynamical transition at the ice edge is probably not well simulated also by the current dynamical system. Could you quantify the impact of this on the evolution of the internal boundary layer? Is the effect of a progressively reduced thickness towards the marginal ice zone negligible compared to the reduction in concentration? I expect this would also changes with the progressing of the freezing season. I think some more details on this in the discussion/conclusion section would be interesting for the reader.
I would like to point out that the using the AMSR2 derived sea ice concentration has also some drawbacks. It certainly comes with a desirable higher resolution because it uses higher frequencies. However, the effect of clouds on the microwave signal at higher frequencies is substantial and can penalize the quality of the retrieval, particularly across the marginal ice zone where clouds are not uncommon.
SPECIFIC COMMENTS
Line 71: I would not consider obs-SSTSIC a real forecast but rather an hindcast or an AMIP type simulation.
Line 75: I think it is worth mentioning here that the sea ice description of OSTIA comes from OSI-SAF. You mention this later in the result section, but I think stating this here would help the reader to understand the verification method.
Line 94: I suggest expressing the typical Arctic resolution of the ORCA025 grid also in km.
Line 96: Do you mean “…that is coupled to the atmosphere is…”?
FIGURES
Fig. 1: I suggest using a different colormap with the white color centered on zero. I don’t like seeing the rectangular domain of the polar stereographic grid in pink. Using red and green for the boxes might not be color friendly.
Fig. 2: Labels and titles are too small. I think it is ok to lose the outlier point in the OSI-SAF timeseries (plots a and c), probably caused by a partial observational coverage on that day.
Fig. 4 and 6: Labels are too small, and I suggest using the scientific notation to improve the readability of the plots.
-
AC1: 'Comment on wcd-2022-5', Jonathan Day, 29 Apr 2022
The comment was uploaded in the form of a supplement: https://wcd.copernicus.org/preprints/wcd-2022-5/wcd-2022-5-AC1-supplement.pdf
Jonathan Day et al.
Jonathan Day et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 352 | 133 | 15 | 500 | 57 | 6 | 7 |
- HTML: 352
- PDF: 133
- XML: 15
- Total: 500
- Supplement: 57
- BibTeX: 6
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1