Evaluating the E3SMv2-MPAS Ocean-Sea Ice Coupled Unstructured Model in the Arctic: Atlantification Processes and Systematic Biases

Lv, Xinyuan; Wang, Huizan; Cao, Yu; Ren, Kaijun; Wang, Yangjun; Ding, Hao

doi:https://doi.org/10.5194/egusphere-2025-2665

Preprints

https://doi.org/10.5194/egusphere-2025-2665

Preprints

10 Jul 2025

| 10 Jul 2025

Evaluating the E3SMv2-MPAS Ocean-Sea Ice Coupled Unstructured Model in the Arctic: Atlantification Processes and Systematic Biases

Xinyuan Lv, Huizan Wang, Yu Cao, Kaijun Ren, Yangjun Wang, and Hao Ding

Abstract. Advancing high-resolution Arctic ocean-sea ice modeling is critical for understanding polar amplification and improving climate projections but faces challenges from computational limits and cross-scale interactions. The simulation capabilities of the ocean-sea ice coupled model (E3SMv2-MPAS) from the Energy Exascale Earth System Model (E3SM) 2.1 for the Arctic sea ocean-sea ice system are systematically evaluated using multi-source observational data and model outputs. A latitudinally varying mesh (60 km in the Southern Hemisphere to 10 km in the Arctic) balances computational efficiency while integrating low-latitude oceanic influences. Unstructured meshes enhance geometric representation of Arctic straits, coupled with a suitable mesoscale eddy transport parameterization to establish a multi-scale simulation framework. Numerical results demonstrate E3SMv2-MPAS's superior Arctic simulation performance: (1) Accurate reproduction of spatial heterogeneity in sea ice concentration, thickness, and sea surface temperature, including their 1995–2020 trend patterns; (2) Successful reconstruction of three-dimensional thermohaline structures within the Atlantic Water layer, capturing Atlantic Water's decadal warming trends and accelerated Atlantification processes – specifically mid-layer shoaling, heat content amplification, and reduced heat transfer lag times in the Eurasian Basin. Persistent systematic biases are identified: 0.5–1 m sea ice thickness overestimation in the Canadian Basin compared to ICESat observations; Coordinated sea surface temperature/salinity underestimation and sea ice concentration overestimation in the Greenland and Barents Seas; Atlantic Water core temperature overestimation; Regional asymmetries in decadal thermohaline field evolution. These systematic biases may be attributed to three principal sources: inadequate representation of eddy dynamics, limitations in mixing parameterizations, and insufficient resolution of cross-scale interactions in key gateways (e.g., Fram Strait).

Received: 05 Jun 2025 – Discussion started: 10 Jul 2025

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Xinyuan Lv, Huizan Wang, Yu Cao, Kaijun Ren, Yangjun Wang, and Hao Ding

Status: final response (author comments only)

CC1:
'Comment on egusphere-2025-2665', Milena Veneziani, 06 Aug 2025
These comments are from the HiLAT-RASM team that authored the paper on E3SM-Arctic-OSI, which described the E3SM-Arctic configuration used here, and which was cited by the authors (Veneziani et al. 2022).
The main comment is about the state of the E3SMv2-MPAS simulations described and evaluated in this manuscript. The authors mentioned (lines 152-156) that they used the file ‘ocean.ARRM60to10.180715.nc’ as the MPAS initial condition for E3SMv2-MPAS. That initial condition came from a very short adjustment run (5 days only) with standalone MPAS-ocean that in turn started from rest and Polar science center Hydrographic Climatology (PHC) climatological temperature and salinity, and therefore does not represent a spun-up ocean state. For E3SM-Arctic-OSI, we ran 3 consecutive JRA-55 cycles to achieve a more adjusted state for the ocean and sea ice models, and we analyzed the climatological results over the third cycle only, specifically over the last 12 or 30 years of the simulation (years 148-177 or 166-177; see Figs. 3, 4, 5a, 6, 10-15, 16c,d in Veneziani et al. 2022). Our understanding is that the E3SMv2-MPAS was run for, and evaluated over only one JRA-55 cycle. If that is the case, we wonder 1) whether the analyzed fields are adjusted or not, and also possibly too close to the PHC climatology for a fair evaluation of model performance, and 2) whether it is fair to compare the E3SMv2-MPAS results with more adjusted model states (as done in Figs. 8-10 and 16).
More specific comments are included in the following.
Line 99: Please add reference to our most recent paper on the E3SM-Arctic fully coupled configuration with E3SMv2.1: Huo et al. 2025 (https://doi.org/10.1029/2024MS004726).

