An ecosystem model for the North Pacific embedded in a general circulation model

Journal of Marine Systems 25 Ž2000. 129–157 www.elsevier.nlrlocaterjmarsys

An ecosystem model for the North Pacific embedded in a general circulation model Part I: Model description and characteristics of spatial distributions of biological variables Michio Kawamiya a,) , Michio J. Kishi b, Nobuo Suginohara a a

Center for Climate System Research, UniÕersity of Tokyo, 4-6-1, Komaba, Meguro-ku, Tokyo 153-8904, Japan b Faculty of Fisheries, Hokkaido UniÕersity 3-1-1, Minato-cho, Hakodate, Hokkaido 041-8611, Japan Received 26 February 1999; accepted 10 January 2000

Abstract An ecosystem model is embedded in an ocean general circulation model ŽOGCM. and the ecosystem–physical combined model is applied to the North Pacific. The OGCM yields realistic physical environments concerning the mixed layer depth ŽMLD. and vertical flow except for MLD off Sanriku coast, Japan, and on both sides of the equator. The modeled nitrate, chlorophyll zooplankton, net primary production, and total organic nitrogen are within the order of magnitude of the observations. The biological activities are low in the subtropical region and high in the subpolar and the equatorial region as observations show. The modeled primary production rate shows a local maximum along the subpolar front, which seems to have a counterpart in the observation. There, the temperature and the nutrient condition are best combined for photosynthesis. The modeled concentrations differ from the observations in some respects: chlorophyll concentration is low in the subpolar region and high in the subtropical and the equatorial region compared with the Coastal Zone Color Scanner observation; the equatorial net primary production rate is higher than the observed. Part of the disagreement may be ascribed to the photoadaptation process, which is not included in the model andror the simple extrapolation of the temperature dependence of photosynthesis. q 2000 Elsevier Science B.V. All rights reserved. Keywords: ecosystem model; North Pacific; GCM simulation; net primary production

1. Introduction How the pelagic ecosystem in the ocean responds to changes in its environments is an urgent problem )

Corresponding author. Current affiliation: Institut fur ¨ Meereskunde Kiel, Universitat Kiel, Dusternbrooker Weg 20, ¨ D-24105 Kiel, Germany. Tel.:q49-431-597-3973. E-mail address: [email protected] ŽM. Kawamiya..

awaiting an answer. The response may lead to variation in strength of the biological pump, and thereby it may have an impact on the climate. However, due to insufficiency of reliable data on the variation of the biological pump, the present ocean models for predicting the future CO 2 change have assumed that there has been no variation in the biological pump over the whole ocean Že.g., Sarmiento and Orr, 1992..

0924-7963r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 4 - 7 9 6 3 Ž 0 0 . 0 0 0 1 2 - 9

130

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

This assumption is partly supported by Falkowski and Wilson Ž1992. who showed that chlorophyll concentration inferred from Secchi disk data collected in the North Pacific does not suggest any significant change over the past 70 years. There are, however, some indications that the ocean biota is indeed experiencing changes on decadal time scales: Venrick et al. Ž1987. suggested that in the subtropical gyre of the North Pacific, chlorophyll concentration may have been doubled since 1968; Brodeur and Ware Ž1992. reported that in the northeastern Pacific doubling took place in the zooplankton biomass between the periods 1956–1962 and 1980–1989; Roemmich and McGowan Ž1995. stated that the zooplankton biomass has decreased by 80% off southern California since 1951. In addition, decadal changes on the higher trophic level Že.g., Beamish and Bouillon, 1993. may be reflecting some changes on the lower trophic level, which may alter strength of the biological pump. It is recognized that, in many cases, causes of ecosystem changes can be attributed to those in physical environments. One of the most important physical factors that may affect the oceanic ecosystem is vertical mixing: it may enhance phytoplankton growth by bringing nutrients up to the surface layer from under the euphotic zone, or it may suppress photosynthesis by carrying phytoplankton away from the surface layer where light is replete. Whether strengthening of vertical mixing acts positively or negatively to the oceanic biota depends on delicate balance amongst various conditions such as nutrient concentration and light intensity. Indeed, many researchers reporting the ecosystem changes have discussed the relation between the ecosystem changes and changes in the mixed layer depth ŽMLD.. For example, Venrick et al. Ž1987. speculated that associated with the decadal climate variation in the North Pacific Že.g., Nitta and Yamada, 1989. the mixed layer deepened and the biological activities in the subtropical Pacific were enhanced. Furthermore, Polovina et al. Ž1995. showed that winter MLD is changing on decadal scales over the North Pacific, and suggested that there is a possibility that the biological activities are deeply affected. A numerical model can be a powerful tool for investigating ecosystem variations. To work on this problem, processes occurring within the mixed layer

should be faithfully represented in the model in both physical and biological aspects. The model of Bacastow and Maier-Reimer Ž1990. is not very suitable for this kind of problem: since their main interest is to reproduce the distribution of tracers in the deep layer, the ecosystem in the surface layer incorporated in their ocean general circulation model ŽOGCM. is extremely simplified so that the relation between MLD and biological activities is only poorly resolved. Improvements were made with their model or other approaches were taken in the subsequent studies Že.g., Bacastow and Maier-Reimer, 1991; Najjar et al., 1992; Anderson and Sarmiento, 1995; Yamanaka and Tajika, 1996; Fasham et al., 1993; Sarmiento et al., 1993; Six and Maier-Reimer, 1996., but many of them are for a better reproduction of tracer distribution in the deep layer, not for a more elaborate representation of the pelagic ecosystem. Among them, however, the approach taken by Fasham et al. Ž1993. and Sarmiento et al. Ž1993. is essentially different from the others. They incorporated ecosystem components such as phytoplankton and zooplankton in an OGCM as explicit model variables so that phytoplankton and zooplankton are directly subject to vertical mixing. This means that the relation between the mixed layer and the ecosystem is much better represented than in other models. They applied their model to the North Atlantic and obtained fairly good results. Although there are many points that should be improved, their model demonstrates great potential for dealing with the issue of ecosystem variations. Six and Maier-Reimer Ž1996. also developed an ecosystem model embedded in an OGCM as an extension of the model of Bacastow and Maier-Reimer Ž1990.; they showed that incorporation of ecosystem dynamics can improve tracer distributions by reducing the magnitude of the undesired subsurface nutrient maximum in the equatorial region Žso-called ‘‘nutrient trapping’’.. In this paper, an ecosystem model, which explicitly includes ecosystem components is embedded in an OGCM. The combined model is applied to the North Pacific, where many researchers have reported ecosystem variations. The ecosystem model was tested with its one-dimensional version by Kawamiya et al. Ž1995, 1997. who used data from Ocean Weather Station ŽOWS. Papa and the Bermuda Atlantic Time-Series ŽBATS. site. They showed that

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

the one-dimensional model bears fairly good results even if it is applied to the totally different oceanic regimes. Distributions of ecosystem variables obtained by the combined model, especially that of primary production, are compared with observations and discussed. This kind of investigation will be able to provide a basis for studying observed distributions. This paper is organized as follows. In Section 2, model description is given and results of the OGCM are compared with observations to confirm that the physical model yields realistic environments for the ecosystem model. In Section 3, we compare results of the combined model with observations mainly on the annual mean basis. We will investigate seasonal variations in the accompanying paper Ž$$Kawamiya et al., 2000, referred to as KKSb hereinafter.. In Section 4, mechanisms determining spatial patterns in the model are discussed with the focus on primary production. Discussion on processes neglected in the model is also given. Lastly, summary and conclusion of this paper are presented in Section 5. Details of the ecosystem model are provided in Appendix A.

2. Model description Here, the OGCM and the ecosystem model are described. The OGCM is the one developed at the Center for Climate System Research, University of Tokyo. Tsujino and Suginohara Ž1998. give detailed analysis on its application to the North Pacific, although the model configurations such as damping time for the surface restoring are different. The ecosystem model is basically the same as that developed by Kawamiya et al. Ž1995, 1997..

131

scheme ŽMellor and Yamada, 1982., which was improved by Kantha and Clayson Ž1994. is used. Isopycnal diffusion is introduced adopting the method proposed by Redi Ž1982. and simplified by Cox Ž1987.. Background diffusivity and viscosity are given in Table 1 along with isopycnal diffusivity. The model domain extends from 1228E to 728W and from 238S to 638N, corresponding to the North Pacific. The horizontal grid interval is 28 = 28. There are 45 levels in the vertical, and the vertical grid spacing is listed in Table 2. From the sea surface to the intermediate depths, the grid intervals are very small compared to those adopted in other modeling studies. This fine resolution is needed to reproduce well the stratification in the surface layer and thus the seasonal variation of the mixed layer, which has a significant impact on the ecosystem model. At the sea surface, the monthly mean wind stress of Hellerman and Rosenstein Ž1983. is imposed as the boundary condition. Surface temperature and salinity are restored to the monthly mean data of Levitus and Boyer Ž1994.. The damping time is 0.2 days. This value is extremely small even if the thinness of the top level Ž3 m. is taken into consideration. The value corresponds to about 3 days for the top level with the thickness of 50 m, while a typical value in OGCMs is about 30 days for that thickness. This value is taken in order to prevent development of MLD from having a large time lag. If we take 30 days, it causes delay of more than 1 month in seasonal variation of MLD at places where winter MLD is relatively deep. This delay should be avoided in this study where the time variation of MLD is of great significance. All model integrations are carried out with seasonal variations. As for time interpolation of the sea surface data, the method of Killworth Ž1996. is used so that the values imposed on the

2.1. Description of the OGCM The OGCM employs the primitive equations with hydrostatic, Boussinesq, and rigid-lid approximations. Convective adjustment is achieved by setting the vertical diffusion coefficient to 1000 cm2rs when a water column is statically unstable. For the mixed layer model, the Mellor–Yamada’s level 2.5 closure

Table 1 Diffusivity and viscosity in OGCM in cm2 rs units Isopycnal diffusivity Horizontal diffusivity Horizontal viscosity Vertical diffusivity Vertical viscosity

1=10 7 3=10 6 8=10 8 0.3 1.0

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

132 Table 2 Vertical grid spacing

For later reference, the formulation of photosynthesis is presented below:

Grid number

Grid interval Žm.

