• For Contributors +
• Journal Search +
Journal Search Engine
ISSN : 1225-6692(Print)
ISSN : 2287-4518(Online)
Journal of the Korean earth science society Vol.42 No.4 pp.390-400
DOI : https://doi.org/10.5467/JKESS.2021.42.4.390

# Impact of Iron Scavenging and Desorption Parameters on Chlorophyll Simulation in the Tropical Pacific within NEMO-TOPAZ

Hyomee Lee1, Byung-Kwon Moon1*, Jong-Yeon Park2, Han-Kyoung Kim1, Hyun-Chae Jung3, Jieun Wie1, Hyo Jin Park4, Young-Hwa Byun5, Yoon-Jin Lim6, Johan Lee7
1Division of Science Education & Institute of Fusion Science, Jeonbuk National University, Jeonju 54896, Korea
2Division of Earth and Environmental Science, Jeonbuk National University, Jeonju 54896, Korea
3Mirae Climate Co., Ltd., Seoul 08511, Korea
4Jeonju Jungang Middle School, Jeonju 54991, Korea
5Innovative Meteorological Research Department, National Institute of Meteorological Sciences, Seogwipo 63568, Korea
6Numerical Modeling Center, Korea Meteorological Administration, Seoul 07062, Korea
7Operational Systems Development Department, National Institute of Meteorological Sciences, Seogwipo 63568, Korea
*Corresponding author: moonbk@jbnu.ac.kr Tel: +82-63-270-2824, Fax: +82-63-270-2802
April 30, 2021 August 15, 2021 August 17, 2021

## Abstract

Ocean biogeochemistry plays a crucial role in sustaining the marine ecosystem and global carbon cycle. To investigate the oceanic biogeochemical responses to iron parameters in the tropical Pacific, we conducted sensitivity experiments using the Nucleus for European Modelling of the Ocean–Tracers of Ocean Phytoplankton with Allometric Zooplankton (NEMO-TOPAZ) model. Compared to observations, the NEMO-TOPAZ model overestimated the concentrations of chlorophyll and dissolved iron (DFe). The sensitivity tests showed that with increasing (+50%) iron scavenging rates, chlorophyll concentrations in the tropical Pacific were reduced by approximately 16%. The bias in DFe also decreased by approximately 7%; however, the sea surface temperature was not affected. As such, these results can facilitate the development of the model tuning strategy to improve ocean biogeochemical performance using the NEMOTOPAZ model.

## Introduction

The latest Earth system models consider the ocean biogeochemistry as an important factor (Eyring et al., 2016;Orr et al., 2017) because the ocean biogeochemical cycle impacts Earth system factors, such as ocean and atmosphere (Charlson et al., 1987, Kim et al., 2018, Jung and Moon, 2019). Specifically, the impacts originate from ocean phytoplankton, which is the primary producer. Furthermore, the studies have shown that chlorophyll feedback by phytoplankton in the eastern equatorial Pacific Ocean affects El Niño- Southern Oscillation (Kang et al., 2017;Park et al., 2014;Park et al., 2017). The eastern equatorial Pacific Ocean is one of the high-nutrient, low-chlorophyll regions; moreover, iron limits the ocean primary production (Libes, 2009;Tagliabue et al., 2017;Williams and Follows, 2011). Therefore, accurate simulations of iron and chlorophyll concentration in the equatorial Pacific are important for the Earth system model.

Various sensitivity tests are needed to improve the Earth system model. Furthermore, it requires many computational resources to perform complicated interactions between coupled factors. Hence, sensitivity tests using coupled biogeochemical ocean general circulation model, decoupled with other factors, can serve to improve the Earth system model. Jung et al. (2020) improved the overestimation of chlorophyll simulations of tropical Pacific in the new coupled biogeochemical ocean general circulation model- Nucleus for European Modelling of the Ocean−Tracers of Ocean Phytoplankton with Allometric Zooplankton (NEMO-TOPAZ)−using changed atmospheric iron deposition inputs. However, NEMO-TOPAZ still remains overestimated for chlorophyll concentration. Tagliabue et al. (2016) showed that the large dissolved iron (DFe) diversity between the numerical models participated in the iron model intercomparison project (FeMIP) and the agreement on DFe simulation of models are mainly dependent on calibration of scavenging rates and organic ligands concentrations, more than inputs for atmospheric dust and sedimentary DFe.

