**3.3.2 Difference between mMA94 and PS77 schemes**

Annual mean chlorophyll-a concentration (SeaWiFS) is shown in Fig. 7a, which was obtained by averaging its monthly mean values. The values are very low in the subtropical circulations in both hemispheres. In contrast, high values are seen near the continental coasts. The corresponding annual mean euphotic layer depth based on the mMA94 scheme (7) (Fig. 7b) exhibits a pattern similar to that of chlorophyll-a. It has large values exceeding 80 m in the center of every subtropical circulation area, and exceeds 100 m in the South Pacific, which is greater than the largest value in IY10 (94 m, Fig. 6b). Very low values of less than 20 m are often observed along the continental coast, corresponding to high chlorophyll-a concentrations. Along the equator chlorophyll-a is relatively high and the euphotic layer depth is relatively shallow (50 - 60 m).

The effective attenuation depth calculated based on zonally averaged annual-mean penetrating irradiance is presented in Fig. 7c, with the zonal-mean euphotic layer depth for PS77, IY10, and mMA94. Figure 7d shows the difference in absorption between mMA94 and PS77 (mMA94 - PS77). In contrast with IY10, the effective attenuation depth (Fig. 7c) is very small (less than 8 m) in the top several-meter layer over the whole range of latitude. This is due to the fact that the attenuation depth *Z*<sup>1</sup> in the mMA94 scheme (7) is a few meters over the whole range of chlorophyll-a concentration. Corresponding to the low effective attenuation depth, the absorption is very large in the top layer in mMA94 compared to that in PS77 (Fig. 7d).

In the deeper layers below, 10 m, where the last term in (7) seems to be dominant, two peaks of dome-shaped distribution of high attenuation depth (low attenuation) correspond to the horizontal pattern of low chlorophyll-a and large euphotic layer depth (Figs. 7a and b). At the equator the two domes are divided, in contrast with IY10, where a single dome is centered at the equator (Fig. 6b). The high concentration of chlorophyll-a along the equator characterizes these optical properties and structures in the deeper part of the surface layer. The zonal-mean euphotic layer depth of mMA10 also has two peaks in magnitude, but much less than that of IY10 as a whole (Fig.7c). The zero line for the absorption difference (Fig. 7d) is at about 7 m over the whole range of latitude. Below that level, it has a vertical structure with a minimum in magnitude of 20 m, but its latitudinal variation seems to be very weak. The magnitude in the absorption difference for (mMA94 - PS77) is one order greater than that for (IY10 - PS77) (Fig. 6d).

#### **3.4 Model and experiments**

The OGCM used in this section is MRI.COM3 (Tsujino et al., 2010) that is, a free-surface, depth-coordinate ocean ice model. The model has a global domain with a tripolar grid (Murray, 1996) that consists of a spherical, latitude-longitude grid south of 64◦N and a bipolar grid with generalized orthogonal coordinates with polar singularities in Siberia (64◦N, 80◦E) and Canada (64◦N, 100◦W).

The horizontal resolution is 1◦ in longitude and 0.5◦ in latitude south of 64◦N. There are 50 vertical levels with a bottom boundary layer (Nakano & Suginohara, 2002). The surface layer thickness is 4 m, and the model has 30 levels in the upper 1000 m.

The mixed layer scheme is based on Noh & Kim (1999). The generalized Arakawa scheme (Ishizaki & Motoi, 1999) was used to calculate the momentum advection terms. A numerical advections scheme based on conservation of second-order moments (SOM) (Prather, 1986) was used for advection of tracers. Isopycnal diffusion (Redi, 1982) and eddy-induced transport parameterized as isopycnals layer thickness diffusion (Gent & McWilliams, 1990) are used as sub-grid-scale mixing.

The surface boundary conditions are based on the surface atmospheric condition by Large & Yeager (2009) and provided as the Coordinate Ocean-ice Reference Experiment (CORE) forcing dataset (CORE.v2). A detrended 59-year interannual forcing dataset from 1948 to 2006 was used for spin-up. More details about the model settings may be found in Tsujino et al. (2011).