Depth of grid point Žm.

1 2 3–21 22 23 24–32 33–35 36 37 38 39 40–42 43–45

3 7 10 20 30 50 80 110 150 200 300 500 600

3 10 200 220 250 700 940 1050 1200 1400 1700 3200 5000

Ž Photosynthesis. s GPP Ž Chl, NH 4 , NO 3 , T , I . s Vmax

½

NO 3 NO 3 q K NO 3 NH 4

q

NH 4 q K NH 4

ž

= 1y

ž

I Iopt

/

5

= exp Ž kT .

Chl,

z

I s I0 exp y

H0 k Ž z . d z

I Iopt

exp

Ž 2.

/

k Ž z . s a 1 q a 2 Chl Ž z . , model are equal to those of the original monthly dataset when averaged over a month. The amplitudes of seasonal variation are, however, taken smaller over the three meridional grids adjacent to the southern boundary so that they vanish at the southernmost grids. At the southern boundary, temperature and salinity are restored to the annual mean values of Levitus and Boyer Ž1994.. The damping time is 50 days as the southernmost grids.

exp Ž yc NH 4 .

,

Ž 3. Ž 4.

where z denotes depth in meter. The full set of the formulation is given in Appendix A. The notation of the variables and the parameters are listed in Table 3 along with the parameter values. The daily and

2.2. Description of the ecosystem model The ecosystem model has six compartments, namely, phytoplankton, zooplankton, nitrate, ammonium, particulate organic nitrogen ŽPON., and dissolved organic nitrogen ŽDON.. Its structure is shown in Fig. 1. Time evolution of each compartment is described as follows: ECi Et

s yn P = Ci q = Ž K= C . q Ž biological term .

Ž 1. where Ci is any one of the six compartments, Õ current velocity, and K diffusion tensor. The mathematical formulations for the biological terms are those described by Kawamiya et al. Ž1995, 1997..

Fig. 1. Schematic of the ecosystem model. Boxes represent nitrogen-based standing stocks and arrows represent nitrogen flows in the ecosystem. Dashed arrows represent the exchange with deep layer through diffusion and advection.

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

133

Table 3 Parameters used in the equation for photosynthesis Chl NH 4 NO 3 T I I0 Vma x k K NO 3 K NH 4 Iopt a1 a2 C

Phytoplankton Žnitrogen-based. Ammonuim Nitrate Temperature Light intensity Light intensity at the sea surface Maximum photosynthetic rate at 08C Temperature coefficient for photosynthetic rate Half saturation coefficient for nitrate Half saturation coefficient for ammonium Optimum light intensity Light dissipation coefficent of sea water Self shading coefficient Ammonium inhibition coefficient

seasonal variation of the surface light intensity Ž I0 . are calculated using astronomical formulae, based on the monthly mean dataset of solar radiation by daSilva et al. Ž1994.. To express the distribution of sinking flux of PON, the formula of Martin et al. Ž1987. is adopted: sinking flux of PON below the depth of 100 m is given as, F Ž z . s F100

z

ž / 100

y0 .858

,

Ž 5.

where F denotes sinking flux of PON, and F100 sinking flux of PON across the 100-m depth. Above that depth, it is assumed that PON sinks with a constant velocity, 20.0 mrday, and PON flux across the 100-m depth calculated with this assumption equals F100 . This value of sinking velocity corresponds to that adopted for the simulation of OWS Papa by Kawamiya et al. Ž1997. and is larger than that for the BATS site. This larger value is necessary to reproduce a realistic nitrate distribution: preliminary experiments showed that, when the value is set to the smaller one used for the BATS site, the high nitrate concentration occurring in the equatorial region spreads out to the subtropical region by the Ekman transport and hence the sea surface nitrate distribution becomes unrealistically large. The half saturation constant for nitrate Ž K NO 3 . is set to 0.03 mmolr1, which is the value adopted in the simulation at the BATS site, i.e., the oligotrophic region. This choice is justified as follows. Nitrate concentration in the eutrophic region often has a

variable variable variable variable variable variable 1.200 0.063 0.03 1.0 0.07 0.035 0.0281 1.5

mmolrl mmolrl mmolrl 8C ly ly rday r8C mmolrl mmolrl lyrmin rm lrmmol N m lrmmol

much higher value than that of K NO 3 while in the oligotrophic region it has a similar or smaller value compared with K NO 3 . In most cases, the photosynthetic rate is well saturated in the eutrophic region with respect to nitrate concentration, and therefore taking the smaller K NO 3 does not affect the nitrate uptake. Sensitivity to change in K NO 3 is far higher in the oligotrophic region than it is in the eutrophic region. Although the small K NO 3 will not do any harm in most of the eutrophic region, it may have some significant effect on the nitrate concentration front between the eutrophic and the oligotrophic region or on the mesotrophic region in the central equatorial region because nitrate concentration in these regions is comparable to the observed K NO 3 . However, it is difficult to incorporate the spatial variation of K NO 3 . One way to include it in the model is to make K NO 3 a function of ambient nitrate concentration so that K NO 3 becomes large when the concentration is high and vice versa, following the suggestion of Harrison et al. Ž1996. that K NO 3 shows an excellent positive correlation with the ambient nitrate concentration. A preliminary experiment showed, however, that it was difficult to prevent the equatorial high nitrate concentration from penetrating into the oligotrophic region when the above K NO 3 variation is incorporated. It is not very difficult to understand the reason: when nitrate concentration becomes larger in the frontal region between the equatorial eutrophic region and the subtropical oligotrophic region, K NO 3 also becomes larger and the uptake rate of nitrate decreases;

134

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

thus, the equatorial high nutrient concentration easily penetrates into the subtropical region. Incorporation of the K NO 3 variation introduces an undesirable positive feedback mechanism. We conclude that it is better to let K NO 3 constant for the present study. The choice of the half saturation constant for ammonium Ž K NH 4 . is not very straightforward because ammonium concentration has a value similar to K NH 4 even in the eutrophic region. Here, it is set to 1.0 mmolrl, which is the value adopted in the simulation of the eutrophic region, OWS Papa. This is based on the consideration that ammonium inhibition process included in Eq. Ž2., which will be more important in the eutrophic region, is dependent on ammonium concentration and that K NH 4 strongly regulates ammonium concentration. Furthermore, the decay rate of DON is taken very small, a constant value of 0.01 dayy1 , compared to the value of Kawamiya et al. Ž1997.. This corresponds to taking the role of semi-refractory DON into consideration ŽKirchamn et al., 1993., while Kawamiya et al. Ž1997. neglected it. The semirefractory DON is incorporated not to spread out the high nutrient concentration from the equatorial region, which results in disappearance of the oligotrophic region in the North Pacific. Preliminary experiments showed that this parameter change affects nutrient concentration only near the equator; it had minor effects on the other variables in other regions. The photosynthetic rate Vmax is increased to 1.2 dayy1 from the value used by Kawamiya et al. Ž1997., 1.0 dayy1 , and the ratio of extracellular excretion to photosynthesis is also increased to 0.3 from 0.135. Nitrate concentration near the southern boundary is restored to the value of Conkright et al. Ž1994. in the same way as for temperature and salinity.

and the OGCM is integrated for 50 years without the acceleration method. The time step is 1 h. In the year 50, the upper layer above a few hundred meters shows a nearly steady cycle. Annual mean temperature averaged over the upper 300 m of the whole domain is changing at a rate as small as ; 10y3 8Cryear. Then, the ecosystem model is integrated together with the OGCM taking the result at the year 50 as the initial condition for the physical field. The time step is 2 h for the ecosystem model and 1 h for the OGCM. The variables in the OGCM averaged over two time steps are supplied to the ecosystem model. The initial value for nitrate is

2.3. Time integration We take two procedures to spin up the OGCM. First, we integrate it using the acceleration method of Bryan Ž1984. for 2000 deep water years. This integration period corresponds to 200 years for the upper 1000 m. For this procedure, vertical diffusivity and viscosity at the depths above 50 m are set to a large constant value, 10 cm2rs, without the mixed layer model. Next, the mixed layer model is incorporated

Fig. 2. Time variation of modeled Ža. nitrate and Žb. chlorophyll in the first layer averaged over the whole domain.

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

taken from Conkright et al. Ž1994.. For other biological variables except phytoplankton and zooplankton, the initial values are set to zero everywhere. The initial value for phytoplankton is 0.053 mmol Nrl Žcorresponding to chlorophyll concentration of 0.1 mgrl using the C:N mol ratio of 133:17 ŽTakahashi et al., 1985. and the common C:Chlorophyll ratio of 50. at the depths above 100 m and zero below; for zooplankton, 0.12 mmol Nrl above 100 m and zero below. The integration period is 6 years. There still exists a trend in nitrate concentration but a steady cycle is established for the other variables in the ecosystem model ŽFig. 2.. Monthly averaged data are created in the model. Results in the sixth year,

135

comprised of 12 datasets for each variable, will be analyzed. For the time evolution of nitrate concentration, vertical diffusion process plays an important role and it has a long time scale of a few thousand years. Consequently, the trend unavoidably exists when the integration is initialized with a climatological data because of inevitable differences between the model and the reality. To obtain the steady state, an integration of a few thousand years is necessary. Such a long integration is not feasible here because the acceleration method of Bryan Ž1984. cannot be used due to the inclusion of the ecosystem and the mixed layer model. As we focus on a rather short time scale

Fig. 3. Zonal sections along 1808 from the sea surface to the depth of 1000 m for Ža. modeled and Žb. observed annual mean temperature, and Žc. modeled and Žd. observed annual mean salinity. For the annual means of the model, results from the last year of integration are used. The observations are taken from Levitus and Boyer Ž1994.. Contour intervals are 18C and 0.1 psu for temperature and salinity, respectively.

