the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Dec 2022
Seasonal Sea Ice Prediction with the CICE Model and Positive Impact of CryoSat-2 Ice Thickness Initialization
Abstract. The Los Alamos sea ice model (CICE) is being tested in standalone mode to identify biases that limit its suitability for seasonal prediction. The prescribed atmospheric forcings to drive CICE are from the NCEP Climate Forecast System Reanalysis (CFSR). A built-in mixed layer ocean model in CICE is used. Initial conditions for the sea ice and the mixed layer ocean are also from CFSR in the control experiments. The simulated sea ice extent is generally in good agreement with observations in the warm season at all lead times up to 12 months both in the Arctic and Antarctic, suggesting that CICE is able to provide useful sea ice edge information for seasonal prediction. However, the Arctic sea ice thickness forecast has a positive bias stemming from the initial conditions, and this bias often persists for more than a season, limiting the model’s seasonal forecast skill. When this bias is reduced by initializing ice thickness using the CryoSat-2 satellite observations while keeping all other initial fields unchanged in the CS2_IC experiments, both simulated ice edge and thickness improve. This confirms the important roles of sea ice thickness initialization in sea ice seasonal prediction seen in many studies.
- Preprint
(1607 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2022-1368', Anonymous Referee #1, 28 Jan 2023
Review of “Seasonal Sea Ice Prediction with the CICE Model and Positive Impact of CryoSat-2 Ice Thickness Initialization” by Sun and Solomon
General comments
This study carries out seasonal sea ice simulation up to 12 months based on CICE, initialized on each month during 2011-2017 with CFSR SIT or CryoSat-2 SIT. The authors compare sea ice metrics, i.e. SIC, SIT, SIE, SIV, in the simulations and in the observations derived from NSIDC, CryoSat-2, PIOMAS. They find that replacing CFSR SIT by CryoSat-2 SIT in the model initial condition has a positive effect on the model trajectory in the following several months, resulting in better sea ice conditions at the end of simulation. They also provide some brief explanation to introduce why there is enhanced melting in the northern Beaufort-Chukchi-East Siberia Seas (Figure 7c, d).
In general, the main finding of this study is the positive effect of SIT “data assimilation”, which has been longly proved in previous studies. That means, this study suffers from a lack of novelty. Secondary, this study suffers from a lack of in-depth mechanism analysis. The authors uses a lot of “might be the results of”, “could be related to”, “another possible source for”, “most likely due to”. CICE is a useful and powerful tool for sea ice modeling, and provides many diagnostic variables both for sea ice thermodynamics and dynamics. The authors could quantitatively identify most of speculations in this study if analyzing the diagnostic terms in-depth.
The paper is generally well-written, but some paragraphs still need editorial works (Part of them are listed below, because the main body needs substantially improvement, I only points out the part of introduction, model setup, and discussion). I hope these comments can help the authors in next submission.
Specific comments:
L16: As a relatively thin
L27: I am less confident on the statement “the predictive skill for sea ice forecasts at INTERANNUAL time scales relies on the accuracy of the sea ice initial conditions”.
L29: it is better to replace “ice-covered area” by “concentration”
L30: delete “to name a few”
L32: spring and early summer atmospheric conditions
L12-47: The “introduction” should be re-organized. The logic of this part is messy. Since this study uses CICE standalone, the introduction should emphasize on previous literature on model initialization skill in standalone sea ice model, and avoid to spend lots of sentences introducing the use of CICE in coupled models
L49-54: This paragraph should expand. More details on model setting on sea ice thermodynamical parameterization are needed since the following analysis is mostly related to sea ice thermodynamics. The authors mention “build-in mixed layer model” several times, but how this build-in mixed layer model is linked to the sea ice basal melting is missing.
L70-72: “despite the large ... were available”.... this sentence is too long.
L80: 1e-6, this threshold is too small. This procedure seems that make no sense.
L85: “Lead times not ending on .5 refer to the month following the stated time”... this sentence is hard to follow.
L99: “the Arctic SIV forecast at 0.5-month lead time is similar to CryoSat-2 observations or PIOMAS reanalysis in the warm season”. There is no CryoSat-2 observations in warm season. This sentence needs to be revised.
L146: a positive SIT bias can still
L209-237: The “Discussion” also needs to be improved. Some points: what is the advantage of using a build-in mixed layer model? What is the difference in the simulated sea ice if includes oceanic heat advection? Why the SIT distribution in the CFSR is so unreasonable?
L245-246: It seems that this paper is a resubmission. If this is the case, I would like to see the previous comments.
Citation: https://doi.org/10.5194/egusphere-2022-1368-RC1 -
RC2: 'Comment on egusphere-2022-1368', Anonymous Referee #2, 05 Feb 2023
The authors presented the seasonal prediction of Arctic and Antarctic sea ice by CICE model, and further to investigate the role of the initialization of satellite sea ice thickness. This CICE model have been used for sea ice forecast in many previous studies, therefore this study looked more like an similiar evaluation of sea ice prediction by CICE model. The role of satellite sea ice thickness initialization or assimilation also have been proved to perform well by the previous studies. The authors used the satellite ice thickness to replace the CFSR ice thickness in the initialization, while didn’t process sea ice concentration and snow depth. Then the questions came. How to ensure the satellite ice thickness agree with the CFSR ice concentration, and snow depth? How large of biases were introduced? Therefore, I suggest to assimilate the satellite ice thickness, ice concentration, even snow depth for a longer time before doing the prediction, which will make this study more reasonable. Above all, I suggest a resubmission or rejection to this manuscript.