We conducted sensitivity tests to calibrate iron scavenging and desorption parameters using NEMOTOPAZ to improve the chlorophyll simulation in the tropical Pacific of the newly developed Earth system model, incorporating NEMO and TOPAZ into the ocean and biogeochemical model. The results were analyzed using the surface chlorophyll concentrations, surface DFe, and sea surface temperature (SST). This paper is structured as follows. Section 2 discusses the experiment methods, model, and reference data. Section 3 describes the results of NEMO-TOPAZ and the sensitivity tests. Finally, Section 4 summarizes and discusses the results.

## Methodology

### Dissolved Iron Parameters

We used the ocean biogeochemistry model NEMOTOPAZ (Jung et al., 2020). TOPAZ implements several iron cycles. It considers iron supply from dust and sediment, iron limitation, food web processing of phytoplankton, and removal and production of DFe and particulate iron (PFe) due to adsorption, desorption, and scavenging (Dunne et al., 2010).

The removal and production of DFe and PFe were expressed as several parameters in NEMO-TOPAZ. We conducted the sensitivity tests of parameters associated with DFe concentrations (Fig. 1), which refers to the form used by phytoplankton. The two phases of iron−DFe and PFe−were transformed into each other through adsorption and desorption processes (Libes, 2009;Tagliabue et al., 2017;Williams and Follows, 2011). In the adsorption process, DFe sticks to the surface of particles; Hence, DFe is removed in seawater onto PFe. Furthermore, PFe sinks to the seafloor; Some PFes are removed from seawater; this process is called scavenging. The iron adsorption, $J F e a d s$, at a time, t, and level, z, is calculated as follows:

$J F e a d s ( t , z ) = F e d ( t , z ) ⋅ min ( k ′ F e max , k ″ F e ⋅ F e d ( t , z ) ) ,$
(1)

where Fed represents the DFe, $k ′ F e max$ denotes the maximum adsorption rate, and $k ″ F e = 1.0 × 10 10$ mol Fe−1 kg d−1 denotes the scavenging parameter, which is tuned to reproduce the iron concentration drop-off at 5° (Johnson et al., 1997; Dunne, 2013).

Desorption is opposite to adsorption, and DFe is released into seawater from PFe. The iron desorption, $J F e d e s$, was calculated as follows:

$J F e d e s ( t , z ) = k F e d e s ⋅ F e p ( t , z ) ,$
(2)

where $J F e d e s$=0.0068 d−1 denotes the desorption rate parameter, which was set using the observed range (0.0041-0.0068 d−1 ) in 12 of the 17 cases of thorium in the ocean presented by Bacon and Anderson (1982) and Dunne (2013). The Fep represents PFe. Hence, DFe was calculated using several variables related to the iron cycle considered in TOPAZ, in which $J F e d e s$ and $J F e a d s$ were added and subtracted, respectively.

### Sensitivity Tests of Iron Parameters

The iron adsorption and desorption rates in the water column vary in reality, and are affected by environmental conditions, such as iron concentration, regions, and depths (Bacon and Anderson, 1982;Johnson et al., 1997;Williams and Follows, 2011; Dunne, 2013; Tagliabue et al., 2017). In the present study, iron parameters of scavenging-associated adsorption and desorption were arbitrarily changed by ±50% to analyze the influence of iron parameters on chlorophyll simulation in the tropical Pacific Ocean (Table 1). Scavenging and desorption were named S and D, respectively. Moreover, the 50% decrease and increase in values were represented 50 and 150, respectively. Therefore, S50 and S150 refer to iron sensitivity experiments with +50% and −50% change in scavenging parameters, respectively. Furthermore, D50 and D150 refer to the same experiments with a +50% and −50% change in desorption parameters. Additionally, the control experiment (CTL) was conducted in which the scavenging and desorption parameters were considered default values.