136

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

of 1 year, it is still informative to analyze the results in the sixth year before the steady state is established. 2.4. Comparison of results of the OGCM with obserÕations Before comparing the results of the ecosystem model with observations, it is helpful to see to what extent the OGCM offers realistic environments to the ecosystem model. Fig. 3 shows annual mean temperature and salinity distributions from the sea surface down to the depth of 1000 m along 1808 for the model and the observation. The modeled temperature distribution bears overall resemblance with the observed but shows that waters at the intermediate depths are warmer than the observed, which is common to other models Že.g., Tsujino and Suginohara, 1998.. The model yields a salinity minimum at the intermediate depths though it is not well developed. The modeled salinity minimum whose center is around 238N and 600 m depth seems to be isolated

but, to the east, it is actually connected to the low salinity water lying in the north. Next, we look at the surface circulation. Fig. 4 shows the annual mean flow field averaged over the upper 100 m. Major features of the surface circulation are reproduced: the Kuroshio and the Oyashio flow along the western boundary; the strongly divergent westward flow exists along the equator; the Alaskan gyre is formed in the northeastern part of the North Pacific. A closer look can, of course, reveal some defects of the flow field: the separation latitude of the Kuroshio is located further north than the observed Že.g., Wyrtki, 1975.; the complicated flows on the northern side of the equator is only partly reproduced, e.g., the Equatorial Counter Current does not exist in the western part. These defects may be inevitable in a coarse resolution model like the present. In practice, the horizontal flow field does not directly affect growth of phytoplankton: contribution from the horizontal advection term in the governing equation for phytoplankton is smaller than other terms such as photosynthesis and grazing.

Fig. 4. Annual mean flow field averaged over the upper 100 m.

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

But it indirectly affects phytoplankton growth through its effect on MLD distribution, vertical flow induced by its divergence and convergence, and temperature dependence of the ecosystem model equations. Thus, we next discuss how well the model reproduces MLD distribution and intensity of vertical flow. MLD needs to be realistically reproduced for the purpose of this study, modeling of the oceanic ecosystem. Here, MLD is defined as the depth at which density difference from the sea surface first exceeds the critical value corresponding to temperature difference of 0.58C at each horizontal grid point. Distributions of the modeled and the observed MLD are shown in Fig. 5 for March and August, when MLD has its extremes at most locations. In March, it is well reproduced that MLD is large in the western

137

part at middle latitudes. The maximum MLD is, however, much larger in the model than in the observation Ž; 500 m in the model while ; 200 m in the observation.. This discrepancy arises from the problem on the separation of the Kuroshio. For a coarse resolution model of this kind, the Kuroshio penetrates much further north than in reality, carrying a warm water further to the north. The water is then cooled by the imposed reference temperature originating from the real Oyashio, thus making MLD too large. Adopting a longer damping time for the surface boundary condition will ease this problem. But we consider that avoiding the lag in MLD variation, as discussed in Section 2.1, is more important because the lag seriously distorts the model phytoplankton growth. The modeled MLD is larger

Fig. 5. Mixed layer depth ŽMLD. for March in Ža. the model and Žb. the observation, and for August in Žc. the model and Žd. the observation. See text for the definition of MLD. The observed MLD is derived from the data of Levitus and Boyer Ž1994.. Contour intervals are 25 m.

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

138

takes the maximum value of ; 250 cmrday near 1408W at the depth of 50 m, and that there is a decreasing trend toward the east. Bryden and Brady Ž1989. also estimated the equatorial upwelling based on mooring current meter data from 1979 to 1981; they showed that the upwelling averaged between 1528W and 1108W takes the maximum value of ; 400 cmrday at the depth of 70 m. The model is well compared with the observations in that intensity of the modeled upwelling is within the range of data, and that its distribution shows the decreasing trend from 1508W toward the east. Fig. 6. Annual mean vertical velocity along the equator. Contour intervals are 50 cmrday. Positive values indicate upwelling.

on both sides of the equator. The reason for this discrepancy is not clear but presumably it is also due to the coarse horizontal resolution: it is recognized that a much finer meridional resolution Ž; 0.38. is needed to realistically reproduce the thermal structure near the equator Že.g., Rosati and Miyakoda, 1988.. In August, the decreasing tendency of MLD toward the north is well reproduced in the model. The unrealistically deep mixed layer near the equator, however, still remains. Vertical velocity near the sea surface is also important for the ecosystem because it contributes to the vertical transport of nutrients. At middle and high latitudes, it is mainly determined by the Ekman velocity w E , which is expressed in the form, t w E s curl Ž 6. f z

3. Comparison of ecosystem model results with observations 3.1. Nitrate Although the integration period of the ecosystem model, 6 years, is not long enough to modify the

ž /

where t is wind stress vector, f is Colioris parameter, and the subscript z denotes the vertical component of curl. As seen in Eq. Ž6., at middle latitudes, vertical velocity is not strongly dependent on model configurations such as horizontal resolution or surface boundary conditions. In the equatorial region, however, it is expected that vertical velocity is highly model dependent. Fig. 6 depicts distribution of vertical velocity along the equator. Upwelling dominates and it takes the maximum value of ; 350 cmrday around 1508W at the depth of 60 m. Halpern et al. Ž1989. calculated the equatorial upwelling between 1408W and 1108W on the basis of mooring current meter data from 1983 to 1984; they reported that it

Fig. 7. Annual mean nitrate concentration averaged over the surface 100 m for Ža. the model and Žb. the observation ŽConkright et al., 1994.. Contour intervals are 2 mmolrl.

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

initial condition of nitrate in the deep ocean, it is sufficient to affect the nitrate distribution at the depths above ; 200 m. Comparison of the modeled nitrate distribution with the observed serves as a test of model performance. Fig. 7 shows the annual mean nitrate concentration averaged over the surface 100 m for the model and the observation ŽConkright et al., 1994.. The model maintains the overall pattern that nitrate is more abundant in the equatorial and the subpolar region, where the Ekman upwelling transports nitrate upward. It also keeps high concentration caused by the coastal upwelling along the California coast. In the northeastern part of the domain, however, concentration is lower in the model. A reason for the disagreement seems to be that the modeled MLD is too shallow in March ŽFig. 5., although it is difficult to identify the ultimate cause because the surface nitrate concentration is determined by combination of various factors such as upwelling, vertical mixing, and sinking of PON. In the eastern part, nitrate concentration is too low at ; 108N. This is presumably because upwelling induced by wind stress curl along this latitude is only poorly resolved due to the coarse horizontal resolution. As shown in Fig. 2, nitrate concentration has not reached a steady state yet. The tongue of high con-

139

centration along the equator is still gradually extending. Also, nitrate in the subpolar gyre is gradually increasing. The trend can be grasped in Fig. 8, where the surface nitrate concentration is plotted for the fifth and the sixth year of integration. As for the trend of the equatorial tongue, the inability of the model to resolve the complex current structure may be responsible, though the equatorial upwelling appears to have an appropriate strength ŽSection 2.4.. For example, Aumont et al. Ž1999. have pointed out that the Equatorial Undercurrent ŽEUC. may play an important role in determining nitrate concentration of the surface equatorial ocean. With the present resolution, it is impossible to reproduce the correct depth and strength of the EUC. 3.2. Chlorophyll 3.2.1. Comparison with satellite data In Fig. 9, the modeled chlorophyll abundance near the sea surface is compared with the satellite observation by Coastal Zone Color Scanner ŽCZCS. during 1978–1986. In this figure, nitrogen based phytoplankton concentration is converted to chlorophyll concentration using the C:N mol ratio of 133:17 and the common C:Chlorophyll weight ratio of 50 Žthe same as those in Section 2.3.. The model data

Fig. 8. Annual mean nitrate concentration averaged over the surface 100 m for the fifth year Ždashed contourlines. and the sixth year Žsolid contourlines. of integration. Contour intervals are 2 mmolrl.

140

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

Fig. 9. Annual mean chlorophyll concentration in mgrl obtained by Ža. the model and Žb. the satellite observations. The model result is averaged over the uppermost 10 m. The observations show the mean for the observations period 1978–1986.

are averaged over the surface 10 m so as to be directly compared with the CZCS data, which can provide averaged information only for the surface layer of ; 10 m Že.g., Gordon et al., 1982..

The model bears the general tendency that concentration is high in the subpolar and the equatorial region and low in the subtropical region corresponding to the nitrate distribution. The modeled concen-

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

tration is high along the California coast due to the coastal upwelling, which is consistent with the observation. There is, of course, disagreement between the model and the observation: the modeled concentration is too high in the equatorial and the subtropical region and too low in the subpolar region; near the equator, the highest abundance occurs along 58 latitude on both sides of the equator, while in the observation it occurs along the equator. The observed surface chlorophyll concentration in the subpolar region is lower than that in the model. This may be partly due to errors in the satellite measurement: CZCS tends to overestimate chlorophyll abundance at latitudes higher than ; 408N in fall and winter ŽYoder et al., 1993.. According to Anderson et al. Ž1977. who summarized surface chlorophyll data at OWS Papa Ž508N, 1458W. during 1959–1976, the annual mean surface concentration is about 0.3 mgrl while it is about 0.6 mgrl in the CZCS observation at this location.

