Long-term variation in dissolved inorganic carbon and ocean acidication indices in the Northwest Pacic from 1993 to 2018: A study of a biogeochemical model with an operational ocean model product and observational evidence

24 The multi-decadal variation in ocean acidification indices in the Northwest Pacific was examined 25 using a biogeochemical model with an operational ocean model product for the period 1993–2018. We 26 found that ocean acidification varied regionally in the Northwest Pacific. The surface ocean (above 27 100 m depth) underwent acidification that progressed more quickly in the subtropical region and the 28 Kuroshio extension than in the subarctic region due to vertical mixing of the dissolved inorganic 29 carbon (DIC) supply exceeding DIC release by air–sea exchange. Below 100 m depth, acidification 30 and alkalinization occurred in the subtropical and subarctic regions, respectively. We attribute these 31 regional differences in acidification and alkalinization to spatially variable biological processes in the 32 upper layer and physical redistribution of DIC, both horizontally and vertically. 33 A biogeochemical model with an operational ocean product was used to examine long-term three- dimensional variation in ocean acidification indices and parameters related to ocean acidification from 2018. The model results indicate an increase in DIC and a progression towards ocean acidification over most of the surface layer, although most of the Northwest Pacific is identified as a 345 source of CO 2 to the atmosphere. The results exhibit distinct contrasts between the subtropical and subarctic regions at increasing depths: ocean acidification is suggested in the subtropical region, while alkalinization appears to occur in the subarctic region. The analysis of the model product indicates that acidification represented by this model is strongly controlled by the physical transport of DIC that occurs due to ocean circulation, with biological activity. Historical DIC observations in the Northwest Pacific show similar patterns across the Kuroshio Extension region. The analysis of the model product indicates that ocean acidification represented by this model is primarily driven by the deeper see


38
The atmospheric partial pressure of CO2 (pCO2, air) has been increasing at a rate of ~1.8 parts per 39 million by volume (ppmv) per year in recent decades as a result of human activities such as fossil fuel 40 burning, deforestation, and cement production (Takahashi et al. 2009;IPCC 2013). During the pre-41 industrial period, the ocean was generally a net source of CO2 emissions to the atmosphere due to in 42 situ production and mineralization of land-derived organic matter and CaCO3 precipitation. However,  The major sink zones for atmospheric CO2 are located at latitudes 40°-60°N and 40°-60°S 50 (Takahashi et al. 2002(Takahashi et al. , 2009(Takahashi et al. , 2014Yasunaka et al. 2013). Conversely, intense CO2 source areas 51 include the eastern equatorial Pacific and the northwestern Arabian Sea (Takahashi et al. 2002(Takahashi et al. , 2009 2014). The tropical Atlantic, Indian Ocean, and northwest subarctic Pacific are also prominent CO2 53 source areas (Takahashi et al. 2002(Takahashi et al. , 2009(Takahashi et al. , 2014. In the sink zones (40°-60°N and 40°-60°S), low-54 pCO2 waters are rapidly formed by the juxtaposition of the cooling of warm waters with biological 55 drawdown of pCO2 in the nutrient-rich subpolar waters, while high wind speeds over these low-pCO2 56 waters further accelerate CO2 uptake (Takahashi et al. 2002). study showed that the rate of ocean acidification was different for each depth, with the rate being twice 76 as high in the subsurface layer (300-400 m depth) than in the surface layer.

77
Analyses using measurements of pHinsitu (pH25) are limited due to their patchy distributions in time 78 and space. The current understanding of ocean acidification relies on pH measurements from time-       Table 1).

135
Temperature, DIC, and ALK drive the annual variation in ocean acidification indices, but DIC exerts 136 the strongest influence on ocean acidification. We used observational profiles to confirm that changes 140°-160°E; Fig. 4a). The DIC decrease becomes stronger below 100 m depth and spreads to the south.

206
A DIC increase is seen in the southern subtropical region (Fig. 4b-d). The strongest DIC increase range between 400 and 1000 m (Fig. 5a). The DIC decrease spreads below 200 m depth between 35° 212 and 50°N, with the strongest decrease occurring in the depth range of 600-800 m near 45°N in the 213 subarctic region (Fig. 5a).

214
The change rate of ALK between 1993 and 2018 is within 1 μmol/kg/yr (Figs. 4e-h, 5b), which is spreads more broadly in the subtropical region along the vertical section (Fig. 5b). The ALK decrease 218 in the subarctic region at 200 m depth connects with the large decrease in the Okhotsk Sea (Fig. 4h).

219
The patterns in ALK variation generally agree with those of DIC (Figs. 4, 5). fraction of the area that contains the DIC increase resides in the subarctic region (Fig. 6a, b, d, e). The 229 areas of DIC increase in the subarctic region are located near the surface layer or close to the Kuroshio 230 Extension (35°-40°N; Fig. 6).

