Synchronous bidecadal periodic changes of oxygen, phosphate and temperature between the Japan Sea deep water and the North Pacific intermediate water

Authors


Abstract

[1] From time series of dissolved oxygen (O2), phosphate (PO4) and water temperature (T) during the last forty years in the Japan Sea Deep Water and the North Pacific Intermediate Water, we found that O2, PO4 and T show a clear bidecadal oscillations of about 18 years superimposed on the linear trends of decrease of 0.47 μmol-O2 kg−1 yr−1 and increase of 0.003 μmol-PO4 kg−1 yr−1 and 0.005°C yr−1 in both two oceanic areas. The changes of O2, PO4 and T in the two oceanic areas were synchronized despite the water formation systems of each area being independent. Both the linear trends and the oscillations of O2, PO4 and T in the two oceanic areas also showed a strong correlation with the anomaly of the sea surface level pressure in the North Pacific that possibly affects the change of surface oceanic condition.

1. Introduction

[2] To understand the extensive long-term changes of physical and biological conditions in the oceanic interior is an essential step for predicting global climate change in the future. Until now, oceanic conditions have been assumed to be constant over time in many models [e.g., Sarmiento et al., 1992]. However, it is possible that the recent oceanic conditions have changed due to the effects of artificial greenhouse warming and/or natural climate change. A recent model calculation predicts that a decrease of water formation rates will be caused as response to global warming in the future [Sarmiento et al., 1998]. It also suggests that the effect might already be occurring in the intermediate-deep water formation area.

[3] Some recent studies have already reported linear increasing T and PO4, and linear decreasing O2 in the North Pacific [e.g., Watanabe et al., 2001]. Similar linear trends of PO4 or O2 in the Japan Sea Deep Water (JSDW) have also been reported [Gamo, 1999; Chen et al., 1999; Kim et al., 2001]. Moreover the decadal changes of salinity and temperature in the other regions have been found [e.g., Bryden et al., 2003]. However, these studies reported only either the linear trend of oceanic properties, or their focus was only on local conditions. On the other hand, recent atmospheric climatic long-term oscillations have been found in the northern hemisphere [e.g., Minobe, 2000], which possibly affects the change of oceanic conditions in extensive areas. The relation between changes in atmospheric conditions and the subsequent changes in oceanic properties, however, are still not well understood.

[4] Thus we need to elucidate whether the changes of oceanic conditions have long-term linear trends and oscillations by the interaction between the atmosphere and the ocean in extensive areas. We here compiled time series data set of O2, PO4 and T during the last forty years in the Japan Sea Deep Water (JSDW) and the North Pacific Intermediate Water (NPIW), and compared the extensive long-term changes in the two oceanic areas.

[5] The North Pacific is the major end point of the deep water conveyer belt produced in both the North Atlantic Ocean and the Southern Ocean [Broecker, 1991]. The North Pacific subpolar region is the only area forming a southward intermediate water called NPIW between 26.8 σθ and 27.4 σθ in the wintertime [e.g., Yasuda, 1997] and this water mass spreads over the entire North Pacific. This region is thus an important area for the oceanic uptake of anthropogenic carbon over the North Pacific [e.g., Tsunogai et al., 1993]. Our research area is suitable to detect evidence of changes in oceanic conditions because it is close to the region producing NPIW.

[6] On the other hand, the Japan Sea is an almost isolated marginal sea and its maximum water depth reaches about 3700 m (Figure 1). The sea water exchange between the Japan Sea and its surrounding seas is restricted by shallow sills of less than 150 m. JSDW is characterized by an extremely homogeneous temperature, salinity and O2 below 200 m. The deep water from 2000 m to the bottom is even more homogeneous with the density of about 27.4 σθ [Gamo et al., 1986], and thus the vertical gradient of O2 is within 1 μmol kg−1. This reflects the active vertical water mixing with the deep water formation process in the surface occurring due to cold weather in the wintertime in the northern Japan Sea [e.g., Sudo, 1986]. This mechanism is similar to the deep water formation process of the global conveyor belt in the Northern Atlantic Ocean and the Southern Ocean [Broecker, 1991], and there is no source for JSDW but the surface water. Thus the changes of oceanic properties in JSDW can give us information on an extent of cooling and warming of the sea surface water in response to changes of atmospheric conditions, and the Japan Sea might be expected to be a miniature model of the global ocean [e.g., Gamo, 1995].

Figure 1.

Map for time-series data of O2, PO4 and T used in this study. The representative areas, the eastern Japan Sea Basin area, the Yamato Basin area and the North Pacific subpolar area, are enclosed by squares.