141

The higher concentration in the subtropical and the equatorial region may be caused by the fact that the model does not incorporate the photoadaptation process. The NrChlorophyll ratio of phytoplankton tends to increase near the sea surface in the subtropical and the equatorial region to adjust to the intense light ŽWinn et al., 1996.. We speculate that the low chlorophyll concentration in the observation does not necessarily mean that the phytoplankton biomass is also low. The chlorophyll maxima on both sides of the equator are caused by vertical mixing there, which is too strong as is found in the MLD distribution ŽFig. 5.. Near the equator, chlorophyll is more abundant at the depth of about 60 m. Off the equator, however, the chlorophyll maximum at that depth is destroyed by the strong vertical mixing so that high concentration appears at the sea surface. When chlorophyll is integrated over the upper 100 m, the maximum exists exactly along the equator Žcf. Figs. 10 and 11..

Fig. 10. Chlorophyll concentration along 1758E obtained by Ža. the model and Žb. the observations conducted as a part of the Northwest Pacific Carbon Cycle Study ŽNOPACCS. from August 7 to October 5 in 1992. The model result is the average for August through September. Contour intervals are 0.1 mgrl.

142

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

Fig. 11. Latitudinal distribution of depth-integrated Ž0–200 m. chlorophyll obtained by the model Žannual mean, solid line. and observations ŽChavez and Smith, 1995, crosses.. The data are collected within the longitudinal range of 160–1008W in various seasons. The observation points are shown in Chavez et al. Ž1995.. The model result is also averaged over the longitudinal range.

the sea surface is presumably photoadaptation as explained above. Chavez and Smith Ž1995. collected chlorophyll data in the eastern Pacific and provided the distribution of integrated chlorophyll abundance. Fig. 11 shows their data and the corresponding model result. The model result lies within the range of data scattering and reproduces the trend that concentration is high in the equatorial and the subpolar region and low in the subtropical region. The overestimation by the model in the subtropical region is not as large as it looks in Fig. 9. The model shows better similarity with the observations in Figs. 10 and 11 than in Fig. 9.

As a whole, the model result is reasonably well compared with the observation: the obtained values are within the order of the observation; the model can reproduce the overall pattern that chlorophyll is abundant in the subpolar and the equatorial region and scarce in the subtropical region. 3.2.2. Comparison with in situ data Chlorophyll along 1758E has been extensively measured by the Northwest Pacific Carbon Cycle Study ŽNOPACCS, Ishizaka and Ishikawa, 1991.. The chlorophyll section in 1992 ŽNOPACCS, 1992. is depicted Fig. 10 with the corresponding model result. The model captures some of the observed features: the depth of the subsurface chlorophyll maximum ŽSCM. in the model agrees with that in the observation; SCM in the model deepens in the subtropical gyre as in the observation; an isolated maximum at the equator exists in both the model and the observation. Moreover, the absolute value of the modeled concentration is well compared with that of the observed. The extremely large value at 408N in the observation is not a permanent feature as it was not observed in another year ŽIshizaka et al., 1994.. On the other hand, the isolated maximum at the equator seems to be persistent except during the period of El Nino ˜ ŽBidigare and Ondrusek, 1996.. The cause for the modeled higher concentration near

Fig. 12. Distribution of zooplankton biomass based on Ža. the model and the data compiled by Žb. Bogorov et al. Ž1968. and Žc. Reid Ž1962.. The model result is averaged over the surface 150 m. The units are mmol Nrl for Ža. and Žb. and in mlrŽ10 3 m3 . for Žc.. Darkest shading in Žc. indicates values approaching 1 mlrm3.

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

3.3. Zooplankton Bogorov et al. Ž1968. and Reid Ž1962. compiled data on the zooplankton biomass from literature as shown in Fig. 12b and c, respectively. The units of the original data of Bogorov et al. Ž1968. Žmg wet weight per m3 . have been converted into mmol Nitrogen per liter Žmmol Nrl. by Glover et al. Ž1994. so that direct comparison can be made with the model result ŽFig. 12a.. The modeled zooplankton biomass is averaged over the surface 150 m, below which no significant biomass is found. Although a conversion factor between the units in Fig. 12a and c has not been firmly established, qualitative discussion is possible and the conversion is not made here. It can be seen that the modeled biomass is of the same order as that in Fig. 12b. In addition, as in the case of chlorophyll, the model reproduces the feature that the biomass abundance is high in the subpolar and the equatorial region and low in the subtropical region. Although the high abundance in the model along ; 428N does not exist in Fig. 12b, it does exist in Fig. 12c.

143

Abundance along the equator looks much higher in the model when Fig. 12a is compared with Fig. 12b, while the discrepancy seems less serious when Fig. 12a is compared with Fig. 12c: abundance along the equator is of the same order as that in the subpolar region in Fig. 12c. Roman et al. Ž1995. measured zooplankton abundance on the equator at 1408W during March and April, and reported a value of ; 0.7 mmol Crl, which is equivalent to Q 0.1 mmol Nrl when the C:N mol ratio of 133:17 is used. Thus, the modeled zooplankton abundance seems high in the equatorial region by a factor of 2–3. This discrepancy may reflect the very high net primary production rate in the equatorial region, which will be shown in Section 3.4. Although the temporal and the spatial extent is limited, a new dataset can be found in Sugimoto and Tadokoro Ž1997., who compiled historical zooplankton data for June and July to the north of 358N in the North Pacific. Their result is presented in wet weight, and is converted into mmol Nrl using a conversion factor used by Wroblewski et al. Ž1988. Ž1 mg wet weight per m3 s 10y4 mmol Nrl.. The data and the corresponding model result in Fig. 13 show similar

Fig. 13. Distribution of zooplankton biomass averaged for June and July, in Ža. the model and Žb. the data compiled by Sugimoto and Tadokoro Ž1997.. The units are mmol Nrl.

144

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

values in the subarctic Pacific. The maximum along 458N extending from 1608E to 1708W in Fig. 13b does not exist in Fig. 13a, although it is found in the annual mean result ŽFig. 12a.. The rough agreement in the zooplankton distribution between the model and the observations may suggest that the modeled phytoplankton is under a proper grazing pressure, because the modeled grazing is calculated by multiplying zooplankton biomass by some factors Žsee Eq. A6.. The grazing is the major sink for the modeled phytoplankton and thus it is very important for the phytoplankton seasonal variation, which will be analyzed extensively by KKSb.

3.4. Net primary productiÕity Fig. 14a shows annual mean net primary productivity ŽNPP. obtained by the model. In the model, NPP is defined as the phytoplankton growth term minus the ‘‘respiration’’ term Žsee Appendix A.. Formation of DON was not taken into account when computing NPP because in the standard technique for measuring NPP Ž14 C technique. dissolved organic matters are filtered out. For the conversion from nitrogen-based value to carbon based one, the same ratio as in Section 2.3 is used. As for observational estimate of NPP, global maps of NPP derived from the CZCS measurement with empirical laws have been proposed by three groups, Longhurst et al. Ž1995., Antoine et al. Ž1996., and Behrenfeld and Falkowski Ž1997.. Although some differences can be found among the three maps, the overall patterns and the values are similar to one another. Detailed comparison among the three maps is made by Behrenfeld and Falkowski Ž1997.. Here, we compare the model result with the map of Behrenfeld and Falkowski Ž1997. ŽFig. 14b. simply because it is the newest one. This map will be referred to as ‘‘observation’’ though in a strict sense it is not based only on the observation. Again, the value is generally high in the subpolar and the equatorial region and low in the subtropical region for both the model and the observation. The high NPP zone between the subpolar and the subtropical region seen in the observation seems to be

reproduced by the model though it does not extend to the eastern boundary in the model. A mechanism forming this NPP distribution will be investigated in Section 3.5. The major discrepancy occurs in the equatorial region. The maximum value in the model exceeds 500 g C my2 yeary1 while in the observation it is only ; 150 g C my2 yeary1 . The too deep modeled MLD shown in Section 2.4 should not be relevant to this overestimation: because the equatorial region is already nutrient rich because of upwelling, deep MLD does not act to increase NPP. One of the possible reasons for this overestimation is simple extrapolation of temperature dependence of the photosynthetic rate: although the dependence is based on the report of Eppley Ž1972., his equation Žincorporated in Eq. 2. seems to become unreliable around R 278C because the number of experimental data is small in this temperature range. In the model, the annual mean temperature above 100 m depth is ; 288C where NPP is much higher than in the observation. Uncertainty of NPP estimation in the equatorial region is, however, quite large because of high variability there. Chavez et al. Ž1996. reported that the mean NPP for the equatorial Pacific from 908W to 1808, and from 58S to 58N, is 330 g C my2 yeary1 , a much larger value than that seen in Fig. 14b. The corresponding average of the modeled NPP is 450 g C my2 yeary1 . Though NPP in the model is still higher than in the observation, it may not be as unrealistic as it looks in Fig. 14. Photosynthesis is the only source term for phytoplankton. Thus, together with the result shown in Section 3.3, it is suggested that both the source and the major sink Žgrazing. may have a proper order.

3.5. Particulate and dissolÕed organic nitrogen Libby and Wheeler Ž1997. measured PON and DON Žtotal organic nitrogen, TON. in the central and the eastern equatorial Pacific. The modeled TON averaged over the surface 40 m ŽFig. 15a. is compared with their data ŽFig. 15b.. The modeled value is similar to the observed although it is slightly underestimated. Two characteristic features in the

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

145

Fig. 14. Annual mean net primary productivity Ža. obtained by the model and Žb. derived from the satellite observation and empirical laws Žredrawn from Behrenfeld and Falkowski, 1997.. The units are g carbon per m2 per year Žg C my2 yeary1 ..