231
Comparing the vertical distribution of the simulated increases and decreases in DIC (Figs. 4a-d, and 232 5) with that of the DIC observations (Fig. 6), we find that the area of alkalinization residing in the 233 subtropical region below the subsurface in the model outputs (Figs. 4a−d, 5a) is not basically found in 234 the observations (Fig. 6). between 1995 and 2018 may include effects of the spin-up process (Fig. 7), but the agreement between 237 distributions of DIC changes from 1995 to 2018 (Fig. 7) and the observations appears to be better than 238 that between DIC changes from 1999 to 2018 (Fig. 5a) and the observations.  Pacific also appear to play a role in the active uptake of CO2.

274
The annual emission rate of CO2 from ocean to atmosphere (Fig. 9)  where the processes controlling acidification and alkalinization are different (Figs. 1a-d, 4a- where the subscripts A, xy_dif, z_dif, Bio, and air-sea represent the time derivatives of DIC induced 293 by advection, horizontal diffusion (i.e., horizontal mixing), vertical mixing, biological processes, and 294 air-sea CO2 exchange, respectively (positive air-sea values indicate a transfer of CO2 from air to sea). 295 We refer to these as the "DIC variation terms", and the DIC variations induced by air-sea CO2 296 exchange are only included for the surface level. The total DIC variation term on the left side of Eq.

297
(2) is the sum of all terms on the right-hand side of the equation. Note that the influencing terms were 40°, and 50°E. The balances at the surface (Fig. 10a-f) indicate that air-sea exchange is generally 305 balanced by vertical mixing (Figs. 10c, e, 13a-d), and that biological processes drive the DIC increase 306 in the Kuroshio Extension and the subtropical region as well as the DIC decrease in the subarctic region 307 north of 50°N (Figs. 10d, 13a-d). The DIC increase in the Kuroshio Extension and the subtropical 308 region can be attributed to remineralization, which occurs due to the spreading of detritus originating 309 from zooplankton (not shown). The DIC decrease in the subarctic region is due to photosynthesis. The  (Fig. 11d). The positive DIC values in the subarctic region north of 40°N (Fig. 11d), however, indicate 318 production of DIC due to remineralization. Such latitudinal differences between the subtropical region 319 and the subarctic region are therefore due to biological processes (Fig. 11d, i), and can be linked with vertical mixing (Fig. 11f, h, j, 13m-s). Horizontal mixing does not appear to play an important role in 324 driving lateral or depth-dependent DIC variations across this study area (Fig. 10-12).

325
Observations of DIC (Fig. 5) suggest that ocean acidification in the subtropical region above 200-326 500 m has advanced more rapidly than the subarctic region. Results from the simulation suggest that 327 vertical mixing balances the ejection of DIC at the surface despite CO2 release to the atmosphere. This 328 likely causes acidification to proceed faster in the subtropical region than in the subarctic region ( Fig.   329 4a). The observations also raise the possibility of alkalinization in the subarctic region below 100 m.

330
The analysis of the model products also suggests that the supply from DIC transport in the subarctic 331 region below 100 m is less than that of the upper layers due to vertical mixing ( Fig. 10-13, Fig.13a, h, 332 l, p). This process is responsible for the DIC decrease in the subarctic region ( Fig. 4b-d, 5a, 10f, 10k, 333 11e, 11j, 13c-d, 13g-h), and likely leads to alkalinization in the shallow waters there (Fig. 1b-d, 1f-h,  (Fig. Aa) fluctuate more widely than those in the deeper layers ( Fig. Ab-d). In the subsurface 378 layer (Fig. Ab), the DIC values decrease in the subarctic region, but increase in the Kuroshio Extension 379 and the subtropical region. In the layers deeper than 500 m (Fig. Ac-d)        Same as Fig. 5a, but between 1995 and 2018.

Figure 10
Horizontal distributions of yearly DIC variation terms generated by advection (a, g), horizontal diffusion (b, h), vertical mixing (c, i), biological processes (e, j) and air-sea CO2 exchange process (e) and total DIC time variation (f, k) at 0 m (a-f) and 50 m (g-k) depth, respectively, on the equation above. Positive and negative values indicate increase and decrease in each term for their DIC variations, respectively. The DIC variation term in uenced by the air-sea CO2 exchange are considered only in the surface, not below it.

Figure 11
Same as Fig. 10, but horizontal distributions of yearly DIC variation terms generated by advection (a, e), horizontal diffusion (b, f), vertical mixing (c, g), biological processes (e, h) and total DIC time variation (d, i) at 100 m (a-d) and 200 m (e-i) depth, respectively.

Figure 12
Same as Fig. 10, but horizontal distributions of yearly DIC variation terms generated by advection (a, e), horizontal diffusion (b, f), vertical mixing (c, g), biological processes (e, h) and total DIC time variation (d, i) at 500 m (a-d) and 1000 m (e-i) depth, respectively.
The yearly mean values calculated within the range of 5-grid (22-45.5 km) from the target region. Color bars indicate the same color showing in uencing processes in Fig. 10.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.