the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 Dec 2023
Assessing the tropical Atlantic biogeochemical processes in the Norwegian Earth System Model
Abstract. State-of-the-art Earth system models exhibit large biases in their representation of the tropical Atlantic hydrography, with potential large impacts on both climate and ocean biogeochemistry projections. This study investigates how biases in model physics influences marine biogeochemical processes in the tropical Atlantic using the Norwegian Earth System Model (NorESM). We assess four different configurations of NorESM: NorESM1 is taken as benchmark (NorESM1-CTL) that we compare against the simulations with (1) a physical bias correction and against (2 and 3) two configurations of the latest version of NorESM with improved physical and biogeochemical parameterizations with low and intermediate atmospheric resolutions, respectively. With respect to NorESM1-CTL, the annual-mean sea surface temperature (SST) bias is reduced largely in the first and comparably third simulations in the equatorial and southeast Atlantic. In addition, the SST seasonal cycle is improved in all three simulations, resulting in more realistic development of the Atlantic Cold Tongue in terms of location and timing. Corresponding to the cold tongue seasonal cycle, the marine primary production in the equatorial Atlantic is also improved and in particular, the Atlantic summer bloom is well represented during June to September in all three simulations. The more realistic summer bloom can be related to the well-represented shallow thermocline and associated nitrate supply from the subsurface ocean at the equator. The climatological intense outgassing of sea-air CO2 flux in the western basin is also improved in all three simulations. Improvements in the climatology mean state also lead to better representation of primary production and sea-air CO2 interannual variability associated with the Atlantic Niño and Niña events. We stress that physical process and its improvement are responsible for modeling the marine biogeochemical process as the first simulations, where only climatological surface ocean dynamics are corrected, provides the better improvements of marine biogeochemical processes.
-
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
(10276 KB)
-
Supplement
(3345 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(10276 KB) - Metadata XML
-
Supplement
(3345 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-2947', Anonymous Referee #1, 19 Feb 2024
General comments
This study evaluate the implications of physical biases on the simulated marine biogeochemical processes in the tropical Atlantic Ocean for 4 different version of a ESM. The models used are different versions of NorESM, an earth system model with different components, with an increasing degree of complexity and resolution. The different results are compared to a base solution, NorESM1, taken as the benchmark.
The main improvement was to decrease the bias of annual mean of SST, giving rise to a realistic development of the Atlantic Cold Tongue (in geographical location and timing), and hence the marine primary production in the Equatorial Atlantic ocean. This shows the clear link between the physical cycles and the biological ones. Consequence of the improvements in the physical representations of the system, is also the improvement of the carbon cycle representations, discussed in the manuscript mainly in terms of air-sea C02 fluxes.
The development of the manuscript start by a broad review of the oceanography of the tropical Atlantic ocean, including it’s links with coastal phenomena (river inputs), the circulation in neighboring tropical systems, and characteristics phenomena of variability in the region (Atlantic Niño’s), and the consequences in terms of anthropogenic and global change effects. The role of ESM is also introduced as key tools, as well as the importance of the physical phenomena on the biogeochemical cycles. Their biases in the physical components clearly decreases its performance downstream regarding the biogeochemical cycles (primary and secondary, oxygen, carbon).
Within this problematic issues, the present manuscript introduce the physical, biological and chemical components of the several versions of the NorESM configurations, and analyze the improvements with relation to the base model, concerning the mean annual, the seasonal and inter-annual time scales.
The NorESM model contributes to CMIP (5 and 6), which provide a degree of general quality and confidence on the results. However, for someone not necessarily familiar with global scale model analysis and its limitations, the large bias reported, even in the most recent (with better performance) versions, give reasons for some degree of concern regarding the confidence for simulations for the recent past / present and mainly the future scenarios.
The structure of the results starts from the comparison with climatological standard data, and the reasons to induce so large bias, primarily associated to wind stresses and air-sea fluxes in the atmospheric components. The improvements of the different versions justify its application, in terms of horizontal and vertical distributions (Figures 1 and 2).
The seasonality is analyzed along the equator in terms of SST, primary production and PCO2 when compared to the climatological values, (Fig 3) and a thorough analysis (although a bit ‘too verbose’) of the differences and the improvements was done in the manuscript.
The next step was to analyze the interannual variability, dominated by Atlantic Niño/Niña phenomena. One wonders if the models are able (or not) to reproduce the actual Niño/a’s years in the recent pass (I think that the response is probably not), as the forcing used in the most advanced models should include the atmospheric mechanisms (wind stress anomalies) to start Niño/a(s). I think that some comment should be done around this issue. The analysis centered the attention around the STD of several fields, (Fig 6 ), composite anomaly differences in the horizontal (Fig 7) and in vertical sections (Fig 8 and 9) for different variables. It seems to me a too technical and specialized explanation section for modelers, while I would expect some comments within the discussion section about this important issue.
