the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Aug 2026
Paleo-NorCPM: A framework toward fully coupled ocean-atmosphere paleoclimate reanalysis over the past centuries
Abstract. We present a new paleo reanalysis based on an adaptation of the Norwegian Climate Prediction Model (NorCPM), in which the Norwegian Earth System Model is equipped with an ensemble Kalman filter to assimilate hundreds of annually resolved proxy records, including tree rings, corals, and ice cores, over the last 500 years. First, an offline 30-member reanalysis is produced. This is followed by a dynamically coupled 10-member reanalysis obtained through atmospheric wind nudging from the offline reconstruction, allowing the ocean to respond to reconstructed atmospheric circulation variability. This approach enables an explicit representation of ocean dynamics and associated atmosphere-ocean feedback, which are typically indirectly represented in existing multi-century reanalyses. The offline reanalysis successfully reproduces large-scale atmospheric variability and hydroclimate variability over the 20th century across the tropics, mid-latitudes, and polar regions, and compares well with existing paleo reconstructions. The wind-nudged ensemble further captures ocean variability and exhibits more pronounced multi-decadal variability, highlighting the role of wind-driven ocean adjustment in shaping low-frequency climate variability. Challenges remain, particularly in the Southern Ocean where the wind-nudged ensemble shows strong sensitivity to wind variability. Nevertheless, these results demonstrate a promising pathway toward fully coupled ocean-atmosphere reanalyses spanning the past centuries, opening new opportunities to investigate the mechanisms underlying low-frequency variability and coupled feedback.
- Preprint
(5164 KB) - Metadata XML
-
Supplement
(5729 KB) - BibTeX
- EndNote
Status: open (until 17 Oct 2026)
- RC1: 'Comment on essd-2026-433', Anonymous Referee #1, 11 Sep 2026 reply
-
RC2: 'Comment on essd-2026-433', Lili Lei, 28 Sep 2026
reply
This manuscript presents a new paleoclimate reanalysis. An offline paleoclimate data assimilation is first performed by updating the priors from ensemble simulations of NorCPM with proxy recodes. The offline three-dimensional atmospheric zonal and meridional wind analyses are then incorporated into the NorCPM through nudging. The wind-nudged ensemble simualtions captures ocean variability and exhibits pronounced multi-decadal variability. These results highlight the role of wind-driven ocean adjustment in shaping low-frequency climate variability.The manuscript is well written. The generated fully copled ocean-atmosphere paleoclimate reanalysis over the past centureis could be a valuable contribution to the climate community. I have several specific comments as below.
- I’d like suggest the authors use the generally adopted terminology of data assimilation. Bold lower-case x is used for state vector, italic and lower-case h for PSM, and bold and upper-case H for the Jacobian of PSM, etc. Please note that h in innovation and H in Kalman gain should be different.
- Lines 120-130, 1) how the inflation value, ‘increased by a factor of two’, is determined? 2) Please give more details of the observation error inflation. When the observation error inflation is applied, and how large the inflation value is?
- Lines 140-145, 1) are the same ensemble priors used for each DA step, or the 30 ensemble priors are used as ensemble priors? 2) The two references used simplified models. But as shown by Lei et al. (2026), the information from analyses can be propagated along with online cycling data assimilation.
- Lines 152-153, how is the flow-dependent background error covariances constructed from the 30-member ensemble different from that constructed from the “analog” ensembles (Sun et al. 2022)? Please give some intuitive expectations.
- Lines 160-165, this procedure is similar to the hybrid gain proposed by Sun et al. (2024). Do you have an explicit sample-estimated P_d? How the hybrid background error covariance matrix P is applied to the solution of ensemble analyses, as shown by equations (3) and (4)?
- Lines 205-206, do you expect wider localization lengthscales for P_s than P_d?
- Table 2, what’s the unit for ‘Min. length’?
- Lines 382-385, observation errors are important for paleoclimate data assimilation. How’s the sample size for each proxy record that is used for PSM calibration. Are there enough samples for this observation error estimation? Are the samples assumed to be independent? If so, is the assumption valid for the same proxy record?
- Lines 400-405, the proxy records with large innovations could contain valuable information, especially for the extreme events that is very difficult to simulate by numerical models. For offline paleoclimate data assimilation, the imbalance or instability is not associated with model simulation, so that is it possible to include such extreme-value proxy records?
- Lines 444-445, are the temperature and SST updated by the proxy? If yes, how about nudge to 3-D wind and mass fields?