### Model Setup and Data

The horizontal resolution of NEMO-TOPAZ was approximately 2° of ORCA2 (182×149 grid points) with approximately 0.5° of equatorial meridional resolution. The vertical resolution was 31 levels with the layer thickness of approximately 10 m from the surface, increasing gradually to approximately 500 m at the bottom layer (5,250m). The initial and boundary conditions are used NEMO reference configuration inputs v3.6-patch (NEMO consortium, 2019) and Australian Research Council’s Centre of Excellence for Climate System Science (ARCCSS, 2018) for physical and biogeochemical fields, respectively. Following Jung et al. (2020), the aeolian dust input prescribed to NEMO-TOPAZ was replaced with another ocean biogeochemistry model−the NEMOPelagic Interactions Scheme for Carbon and Ecosystem Studies (NEMO-PISCES). Moreover, Jung et al. (2020) showed that the dust input of PISCES was smaller than TOPAZ and reduced the overestimation of the chlorophyll concentration in the equatorial Pacific Ocean using it. The Coordinated Ocean-ice Reference Experiments−Phase II (CORE-II; Large and Yeager., 2009) was used for atmospheric forcing. The spin-up was performed for 300 years using CORE-II climatology. Subsequently, the model was employed for 62 years using CORE-II from 1948 to 2009. Finally, the data were analyzed for the last 11 years (1998-2008) owing to the limited observation period of chlorophyll.

The reference data were used to compare the experimental results. Specifically, surface chlorophyll was compared with satellite chlorophyll data from Sea-viewing Wide Field-of-view Sensor (SeaWiFS; McClain et al., 1998) between 1998 and 2008. The reference data of DFe used the median of the 13 models participating in the FeMIP (Tagliabue et al., 2016). The FeMIP dataset was used as reference data for model assessment (Orr et al., 2017;Eyring et al., 2016;Tagliabue et al., 2017). The mean biases in the ocean iron concentration of FeMIP compared to observational data ranged from −0.02 to −0.48 nM (Tagliabue et al., 2016). Although FeMIP differs from the observations and has large inter-model diversity, it serves as reference data to overcome the observational limitation that there is no global three-dimensional observation DFe dataset. The SST reference was extended reconstructed sea surface temperature version 4 (ERSSTv4; Huang et al., 2015) in 1998-2008.

## Results

The annual surface chlorophyll concentration in the tropical Pacific Ocean of CTL was compared with those of SeaWiFS (Fig. 2) to analyze the default chlorophyll simulation performance of the NEMOTOPAZ. The surface chlorophyll concentrations of the model experiments were averaged from the surface to 20 m because the SeaWiFS-based chlorophyll concentration mainly represents the surface value owing to light backscattering (Jochum et al., 2009, Jung et al., 2020, Park et al., 2014). Chlorophyll concentration was high on the coast of the eastern equatorial Pacific Ocean owing to the upwelling of nutrients (Williams and Follows, 2011). Moreover, this high concentration pattern expanded to the western Pacific Ocean in a narrow latitudinal range (Fig. 2a). These observed chlorophyll patterns were well reproduced in CTL (Fig. 2b). However, CTL overestimated chlorophyll concentration within the range of 0-20 °N, including the Niño 3 region (>0.3 mg kg−1 ) in the eastern equatorial Pacific Ocean, and expanded further from high concentration region to western and high latitude region (Fig. 2c).

