Repeat trans-Pacific hydrographic observations along the pathway of Lower Circumpolar Deep Water (LCDW) reveal that bottom water has warmed by about 0.005 to 0.01°C in recent decades. The warming is probably not from direct heating of LCDW, but is manifest as a decrease of the coldest component of LCDW evident at each hydrographic section. This result is consistent with numerical model results of warming associated with decreased bottom water formation rates around Antarctica.
 The role of the Thermo-Haline Circulation (THC) in climate is an important aspect of the planetary response to global warming. Model studies suggest that the THC in the Atlantic Ocean is sensitive to anthropogenic climate change [Cubash and Meehl, 2001]. Recently Bryden et al.  reported that the Atlantic meridional circulation had slowed by about 30% between 1957 and 2004, based on five sets of repeated trans-Atlantic observations along 25°N. The warming trend of the global ocean [Levitus et al., 2000], decreases in the signature of North Atlantic Deep Water (NADW) in the South Pacific [Johnson et al., 1994], and the warming at mid-depths in the Southern Ocean [Gille, 2002] could all potentially affect the THC in the Pacific Ocean.
 Lower Circumpolar Deep Water (LCDW) formed in the Southern Ocean flows along the bottom in the Pacific Ocean as the northward component of the THC. It enters the Pacific east of New Zealand and flows northward to the North Pacific through the Samoan Passage. It upwells in the North Pacific and returns southward as modified North Pacific Deep Water (mNPDW) [Schmitz, 1996]. Repeated trans-Pacific surveys along 47°N show that the deepest waters of the North Pacific Ocean have warmed significantly owing to a decrease in the volume of the colder portion of modified NADW, which is the upper part of LCDW [Fukasawa et al., 2004], but the relationship between this warming and reported decreases in the NADW signature in the South Pacific Ocean [Johnson et al., 1994; Johnson and Orsi, 1997] is not clear. Here we analyze data collected between 2003 and 2006 by trans-Pacific surveys along 32°S, 149°E, 24°N, and 30°N. These surveys were designed to revisit the hydrographic stations previously occupied during the World Ocean Circulation Experiment (WOCE) and thus improve our understanding of temperature changes in the deep and bottom water of the Pacific Ocean.
2. Data Sets and Methods
 Highly accurate observations and careful data processing are crucial for detecting any change of temperature in the deep and bottom water. The WOCE Hydrographic Program (WHP) collected data along dozens of trans-Pacific hydrographic sections, mostly during the 1990s, with the highest accuracy and precision possible at that time. We used the data from hydrographic sections along 32°S (section WHP-P6; 1992), 149°E (P10; 1993), 24°N (P3; 1985), and 30°N (P2; 1994), available from the CLIVAR(Climate Variability and Predictability)/Carbon Hydrographic Data Office(CCHDO; http://whpo.ucsd.edu/). Accuracy of temperature and salinity measurements from these cruises was about 0.001°C and a few parts per million, respectively.
 Section P6 was revisited during July and August 2003 as a part of the “BEAGLE 2003” cruise of the R/V Mirai of the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) [Uchida and Fukasawa, 2005]. The R/V Mirai revisited section P10 (124 stations) in June 2005, and section P3 (224 stations plus 11 stations across Wake Island Passage) between October 2005 and January 2006. At each station along sections P10 and P3, a full-depth CTD profile was conducted and up to 36 water samples were collected and analyzed. In situ temperature of these three cruises was corrected against the value measured at water sampling layers by a deep ocean standard thermometer (model SBE35; Sea-Bird Electronics Inc., Bellevue, WA, USA) calibrated by the manufacturer in water triple-point and gallium melting-point cells following the methodology used for a standard platinum resister thermometer. The uncertainty of temperature measurements from these three cruises was less than 0.001°C (H. Uchida et al., In-situ calibration of Sea-Bird 9 plus CTD thermometer, submitted to Journal of Atmospheric and Oceanic Technology, 2006). CTD salinity was calibrated with sampled water salinity, measured by a Guildline Autosal 8400B standardized using IAPSO standard seawaters (SSW). The analysis of differences between calibrated CTD salinity and sampled water salinity yielded an estimated maximum uncertainty of 0.001 in PSS-78 with reference to SSW. The R/V MELVILLE revisited section P2 in 2004, and the data and cruise report are available from CCHDO. Accuracy of salinity measurements below 1000 dbar was reported at about 0.002 in PSS-78. Since the temperature calibration was carefully made by a method similar to that used for sections P6, P3, and P10 in this study, we have assumed an equivalent accuracy for section P2 in 2004. The SSW batch correction for WOCE expeditions led to a significant reduction in salinity variance at crossovers of WHP sections in the Pacific Ocean [Aoyama et al., 2002]; therefore, we applied the latest batch correction [Kawano et al., 2006] to salinity data in this study. The batch used for WOCE P2 (revisit), P3 (revisit), P6 (revisit) and P10 (revisit) was P121/P123 (P144), P96 (P145), P116 (P142) and P114 (P145), respectively. Temperature was scaled with the international temperature scale of 1990 [Preston-Thomas, 1990].