Otherwise the manuscript are well organized and well written, and deserves to be published in my opinion.
Specific comments
The description of the different versions of NorESM model is rather difficult to follow for someone that does not know the NorESM* system, and a table containing the four versions and main features would help to the reader better identify the common points and differences between models.
From my point of view the way how the Figure 4 , containing Taylor diagrams of the SST, PP and CO2 fluxes was done, should be better explained.
Citation: https://doi.org/10.5194/egusphere-2023-2947-RC1 - AC1: 'Reply on RC1', Shunya Koseki, 30 Apr 2024
-
RC2: 'Comment on egusphere-2023-2947', Anonymous Referee #2, 12 Mar 2024
his study analyses the skills of the ESM NorESM at reproducing the physical and biogeochemical characteristics of the tropical Atlantic Ocean. A set of 4 model configurations is compared: the NORESM1 configuration, the NORESM1 configuration with flux correction based on observations, and two configurations based on NORESM2 (coarse and medium spatial resolution of the atmospheric component). The standard NORESM1 setup exhibits strong biases both on ocean dynamics and biogeochemistry which are improved with flux correction or a higher resolution of the atmospheric configuration. The low resolution version of NORESM1 also shows some improvements which suggests that the new parameterizations and calibration in the version 2 also contribute to the improved skills. Another important conclusion of the study is that biases in the simulated ocean dynamics have a strong imprint on the simulated ocean biogeochemistry (nutrients, NPP and pCO2). And this concerns the mean state, the seasonality and interannual variability.
I have several general concerns:
1) The description of the different model configurations and the main differences between them is rather short and is thus very difficult to follow for someone who is not an expert of NorESM. I think that some additional information is necessary such as more details on the flux correction technique, on the main differences between the different model components that are relevant for the study.
2) The different configurations are quite well designed to illustrate the improvements to be expected, at least from a better representation of the atmospheric state (flux correction, higher spatial resolution). However, I find that the attribution and the mechanistic understanding are rather too vague. For sure, some changes in the model components explain part of the improvement since NorESM2-LM performs better than NorESM1-CTL, but there is no discussion on these changes and what role they play in the improvement. For instance, what is explained by changes in the atmopheric, oceanic and biogeochemical components respectively? This would probably require additional experiments such as a NorESM2-AC. We also see that the upper thermocline is much more stratified in NorESM1 than in NorESM2: why and what are the consequences? Winds and evaporation minus precipitation are the main players but we have no idea of the biases they exhibit in the non corrected model configurations. A consequence of this general concern is the discussion that is really vague and not very informative, to my opinion.
3) Marine biogeochemistry is evaluated by inspecting nutrients, PP and pCO2. pCO2 is very sensitive to the dynamics (as mentioned and shown in the study) and it is thus not very surprising that any improvement in the representation of ocean dynamics has a strong impact on it. It is not a very good tracer of the ecosystem component of the biogeochemical model. PP is not observed but reconstructed from some algorithms both for chlorophyll and PP itself which are known to have significant issues (different algorithms can give very different results). I would have liked to see a comparison to chlorophyll satellite data which are much more direct and with less uncertainties.
In summary, I think that this study needs some major revisions addressing my general concerns before it can deserve publication. A crucial point is a more thorough investigation of the features that explain the improvements obtained in the different model configurations. Finally, the model performs quite bad in terms of PP and pCO2, even in the best configurations depite what the authors state sometimes in the study. However, this is not a concern for me because ESM but also quite coarse ocean-only models tend to behave quite badly in this basin. However, I would be curious to see Chlorophyll.
Minor comments:
on the manuscript as a whole: Obviously I'm not a native English speaker, but I think the English can be improved. In addition, there are typos and formatting problems with references throughout the manuscript that should be corrected.Section 2.3: You don't explain what MPI SOM-FEM is.
Line 176: NorESM2 has a warmer subsurface and a less stratified upper thermocline (seen also on the nutrient vertical distribution), why? It relates to my general concern 2.
Section 3.2 and figure 3: The ACT is clearly improved, especially in NorESM2-MM but also in NorESM2-LM and is better (at least from what I can see) than NorESM1-CTL. Thus, part of the improvement is not related to the increased atmospheric resolution but to changes in the ingredients of the physical components. Any clue on what they are. Furthermore from January to June, NorESM2 is not that good and worse than NorESM1-CTL. It is significantly warmer and with two maxiam close to the African coast and 30-35W. It should be mentionend and ideally commented.