- Lines 454-456, how about the nudging coefficient in equation (10)? Would a trapezoidal nudging coefficient help with synchronization?
- Equation (13), is xi here the ensemble mean?
- Figure 3, between 1500 and 1600, paleo-NorCPM1 and paleo-NorCRM1 wind nudging are very similar to each other. Does this indicate that the spin-up period for online data assimilation is around 100 years?
- Figure 5, do you have explanations for the discrepancies between paleo-NorCMP1 wind nudging and LMR online before 1800?
References:
Lei, L., H. Sun, Z.‐M. Tan, L. Ning, and Z. Liu, 2026: Strongly coupled data assimilation for paleoclimate reconstruction using deep learning‐based models. J. Adv. Model. Earth Syst., 18, e2025MS005686. https://doi.org/10.1029/2025MS005686
Sun, H., L. Lei, Z. Liu, L. Ning, and Z.-M. Tan, 2024: A hybrid gain analog offline EnKF for paleoclimate data assimilation. J. Adv. Model. Earth Syst., 16, e2022MS003414, https://doi.org/10.1029/2022MS003414
Sun, H., L. Lei, Z. Liu, L. Ning, and Z.-M. Tan, 2022: An analog offline EnKF for paleoclimate data assimilation. J. Adv. Model. Earth Syst., 14, e2021MS002674, https://doi.org/10.1029/2021MS002674
Citation: https://doi.org/10.5194/essd-2026-433-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 262 | 139 | 122 | 523 | 61 | 186 | 122 |
- HTML: 262
- PDF: 139
- XML: 122
- Total: 523
- Supplement: 61
- BibTeX: 186
- EndNote: 122
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This is an important data paper describing a valuable product that enriches the existing toolkit. The main question is whether we need another dataset if it is generated using a different model but a more or less similar method. In other words, how much can this new dataset add to the current toolkit? I think it does offer something new, especially the nudging experiment. Although some issues related to the nudging approach are not yet fully understood, I consider it the main contribution of this new dataset. This is also a very challenging task because uncertainties from different sources, including psm, climate models, and original records, will all accumulate and leave a strong imprint on the final product. It is difficult to determine which source degrades the quality of the dataset. Users often have little choice but to use all data they can obtain. This is a common situation in current paleoclimate research and probably reflects the nature of reconstruction work itself. Another difficulty is that we do not know the truth. At present, perhaps the best situation we can expect is that a new dataset falls somewhere within the range of estimates from existing datasets. Based on this paper and other recent work on similar topics, this seems to be how the field is moving forward.
I do not have any major concerns about the paper. Given its scope, depth, and the details required to explain how the dataset was generated, the authors did a very good job of including the necessary information ( so it is a very long paper actually). The work mainly includes two parts, the methodology and the evaluation. I cannot comment too much on the methodology, but I think the nudging component is the most novel part of the study. The authors may need to discuss more clearly what they plan to do along this direction in the future. From what I can see, the other method, offline DA, is already quite well developed. No matter how it is redone or retested, it may not lead to a major improvement. The nudging approach may have more room for further development. The evaluation was also conducted well, although it is difficult because no true ground truth is available.
The most valuable part of this paper is in lines 420–490. The other parts basically represent what we already know or what has been discussed in LMR and other reconstruction papers. I still question how reliable the wind fields generated by offline DA are when they are further used to force a fully coupled model. This is an interesting step, but it seems that monthly winds +artificially generated 6-h winds+ q-flux are all used. The authors go over these steps very quickly without discussing the advantages and disadvantages of each one. For example, are the artificially generated 6-h winds too strong or too weak? Should these 6h winds also reflect some changes in the monthly wind? I also wonder why the authors do not nudge u, v, &t above the PBL if the goal is to produce realistic climate changes. All these issues should be discussed extensively because the ocean response to realistic daily wind variability can be very different from its response to artificially generated 6-h winds. How are freshwater fluxes, particularly precipitation, related to these wind patterns? Freshwater forcing may be critical for explaining some of the biases that are now probably compensated for by the q-flux. This may be especially important over the Southern Ocean, where wind stress curl/fresh water can play an important role in determining the ocean circulation and thermal structure.
When introducing the nudging approach, the authors should cite the original work that implemented this method in CAM5. That paper discusses in detail how the nudging method was implemented in CAM5. Many follow-up studies have not given enough credit to this relatively early and original application. https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2012JD018588
The literature review in the intro is quite narrow and does not include much recent work. I feel that the authors should make effort to improve this part. Many relevant studies have been published, and previous work should be properly recognized and credited.