3. LCDW Pathway
 LCDW consists of two cold water masses, each with relatively high salinity, high oxygen concentration, and low silicic acid (silicate) concentration. The upper portion of LCDW is modified North Atlantic Deep Water (mNADW), and the lower portion is influenced by Antarctic Bottom Water (modified AABW) [Johnson et al., 1994].
 Hydrographic sections P2, P3, P6, and P10 sample the path of LCDW throughout much of the Pacific Ocean (Figure 1). LCDW enters the Pacific Ocean mainly from the Southwest Pacific Basin and is transported by a deep western boundary current off the Tonga-Kermadec Ridge [Whitworth et al., 1999]. Since a strong silicate front was observed around 162°W at 4000 m depth in section P6 [Uchida and Fukasawa, 2005], LCDW can be considered to flow mainly between this longitude and the Tonga-Kermadec Ridge. From the Southwest Pacific Basin, LCDW flows northward through the Samoan Passage, but the local salinity maximum and silicate minimum associated with the core of mNADW disappear north of the passage [Johnson et al., 1994]. Only the upper portion of the LCDW, composed of mNADW, thus flows into the central Pacific Basin [Fukasawa et al., 2004], where it splits into three branches. The western branch enters the East Mariana Basin, the northern branch proceeds into the Northwest Pacific Basin through the Wake Island Passage, and the eastern branch flows toward the Northeast Pacific Basin [Mantyla and Reid, 1983].
 The 1.2°C potential isotherm has been used to define the boundary between LCDW and NPDW in the North Pacific [Johnson and Toole, 1993; Fukasawa et al., 2004]. At Wake Island Passage, water with potential temperature lower than 1.2°C had silicate concentrations less than about 150 μmol/kg (Mantyla  and P3 in this study, not shown here), and we considered water with these properties to be mNADW and indicative of LCDW. In the North Pacific, this water was found from 140°E to 150°W along section P2. Along section P3, this water was found from 142°E to 168°W and from 155°W to 132°W. The mNADW water found in the Northeast Pacific Basin was from the eastern branch of the LCDW inflow, which originates north of the Samoan Passage and flows through the Clarion Passage [Mantyla, 1975]. In section P10, mNADW water was found north of 10°N. The water around 12°N was from the western branch of the LCDW inflow [Wijffels et al., 1998; Kawabe et al., 2003], which originates north of the Samoan Passage. The mNADW water observed north of 20°N along section P10 arrived through the Wake Island Passage, as did that observed along sections P2 and P3 in the Northwest Pacific Basin (Figure 1).
 We calculated potential temperature differences below 2000 dbar along each hydrographic section by subtracting earlier (WHP) values from more recent ones (Figure 2). We considered absolute differences less than 0.003°C to be insignificant from the standpoint of instrumental accuracies. Potential temperature increased by about 0.005 to 0.01°C in almost all regions where the LCDW was seen. We considered a salinity difference of less than 0.003 in PSS-78 as insignificant and found no significant salinity differences in these regions (data not shown). For each hydrographic section we calculated an integrated area difference starting from the lowest temperature [Fukasawa et al., 2004]. By computing a series of areas for potential temperature classes with a bin size of 0.01°C, we calculated the difference in area for each potential temperature class. Then differences were integrated from the coldest bin to warmer ones. Although the temperature of the bottom water was increasing to the north and the signal was relatively small in P2 and P10, the area of the coldest portion of the LCDW decreased and the area of the warmer portion increased in all hydrographic sections (Figure 3).
 Potential temperature below 1.2°C was averaged over the area where the LCDW was seen in each section (Table 1). In the North Pacific (sections P1, P2, P3 and P10), the area-averaged temperature increased by 0.001°C to 0.004°C, while it increased by 0.008°C in the South Pacific (section P6). Time intervals between section repeats were different. The required heat input per unit volume for each section in Table 1 was calculated as a function of in situ density (ρ), heat capacity (Cp), and potential temperature (θ) as follows [Bacon and Fofonoff, 1996];
where, subscript 1 and 2 denotes the first and second visit, respectively. H is heat content and t is time. X and z is the horizontal and vertical distance, respectively. Area of the layer colder than 1.2°C for each visit was averaged to give mean area, S.