The model was integrated for about 1,350 years (23 cycles) from an initial state by using the detrended CORE data, and reached a quasi-steady state. In the spin-up period, we used the PS77 scheme as the absorption scheme of solar radiation.

Three experiments were then carried out to examine the impact of the three absorption schemes described in the previous subsection. The first experiment used the conventional PS77 scheme and was called "CTL". The second experiment used the IY10 scheme and was called "SLR". The third experiment used the chlorophyll-a dependent scheme based on mMA94 scheme and was called "CHL". In the third experiment, chlorophyll-a data was derived from the monthly mean satellite-based observation (SeaWiFS: Fig. 7a). Each experiment started from a quasi-steady state and was integrated for five additional cycles (295 years) using the detrended CORE data. The yearly mean data over the fifth cycle were used for analysis.

### **3.5 Results**

14 Will-be-set-by-IN-TECH

in both hemispheres. In contrast, high values are seen near the continental coasts. The corresponding annual mean euphotic layer depth based on the mMA94 scheme (7) (Fig. 7b) exhibits a pattern similar to that of chlorophyll-a. It has large values exceeding 80 m in the center of every subtropical circulation area, and exceeds 100 m in the South Pacific, which is greater than the largest value in IY10 (94 m, Fig. 6b). Very low values of less than 20 m are often observed along the continental coast, corresponding to high chlorophyll-a concentrations. Along the equator chlorophyll-a is relatively high and the euphotic layer depth is relatively

The effective attenuation depth calculated based on zonally averaged annual-mean penetrating irradiance is presented in Fig. 7c, with the zonal-mean euphotic layer depth for PS77, IY10, and mMA94. Figure 7d shows the difference in absorption between mMA94 and PS77 (mMA94 - PS77). In contrast with IY10, the effective attenuation depth (Fig. 7c) is very small (less than 8 m) in the top several-meter layer over the whole range of latitude. This is due to the fact that the attenuation depth *Z*<sup>1</sup> in the mMA94 scheme (7) is a few meters over the whole range of chlorophyll-a concentration. Corresponding to the low effective attenuation depth, the absorption is very large in the top layer in mMA94 compared to that in PS77 (Fig.

In the deeper layers below, 10 m, where the last term in (7) seems to be dominant, two peaks of dome-shaped distribution of high attenuation depth (low attenuation) correspond to the horizontal pattern of low chlorophyll-a and large euphotic layer depth (Figs. 7a and b). At the equator the two domes are divided, in contrast with IY10, where a single dome is centered at the equator (Fig. 6b). The high concentration of chlorophyll-a along the equator characterizes these optical properties and structures in the deeper part of the surface layer. The zonal-mean euphotic layer depth of mMA10 also has two peaks in magnitude, but much less than that of IY10 as a whole (Fig.7c). The zero line for the absorption difference (Fig. 7d) is at about 7 m over the whole range of latitude. Below that level, it has a vertical structure with a minimum in magnitude of 20 m, but its latitudinal variation seems to be very weak. The magnitude in the absorption difference for (mMA94 - PS77) is one order greater than that for (IY10 - PS77)

The OGCM used in this section is MRI.COM3 (Tsujino et al., 2010) that is, a free-surface, depth-coordinate ocean ice model. The model has a global domain with a tripolar grid (Murray, 1996) that consists of a spherical, latitude-longitude grid south of 64◦N and a bipolar grid with generalized orthogonal coordinates with polar singularities in Siberia (64◦N, 80◦E)

The horizontal resolution is 1◦ in longitude and 0.5◦ in latitude south of 64◦N. There are 50 vertical levels with a bottom boundary layer (Nakano & Suginohara, 2002). The surface layer

The mixed layer scheme is based on Noh & Kim (1999). The generalized Arakawa scheme (Ishizaki & Motoi, 1999) was used to calculate the momentum advection terms. A numerical advections scheme based on conservation of second-order moments (SOM) (Prather, 1986) was used for advection of tracers. Isopycnal diffusion (Redi, 1982) and eddy-induced transport parameterized as isopycnals layer thickness diffusion (Gent & McWilliams, 1990)

thickness is 4 m, and the model has 30 levels in the upper 1000 m.

shallow (50 - 60 m).