The eastern equatorial Pacific Ocean is known to have a high-nutrient low-chlorophyll (HNLC) region, which has limited marine productivity due to DFe (Aumont and Bopp, 2006, Hamilton et al., 2020, Ito et al., 2019, Schneider et al., 2008). Moreover, Jung et al. (2020) diminished the overestimation error of chlorophyll concentration in the equatorial Pacific Ocean of NEMO-TOPAZ by prescribing atmospheric dust input with low concentration. Therefore, we compared the surface DFe distribution of CTL with FeMIP (Fig. 3) to analyze the impact of DFe on the differences in chlorophyll concentrations between CTL and SeaWiFS. The feature of DFe distribution was similar to chlorophyll concentration. FeMIP showed DFe concentrations of >0.05 nmol kg−1 in the equatorial Pacific Ocean; this high concentration area expanded westward (Fig. 3a). Moreover, CTL also simulated the DFe distribution well compared to FeMIP (Fig. 3b); however, concentrations (>0.12 nmol kg−1 ) and spatial ranges were overestimated (Fig. 3c) like the result of chlorophyll concentration (Fig. 2c). This result is in accord with the results of Tagliabue et al. (2016) that demonstrated the TOPAZ coupled with another physical ocean model (i.e., Modular Ocean Model) shows larger DFe concentration than FeMIP models. Therefore, the results imply that chlorophyll error is affected by DFe error, and improvements in the marine iron cycle process can improve chlorophyll concentration.

We analyzed the SST simulation performance of NEMO-TOPAZ by comparing the annual SST of CTL with ERSSTv4 (Fig. 4) in the tropical Pacific, which was later used to analyze the impact of iron parameter changes on SST. CTL reasonably simulated the observed SST distributions with the high SST (>30 °C) in the western Pacific warm pool and the low SST (~24 °C) in the eastern equatorial Pacific cold tongue (Fig. 4a, b). However, the simulated SST values were lower in the narrow regions near the equator than ERSSTv4 (Fig. 4c). These biases arise owing to the simulation of excessive and narrow cold tongue expansion only by physical ocean model NEMO. Breivik et al. (2015) found similar simulated patterns with NEMO, and these errors were partly because of the low resolution of the model (Graham, 2014, Liu et al., 2004). Vannière et al. (2013) revealed that zonal wind errors in the equatorial Pacific can lead to cold tongue bias because these errors can generate strong upwelling in coupled atmosphere-ocean circulation and ocean-only models. Moreover, several models have shown that the underestimation of the mixed layer can intensify the SST error in equatorial Pacific.

Figure 5 shows the distributions of DFe differences between each experiment and CTL. The DFe concentrations in S50 and D150 increased, and those in S150 and D50 decreased than in CTL. This relationship between DFe concentrations and iron parameters implies the DFe removal and increment in the water column due to scavenging and desorption, as mentioned in Section 2. The DFe concentrations in the Niño 3 region from FeMIP, CTL, S50, S150, D50, and D150 were 0.0591, 0.1140, 0.1282, 0.1063, 0.1134, and 0.1146 nmol kg−1 , respectively (Fig. 6a). The CTL and sensitivity tests overestimated DFe concentrations by approximately twice that of FeMIP. However, this bias was reduced in S150 approximately by 7% (~0.008 nmol kg−1 ) than in CTL (Fig. 6b). Moreover, the correlation coefficient of the pattern also improved slightly in S150 (Fig. 6c). However, it was not significant owing to its small value (0.001). S50 shows a significant increase in DFe concentration and errors; moreover, the differences of D50 and D150 for CTL were smaller than S50 and S150 tests. Therefore, the impact of the scavenging parameter on the DFe concentration is greater than that of the desorption parameter of NEMO-TOPAZ.

