the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
An along-track biogeochemical Argo modelling framework, a case study of model improvements for the Nordic Seas
Veli Çağlar Yumruktepe
Erik Askov Mousing
Jerry Tjiputra
Annette Samuelsen
Abstract. We present a framework that links in situ observations from the biogeochemical-Argo (BGC-Argo) array to biogeochemical models. The framework allows a minimized technical effort to construct a Lagrangian type 1D modelling experiment along BGC-Argo tracks. We utilize the Argo data in two ways; (1) drive the model physics, (2) evaluate the model biogeochemistry. BGC-Argo physics data is used to nudge the model physics closer to observations to reduce the errors in biogeochemistry stemming from physics errors. This allows us to target model biogeochemistry and by using the Argo biogeochemical dataset, we identify potential sources of model errors, introduce changes to model formulation, and validate model configurations. We present experiments for the Nordic Seas and showcase how we identify potential BGC-Argo buoys to model, prepare forcing, design experiments and approach model improvement and validation. We used ECOSMO II(CHL) model as the biogechemical component and focused on chlorophyll a. The experiments revealed that ECOSMO II(CHL) required improvements during low-light conditions, as the comparison to BGC-Argo reveals that ECOSMO II(CHL) simulates a late spring bloom and does not represent the deep chlorophyll maximum formation in summer periods. We modified the productivity and chlorophyll a relationship and statistically documented decreased bias and error in the revised model using BGC-Argo data. Our results reveal that nudging the model T and S closer to BGC-Argo data reduces errors in biogeochemistry, and we suggest a relaxation time-period of 1–10 days. The BGC-Argo data coverage is ever growing and the framework is a valuable asset for improving models in 1D-model efficiently and transfer the configurations to 3D-model with a wide range of focus from operational, regional/global and climate scale.
- Preprint
(5959 KB) -
Supplement
(18501 KB) - BibTeX
- EndNote
Veli Çağlar Yumruktepe et al.
Status: final response (author comments only)
-
RC1: 'Comment on gmd-2023-25', Anonymous Referee #1, 15 May 2023
The authors present a study that use BGC Argo data and model experiments in tandem to drive and evaluate the model. Argo data are used to both help simulate a realistic physical environment and interpret biogeochemistry. The framework presented is intended to have minimal technical effort and set up ideas for future synthetic modeling activities in the future as the Argo fleet grows. The article focuses on chlorophyll, which is known to be highly variable and not the best metric for biomass. The model itself is not as simple as advertised, as it includes assumptions about the partitioning of biomass into different groups and many more parameters than in the simplest possible case if one if mainly interested in phytoplankton stocks or physiology (which seems like the goal, given that chl is emphasized). Other high quality metrics are available, which could allow model simplification and interpretation beyond what was considered herein. Argo floats also have many more biogeochemical variates that are not well ultilized in this study, including oxygen (which can be used to assess zooplankton in some capacity) and bbp (which can be used to assess accumulation rates directly rather than through a model). I suggest either simplifying the model so that improved interpreation can be done, or adding complexity to the analysis to better constrain the model in its current construction (ie, incorporating Argo nutrients and metrics for zooplankton). More detailed comments follow.
Satellite merged products have known issues (see van Oostende et al 2022). It may be better to have a consistent mission value to evaluate the BGC-Argo data, as OC CCI is essentially a modeled product.
Section 2.2 How many matchups did you obtain? What was the time separation between samples? 2km may be good for +/-a few hours of a matchup, but if more time is used, a great physical distance should be used.
Line 140: What about parameterizing C:Chl variability with temperature or nutrient stress as well?
Table 2: Why is grazing rate held constant rather than fluctuating with standing stock of phytoplankton? Should the table be revised to say ‘max grazing rate?’
Table 2: shouldn’t mortality rate be a function of growth rate or concentration (viruses) than a fixed number held constant with any concentration?
Line 150-155. The model has so many parameters that many can be tuned in different combinations to match the observations. Given that a goal is biogeochemical interpretation, how can the number of free parameters be justified? A simpler model may be a better place to start before adding phytoplankton and zooplankton groups.
Line 160: What resolution is required specifically for temporal variations?
Line 163: What is meant by ‘relatively close’ in a quantitative sense?