146

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

4. Discussion 4.1. Distribution of primary productiÕity

Fig. 15. Total organic nitrogen Žparticulate and dissolved organic nitrogen. averaged over the surface 40 m, in Ža. the model and Žb. the observation by Libby and Wheeler Ž1997.. The model result shows the average for September through November because the observation was carried out during this period. Solid circles in Žb. show observation points. Contour intervals are 0.5 mmolrl.

data can be found in the model result, i.e., the minimum along the equator and the tendency to decrease toward the east. The equatorial minimum is formed because TON is biologically produced while a watermass is exported away from the equator. However, the cause for the eastward decrease in the observation is not as clear as in the case of the equatorial minimum. In this model, the tendency is caused by the fact that vertical mixing can penetrate more deeply in the western region. Because the optimal depth for photosynthesis is deeper than 40 m at this location, the surface TON concentration becomes higher if mixing reaches the optimal depth. The observed tendency that the subsurface TON maximum becomes more distinct eastward ŽFig. 3 of Libby and Wheeler Ž1997.. is not inconsistent with what is occurring in the present model.

Here, we analyze spatial patterns of NPP obtained by the model. An interesting feature in Fig. 14a is the high NPP belt extending to the east from the coast of Japan along ; 358N. Because it seems to have a counterpart in the observation ŽFig. 14b., it is very intriguing to explore the reason of this modeled high NPP. First, we investigate why the high NPP belt exists along this latitude. The term for photosynthesis in Eq. Ž2. is decomposed into three parts, namely, dependence on nutrient ŽwNO 3r NO 3 q K NO 3 4xexp yc NH 4 4 q NH 4r NH 4 q K NH 4 4., temperature Žexp kT 4., and light Ž IrIopt 4 exp 1 y IrIopt 4.. Fig. 16 shows the three conditions and their product along 1698E. It is clearly seen that the nutrient condition is favorable for photosynthesis to the north and the temperature condition is so to the south. The light condition does not have a distinct meridional gradient. After all, as can be seen in the product ŽFig. 16d., the overall condition becomes most suitable for photosynthesis at ; 358N where both the nutrient and the temperature condition are modestly good. In other words, this is the location where the nutrient and the temperature condition are best combined for photosynthesis. Because of the most suitable condition for phytoplankton growth, the annual mean standing stock of phytoplankton also has a local maximum along this latitude Žcf. Fig. 2 of KKSb.. Thus, the productivity has its maximum there. Next, we examine a mechanism forming the zonal distribution on this high NPP belt. The maximum at ; 1708E occurs mainly because nitrate is vigorously transported from under the euphotic zone even in summer at this location. To see this, the photosynthetic term is again decomposed, and the nutrient condition in August along 348N is plotted in Fig. 17a. Around 1708E, the nutrient condition is favorable for photosynthesis near the surface. The reason for this can be found in Fig. 17b, where the vertical diffusion coefficient in August is depicted. Right under the euphotic zone Žto the depth of ; 100 m., a watermass with large diffusivity is present at ; 1708E and thus nitrate is easily brought up to the

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

147

Fig. 16. Decomposition of the term for photosynthesis in Eq. Ž2. along 1698E. The upper three figures represent dimensionless dependence on Ža. nutrient, Žb. temperature, and Žc. light, respectively. See text for details. The product of the three parts is given in Žd.. The model results are taken from January as an example.

148

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

Fig. 17. Ža. Nutrient dependence Ždimensionless. for photosynthesis based on the decomposition of the term for photosynthesis in Eq. Ž2., and vertical diffusion coefficient in cm2rs expressed in common logarithm for Žb. August and Žc. March. The section is along 348N.

surface. The effect of this large diffusivity is visible only at the depths shallower than ; 100 m because

below this depth the nitrate concentration is well above the half saturation constant Ž K NO 3 in Eq. 2..

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

This watermass with the large diffusivity is, in turn, formed from winter to early spring when surface cooling induces weak stratification and thus leads to large diffusivity. By comparing Fig. 17b with Fig. 17c, where the vertical diffusion coefficient in March is depicted, it is understood that the watermass with large diffusivity in Fig. 17b is a remnant of that with weak stratification in Fig. 17c. Concerning the mechanism determining the zonal structure along ; 358N, the deep, vigorous vertical mixing from winter to early spring is important. However, as can be seen in Fig. 5, the depth to which the intense vertical mixing penetrates is much shallower in the observation than in the model: in March, it is only ; 150 m in the observation while it is as deep as ; 300 m in the model. Thus, it is unlikely that this mechanism is also functioning in the real world though the possibility cannot be completely denied. Meridionally, on the other hand, the high NPP belt in the model seems to exist also in the observation ŽFig. 14b. as mentioned above. For creating Fig. 14b, Behrenfeld and Falkowski Ž1997. have adopted an empirical relation between temperature and photosynthetic rate, which is similar to that in our model, i.e., the photosynthetic rate increases rapidly with temperature. In the relation adopted by Behrenfeld and Falkowski Ž1997., the increase of the photosynthetic rate ceases at ; 208C. Nevertheless, it is possible that the same mechanism as in our model works also in their calculation because the annual mean SST north of 358N is less than 208C. Therefore, the high NPP belt in Fig. 14b may not be independent of that in our model. The empirical relation of Behrenfeld and Falkowski Ž1997. is, however, based on compilation of in situ data unlike that exploited in our model ŽEppley, 1972., which is derived from laboratory experiments. In that sense, the agreement between Fig. 14a and b on the existence of the high NPP belt provides an observational basis for the mechanism forming the belt in the model. An area of high NPP is also found in the central equatorial region. The NPP of this area is, however, significantly higher than that in the observation. Thus, it will not be examined in detail. Here, we will suffice it to say that the same mechanism forming the meridional distribution around 358N is also in

149

effect. The temperature condition is more suitable for photosynthesis to the west because water temperature is higher in the west than in the east, and the nutrient condition is so to the east as seen in Fig. 7. Around 1808, the two conditions are best combined for photosynthesis. 4.2. Neglected processes The structure of the present model is very simple compared with that of real ecosystem. Many processes that exist in reality are not explicitly included in the model. Among them, importance of iron limitation and that of nitrogen fixation are often pointed out by recent papers. Here, we discuss the implication of neglecting these processes. 4.2.1. Iron limitation Some recent papers Že.g., Martin and Fitzwater, 1988. state that iron limitation is essential for maintaining High Nutrient Low Chlorophyll ŽHNLC. in the subpolar andror equatorial region of the North Pacific. In this study, HNLC is basically reproduced. One of the reasons of this success is that we adopted relatively small value for photosynthetic rate Ž Vmax in Table 4.. With this small value, iron limitation is implicitly taken into account as Chai et al. Ž1996. have pointed out. Boyd et al. Ž1998. suggested that production in the northeast Pacific may be temporarily enhanced through injections of iron transported by wind. If these injections occur randomly, the enhancement is also implicitly expressed on an annual time scale through parameter tuning. However, if the frequency of the infections varies systematically on an annual time scale, the present model cannot express such an effect; it may be necessary to include iron as an explicit variable as in the model of Loukos et al. Ž1998.. It is an important task to determine the characteristic time scale of this effect by observations. 4.2.2. Nitrogen fixation Karl et al. Ž1997. suggested that nitrogen fixation significantly contributes to nitrogen budget in the subtropical Pacific. The present model does not take this process into consideration. If we incorporate it in the model, that would of course result in increase of

150

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

Table 4 Notation and parameter values used in the ecosystem model. Parameters in Table 3 are also listed Vma x k K NO 3 K NH 4 Iopt a1 a2 C g R0 kR M P0 kMP GR ma x kg l Chl ) a b MZ 0 kMZ V PI 0 V PI T V PD 0 V PD T V DI 0 V DI T k N0 k NT S

Maximum photosynthetic rate at 08C Temperature coefficient for photosynthetic rate Half saturation coefficient for nitrate Half saturation coefficient for ammonium Optimum light intensity Light dissipation coefficent of sea water Self shading coefficient Ammonium inhibition coefficient Ratio of extracellular excretion to photosynthesis Respiration rate at 08C Temperature coefficient for respiration Phytoplankton mortality rate at 08C Temperature coefficient for phytoplankton mortality Maximum grazing rate at 08C Temperature coefficient for grazing Ivlev constant Threshold value for grazing Assimilation efficiency of zooplankton Growth efficiency of zooplankton Zooplankton mortality rate at 08C Temperature coefficient for zooplankton mortality PON decomposition rate at 08C Žto inorganic nitrogen. Temperature coefficient for PON decomposition Žto inorganic nitrogen. PON decomposition rate at 08C Žto DON. Temperature coefficient for PON decomposition Žto DON. DON decomposition rate at 08C Temperature coefficient for DON decompotion Nitrification rate at 08C Temperature coefficient for nitrification Sinking velocity of PON

NPP. Indeed, the modeled NPP in the subtropical Pacific Žsee Fig. 14. is near the lower limit of the NPP estimated by Letelier et al. Ž1996. Ž46–385 g C my2 yeary1 . in the subtropical Pacific. Currently, however, data are not enough for nitrogen fixation to be justifiably incorporated in an ecosystem model. Furthermore, if nitrogen fixation is incorporated, it may be also necessary to incorporate denitrification process to keep the balance of source and sink for nitrogen. This will also require much effort since none of the existing ecosystem model for a basin Žor larger. scale includes denitrification process explicitly. McGillicuddy et al. Ž1998., on the other hand, showed the possibility that nitrogen transport by mesoscale eddies is more important than nitrogen fixation for the nitrogen budget. The importance of nitrogen fixation is not firmly established yet. Deter-