We confirmed the effect of changing DFe in sensitivity tests on chlorophyll concentrations (Figs. 7 and 8). We found that chlorophyll concentrations of S50 and D150 in 30 °S-15 °N increased and that of S150 and D50 decreased. The distributions of changed chlorophyll values are consistent with distributions of changed DFe values, as shown in Fig. 5. The chlorophyll concentrations in the Niño 3 region from SeaWiFS, CTL, S50, S150, D50, and D150 were 0.1978, 0.4350, 0.5015, 0.3991, 0.4320, and 0.4374 mg kg−1 , respectively (Fig. 8a). The overestimated chlorophyll concentrations in the Niño 3 region of CTL greatly reduced in S150 by approximately 8% and increased in S50 by approximately 15%. Moreover, the differences of D50 and D150 for CTL were smaller than that of scavenging parameter experiments. S150 showed significant reduction in bias by approximately 16% (Fig. 8b; 0.244 to 0.205), in root mean square error by approximately 13% (Fig. 8c; 0.289 to 0.249), and in standard deviation (STDV) by approximately 24% (Fig. 8e; 0.037 to 0.028) than CTL, while correlation coefficient decreased slightly (Fig. 8d; 0.171 to 0.167). The results show that changing DFe to control scavenging parameters greatly impacts chlorophyll concentration. Moreover, prescribing appropriate scavenging parameters can lead to reasonable primary productivity simulations related with chlorophyll in an ocean biogeochemical model in the tropical Pacific Ocean.

Figure 9 shows the distributions of SST differences between sensitivity tests and CTL. Moreover, NEMOTOPAZ implemented optical feedback, which increases marine temperature due to absorption of the solar radiation by phytoplankton. However, the changes in SST in sensitivity tests are insignificant compared to those in CTL. For example, improvement of STDV bias for chlorophyll concentration in S150 (Fig. 8e) did not significantly impact SST STDV bias related to El Niño-Southern Oscillation variability; it only reduced from 0.283 to 0.281.

## Summary and Discussion

We investigated the influence of iron scavenging and changes in desorption parameters on chlorophyll concentration in the tropical ocean using the NEMOTOPAZ coupled biogeochemical ocean general circulation model. The CTL experiment simulated distributions of DFe, chlorophyll concentration, and SST well compared to reference data. However, simulated values of DFe and chlorophyll were overestimated, and that of SST was underestimated in the tropical Pacific.

Sensitivity experiments showed clear relationships between iron parameters and chlorophyll concentrations: increasing iron scavenging parameter or decreasing desorption parameter decreases chlorophyll concentration in tropical Pacific and vice versa. Chlorophyll is largely affected by changes in scavenging parameter than desorption parameter (Figs. 5-8) because DFe is largely controlled by scavenging parameter. In the Niño 3 region, when the scavenging parameter increased by 50%, the DFe concentrations decreased by ~6% (from 0.1140 to 0.1063 nmol kg−1 ). It led to reduction in chlorophyll concentration by ~8% (from 0.44 to 0.40 mg kg−1 ). Moreover, SST changes in sensitivity tests were insignificant, which will be considered in the future study.

We demonstrated that tuning a scavenging parameter in NEMO-TOPAZ affects DFe and chlorophyll concentrations in a coupled biogeochemical ocean general circulation model. In the additional experiments, compared to CTL, the bias of chlorophyll concentrations decreased by approximately 29% and 61% when the scavenging parameter was increased to 200% and 500% (i.e., S200 and S500), respectively. Further careful sensitivity tests are required to determine the appropriate scavenging parameters. Moreover, other variables also need to be verified because chlorophyll concentration is related to various tracers, such as nutrients. The iron sources considered in NEMOTOPAZ are dust deposition and sedimentary sources (Dunne et al., 2010;Tagliabue et al., 2016). Furthermore, the effects of the scavenging parameters of DFe were not sufficient to reduce the error in chlorophyll concentrations. Jung et al. (2020) found an improvement in chlorophyll concentration by changing the iron deposition data to a reduced value. Sensitivity tests for iron parameters and iron sources are needed to improve chlorophyll simulation.

Examining the biogeochemical cycles and feedbacks in Earth system model are important issue in CMIP6 (Eyring et al., 2016), which may be achieved based on the reasonable performance for primary production of ocean in each model. Therefore, the application of calibrated scavenging parameters of the Earth system model using NEMO-TOPAZ as the coupled biogeochemical ocean general circulation model can improve the simulation of ocean primary productivity.