Line 126: it seems that E3SMv2-MPAS uses the same MPAS mesh configuration as in Veneziani et al. 2022. Here, it would be good to clarify whether that is the case or whether the mesh is different.

Line 141: the ‘spatially varying GM’ was actually implemented prior to the development of the E3SM-Arctic-OSI configuration. Nevertheless, the current manuscript does not set GM=0 in the Arctic as we did in Veneziani et al. 2022, so maybe here one could say: “Similarly to what was done in Veneziani et al. 2022, we adopt a spatially varying…” and then include the sentence that is now on lines 212-215.

Line 146: Huo et al. 2025 mentioned above could also be cited here.

Line 159: Please clarify how the simulation was restarted after the 1981–1994 gap. What initial condition was used?

Line 205 (and more generally for the manuscript): the comparison with CMIP6 experiments seems unfair to us because those simulations are fully coupled. In addition, the comparison with the OMIP models and E3SM-Arctic-OSI should be done over similar (more adjusted) oceanic states. Finally, for the Veneziani et al. 2022 paper, we also provided data from a E3SMv1-LR-OSI simulation, which could be used for comparison to the E3SMv2-MPAS simulations here.

Line 230 (and Figs. 3-4): We assume these are annual sea ice quantities. Wouldn’t it be better to show seasonal sea ice climatologies?

As mentioned above, the comparison with E3SM-Arctic-OSI in section 3.3 (Fig. 10) should be over the same time period. For the Veneziani et al. 2022 paper, we provided climatologies for both E3SM-Arctic-OSI and E3SM-LR-OSI over years 148-177 and 166-177 (end of third JRA-55 cycle). A similar time frame should be used for E3SMv2-MPAS.

Similarly to other reanalysis products, JRA55 is known to overestimate surface air temperatures over Arctic sea ice (e.g., Batrak and Müller 2019, https://doi.org/10.1038/s41467-019-11975-3; see their Fig. 3d). These warm biases can propagate into the ocean component by modifying surface fluxes—particularly enhancing downward longwave radiation and reducing sensible heat loss—potentially leading to underestimated sea ice growth and overestimated upper-ocean temperatures. A brief discussion on how such forcing biases might influence ocean stratification, mixed layer depth, or Atlantification could be included in the Conclusions section.
Citation: https://doi.org/10.5194/egusphere-2025-2665-CC1
- AC4: 'Reply on CC1', Xinyuan Lv, 24 Sep 2025
  
  We sincerely thank the reviewer for the expert assessment and valuable comments. We have carefully considered all suggestions and have revised the manuscript accordingly to improve the clarity and accuracy of our work.
  A point-by-point response to all comments is included in the attached PDF file.
  
  Citation: https://doi.org/10.5194/egusphere-2025-2665-AC4
CC2:
'Comment on egusphere-2025-2665', Hailong Liu, 10 Aug 2025

This study involves much work for the model evaluation. But I do have several concerns.
1) Why do the authors particularly reply on the two time periods of 1960-1980 and 1995-2000? Line 160: I dont understand how it can be verified through overlapping period consistency checks (1995–2020) .
2) Line 187: any SSS data below sea ice? Is there any justification?
3) The major issue is the explanation of the causal analysis throughout the manuscript . For example, the major conclusion of "These systematic biases may be attributed to three principal sources: inadequate representation of eddy dynamics, limitations in mixing parameterizations, and insufficient resolution of cross-scale interactions in key gateways (e.g., Fram Strait) " is not convincing. Any sensitivity experiments can be considered to support the findings?

Citation: https://doi.org/10.5194/egusphere-2025-2665-CC2
- AC3: 'Reply on CC2', Xinyuan Lv, 24 Sep 2025
  
  Thank you for your thoughtful comments and for acknowledging the substantial work involved in the model evaluation. To address the concerns you raised, we have prepared point-by-point responses.
  A detailed document is included in the attached PDF file.
  