1.2 0.063 0.03 1.0 0.07 0.035 0.0281 1.5 0.3 0.03 0.0519 0.0281 0.069 0.3 0.0693 1.4 0.043 0.7 0.3 0.0585 0.0693 0.05 0.0693 0.05 0.0693 0.01 0.0 0.03 0.0693 20

rday r8C mmolrl mmolrl lyrmin rm lrmmol N m lrmmol rday r8C lrmmol N day r8C rday r8C lrmmol N mmol Nrl

1rmmol N day r8C rday r8C rday r8C rday r8C rday r8C mrday

mination of nitrogen budget in the subtropical regions is critical for future development of an ecosystem model for a basin Žor larger. scale. 4.3. Export production and its ratio to NPP Fig. 18a shows export production flux across 200 m, that is, sinking flux of particulate organic carbon ŽPOC. at that depth. Here, the flux of PON obtained by the model is converted to that of POC using the C:N ratio in Section 2.3. Basically, the structure of the distribution reflects that of NPP ŽFig. 14a.. In the equatorial region, however, there is relatively a large difference between the two distributions due to the effect of strong current. According to Karl et al. Ž1996., the annual mean POC flux at 200 m depth at OWS ALOHA Ž228N, 1588W. is ; 20 mg C my2 dayy1 . At OWS Papa, Honjo Ž1996. reports the

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

Fig. 18. Ža. Annual mean export production flux across 200 m. The PON flux obtained by the model is converted to the POC flux using the ratio given in Section 2.3. The contour intervals are 10 mg C my2 dayy1 . Žb. Ratio of the annual mean export production to the NPP in Fig. 14a. The contour intervals are 0.05.

results of POC flux observations during March– August, 1985. According to his Fig. 7.8, the POC flux at 1000 m depth averaged over this period is ; 8 mg C my2 dayy1 . If this value is converted to that of 200 m depth using the formula of Martin et al. Ž1987. ŽEq. 5., it gives the value of 32 mg C my2 dayy1 . Also, Honjo et al. Ž1995. observed POC flux in the equatorial region along 1458W. Based on their data, the annual mean POC flux at 880 m depth on the equator can be calculated as 4.6 mg C my2 dayy1 . This corresponds to 16 mg C my2 dayy1 at 200 m depth when it is converted in the same manner. When the above values are compared with those in Fig. 18a, it is seen that the model generally bears larger values, especially in the equatorial region. This may reflect the high NPP of the model in this region ŽFig. 14a..

151

The ratio of the export production to NPP is shown in Fig. 18b. Here, we refer to this ratio as the e-ratio, following Honjo Ž1996.. It may be considered as corresponding to the f-ratio. The f-ratio itself is not shown here because the f-ratio in this particular model is not a very informative value due to the fact that nitrification process is included even in the ocean surface. The verification for this is given in Appendix A. In Fig. 18b, the e-ratio is low in the equatorial region because the decomposition rate of PON is higher here due to high temperature. The reason why the e-ratio is high in the subtropical region is not clear, but this may be caused by the definition of this ratio. Namely, because NPP is high at deeper depths in the subtropical region due to the formation of SCM ŽFig. 10a., PON in this region is subject to less decomposition before it reaches 200 m depth thereby resulting in a higher e-ratio. The maximum of e-ratio is found in the northeast of the subtropical region probably because the two conditions, that is, SCM formation and low temperature, are both favorable for producing a high e-ratio. Sarmiento et al. Ž1993. showed the opposite tendency for the f-ratio that it is quite low in the subtropical region. This difference may be caused, at least partly, by the difference in definition. For example, in the definition of the f-ratio, the new production includes the production using nitrate transported to the ocean surface by upwelling immediately after nitrification occurring just below the euphotic layer Žabove 200 m depth.. The export production shown here does not include such production.

5. Summary and conclusion We have developed a three-dimensional ecosystem model embedded in an OGCM. The ecosystem model is the one that was tested using its one-dimensional version at a couple of OWSs ŽKawamiya et al., 1995, 1997.. The vertical resolution of the OGCM from the sea surface to the intermediate depths is very fine compared with other OGCMs. A mixed layer model is incorporated. The OGCM provides realistic environments for the ecosystem model concerning MLD and vertical velocity. The overall MLD distribution in the model

152

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

is similar to that in the observation. Intensity of the equatorial upwelling is within the range of data and it shows the decreasing trend from 1508W toward the east as seen in the observation. The major defects of the model are MLD off Sanriku, Japan, which is too deep in March and also MLD on both sides of the equator throughout the year. The former is caused by the model’s inability to correctly reproduce the separation of the Kuroshio due to the coarse horizontal resolution. The cause of the latter is not very clear but again the coarse horizontal resolution may be relevant. The results of the ecosystem model have been compared with the observations. The model bears basin-wide distributions of nitrate, chlorophyll, zooplankton, and NPP similar to those in the observations, in that their values are of the same order as the observed, and that the general feature common to these variables is reproduced: the values are small in the subtropical region and large in the subpolar and the equatorial region. There are, however, notable discrepancies between the model and the observations. Chlorophyll concentration at the sea surface in the model is lower in the subpolar region and higher in the subtropical and the equatorial region than it is in the CZCS observation. The former can be ascribed at least partly to the error in the CZCS data, while the latter to the neglect of the photoadaptation process in the model. The modeled high NPP in the equatorial region may be also due to that the temperature dependence of the photosynthetic rate is simply extrapolated to the equatorial region. The mechanisms determining the distribution of the modeled NPP have been examined. The high NPP belt along ; 358N in the western North Pacific is formed because at this latitude the nutrient and the temperature condition are best combined for photosynthesis. The high NPP belt is also seen in the observation, and it is possible that the mechanism is applicable to the real ocean. The modeled NPP takes the highest value at ; 1708E in this belt because nitrate is vigorously transported by vertical mixing even in summer at this location. However, it is unlikely that the active nutrient transport is also occurring in the real ocean. On the whole, the model results are acceptable in the present state of the art in ecosystem modeling on the basin scale. The similarity of the model results to

the observations in the present study is not worse than that in, for example, Sarmiento et al. Ž1993.. We recognize the model results as worth analyzing in detail. The seasonal variations in the model are extensively examined in the accompanying paper, KKSb.

Acknowledgements We would like to thank T. Sugimoto, I. Koike, M. Kawabe, M. Takahashi, and I. Yasuda for helpful comments. Thanks are extended to M. Miki for helping produce several figures. We also appreciate the comments of the three anonymous reviewers, which contributed greatly to the improvement of the manuscript.

Appendix A. Governing Equations of the Ecosystem Model A.1. Structure The model is nitrogen based, and composed of following six compartments: phytoplankton ŽChl., zooplankton ŽZOO., nitrate ŽNO 3 ., ammonium ŽNH 4 ., PON, and DON. Bacterial biomass and size distribution of planktons are neglected. This is partly due to difficulty in obtaining reliable data for constructing a model in which those effects are taken into account. The role of microbial loop is not explicitly included in our model, but it can be implicitly expressed by tuning zooplankton mortality because microbial loop means material flux from bacteria to zooplankton. Another important role of microbial community, that is, remineralization, is also implicitly included as decomposition of organic matters as shown later. A.2. Formulation of each process A.2.1. Photosynthesis Photosynthesis is assumed to be a function of phytoplankton concentration, temperature, nutrient concentration, and intensity of light. For the dependence on nutrient concentration, Michaelis–Menten formula is adopted. Ammonium inhibition is taken

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

153

into account Wroblewski Ž1977.. To express the dependence on light intensity, the formula used by Steele Ž1962., by which light inhibition can be expressed, is employed. As for the dependence on temperature, it is assumed that the photosynthetic rate is nearly doubled when temperature increases by 108C ŽEppley, 1972.. Similar assumption is adopted for other processes that depend on temperature. The formulation is shown in the text as Eqs. Ž2. – Ž4.. Here, the ratio of the nitrate uptake to the nitrogen uptake, R NO 3 , is defined for the simplicity of later description as follows:

dependent on temperature. Then, respiration can be expressed by:

R NO 3

Ž Mortality of phytoplankton. s M P0 exp Ž k M P T . Chl 2 ; Ž A4.

NO 3 s

NO 3 q K NO 3 NO 3 NO 3 q K NO 3

exp Ž yc NH 4 .

exp Ž yc NH 4 . q

.

NH 4

Ž Respiration of phytoplankton. s R 0 exp Ž k R T . Chl. Ž A3. The parameter value for this respiration is assumed to be similar to that of the real respiration. A.2.4. Mortality Following Steele and Henderson Ž1992., mortality of phytoplankton and zooplankton is assumed to be proportional to the second power of plankton concentration and be dependent on temperature:

Ž Mortality of zooplankton. s MZ 0 exp Ž k M Z T . ZOO 2 . Ž A5.

NH 4 q K NH 4

Ž A1.

A.2.2. Extracellular excretion Extracellular excretion is assumed to be proportional to photosynthesis,

Ž Grazing. s GR Ž T ,Chl,ZOO . s Max  0, GR max exp Ž k g T .

Ž Extracellular excretion. s g GPP Ž Chl, NH 4 ,NO 3 ,T , I . .

A.2.5. Grazing Grazing is expressed as a function of temperature, phytoplankton concentration, and zooplankton concentration:

Ž A2.