Section 2.4.2 How are uncertainties in Argo values incorporated into the model (for example, chl, which even when corrected, can have errors, e.g., your figure 4? RMSE of ~ 0.27 or 0.29 in log10 space or by itself (not clear from the figure legend if log10 was applied to obs) is nontrivial)
Why is chlorophyll used from BGC Argo rather than bbp, which is shown to be more reliable with satellite data and also with phytoplankton biomass? Using bbp allows one to calculate both standing stocks (Graff et al 2015) and accumulation rates, which may be compared to the model and allow physiological model errors (Chl:C ratios) to be irrelevant. I know that the authors list bbp and other Argo variates in the concluding remarks. However, bbp may be a simpler case study for the authors to examine. The other Argo variates could be used or discussed for creative model development. For example, zooplankton can be parameterized from Argo data and that is not mentioned well in the text.
Finally, I'm curious about the choice of the model. I'm not surprised that the model can be tuned to match the obserations because it has so many free parameters. The goal is eventually moving beyond an accurate deterministic model and being able to interpret and attribute changes to biogeochemical function. How can that be accomplished given the number of assumptions listed herein? More text on that point will help the reader hoping to employ a similar analysis. I'm not clear on how the model can be used to extend interpretation beyond what can be done from the Argo observations alone.
Some relevant and additional reading:
Yang, B., Fox, J., Behrenfeld, M. J., Boss, E. S., Haëntjens, N., Halsey, K. H., ... & Doney, S. C. (2021). In situ estimates of net primary production in the western North Atlantic with Argo profiling floats. Journal of Geophysical Research: Biogeosciences, 126(2), e2020JG006116.
Lacour, L., Llort, J., Briggs, N., Strutton, P. G., & Boyd, P. W. (2023). Seasonality of downward carbon export in the Pacific Southern Ocean revealed by multi-year robotic observations. Nature Communications, 14(1), 1278.
Bisson, K. M., Boss, E., Westberry, T. K., & Behrenfeld, M. J. (2019). Evaluating satellite estimates of particulate backscatter in the global open ocean using autonomous profiling floats. Optics express, 27(21), 30191-30203.
van Oostende, M., Hieronymi, M., Krasemann, H., Baschek, B., & Röttgers, R. (2022). Correction of inter-mission inconsistencies in merged ocean colour satellite data. Frontiers in Remote Sensing, 3, 74.
Haëntjens, N., Boss, E., & Talley, L. D. (2017). Revisiting O cean C olor algorithms for chlorophyll a and particulate organic carbon in the S outhern O cean using biogeochemical floats. Journal of Geophysical Research: Oceans, 122(8), 6583-6593.
Arteaga, L. A., Behrenfeld, M. J., Boss, E., & Westberry, T. K. (2022). Vertical Structure in Phytoplankton Growth and Productivity Inferred From Biogeochemical‐Argo Floats and the Carbon‐Based Productivity Model. Global Biogeochemical Cycles, 36(8), e2022GB007389.
Graff, J. R., Westberry, T. K., Milligan, A. J., Brown, M. B., Dall’Olmo, G., van Dongen-Vogels, V., ... & Behrenfeld, M. J. (2015). Analytical phytoplankton carbon measurements spanning diverse ecosystems. Deep Sea Research Part I: Oceanographic Research Papers, 102, 16-25.
Haëntjens, N., Della Penna, A., Briggs, N., Karp‐Boss, L., Gaube, P., Claustre, H., & Boss, E. (2020). Detecting mesopelagic organisms using biogeochemical‐Argo floats. Geophysical Research Letters, 47(6), e2019GL086088.
Mignot, A., Claustre, H., Cossarini, G., D'Ortenzio, F., Gutknecht, E., Lamouroux, J., ... & Teruzzi, A. (2023). Using machine learning and Biogeochemical-Argo (BGC-Argo) floats to assess biogeochemical models and<? xmltex\break?> optimize observing system design. Biogeosciences, 20(7), 1405-1422.
Citation: https://doi.org/10.5194/gmd-2023-25-RC1 -
CC1: 'Comment on gmd-2023-25', Martí Galí, 24 May 2023
The article of Yumruktepe and colleagues presents a new approach that allows using BGC-Argo data to improve biogeochemical models. By relaxing the 1D model physics towards Argo float observations, the authors can subsequently focus on improving the biogeochemical model. I read with great interest the article and I believe this approach represents a step forward in this developing field of research.
As a general note, one potential flaw of "along-track" approaches, whereby 1D models are matched to observations made by individual Argo floats, is the assumption of float Lagrangianity. Given that floats profile between 1000 m (sometimes 2000 m) and the surface, it is unlikely that they can track the same water masses at all depths over an extended period. Therefore, very strong relaxation towards observations may force the 1D model physics beyond what is physically reasonable.