7d).

(Fig. 6d).

**3.4 Model and experiments**

and Canada (64◦N, 100◦W).

are used as sub-grid-scale mixing.

This subsection describes the impact of the three absorption schemes on ocean simulations. In particular, we focus on the oceanic structures of the tropical Pacific, where the impact of those schemes is most clearly found.

Figure 8a shows the SST and surface current differences between SLR and CTL. When the IY10 scheme was introduced, the SST increased slightly to about 0.1◦C in the western tropical Pacific, the Indian Ocean and the subtropics. In contrast, the SST decreased to about 0.3◦C in the central and eastern equatorial Pacific east of 175◦E within the north-south 5 degrees band, together with the coastal areas. This SST contrast in the tropical Pacific Ocean has already been reported by IY10. Introducing the solar angle shifted the locus of radiation absorption upward, resulting in warming in the SST in all regions, except the eastern equatorial Pacific, where the indirect effect led to the cooling in the SST. When the mMY94 scheme was introduced (Fig. 8c), the pattern of the SST contrast in the tropical Pacific was almost the same, but the magnitude increased. The SST decreased in the eastern equatorial Pacific reaching even about 1◦C around 120◦W. The impact of the chlorophyll-a concentration (CHL) on the SST in the equatorial Pacific was about three times greater than that of the solar angle (SLR).

The IY10 scheme caused the westward current surface anomalies in the equatorial Pacific (Fig. 8a). The direction of the surface current anomalies turned pole ward apart from the equatorial region, corresponding to a divergent flow. This result is consistent with the impact of the IY10 scheme as described by IY10. These surface current anomalies are associated with increased upwelling and changes in the equatorial current system. Introduction of the chlorophyll-a concentration (CHL) produced effects similar to those of SLR, but with greater amplitude of the current anomaly (Fig. 8b). The amplitude is almost twice that of SLR, and the direction of the surface current is more pole ward. Thus, the divergent flow in the eastern equatorial Pacific is more enhanced in the CHL run than in the SLR run.

Fig. 8. (a) SST [0.1◦C] (color) and surface current [cm/s] (vector) differences between SLR and CTL. (b) Mixed layer depth [m] difference between SLR and CTL (shaded) and mixed layer depth [m] for SLR (contour) . (c)-(d) Same as (a)-(b) but for CHL.

Introduction of the solar angle changed the vertical profile heating, due to solar radiation. The surface layer received more heating, while the subsurface layer received less (Fig. 6d, Fig. 7d). This vertical contrast of heating made the MLD shallower, over most regions (Fig. 8c). In the tropical Pacific, the MLD decreased to about 5 m in the central equatorial Pacific where the mean MLD was large. This situation was strengthened when the mMA94 scheme was used (Fig. 8d). Introducing the chlorophyll-a concentration made the MLD more than 20 m shallower. High chlorophyll-a concentration along the equator characterizes optical properties and structures in the deeper part of the surface layer, as mentioned in 3.3.2. The decrease in the MLD was also significant in the Arabian Sea, where chlorophyll-a concentration was relatively high (Fig. 7a). These changes in the MLD reflected the absorption differences among the mMA94, IY10, and PS77 schemes (Fig. 7d).