A.2.3. Respiration of phytoplankton In this model, the respiration process is included although the model is based on nitrogen, which is not exchanged directly through this process. This ‘‘respiration’’ does not have a counterpart in reality, but is included just to prevent the situation where NPP has a positive sign even at a depth, which is obviously below the compensation depth. The same kind of process is also included in the model of Horiguchi and Nakata Ž1995.. ‘‘Respiration’’ of phytoplankton is assumed to be proportional to phytoplankton concentration and be

=  1 y exp Ž l Ž Chl ) y Chl . . 4 ZOO 4 , Ž A6. where Max a,b4 equals to the larger value of a and b. In this formulation, the grazing rate is saturated when phytoplankton concentration is sufficiently large, while no grazing occurs when phytoplankton concentration is lower than the critical value Chl ) . A.2.6. Excretion and digestion by zooplankton Excretion and digestion are assumed to be proportional to grazing:

Ž Excretion. s Ž a y b . GR Ž T ,Chl,ZOO . ;

Ž A7.

Ž Digestion. s Ž 1 y a . GR Ž T ,Chl,ZOO . .

Ž A8.

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

154

A.2.7. Decomposition of organic matters, nitrification Decomposition rates of PON and DON and nitrification rate are assumed to depend on temperature:

dNH 4 dt

= Ž 1 y R NO 3 . y Ž Nitrification. q Ž Decomposition of PON into ammonium.

Ž Decompostion of PON into ammonium. s VPI 0 exp Ž VPI T T . PON;

q Ž Decomposition of DON into ammonium.

Ž A9.

q Ž Excretion. q Dif Ž NH 4 . q Adv Ž NH 4 . ; Ž A15.

Ž Decomposition of PON into DON. s VPD 0 exp Ž VPD T T . PON;

Ž A10.

dChl dt

Ž Decomposition of DON into ammonium. s VDI 0 exp Ž V DI T T . DON;

s y  Ž Photosynthesis. y Ž Respiration. 4

s Ž Photosynthesis. y Ž Respiration. y Ž Extracellular Excretion.

Ž A11.

y Ž Mortality of phytoplankton.

Ž Nitrification. s k N 0 exp Ž k N T T . NH 4 .

Ž A12.

In many models, it is assumed that, unlike in this one, nitrification does not occur in the mixed layer. But Ward Ž1987. suggested that vigorous nitrification occurs near the bottom of the mixed layer, where significant photosynthesis is still occurring. It is not very unrealistic to assume that modest nitrification is occurring throughout the water column.

y Ž Grazing. q Dif Ž Chl . q Adv Ž Chl . ; Ž A16. dPON dt s Ž Mortality of phytoplankton. q Ž Mortality of zooplankton. q Ž Digestion. q Ž Sinking of PON .

A.2.8. Sinking of PON Letting S be the sinking velocity of PON, sinking of PON can be written in the form,

y Ž Decomposition of PON into ammonium. y Ž Decomposition of PON into DON .

E

Ž Sinking of PON. s y

Ez

Ž S P PON. .

q Dif Ž PON . q Adv Ž PON . ;

Ž A13.

Ž 17 .

dDON dt s Ž Extracellular excretion.

A.3. GoÕerning equations

q Ž Decomposition of PON into DON .

Combining those formulated processes, how each compartment evolves with time t can be described as follows Ž‘‘Diff’’ and ‘‘Adv’’. represent diffusion and advection process, respectively.: dNO 3 dt

y Ž Decomposition of DON into ammonium. q Dif Ž DON . q Adv Ž DON . ; dZOO

s y  Ž Photosynthesis. y Ž Respiration. 4 =R NO 3 q Ž Nitrification. q Dif Ž NO 3 . q Adv Ž NO 3 . ;

Ž A14.

dt

Ž A18.

s Ž Grazing. y Ž Digestion. y Ž Excretion. y Ž Mortality of zooplankton. q Dif Ž ZOO . q Adv Ž ZOO . .

The parameter values are given in Table 4.

Ž A19.

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

References Anderson, G.C., Lam, R.K., Booth, B.C., Glass, J.M., 1977. A description and numerical analysis of the factors affecting the processes of production in the Gulf of Alaska. University of Washington, Department of Oceanography Special Report 76 ŽReference M-77-40. 231 pp. Anderson, L.A., Sarmiento, J.L., 1995. Global ocean phosphate and oxygen simulations. Global Biogeochem. Cycles 9, 621– 636. Antoine, D., Andre, ´ J.-M., Morel, A., 1996. Oceanic primary production: 2. Estimation at global scale from satellite Žcoastal zone color scanner. chlorophyll. Global Biogeochem. Cycles 10, 57–69. Aumont, O., Orr, J.C., Monfray, P., Madec, G., 1999. Nutrient trapping in the equatorial Pacific: the ocean circulation solution. Global Biogeochem. Cycles, in press. Bacastow, R., Maier-Reimer, E., 1990. Ocean-circulation model of the carbon cycle. Clim. Dyn. 4, 95–125. Bacastow, R., Maier-Reimer, E., 1991. Dissolved organic carbon in modeling oceanic new production. Global Biogeochem. Cycles 5, 71–85. Beamish, R.J., Bouillon, D.R., 1993. Pacific salmon production trends in relation to climate. Can. J. Fish. Aquat. Sci. 50, 1002–1016. Behrenfeld, M.J., Falkowski, P.G., 1997. Photosynthetic rates derived from satellite-based chlorophyll concentration. Limnol. Oceanogr. 42, 1–20. Bidigare, R.R., Ondrusek, M.E., 1996. Spatial and temporal variability of phytoplankton pigment distributions in the central equatorial Pacific Ocean. Deep-Sea Res., Part II 43, 809–833. Bogorov, V., Vinogradov, M.D., Voronina, N.M., Kanaeva, I.P., Svetova, I.A., 1968. Distribution of zooplankton biomass within the surficial layer of the world ocean. Dokl.-Akad. Nauk, 1205–1207. Boyd, P.W., Wong, C.S., Merrill, J., Whitney, F., Snow, J., Harrison, P.J., Gower, J., 1998. Atmospheric iron supply and enhanced vertical carbon flux in the NE subarctic Pacific: is there a connection? Global Biogeochem. Cycles 12, 429–441. Brodeur, R.D., Ware, D.M., 1992. Long-term variability in zooplankton biomass in the subarctic pacific ocean. Fish. Oceanogr. 1, 32–38. Bryan, K., 1984. Accelerating the convergence to equilibrium of ocean-climate models. J. Phys. Oceanogr. 14, 666–673. Bryden, H.L., Brady, E.C., 1989. Eddy momentum and heat fluxes and their effects on the circulation of the equatorial Pacific Ocean. J. Mar. Res. 47, 55–79. Chai, F., Lindley, S.T., Barber, R.T., 1996. Origin and maintenance of a high nitrate condition in the equatorial Pacific. Deep-Sea Res., Part II 43, 1031–1064. Chavez, F.P., Smith, S.L., 1995. Biological and chemical consequences of open ocean upwelling. In: Summerhayes, C.P., Emeis, K.-C., Angel, M.V., Smith, R.L., Zeitzschel, B. ŽEds.., Upwelling in the Ocean: Modern Processes and Ancient Records. Wiley, Chichester, pp. 149–169. Chavez, F.P., Buck, K.R., Bidigare, R.R., Karl, D.M., Hebel, D., Latasa, M., Campbell, L., Newton, J., 1995. On the chloro-

155

phyll a retention properties of glass-fiber GFrF filters. Limnol. Oceanogr. 40, 428–433. Chavez, F.P., Buck, K.R., Service, S.K., Newton, J., Barber, R.T., 1996. Phytoplankton variability in the central and eastern tropical Pacific. Deep-Sea Res., Part II 43, 835–870. Conkright, M.S., Levitus, S., Boyer, T.P., 1994. World Ocean Atlas 1994, Volume 1 of NOAA Atlas NESDIS. NODC. Cox, M.D., 1987. Isopycnal diffusion in a z-coordinate ocean model. Ocean Model. 74, 1–5. daSilva, A.M., Yong, C.C., Levitus, S., 1994. Atlas of Surface Marine Data. NOAA. Eppley, R.W., 1972. Temperature and phytoplankton growth in the sea. Fish. Bull. 70, 1063–1085. Falkowski, P.G., Wilson, C., 1992. Phytoplankton productivity in the North Pacific ocean since 1900 and implications for absorption of anthropogenic CO 2 . Nature 358, 741–743. Fasham, M.J.R., Sarmiento, J.L., Slater, R.D., Ducklow, H.W., Williams, R., 1993. Ecosystem behavior at Bermuda Station S and Ocean Weather Station INDIA: an observational analysis. Global Biogeochem. Cycles 7 Ž2., 379–415. Glover, D.M., Wroblewski, J.S., McClain, C.R., 1994. Dynamics of the transition zone in coastal zone color scanner-sensed ocean color in the North Pacific during oceanographic spring. J. Geophys. Res. 99, 7501–7511. Gordon, H.R., Clark, D.K., Brown, J.W., Evans, R.H., 1982. Satellite measurements of the phytoplankton pigment concentration in the surface waters of a warm core Gulf Stream ring. J. Mar. Res. 40, 491–502. Halpern, D., Knox, R.A., Luther, D.S., Philander, S.G.H., 1989. Estimates of equatorial upwelling between 1408 and 1108W during 1984. J. Geophys. Res. 94, 8018–8020. Harrison, W.G., Harris, L.R., Irwin, B.D., 1996. The kinetics of nitrogen utilization in the oceanic mixed layer: nitrate and ammonium interactions at nanomolar concentrations. Limnol. Oceanogr. 41, 16–32. Hellerman, S., Rosenstein, M., 1983. Normal monthly wind stress over the world ocean with error estimates. J. Phys. Oceanogr. 13, 1093–1104. Honjo, S., 1996. Fluxes of particles to the interior of the open oceans. In: Ittekkot, V., Schaefer, ¨ P., Honjo, S., Depetris, P.J. ŽEds.., Particle Flux in the Ocean, Number 57 in SCOPE. Wiley, Chichester, pp. 91–154. Honjo, S., Dynamics, J., Collier, R., Manganini, S.J., 1995. Export production of particles to the interior of the equatorial Pacific Ocean during the 1992 EqPac experiment. Deep-Sea Res. 42, 831–870. Horiguchi, F., Nakata, K., 1995. Water quality analysis of Tokyo Bay using mathematical model. J. Adv. Mar. Sci. Technol. Soc. 1, 71–92, in Japanese with English abstract. Ishizaka, J., Ishikawa, K., 1991. Northwest Pacific Carbon Cycle Study ŽNOPACCS. - MITI. La Mer 29, 152–154. Ishizaka, J., Kiyosawa, H., Ishida, K., Ishikawa, K., Takahashi, M., 1994. Meridional distribution and carbon biomass of autotrophic picoplankton in the Central North Pacific Ocean during Late Northern Summer 1990. Deep-Sea Res. 41, 1745– 1766. Kantha, L.H., Clayson, C.A., 1994. An improved mixed layer