## Acknowledgments

This work was funded by the Korea Meteorological Administration Research and Development Program under Grant KMI (KMI2018-03513) and the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2019R1A2C 1008549). The main calculations were performed by using the supercomputing resource of the Korea Meteorological Administration (National Center for Meteorological Supercomputer). This work is part of the Ph.D. thesis of H. Lee.

## Figure

Schematic diagram of iron cycle in NEMO-TOPAZ associated with sensitivity tests for iron scavenging (k″Fe) and desorption (kFeads) parameters (yellow box). Red and blue arrows represent production and removal of dissolved iron (DFe), respectively. Sinking to seafloor of particulate iron (PFe) is represented with blue arrow because DFe is removed by scavenging.

Annual surface chlorophyll concentration (mg kg−1 ) in tropical Pacific of (a) SeaWiFS, and (b) CTL in 1998-2008. (c) Differences between (b) CTL and (a) SeaWiFS. Purple box indicates the Niño 3 region.

Annual surface dissolved iron (DFe) concentration (nmol kg−1 ) in tropical Pacific of (a) FeMIP, and (b) CTL (1998- 2008). (c) Differences between (b) CTL and (a) FeMIP.

Annual sea surface temperature (SST; °C) in tropical Pacific in 1998-2008 of (a) ERSSTv4, and (b) CTL. (c) Differences between (b) CTL and (a) ERSSTv4.

Differences of DFe concentration (nmol kg−1 ) between sensitivity tests and CTL in 1998-2008; (a) S50-CTL, (b) S150- CTL, (c) D50-CTL, and (d) D150-CTL.

(a) The annual mean DFe concentration (nmol kg−1 ) of FeMIP (black), CTL (purple), S50 (red), S150 (blue), D50 (sky blue), and D150 (yellow) for Niño 3 region. Model experiments were averaged in 1998-2008. (b) Bias, (c) pattern correlation coefficient between each model experiment and that of FeMIP.

Differences of surface chlorophyll concentration (mg kg−1 ) between sensitivity tests and CTL in 1998-2008; (a) S50- CTL, (b) S150-CTL, (c) D50-CTL, and (d) D150-CTL.

(a) The annual mean sea surface chlorophyll concentration (mg kg−1 ) of SeaWiFS (black), CTL (purple), S50 (red), S150 (blue), D50 (sky blue), and D150 (yellow) for Niño 3 region in 1998-2008. (b) Bias of annual mean, (c) root mean square error (RMSE), (d) correlation coefficient, and (e) bias of standard deviation between anomalies of each model experiment and that of SeaWiFS.

Differences of SST (°C) between sensitivity tests and CTL in 1998-2008; (a) S50-CTL, (b) S150-CTL, (c) D50-CTL, and (d) D150-CTL.

## Table

List of sensitivity tests for iron parameters. The tests were performed to analyze the impacts of scavenging (S) and desorption (D) parameters on surface chlorophyll concentration in tropical Pacific. The default values refer to the values used in the CTL experiment. The parameter information within parentheses of second row represent the variable name in Eqs. (1) and (2) and its unit.