2. Data and Method

[7] We here compiled time series of T and, O2 and PO4, by the Winkler method and the Molybdenum photometric method as classical procedures during the last forty years in the Japan Sea and the North Pacific subpolar regions (Figure 1) [Gamo, 1999; Ono et al., 2001; Hokkaido Univ., 2002; Japan Oceanographic Data Center, 2003]. Assuming O2 and PO4 to be unchanged in deep water in the North Pacific Ocean and the Okhotsk Sea, and comparing time series of deep water data sets (>27.5 σθ) in the two oceanic regions obtained by the same laboratories measuring the data set in the Japan Sea, we estimated the offsets in O2 and PO4 to be within 5 μmol-O2 kg−1 and 0.1 μmol-PO4 kg−1 for the entire time series data set in this study. We here used these data without any correction of the offsets in this study.

[8] To evaluate whether the oceanic properties have a long-term linear trend and oscillation, we also applied an equation of the Fourier sine expansion to the time series data sets of O2, PO4 and T in the two oceanic areas: (X = −a · y + b + c · sin {2π (y − d)/e}, where ‘X’ refers one within O2, PO4 and T. ‘y’ is the calendar year. ‘a’, ‘b’, ‘c’, ‘d’ and ‘e’ are constants. That is, X = linear trend component + oscillation component).

3. Results and Discussion

[9] We found that JSDW in two basins have shown an averaged linear decrease of 0.46 ± 0.004 (standard deviation, SD) μmol-O2 kg−1 yr−1 with a clear periodicity (periodicity = 18.8 ± 0.2 yr, amplitude = 4.8 ± 1.0 μmol-O2 kg−1, correlation coefficient (r) = 0.94 ± 0.001) (Figures 2a and 2b. Moreover, we also found an averaged linear increase of 0.002 ± 0.0001 μmol-PO4 kg−1 yr−1 with a similar periodicity despite a slightly smaller coefficient compared to O2 (periodicity = 15.0 ± 0.9 yr, amplitude = 0.05 ± 0.03 μmol-PO4 kg−1, r = 0.58 ± 0.13). In addition, we found that the bidecadal periodicity of T (periodicity = 16.6 ± 2.8 yr, amplitude = 0.014 ± 0.006°C, r = 0.57 ± 0.11) despite the linear increase of T not being significant at the 90% confidence level. The averaged density of JSDW also decreased with 0.0002 ± 0.00004 σθ yr−1 and no significant periodicity at the 90% confidence level (r = 0.55 ± 0.12, p < 0.01 for LT, data not shown), suggesting that water density in JSDW became lighter with time. Both the linear trend and oscillation of O2 have shown a strong negative correlation with those of PO4 and T. In both the trend and the amplitude, the change ratio of PO4 to O2 in JSDW closely agreed with the stoichiometric ratios of PO4 to O2 estimated in the North Pacific [Anderson and Sarmiento, 1994]. The oscillations of O2 in the intermediate water of the two basins in the Japan Sea (1000 m ∼ 1500 m) were also found to be the same as that in the deep water despite both the linear trends being smaller than that in the deep water (decrease rate = 0.16 ± 0.05 μmol-O2 kg−1 yr−1, periodicity = 20.5 ± 1.3 yr, amplitude = 3.5 ± 0.4 μmol-O2 kg−1, r = 0.78 ± 0.04, p < 0.01 for LT, data not shown), representing that the sinking surface water has been recently reaching only an intermediate depth shallower than 2000 m and not the bottom.

Figure 2.