Changes in SSTs and surface currents of the equatorial Pacific due to the ocean radiant schemes were associated with a change in the shallow meridional circulation of the tropical Pacific, called the subtropical cell (STC). The STC played an important role in connecting subduction regions of the subtropical gyre with upwelling regions in the tropics. When the solar angle was introduced (Fig. 9a), the pole ward surface current in the upper 30 m, and

Fig. 8. (a) SST [0.1◦C] (color) and surface current [cm/s] (vector) differences between SLR and CTL. (b) Mixed layer depth [m] difference between SLR and CTL (shaded) and mixed

Introduction of the solar angle changed the vertical profile heating, due to solar radiation. The surface layer received more heating, while the subsurface layer received less (Fig. 6d, Fig. 7d). This vertical contrast of heating made the MLD shallower, over most regions (Fig. 8c). In the tropical Pacific, the MLD decreased to about 5 m in the central equatorial Pacific where the mean MLD was large. This situation was strengthened when the mMA94 scheme was used (Fig. 8d). Introducing the chlorophyll-a concentration made the MLD more than 20 m shallower. High chlorophyll-a concentration along the equator characterizes optical properties and structures in the deeper part of the surface layer, as mentioned in 3.3.2. The decrease in the MLD was also significant in the Arabian Sea, where chlorophyll-a concentration was relatively high (Fig. 7a). These changes in the MLD reflected the absorption

Changes in SSTs and surface currents of the equatorial Pacific due to the ocean radiant schemes were associated with a change in the shallow meridional circulation of the tropical Pacific, called the subtropical cell (STC). The STC played an important role in connecting subduction regions of the subtropical gyre with upwelling regions in the tropics. When the solar angle was introduced (Fig. 9a), the pole ward surface current in the upper 30 m, and

layer depth [m] for SLR (contour) . (c)-(d) Same as (a)-(b) but for CHL.

differences among the mMA94, IY10, and PS77 schemes (Fig. 7d).

Fig. 9. (a) Meridional mass transport [Sv] zonally averaged in the Pacific Ocean for SLR (contour). The shaded area denotes the difference between SLR and CTL. (b) Same as (a) but for CHL.

the equator ward surface current at depths of 30 to 60 m were enhanced. This results in enhanced meridional circulation up to about 2.5 Sv (1Sv=106 m3/s) in the North Pacific. Since the maximum transport of the STC is about 35 Sv in the mean state, the STC was strengthened by about 7 %. When the effect of chlorophyll-a concentration was introduced (Fig. 9b), similar current changes occured in the upper 70 m; however, the STC was further enhanced by about 7 Sv, corresponding to more than 20 % of the mean state. Thus, a more advanced ocean radiant scheme leads to more enhanced STC.

These results are understood from the following. In the tropical Pacific, the upper meridional transport (*My*) is expressed as (Sweeney et al., 2005):

$$M\_{\!\!\!V} = \int\_{D\_{\!\!\!ML}}^{\eta} -\frac{1}{\rho\_0 f} \frac{\partial p}{\partial \mathbf{x}} dz + \frac{\tau\_{\mathbf{x}}}{\rho\_0 f} \tag{8}$$

where *f* is a Coriolis parameter, *τ<sup>x</sup>* is a zonal wind stress, *∂p*/*∂x* is a zonal pressure gradient, *ρ*<sup>0</sup> is sea water density, *η* is a surface elevation, and *DML* is a MLD. This equation means that the upper meridional transport is given by the difference between the pole ward Ekman transport and the equatorward geostrophic transport. The MLD is reduced by the introduction of the solar angle, or the effect of the chlorophyll-a distribution. The Ekman transport is the same in the three runs, because the employed surface wind stress is the same. Also, the difference between zonal pressure gradients in each run is small (Sweeney et al., 2005). Hence, the decreased MLD leads to the reduced meridional geostrophic transport. As a result, the pole ward transport increases, leading to an enhanced STC. The enhanced STC produces a divergent flow at the surface and strengthens the equatorial upwelling, and the resulting cold water from the deep layer cools the SST in the eastern equatorial Pacific.

To summarize, the impact of the changes in absorption schemes of the solar radiation on the tropical Pacific occurs not only due to the local heating of the solar radiation itself as a direct effect, but also by the dynamical response as an indirect effect. Introducing the chlorophyll-a distribution and varying the solar angle enhances the shallow meridional circulation (STC), which leads to nontrivial changes in the tropical oceanic structure.