  Citation: https://doi.org/10.5194/egusphere-2025-2665-AC3
RC1:
'Comment on egusphere-2025-2665', Xi Liang, 13 Aug 2025

This study evaluates the performance of E3SMv2-MPAS on the Arctic simulation, with emphasize on sea ice concentration, sea ice thickness, SST/SSS, AW layer temperature and depth. The authors find that the E3SMv2-MPAS is superior on some aspects and also some limitations are identified. In general, the manuscript is well written, easy to follow, and the scientific significance of the manuscript is guaranteed. I hope the following comments are useful for the authors to revise their manuscript, most of them are minor, major comments are marked with ***.
L11: change “Arctic sea ocean-sea ice system” to “Arctic ocean-sea ice system”

L14: please clarify what “multi-scale” refers to.

L23-24: “These systematic biases may be attributed to three principal sources ...... in key gateways”. This speculation should be supported by some evidences in the main text.

L27: change “components” to “area”

L37-38: “Liang and Losch, 2018; Tian et al., 2022” are based on regional ice-ocean model, not climate model.

L48: change “and” to “to”

L53: change “seafloor regions” to “seafloor”

L60: change “shifts” to “shift”

L79: change “temporal” to “spatial” ?

L95: change “FESOM’s” to “FESOM”

L96: delete “then”

L130-132: “The North Atlantic sector ...... in the Gulf Stream extension region”. Please rephrase this sentence.

L134: “subpolar North Pacific sector adjacent to the Arctic Ocean”?

L139: how long is the sea ice dynamic step? The same to ocean dynamic step?

L157: rapidly

***L159-161: “The simulation periods ...... period consistency checks (1995–2020)”. If I understand correctly, you derive the simulation of 1995-2020 using JRA55 forcing during 1995-2020 but initialized at the latest model state of 1980. It is seldom to see such design of model simulation. As the intermediate years only span 15 years, I suggest the authors conduct a continuously simulation from 1960 to 2020.

***Section 2.1: there is no detailed information of sea ice model provided here. Please specify sea ice thermodynamics and dynamics in this configuration. As section 3.1 relates to sea ice validation, sea ice model description is necessary.

L197: EN.4.2.2 dataset

L244: from Figure 3a, “systematic winter overestimation” is caused partly by positive sea ice bias in the southern Greenland Sea and south extension of sea ice cover in the Barents Sea, suggesting upper ocean temperature bias in these regions. “moderate summer underestimation” may be related to inaccurate ice-albedo feedback and melt pond dynamics.

L249: “overestimated seasonal variability amplitudes” relates to sea ice thermodynamics, as no sea ice thermodynamics information is provided, it is hard to judge its causality.

***L258-269, 279-281: The modeled SIT has large biases in the Beaufort Gyre region, suggesting potential upper ocean thermal biases in the Beaufort Gyre. The author could check whether the ocean-ice heat flux over the Beaufort Gyre region is reasonable.

Figure 4d: given the known biases of the PIOMAS SIT, I suggest the author additionally validate the modeled sea ice volume against that derived from the CS2SMOS SIT from 2012.

L347: Please specify the define of AW layer thickness.

L347-349: This sentence needs to be rephrased as “Amerasian Basin” is not in Figure 8.

L362-363: “E3SMv2-MPAS maintains systematic temperature overestimation (~0.5C average)”. This statement is not appropriate for the other three regions.

L364-365: systematic salinity underestimation only occurs in western and eastern Eurasian Basin.

L390: ~ 0.5 C at 800 m ?

L417: I understand from section 2 that both the E3SMv1 and E3SMv2 use the same KPP. Why the vertical mixing scheme in E3SMv2 is refined?

Figure 11: please clarify this figure is conducted over the whole Arctic Basin or Eurasian Basin or Amerasian Basin?

***L466-467, 469-470: “likely modulated by differential ocean-ice feedbacks and cross-basin transport dynamics”, “indicating limitations inAW transport pathways and heat redistribution”, such speculations are too arbitrary, it’s better to avoid using such speculations.

L488-490: see the previous comment.

L519: delete “the” before “Fram Strait”. “Similar negative deviations (-0.5 C)”

L527-530: This conclusion can not derived from the observations STRICTLY.

L582: a detailed description of the thermal linkage framework between the upper and intermediate ocean layers is needed here.