Time series of O2, PO4 and T in JSDW and NPIW, and NPI. The standard errors (SE) of our compiled data in each year were ±1.48 μmol kg−1 for O2, ±0.037 μmol kg−1 for PO4 and ±0.0077°C for T, respectively. (a) O2, PO4 and T in JSDW below 2000 m in the eastern Japan Sea Basin. The symbols, open circle, solid circle and plus are O2, PO4 and T, respectively. The fitted curve equations are as follows; O2 = −0.47y + 1148 + 5.50 sin{2π (y − 1967)/19.0} (SE = 1.09 μmol-O2 kg−1, r = 0.94, p < 0.01 for the linear trend (LT)), PO4 = 0.002y − 2.2 + 0.075 sin{2π (y − 1969)/14.4} (SE = 0.014 μmol-PO4 kg−1, r = 0.68, p < 0.05 for LT) and T = 0.0001y − 0.03 + 0.0181 sin{2π (y − 1970)/14.6} (SE = 0.0027°C, r = 0.65, p > 0.10 for LT). Solid and dashed lines indicate the fitted curves and the linear trends, respectively. ‘r’ is the correlation coefficient for the fitted curve equation. (b) O2, PO4 and T in JSDW below 2000 m in the Yamato Basin. The symbols, open triangle, solid triangle and cross are O2, PO4 and T, respectively. The fitted curve equations are as follows; O2 = − 0.46y + 1130 + 4.02 sin{2π (y − 1967)/18.8} (SE = 1.05 μmol-O2 kg−1, r = 0.94, p < 0.01 for LT), PO4 = 0.002y − 2.6 + 0.028 sin{2π (y − 1973)/15.7} (SE = 0.012 μmol-PO4 kg−1, r = 0.49, p < 0.05 for LT) and T = 0.0001y − 0.08 + 0.0085 sin{2π (y − 1969)/18.5} (SE = 0.0020°C, r = 0.49, p > 0.10 for LT). These data are based on historical data below 2000 m from 1961 to 2001 [Gamo, 1999; Hokkaido University, 2002; Japan Oceanographic Data Center, 2003]. (c) O2, PO4 and T in NPIW between 26.8 σθ and 27.4 σθ in the North Pacific subpolar area. The symbols, open square, solid square and open triangle are O2, PO4 and T, respectively. We here show time-series data sets of O2, PO4 and T on 26.8 σθ as typical data because all periodicity of O2, PO4 and T were the same despite the linear trends to be gradually smaller with increasing density. The relations were based on the historical data sets during 1968 to 1998 (averaged water depth: 260 ∼ 1030 m) [Ono et al., 2001]. The fitted equations are as follows; O2 = −0.73y + 1617 + 14.99 sin{2π (y − 1967)/18.3} (SE = 2.15 μmol-O2 kg−1, 26.8 σθ, r = 0.83, p < 0.01 for LT), PO4 = 0.003 y − 3.133 − 0.087 sin{2π (y − 1966)/19.1} (SE = 0.021 μmol-PO4 kg−1, r = 0.60, p < 0.10 for LT) and T = 0.005y − 7.610 (SE = 0.005°C, r = 0.16, p > 0.10 for LT). The averaged linear trend and periodicity of O2, PO4 with SD between 26.8 σθ and 27.4 σθ were −0.47 ± 0.17 μmol-O2 kg−1 yr−1 and 18.5 ± 1.0 yr for O2, and 0.003 ± 0.001 μmol-PO4 kg−1 yr−1and 19.7 ± 1.4 yr for PO4, respectively. The averaged linear trend of T with SD was 0.0050 ± 0.0012°C yr−1(averaged p < 0.1 for LT). (d) NPI in the wintertime. We here showed the 10–80 year band-pass filtered time series of NPI after 1960 based on Minobe [2000]. The fitted curve equation was NPI = −0.07y + 139 + 1.58 sin{2π (y − 1966)/20.1} (SE = 0.04 hPa, r = 0.98, p < 0.01 for LT).

[10] On the other hand, we found the linear trend of O2 decrease of 0.73 μmol-O2 kg−1 yr−1 on 26.8 σθ in NPIW (Figure 2c). Between 26.8 σθ and 27.4 σθ, the averaged linear trend of O2 decrease of 0.47 ± 0.17 μmol-O2 kg−1 yr−1 was equal to that in JSDW. The decrease of O2 over NPIW was extensively found during the past two decades [e.g., Watanabe et al., 2001], and its decreasing rate of O2 is also almost equal to those in both JSDW and NPIW in this study. Moreover the averaged linear increase of PO4 of 0.003 ± 0.001 μmol-PO4 kg−1 yr−1 was found above the density of 27.2 σθ, which is consistent with that in JSDW. We also found the averaged linear increase of T of 0.005 ± 0.001°C yr−1 despite the bidecadal periodicity not being significant at the 90% confidence level. The decrease of 0.035 σθ yr−1 in the surface was reported in this region [Ono et al., 2001]. These suggested that water mass sinking from the surface to the ocean interior has been recently decreasing in this area.

[11] In addition, NPIW above 27.4 σθ had the same bidecadal periodicity of O2 and PO4 as those in JPDW, superimposed on the linear decrease of O2 (periodicity = 18.5 ± 1.0 yr, amplitude = 8.5 ± 4.5 μmol-O2 kg−1, r = 0.91 ± 0.06) and a linear increase of PO4 (periodicity = 19.7 ± 1.4 yr, amplitude = 0.07 ± 0.02 μmol-PO4 kg−1, r = 0.51 ± 0.08). These first findings indicate that the changes of O2 and PO4 in JSDW synchronize with those in NPIW down to water density of 27.4 σθ despite the fact that the two water formation systems are independent of one another.