Citation: https://doi.org/10.5194/egusphere-2022-1368-RC2
Status: closed
-
RC1: 'Comment on egusphere-2022-1368', Anonymous Referee #1, 28 Jan 2023
Review of “Seasonal Sea Ice Prediction with the CICE Model and Positive Impact of CryoSat-2 Ice Thickness Initialization” by Sun and Solomon
General comments
This study carries out seasonal sea ice simulation up to 12 months based on CICE, initialized on each month during 2011-2017 with CFSR SIT or CryoSat-2 SIT. The authors compare sea ice metrics, i.e. SIC, SIT, SIE, SIV, in the simulations and in the observations derived from NSIDC, CryoSat-2, PIOMAS. They find that replacing CFSR SIT by CryoSat-2 SIT in the model initial condition has a positive effect on the model trajectory in the following several months, resulting in better sea ice conditions at the end of simulation. They also provide some brief explanation to introduce why there is enhanced melting in the northern Beaufort-Chukchi-East Siberia Seas (Figure 7c, d).
In general, the main finding of this study is the positive effect of SIT “data assimilation”, which has been longly proved in previous studies. That means, this study suffers from a lack of novelty. Secondary, this study suffers from a lack of in-depth mechanism analysis. The authors uses a lot of “might be the results of”, “could be related to”, “another possible source for”, “most likely due to”. CICE is a useful and powerful tool for sea ice modeling, and provides many diagnostic variables both for sea ice thermodynamics and dynamics. The authors could quantitatively identify most of speculations in this study if analyzing the diagnostic terms in-depth.
The paper is generally well-written, but some paragraphs still need editorial works (Part of them are listed below, because the main body needs substantially improvement, I only points out the part of introduction, model setup, and discussion). I hope these comments can help the authors in next submission.
Specific comments:
L16: As a relatively thin
L27: I am less confident on the statement “the predictive skill for sea ice forecasts at INTERANNUAL time scales relies on the accuracy of the sea ice initial conditions”.
L29: it is better to replace “ice-covered area” by “concentration”
L30: delete “to name a few”
L32: spring and early summer atmospheric conditions
L12-47: The “introduction” should be re-organized. The logic of this part is messy. Since this study uses CICE standalone, the introduction should emphasize on previous literature on model initialization skill in standalone sea ice model, and avoid to spend lots of sentences introducing the use of CICE in coupled models
L49-54: This paragraph should expand. More details on model setting on sea ice thermodynamical parameterization are needed since the following analysis is mostly related to sea ice thermodynamics. The authors mention “build-in mixed layer model” several times, but how this build-in mixed layer model is linked to the sea ice basal melting is missing.
L70-72: “despite the large ... were available”.... this sentence is too long.
L80: 1e-6, this threshold is too small. This procedure seems that make no sense.
L85: “Lead times not ending on .5 refer to the month following the stated time”... this sentence is hard to follow.
L99: “the Arctic SIV forecast at 0.5-month lead time is similar to CryoSat-2 observations or PIOMAS reanalysis in the warm season”. There is no CryoSat-2 observations in warm season. This sentence needs to be revised.
L146: a positive SIT bias can still
L209-237: The “Discussion” also needs to be improved. Some points: what is the advantage of using a build-in mixed layer model? What is the difference in the simulated sea ice if includes oceanic heat advection? Why the SIT distribution in the CFSR is so unreasonable?
L245-246: It seems that this paper is a resubmission. If this is the case, I would like to see the previous comments.
Citation: https://doi.org/10.5194/egusphere-2022-1368-RC1 -
RC2: 'Comment on egusphere-2022-1368', Anonymous Referee #2, 05 Feb 2023
The authors presented the seasonal prediction of Arctic and Antarctic sea ice by CICE model, and further to investigate the role of the initialization of satellite sea ice thickness. This CICE model have been used for sea ice forecast in many previous studies, therefore this study looked more like an similiar evaluation of sea ice prediction by CICE model. The role of satellite sea ice thickness initialization or assimilation also have been proved to perform well by the previous studies. The authors used the satellite ice thickness to replace the CFSR ice thickness in the initialization, while didn’t process sea ice concentration and snow depth. Then the questions came. How to ensure the satellite ice thickness agree with the CFSR ice concentration, and snow depth? How large of biases were introduced? Therefore, I suggest to assimilate the satellite ice thickness, ice concentration, even snow depth for a longer time before doing the prediction, which will make this study more reasonable. Above all, I suggest a resubmission or rejection to this manuscript.
Citation: https://doi.org/10.5194/egusphere-2022-1368-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 233 | 105 | 16 | 354 | 7 | 6 |
- HTML: 233
- PDF: 105
- XML: 16
- Total: 354
- BibTeX: 7
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1