## Reference

1. ARCCSS,2018, MOM5 input data, Australian Research Council’s Centre of Excellence for Climate System Science. https://climate-cms.unsw.wikispaces.net/Data. Accessed 22 November 2018.
2. Aumont, O. and Bopp, L. ,2006, Globalizing results from ocean in situ iron fertilization studies. Global Biogeochemical Cycles, 20(2), GB2017,
3. Bacon, M. P. and Anderson, R. F. ,1982, Distribution of thorium isotopes between dissolved and particulate forms in the deep sea. Journal of Geophysical Research: Oceans, 87(C3), 2045-2056.
4. Breivik, Ø. , Mogensen, K. , Bidlot, J. R. , Balmaseda, M. A. , and Janssen, P. A. ,2015, Surface wave effects in the NEMO ocean model: Forced and coupled experiments. Journal of Geophysical Research, Oceans, 120(4), 2973-2992,
5. Charlson, R. J. , Lovelock, J. E. , Andreae, M. O. , Warren, S. G. ,1987, Oceanic phytoplankton, atmospheric sulphur, cloud albedo and climate. Nature, 326, 655- 661,
6. Dunne, J. P. , Gnanadesikan, A. , Sarmiento, J. L. , and Slater, R. D. ,2010, Technical description of the prototype version (v0) of Tracers of Phytoplankton with Allometric Zooplankton (TOPAZ) ocean biogeochemical model as used in the Princeton IFMIP model. Biogeosciences, 7(11), 3593.
7. Eyring, V. , Bony, S. , Meehl, G. A. , Senior, C. A. , Stevens, B. , Stouffer, R. J. , and Taylor, K. E. ,2016, Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geoscientific Model Development, 9, 1937-1958, 10.5194/gmd-9-1937-2016.
8. Graham,2014, The importance of eddy permitting model resolution for simulation of the heat budget of tropical instability waves. Ocean Modelling, 79, 21-32,
9. Hamilton, D. S. , Moore, J. K. , Arneth, A. , Bond, T. C. , Carslaw, K. S. , Hantson, S. , Ito, A. , Kaplan, J. O. , Lindsay, K. , Nieradzik, L. , Rathod, S. D. , Scanza, R. A. , and Mahowald, N. M. ,2020, Impact of changes to the atmospheric soluble iron deposition flux on ocean biogeochemical cycles in the Anthropocene. Global Biogeochemical Cycles, 34(3), e2019GB006448,
10. Huang, B. , Banzon, V. F. , Freeman, E. , Lawrimore, J. , Liu, W. , Peterson, T. C. , Smith, T. M. , Thorne, P. W. , Woodruff, S. D. , and Zhang, H.-M. ,2015, Extended Reconstructed Sea Surface Temperature Version 4 (ERSST.v4). Part I: Upgrads and Intercomparisons. Journal of Climate, 28(3), 911-930,
11. Ito, A. , Ye, U. , Yamamoto, A. , Watanabe, M. , and Aita, M. N. ,2019, Responses of ocean biogeochemistry to atmospheric supply of lithogenic and pyrogenic ironcontaining aerosols. Geological Magazine, 157(5), 741- 766,
12. Jochum, M. , Yeager, S. , Lindsay, K. , Moore K. , and Murtugudde, R. ,2009, Quantification of the feedback between phytoplankton and ENSO in the community climate system model. Journal of Climate, 23, 2916- 2925,
13. Johnson, K. S. , Gordon, R. M. , and Coale, K. H. ,1997, What controls dissolved iron concentrations in the world ocean? Marine chemistry, 57(3-4), 137-161.
14. Jung, H.-C. and Moon, B.-K ,2019. Impacts of Albedo and Wind Stress Changes due to Phytoplankton on Ocean Temperature in a Coupled Global Ocean-biogeochemistry Model. Journal of the Korean Earth Science Society, 40(4), pp. 392-405,
15. Jung, H.-C. , Moon, B.-K. , Lee, H. , Choi, J.-H. , Kim, H.-K. , Park, J.-Y. , Byun, Y.-H. , Lim, Y.-J. , and Lee, J ,2020. Development and Assessment of NEMO(v3.6)- TOPAZ(v2), a Coupled Global Ocean Biogeochemistry Model. Asia-Pacific Journal of Atmospheric Sciences, 56, 411-428,