Table 1. Area-Averaged Temperature and Required Heat Input
Averaged Temperature Below 1.2°C, deg
Required Heat Input Per Unit Volume Below Mean 1.2°C Isotherm,bμW/m3
Required Heat Input Along Section Below Mean 1.2°C Isotherm,bμW/m2
 The required heat input along each section (also in Table 1) was the product of the averaged thickness of the layer colder than 1.2°C and dH/dt. The results in the North Pacific are close to the geothermal heat flux of 0.042 W/m2 for old oceanic crust [Sclater and Parsons, 1981] or 0.05 W/m2 estimated from observations in the sub-arctic Pacific Ocean [Joyce et al., 1986], while the value of heat input in the South Pacific (P6) is larger by a factor of five. Fukasawa et al.  estimated the heat input required for the observed bottom water warming along P1 to be 0.080 W/m2 and concluded that the observed warming cannot be explained by geothermal heating but is the result of change in oceanic circulation because the observed heat input was almost 200% of the geothermal heat flux and because no significant change in the relationship between potential temperature and dissolved oxygen occurred. We performed the same analysis using dissolved oxygen from each section (Table 2). The accuracy of dissolved oxygen measurements from WOCE P2 (1994), P3 (1985), P6 (1992) and P10 (1993) was approximately 0.064 ml/l, from 0.021 to 0.022 ml/l, 0.02 ml/l and 0.02 ml/l, respectively [You et al., 2000; Roemmich et al., 1991; cruise reports for P10; Whitworth et al., 1999]. These values are equivalent to 3 μmol/kg for P2 and 1 μmol/kg for P3, P6 and P10. The accuracy from cruises revisiting these sections was estimated at 1 μmol/kg. It is notable here that neither significant increase in zonally averaged dissolved oxygen nor significant apparent change in the relationship between potential temperature and dissolved oxygen were observed along all sections, moreover the larger heat input was estimated along the southernmost section P6. These points suggest that changes in oceanic circulation are more important than geothermal heating to explain the warming observed in present study as in the case of Fukasawa et al. .
Table 2. Averaged Dissolved Oxygen for Each Section
Dissolved oxygen is averaged over the layer of which potential temperature falls in the range on the left-most column in the table.
1.1 < θ < 1.2
1.0 < θ < 1.1
0.9 < θ < 1.0
0.8 < θ < 0.9
0.7 < θ < 0.8
0.6 < θ < 0.7
0.5 < θ < 0.6
 The mechanism of the warming phenomena is not clear. It is possible that LCDW was warmed in the region of formation and then transported by the THC. If so, the LCDW warming should extend back over the time scale of the THC, that is, about 1000 years. This is not likely since reconstructed climate data for this millennium suggest that it has only been since 1981 that average surface temperatures (near surface air temperatures over land, and sea surface temperatures) have shown rapid increases in the Northern Hemisphere [Mann et al., 1999]. Conditions in the Southern Hemisphere are less well known because of fewer data. Analysis of numerical model results suggests another possible explanation [Nakano and Suginohara, 2002]. In this model, 50 years after cessation of bottom water formation off the Adèlie Coast of Antarctica, bottom water at 30°N in the Pacific Ocean warmed by 0.02°C after 50 years, with a salinity change of less than 0.005. The warming rate and salinity changes in their model resemble the observations in this study, that is, warming of about 0.005 to 0.01°C for 10 to 20 years along sections P2 and P3 with no significant change in salinity. Modeled temperature increases in the Pacific are not due to the direct warming of LCDW but rather to changes in the pressure field along the pathway of LCDW caused by Kelvin and Rossby wave propagation. This means that, according to their model, changes occurring in the region of deep water formation influenced conditions to the north on a time scale much shorter than that of the THC.
 Our results may reflect changes occurring in the region of deep water formation, extended to the North Pacific through the processes of wave propagation described by the models [Nakano and Suginohara, 2002; Suginohara and Fukasawa, 1988]. If so, this could suggest that the rate of bottom water formation in the Antarctic region has decreased in recent decades.
 The authors thank Nobuo Suginohara of JAMSTEC for his suggestions and discussion on the interpretation of results from numerical models. The authors are indebted to Gregory Johnson of NOAA/PMEL for valuable comments for us to prepare the manuscript. The authors are also grateful to Captains Masaharu Akamine and Takaaki Hashimoto, the crew of R/V Mirai and technicians from Marine Works Japan Co. Ltd. for their cooperation. The 2004 P2 reoccupation is part of the NOAA/NSF funded U.S. CLIVAR/CO2 Repeat Hydrography Program.