Citation: https://doi.org/10.5194/egusphere-2025-2665-RC1
- AC2: 'Reply on RC1', Xinyuan Lv, 24 Sep 2025
  
  Thank you very much for your thorough review and positive feedback on our manuscript. We are grateful for your insightful comments and suggestions, which have greatly helped us improve the quality of our paper. We have carefully addressed each of your points in the revised manuscript.
  A point-by-point response to all comments is included in the attached PDF file.
  
  Citation: https://doi.org/10.5194/egusphere-2025-2665-AC2
RC2:
'Comment on egusphere-2025-2665', Qi Shu, 25 Aug 2025

The manuscript titled ‘Evaluating the E3SMv2-MPAS Ocean-Sea Ice Coupled Unstructured Model in the Arctic: Atlantification Processes and Systematic Biases’ presents a coupled ocean–sea ice model based on the E3SMv2-MPAS framework from the Energy Exascale Earth System Model, designed for Arctic sea ice and ocean simulations. The model features a high resolution of 10 km in the Arctic Ocean. According to the model validation, it demonstrates good performance in simulating Arctic sea ice and ocean conditions. Utilizing these simulations, the authors further identify a fundamental regime shift in intermediate-to-surface thermal coupling mechanisms in the Arctic under climate warming. In my assessment, the manuscript requires major revision. I offer the following suggestions to enhance its quality:

1. I concur with CC1’s comments. It is not entirely appropriate to compare the E3SMv2-MPAS results with those from E3SM-Arctic-OSI, CMIP6, and OMIP models, due to significant differences in integration lengths and associated model drifts. These discrepancies undermine the fairness of direct comparison. I recommend that the authors remove the model intercomparison from the main text and instead address relevant points briefly in the discussion section.

2. The model configuration requires further clarification. As the simulations do not adhere to the standard OMIP protocol, more detailed information regarding the model integration setup should be provided.

3. Regarding sea ice validation, while sea ice concentration and thickness are evaluated, I encourage the authors to also include assessments of sea ice extent, volume, and their long-term trends.

4. Given that this is a model evaluation study, I suggest a more comprehensive evaluation of the Arctic Ocean simulations. Key metrics should include Arctic Ocean freshwater content and its trend, as well as volume, heat, and freshwater fluxes through major Arctic gateways. These aspects are critical for assessing the model’s performance in simulating Arctic Ocean climate.

Citation: https://doi.org/10.5194/egusphere-2025-2665-RC2
- AC1: 'Reply on RC2', Xinyuan Lv, 24 Sep 2025
  
  We sincerely appreciate the thorough and constructive review of our manuscript. We are grateful for the positive recognition, as well as for the insightful comments aimed at further improving the study. We have carefully considered all the suggestions provided and have revised the manuscript accordingly to address each point raised.
  A point-by-point response to all comments is included in the attached PDF file.
  
  Citation: https://doi.org/10.5194/egusphere-2025-2665-AC1

Xinyuan Lv, Huizan Wang, Yu Cao, Kaijun Ren, Yangjun Wang, and Hao Ding

Viewed

Total article views: 1,884 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
1,804	55	25	1,884	13	20

HTML: 1,804
PDF: 55
XML: 25
Total: 1,884
BibTeX: 13
EndNote: 20

Views and downloads (calculated since 10 Jul 2025)

Month	HTML	PDF	XML	Total
Jul 2025	81	16	5	102
Aug 2025	382	19	9	410
Sep 2025	1,296	18	11	1,325
Oct 2025	45	2	0	47

Cumulative views and downloads (calculated since 10 Jul 2025)

Month	HTML	PDF	XML	Total
Jul 2025	81	16	5	102
Aug 2025	382	19	9	410
Sep 2025	1,296	18	11	1,325
Oct 2025	45	2	0	47

Viewed (geographical distribution)

Total article views: 1,828 (including HTML, PDF, and XML) Thereof 1,828 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 06 Oct 2025

Short summary

This study evaluates the performance of the varying-resolution mesh model (E3SMv2-MPAS) in simulating Arctic processes. We show this model avoids significant errors found in older models when representing key ocean layers. We also pinpoint specific areas where the model still struggles and uncover why these problems might occur. This exciting progress means scientists can now use this tool more confidently to understand how Arctic ocean layers work and change.


Total:	0
HTML:	0
PDF:	0
XML:	0