16. Kang, X. , Zhang, R. H. , Gao, C. , and Zhu, J. ,2017, An improved ENSO simulation by representing chlorophyllinduced climate feedback in the NCAR Community Earth System Model. Scientific Reports., 7, 1-9,
17. Kim, A.-H. , Yum, S. S. , Lee, H. , Chang, D. Y. , Shim, S. ,2018, Polar cooling effect due to increase of phytoplankton and dimethyl-sulfide emission. Atmosphere, 9, 384,
18. Large, W.G. and Yeager, S.G. ,2009, The global climatology of an interannually varying air-sea flux data set. Climate Dynamics, 33, 341-364,
19. Libes, S. M. ,2009, Introduction to Marine Biogeochemistry. Academic Press, New York, USA, 909 p.
20. Liu, H. , Zhang, X. , Li, W. , Yu, Y. , and Yu, R. ,2004, An Eddy-Permitting Oceanic General Circulation Model and Its Preliminary Evaluation. Advances in Atmospheric Sciences, 21(5), 675-690,
21. McClain, C. R. , Cleave, M. L. , Feldman, G. C. , Gregg, W. W. , Hooker, S. B. , and Kuring, N. ,1998, Science quality seaWIFS data for global biosphere research. Sea Technology, 39, 10-16.
22. NEMO Consortium. (2019). NEMO Reference configurations inputs (Version v3.6-patch) [Data set]. Zenodo.
23. Orr, J. C. , Najjar, R. G. , Aumont, O. , Bopp, L. , Bullister, J. L. , Danabasoglu, G. , Doney, S. C. , Dunne, J. P. , Dutay, J.-C. , Graven, H. , Griffies, S. M. , John, J. G. , Joos, F. , Levin, I. , Lindsay, K. , Matear, R. J. , McKinley, G. A. , Mouchet, A. , Oschlies, A. , Romanou, A. , Schlitzer, R. , Tagliabue, A. , Tanhua, T. , and Yool, A. ,2017, Biogeochemical protocols and diagnostics for the CMIP6 Ocean Model Intercomparison Project (OMIP). Geoscientific Model Development, 10, 2169-2199,
24. Park, H.-J. , Moon, B.-K. , Wie, J. , Kim, K.-Y. , Lee, J. , and Byun, Y.-H. ,2017, Biophysical effects simulated by an ocean general circulation model coupled with a biogeochemical model in the tropical Pacific. Journal of Korean Earth Science Society, 38, 469-480,
25. Park, J.-Y. , Kug, J.-S. , Seo, H. , and Bader, J. ,2014, Impact of bio-physical feedbacks on the tropical climate in coupled and uncoupled GCMs. Climate Dynamics, 43, 1811-1827,
26. Schneider, B. , Bopp, L. , Gehlen, M. , Segschneider, J. , Frölicher, T. L. , Cadule, P. , Friedlingstein, P. , Doney, S. C. , Behrenfeld, M. J. , and Joos, F. ,2008, Climateinduced interannual variability of marine primary and export production in three global coupled climate carbon cycle models. Biogeosciences, 5, 597-614,
27. Tagliabue, A. , Aumont, O. , DeAth, R. , Dunne, J. P. , Dutkiewicz, S. , Galbraith, E. , Misumi, K. , Moore, J. K. , Ridgwell, A. , Sherman, E. , Stock, C. , Vichi, M. , Völker, C. , and Yool, A. ,2016, How well do global ocean biogeochemistry models simulate dissolved iron distributions? Global Biogeochemical Cycles, 30(2), 149-174,
28. Tagliabue, A. , Bowie, A. R. , Boyd, P. W. , Buck, K. N. , Johnson, K. S. , and Saito, M. A. ,2017, The integral role of iron in ocean biogeochemistry. Nature, 543 (7643), 51-59,
29. Vannière, B. , Guilyardi, E. , Madec, G. , Doblas-Reyes, F. J. , and Woolnough, S. ,2013, Using seasonal hindcasts to understand the origin of the equatorial cold tongue bias in CGCMs and its impact on ENSO. Climate Dynamics, 40(3-4), 963-981.
30. Williams, R. G. and Follows, M. J. ,2011, Ocean dynamics and the carbon cycle: Principles and mechanisms. Cambridge University Press, New York, USA, 404 p.