156

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157

model for geophysical applications. J. Geophys. Res. 99, 25235–25266. Karl, D.M., Christian, J.R., Dore, J.E., Hebel, D.V., Letelier, R.M., Tupas, L.M., Winn, C.D., 1996. Seasonal and interannual variability in primary production and particle flux at Station ALOHA. Deep-Sea Res. 43, 539–568. Karl, D., Letelier, R., Tupas, L., Dore, J., Christian, J., Hebel, D., 1997. The role of nitrogen fixation in biogechemical cycling in the subtropical North Pacific Ocean. Nature 388, 533–538. Kawamiya, M., Kishi, M.J., Yamanaka, Y., Suginohara, N., 1995. An ecological-physical coupled model applied to Station Papa. J. Oceanogr. 51, 635–664. Kawamiya, M., Kishi, M.J., Yamanaka, Y., Suginohara, N., 1997. Obtaining reasonable results in different oceanic regimes with the same ecological-physical coupled model. J. Oceanogr. 53, 397–402. Kawamiya, M., Kishi, M., Suginohara, S., 2000. An ecosystem model for the North Pacific embedded in a general circulation model: Part II. Mechanisms forming seasonal variations of chlorophyll. J. Mar. Syst. 25, 159–178. Killworth, P.D., 1996. Time interpolation of forcing fields in ocean models. J. Phys. Oceanogr. 26, 136–143. Kirchamn, D.L., Lancelot, C., Fasham, M., Legendre, L., Radach, G., Scott, M., 1993. Dissolved oragnic matter in biogeochemical models of the ocean. In: Evans, G.T., Fasham, M.J.R. ŽEds.., Towards a Model of Ocean Biogeochemical Processes. Springer, Berlin, pp. 209–225. Letelier, R.M., Dore, J.E., Karl, D.M., 1996. Seasonal and interannual variations in photosynthetic carbon assimilation at Station ALOHA. Deep-Sea Res., Part II 43, 467–490. Levitus, S., Boyer, T.P., 1994. World Ocean Atlas 1994, Volume 4 of NOAA Atlas NESDIS. NODC. Libby, P.S., Wheeler, P.A., 1997. Particulate and dissolved organic nitrogen in the central and eastern equatorial Pacific. Deep-Sea Res. 44, 345–361. Longhurst, A., Sathyendranath, S., Platt, T., Caverhill, C., 1995. An estimate of global primary production in the ocean from satellite radiometer data. J. Plankton Res. 17, 1245–1271. Loukos, H., Frost, B., Harrison, D.E., Murray, J.W., 1998. An ecosystem model with iron limitation of primary production in the equatorial pacific at 1408W. Deep-Sea Res., Part II 44, 2221–2249. Martin, J.H., Fitzwater, S.E., 1988. Iron deficiency limits phytoplankton growth in the north-east Pacific subarctic. Nature 331, 341–343. Martin, J.H., Knauer, G.A., Karl, D.M., Broenkow, W.W., 1987. VERTEX: Carbon cycling in the northeast Pacific. Deep-Sea Res. 34, 267–285. McGillicuddy, D.J., Robinson, A.R., Siegel, D.A., Jannasch, H.W., Johnson, R., Dickey, T.D., McNeil, J., Michaels, A.F., Knap, A.H., 1998. Influence of mesoscale eddies on new production in the Sargasso Sea. Nature 394, 263–266. Mellor, G.L., Yamada, T., 1982. Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys. Space Phys. 20 Ž4., 851–875. Najjar, R.G., Sarmiento, J.L., Togweiler, J.R., 1992. Downward transport and fate of organic matter in the ocean: simulation

with a general circulation model. Global Biogeochem. Cycles 6 Ž1., 45–76. Nitta, T., Yamada, S., 1989. Recent warming of tropical sea surface temperature and its relationship to the northern hemisphere circulation. J. Meteorol. Soc. Jpn. 67, 375–383. NOPACCS, 1992. Cruise report of the Hakurei Maru cruise NH-02-2. Technical report, New Energy and Industrial Technology Development Organization. Polovina, J.J., Mitchum, G.T., Evans, G.T., 1995. Decadal and basin-scale variation in mixed layer depth and the impact on biological production in the Central and North Pacific, 1960– 88. Deep-Sea Res. 42, 1701–1716. Redi, R.H., 1982. Oceanic isopycnal mixing by coordinate rotation. J. Phys. Oceanogr. 12, 1154–1158. Reid, J.L., 1962. On curculation, phosphate-phosphorus content, and zooplankton volumes in the upper part of the Pacific ocean. Limnol. Oceanogr. 7, 287–306. Roemmich, D., McGowan, J., 1995. Climatic warming and the decline of zooplankton in the California Current. Science 267, 1324–1325. Roman, M.R., Dam, H.G., Gauzens, A.L., Urban-Rich, J., Foley, D.G., Dickey, T.D., 1995. Zooplankton variability on the equator at 1408W during the JGOFS EqPac study. Deep-Sea Res., Part II 42, 673–693. Rosati, A., Miyakoda, K., 1988. A general circulation model for upper ocean simulation. J. Geophys. Res. 18, 1601–1626. Sarmiento, J.L., Orr, J.C., 1992. A perturbation simulation of CO 2 uptake in an ocean general circulation model. J. Geophys. Res. 97, 3621–3645. Sarmiento, J.L., Slater, R.D., Fasham, M.J.R., Ducklow, H.W., Toggweiler, J.R., Evans, G.T., 1993. A seasonal three-dimensional ecosystem model of nitrogen cycling in the North Atlantic euphotic zone. Global Biogeochem. Cycles 7 Ž2., 417–450. Six, K.D., Maier-Reimer, E., 1996. Effects of plankton dynamics on seasonal carbon fluxes in an ocean general circulation model. Global Biogeochem. Cycles 10 Ž4., 559–583. Steele, J.H., 1962. Environmental control of photosynthesis in sea. Limnol. Oceanogr. 7, 137–150. Steele, J.H., Henderson, E.W., 1992. The role of predation in plankton models. J. Plankton Res. 14 Ž1., 157–172. Sugimoto, T., Tadokoro, K., 1997. Interannual-interdecadal variations in zooplankton biomass, chlorophyll concentration and physical environment in the sub-arctic Pacific and Bering Sea. Fish. Oceanogr. 6, 74–93. Takahashi, T., Broecker, W.S., Langer, S., 1985. Redfield ratio based on chemical data from isopycnal surfaces. J. Geophys. Res. 90, 6907–6924. Tsujino, H., Suginohara, N., 1998. Thermohaline effects on upper layer circulation of the North Pacific. J. Geophys. Res. 103, 18665–18679. Venrick, E.L., McGowan, J.A., Cayan, D.R., Hayward, T.L., 1987. Climate and chlorophyll a: long-term trends in the central North Pacific Ocean. Science 238, 70–72. Ward, B.B., 1987. Nitrogen transformations in the Southern California Bight. Deep-Sea Res. 34, 785–805. Winn, C.C., Campbell, J.R., Christian, R.M.L., Hebel, D.V.,

M. Kawamiya et al.r Journal of Marine Systems 25 (2000) 129–157 Dore, J.E., Fujieki, L., Karl, D.M., 1996. Seasonal variability in the phytoplankton community of the North Pacific Subtropical Gyre. Global Biogeochem. Cycles 9, 605–620. Wroblewski, J.S., 1977. A model of phytoplankton plume formation during Oregon upwelling. J. Mar. Res. 35, 357–394. Wroblewski, J.S., Sarmiento, J.L., Flierl, G.R., 1988. An ocean basin scale model of plankton dynamics in the north atlantic: 1. Solution for the climatological oceanographic conditions in May. Global Biogeochem. Cycles 2 Ž3., 199–218.

157

Wyrtki, K., 1975. Fluctuations of the dynamic topography in the Pacific Ocean. J. Phys. Oceanogr. 5, 450–459. Yamanaka, Y., Tajika, E., 1996. The role of the vertical fluxes of particulate organic matter and calcite in the oceanic carbon cycle: studies using an ocean biogeochemical general circulation model. Global Biogeochem. Cycles 10, 361–382. Yoder, J.A., McClain, C.R., Feldman, G.C., Esaias, W.E., 1993. Annual cycles of phytoplankton chlorophyll concentrations in the global ocean. Global Biogeochem. Cycles 7 Ž1., 181–193.