Lines 262-264: what are these improvements? Very vague.
Section 3.3: Why using different types of analysis for SST and PP? Is there any reason behind the differential treatment?
Why not a taylor plot for PP (or even better Chlorophyll) similar to what is done with SST.Lines 380-382: the ingassing bias between 8S and 10S along Africa is quite strong in NorESM1. What is the cause of that sink. PP does not seem to be very very high at the specific location according to Figure S4. In lines 386-387, it is stated that it might be biogeochemical issues or riverine input. This is really vague and does not say anything.
Citation: https://doi.org/10.5194/egusphere-2023-2947-RC2 - AC2: 'Reply on RC2', Shunya Koseki, 30 Apr 2024
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-2947', Anonymous Referee #1, 19 Feb 2024
General comments
This study evaluate the implications of physical biases on the simulated marine biogeochemical processes in the tropical Atlantic Ocean for 4 different version of a ESM. The models used are different versions of NorESM, an earth system model with different components, with an increasing degree of complexity and resolution. The different results are compared to a base solution, NorESM1, taken as the benchmark.
The main improvement was to decrease the bias of annual mean of SST, giving rise to a realistic development of the Atlantic Cold Tongue (in geographical location and timing), and hence the marine primary production in the Equatorial Atlantic ocean. This shows the clear link between the physical cycles and the biological ones. Consequence of the improvements in the physical representations of the system, is also the improvement of the carbon cycle representations, discussed in the manuscript mainly in terms of air-sea C02 fluxes.
The development of the manuscript start by a broad review of the oceanography of the tropical Atlantic ocean, including it’s links with coastal phenomena (river inputs), the circulation in neighboring tropical systems, and characteristics phenomena of variability in the region (Atlantic Niño’s), and the consequences in terms of anthropogenic and global change effects. The role of ESM is also introduced as key tools, as well as the importance of the physical phenomena on the biogeochemical cycles. Their biases in the physical components clearly decreases its performance downstream regarding the biogeochemical cycles (primary and secondary, oxygen, carbon).
Within this problematic issues, the present manuscript introduce the physical, biological and chemical components of the several versions of the NorESM configurations, and analyze the improvements with relation to the base model, concerning the mean annual, the seasonal and inter-annual time scales.
The NorESM model contributes to CMIP (5 and 6), which provide a degree of general quality and confidence on the results. However, for someone not necessarily familiar with global scale model analysis and its limitations, the large bias reported, even in the most recent (with better performance) versions, give reasons for some degree of concern regarding the confidence for simulations for the recent past / present and mainly the future scenarios.
The structure of the results starts from the comparison with climatological standard data, and the reasons to induce so large bias, primarily associated to wind stresses and air-sea fluxes in the atmospheric components. The improvements of the different versions justify its application, in terms of horizontal and vertical distributions (Figures 1 and 2).
The seasonality is analyzed along the equator in terms of SST, primary production and PCO2 when compared to the climatological values, (Fig 3) and a thorough analysis (although a bit ‘too verbose’) of the differences and the improvements was done in the manuscript.
The next step was to analyze the interannual variability, dominated by Atlantic Niño/Niña phenomena. One wonders if the models are able (or not) to reproduce the actual Niño/a’s years in the recent pass (I think that the response is probably not), as the forcing used in the most advanced models should include the atmospheric mechanisms (wind stress anomalies) to start Niño/a(s). I think that some comment should be done around this issue. The analysis centered the attention around the STD of several fields, (Fig 6 ), composite anomaly differences in the horizontal (Fig 7) and in vertical sections (Fig 8 and 9) for different variables. It seems to me a too technical and specialized explanation section for modelers, while I would expect some comments within the discussion section about this important issue.
Otherwise the manuscript are well organized and well written, and deserves to be published in my opinion.
Specific comments
The description of the different versions of NorESM model is rather difficult to follow for someone that does not know the NorESM* system, and a table containing the four versions and main features would help to the reader better identify the common points and differences between models.
From my point of view the way how the Figure 4 , containing Taylor diagrams of the SST, PP and CO2 fluxes was done, should be better explained.
Citation: https://doi.org/10.5194/egusphere-2023-2947-RC1 - AC1: 'Reply on RC1', Shunya Koseki, 30 Apr 2024
-
RC2: 'Comment on egusphere-2023-2947', Anonymous Referee #2, 12 Mar 2024
his study analyses the skills of the ESM NorESM at reproducing the physical and biogeochemical characteristics of the tropical Atlantic Ocean. A set of 4 model configurations is compared: the NORESM1 configuration, the NORESM1 configuration with flux correction based on observations, and two configurations based on NORESM2 (coarse and medium spatial resolution of the atmospheric component). The standard NORESM1 setup exhibits strong biases both on ocean dynamics and biogeochemistry which are improved with flux correction or a higher resolution of the atmospheric configuration. The low resolution version of NORESM1 also shows some improvements which suggests that the new parameterizations and calibration in the version 2 also contribute to the improved skills. Another important conclusion of the study is that biases in the simulated ocean dynamics have a strong imprint on the simulated ocean biogeochemistry (nutrients, NPP and pCO2). And this concerns the mean state, the seasonality and interannual variability.