With this comment, besides acknowledging the value of the current study, I just wanted to point the authors to other papers published by our group and by others, where different approaches to a similar problem were proposed.
In a recent paper, we proposed an approach whereby along-track BGC-Argo annual time series were matched to pre-computed vertical mixing fields, obtained from 3D simulations. The best-matching dynamical fields were used to force 1D biogeochemical simulations offline, restoring nutrient towards the climatology only below 300 m depth. Finaly, we evaluated the model's ability to capture the dynamics of particulate backscattering observed by BGC-Argo floats across ocean biomes.
Galí, M., Falls, M., Claustre, H., Aumont, O., & Bernardello, R. (2022). Bridging the gaps between particulate backscattering measurements and modeled particulate organic carbon in the ocean. Biogeosciences, 19(4), 1245-1275. https://doi.org/10.5194/bg-19-1245-2022
In a companion paper, we used this 1D modelling framework and a genetic algorithm to optimize biogeochemical parameters that control POC cycling and export, focusing on the subpolar North Atlantic:
Falls, M., Bernardello, R., Castrillo, M., Acosta, M., Llort, J., & Galí, M. (2022). Use of genetic algorithms for ocean model parameter optimisation: a case study using PISCES-v2_RC for North Atlantic particulate organic carbon. Geoscientific Model Development, 15(14), 5713-5737. https://doi.org/10.5194/gmd-15-5713-2022Below I am listing other articles that the authors might find interesting to provide further context to their study:
Hemmings, J. C. P., Challenor, P. G., & Yool, A. (2015). Mechanistic site-based emulation of a global ocean biogeochemical model (MEDUSA 1.0) for parametric analysis and calibration: an application of the Marine Model Optimization Testbed (MarMOT 1.1). Geoscientific Model Development, 8(3), 697-731. https://doi.org/10.5194/gmd-8-697-2015
Kaufman, D. E., Friedrichs, M. A., Hemmings, J. C., & Smith Jr, W. O. (2018). Assimilating bio-optical glider data during a phytoplankton bloom in the southern Ross Sea. Biogeosciences, 15(1), 73-90. https://doi.org/10.5194/bg-15-73-2018
Shu, C., Xiu, P., Xing, X., Qiu, G., Ma, W., Brewin, R. J., & Ciavatta, S. (2022). Biogeochemical Model Optimization by Using Satellite-Derived Phytoplankton Functional Type Data and BGC-Argo Observations in the Northern South China Sea. Remote Sensing, 14(5), 1297. https://doi.org/10.3390/rs14051297
Best of luck with your submission!
Citation: https://doi.org/10.5194/gmd-2023-25-CC1 -
RC2: 'Comment on gmd-2023-25', Anonymous Referee #2, 29 May 2023
This paper presents a designed framework linking Biogeochemical-Argo (BGC-Argo) observations to biogeochemical models in the Nordic Seas. The BGC-Argo and satellite surface temperature were used to evaluate the simulated temperature, salinity, and mixed layer depth. The Modeled chlorophyll a (chl-a) was evaluated/compared against the BGC-Argo chl-a along the BGC-Argo trajectory. The differences between modelled chl-a and BGC-Argo chl-a (Figure 9) indicated that (1) the model failed to reproduce the deep chlorophyll maxima throughout June to September in the 20-50m depth range, and (2) the timing of spring bloom initiation is late from the model. To address these differences, the authors tried to improve the model, but in phytoplankton growth formulation and parameterization section, there are lots of hypotheses, will the hypotheses affect the model setup and outputs? It is hard to see which result is better when compared Fig. 11c (chl-a difference between improved model and BGC-Argo) with Fig. 9e (chl-a difference between model and BGC-Argo).
Also, in the paper, I did not see the discussion on uncertainty of BGC-Argo observations, such as what’s the uncertainty of chlorophyll a, temperature, and salinity from BGC-Argo? Will the uncertainty affect the comparisons between the modelled results and BGC-Argo results?
Citation: https://doi.org/10.5194/gmd-2023-25-RC2
Veli Çağlar Yumruktepe et al.
Veli Çağlar Yumruktepe et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 250 | 58 | 12 | 320 | 16 | 4 | 5 |
- HTML: 250
- PDF: 58
- XML: 12
- Total: 320
- Supplement: 16
- BibTeX: 4
- EndNote: 5
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1