[12] The decrease of O2 with time in the ocean interior could be caused by a reduction of the water formation rate or an increase in the biological production in the surface mixed layer. We here estimated the decrease of O2 inventory in JSDW (∫ΔO2 dz) to be 29 mol-O2 m−2 from 1961 to 2001. From a previous study in the Japan Sea [Berger et al., 1987], an export flux of biological organic carbon from the surface where oceanic biota mainly exists was estimated to range from 1.6 to 2.8 mol-C m−2 yr−1 (average: 2.2 mol-C m−2 yr−1) (F). The vertical export flux from the surface to the deep water was estimated to be only 0.07 (7%) of the initial export flux (E) due to the decomposition of biological organic matter [Martin et al., 1987]. If the biological activity changed with the water formation rate to be constant during the past forty years, the recent increase of biological activity (ΔB) has to become five times as large as that in the beginning of 1960s by using the equation of ΔB = ∫ΔO2 dz / (E·F·RO2/C) with the stoichiometric ratios of carbon to oxygen (RO2/C) [Redfield et al., 1963; Anderson and Sarmiento, 1994]. Similarly, the biological activity in the North Pacific subpolar region would also has to increase by two times relative to that in the beginning of 1960s. Since there are no reports of a significant increase in the biological activity in the two oceanic areas, it is reasonable to assume that the export flux from the surface mixed layer has been relatively constant over the past forty years. Therefore, it suggests that the reductions of water formation rate have caused the changes of oceanic properties, and that these changes in the two oceanic areas have continuously synchronized down to the density of 27.4 σθ with the bidecadal oscillation at least during the past forty years.

[13] What has caused the synchronous extensive change of water formation rate and consequently has produced the bidecadal oscillation of O2, PO4 and T superimposed on the linear trends in JSDW and NPIW? Taking notice of the interaction between the atmosphere and the ocean in extensive areas, we here focused on the North Pacific Index (NPI) as an anomaly of the sea surface level pressure in the wintertime of the North Pacific (160°E–140°W, 30°N–65°N) [Minobe, 2000]. It is possible that NPI shows the changes of heat flux and wind stress causing the change of surface oceanic condition. From 1960 to 2000, NPI exhibited a linear decrease of 0.07 hPa yr−1 with a sinusoidal trend (periodicity = 20.1 yr, amplitude = 1.6 hPa) although the linear trend of NPI was not found through 1900 to 2000 (Figure 2d). In both oceanic regions, O2 decreased and PO4 increased with a drop of the sea surface pressure. All the oscillations of O2 and PO4 in JSDW and NPIW synchronized with that of NPI (JSDW: r = 0.96 ± 0.01; NPIW: r = 0.87 ± 0.07). When the sea surface pressure drops, the reduction of water formation rate possibly occurs due to the warming of sea surface water (Figures 2a–2c) and consequently the supply of O2-rich water could decrease from the surface to the ocean interior and PO4 could increase.

[14] One possibility to explain the synchronous changes between the air and the intermediate-deep sea in extensive areas is that the atmospheric change causes the change of the water formation rate in the ocean surface, and affects the movement of its related intermediate-deep water along an isopycnal surface or vertically. Only the change of upstream water movement and not new water itself may be conveyed downstream one after another in the water formation system without the time lag even if the water ventilation time is long.

[15] In the summer 2001, O2, salinity and chlorofluorocarbons significantly have increased over a wide area of JSDW due to the severe cold winter of 2000–2001 [e.g., Tsunogai et al., 2003]. This fact suggests that new deep water was formed because of an intense heat flux from the sea surface to the atmosphere, and the change of deep water movement due to the change in water formation rate spread over the entire Japan Sea without the time lag. Similarly, in NPIW, the change of intermediate water movement due to the change in water formation rate would spread in response to atmospheric change.

[16] The relations among the changes of O2, PO4, T and NPI may be evidence that atmospheric oscillations and/or the artificial greenhouse warming effects simultaneously cause extensive rapid changes of oceanic conditions in respective independent oceanic areas. The change of oceanic conditions in JSDW may serve as an indicator of the atmosphere-ocean coupled climate change including the artificial greenhouse warming effects at least in the North Pacific and/or the global oceans.

Acknowledgments

[17] We would like to express our gratitude to the many scientists and technicians for assistance with the collection of hydrographic data in the Japan Sea and the North Pacific subpolar regions for their dedicated works of long-term observations. We also extend our profound thanks to Dr. C. T. A. Chen and anonymous two reviewers for their many fruitful comments.

Ancillary