I have several general concerns:
1) The description of the different model configurations and the main differences between them is rather short and is thus very difficult to follow for someone who is not an expert of NorESM. I think that some additional information is necessary such as more details on the flux correction technique, on the main differences between the different model components that are relevant for the study.
2) The different configurations are quite well designed to illustrate the improvements to be expected, at least from a better representation of the atmospheric state (flux correction, higher spatial resolution). However, I find that the attribution and the mechanistic understanding are rather too vague. For sure, some changes in the model components explain part of the improvement since NorESM2-LM performs better than NorESM1-CTL, but there is no discussion on these changes and what role they play in the improvement. For instance, what is explained by changes in the atmopheric, oceanic and biogeochemical components respectively? This would probably require additional experiments such as a NorESM2-AC. We also see that the upper thermocline is much more stratified in NorESM1 than in NorESM2: why and what are the consequences? Winds and evaporation minus precipitation are the main players but we have no idea of the biases they exhibit in the non corrected model configurations. A consequence of this general concern is the discussion that is really vague and not very informative, to my opinion.
3) Marine biogeochemistry is evaluated by inspecting nutrients, PP and pCO2. pCO2 is very sensitive to the dynamics (as mentioned and shown in the study) and it is thus not very surprising that any improvement in the representation of ocean dynamics has a strong impact on it. It is not a very good tracer of the ecosystem component of the biogeochemical model. PP is not observed but reconstructed from some algorithms both for chlorophyll and PP itself which are known to have significant issues (different algorithms can give very different results). I would have liked to see a comparison to chlorophyll satellite data which are much more direct and with less uncertainties.
In summary, I think that this study needs some major revisions addressing my general concerns before it can deserve publication. A crucial point is a more thorough investigation of the features that explain the improvements obtained in the different model configurations. Finally, the model performs quite bad in terms of PP and pCO2, even in the best configurations depite what the authors state sometimes in the study. However, this is not a concern for me because ESM but also quite coarse ocean-only models tend to behave quite badly in this basin. However, I would be curious to see Chlorophyll.
Minor comments:
on the manuscript as a whole: Obviously I'm not a native English speaker, but I think the English can be improved. In addition, there are typos and formatting problems with references throughout the manuscript that should be corrected.Section 2.3: You don't explain what MPI SOM-FEM is.
Line 176: NorESM2 has a warmer subsurface and a less stratified upper thermocline (seen also on the nutrient vertical distribution), why? It relates to my general concern 2.
Section 3.2 and figure 3: The ACT is clearly improved, especially in NorESM2-MM but also in NorESM2-LM and is better (at least from what I can see) than NorESM1-CTL. Thus, part of the improvement is not related to the increased atmospheric resolution but to changes in the ingredients of the physical components. Any clue on what they are. Furthermore from January to June, NorESM2 is not that good and worse than NorESM1-CTL. It is significantly warmer and with two maxiam close to the African coast and 30-35W. It should be mentionend and ideally commented.
Lines 262-264: what are these improvements? Very vague.
Section 3.3: Why using different types of analysis for SST and PP? Is there any reason behind the differential treatment?
Why not a taylor plot for PP (or even better Chlorophyll) similar to what is done with SST.Lines 380-382: the ingassing bias between 8S and 10S along Africa is quite strong in NorESM1. What is the cause of that sink. PP does not seem to be very very high at the specific location according to Figure S4. In lines 386-387, it is stated that it might be biogeochemical issues or riverine input. This is really vague and does not say anything.
Citation: https://doi.org/10.5194/egusphere-2023-2947-RC2 - AC2: 'Reply on RC2', Shunya Koseki, 30 Apr 2024
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 300 | 76 | 39 | 415 | 44 | 24 | 24 |
- HTML: 300
- PDF: 76
- XML: 39
- Total: 415
- Supplement: 44
- BibTeX: 24
- EndNote: 24
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Lander R. Crespo
Jerry Tjiputra
Filippa Fransner
Noel S. Keenlyside
David Rivas
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(10276 KB) - Metadata XML
-
Supplement
(3345 KB) - BibTeX
- EndNote
- Final revised paper