Warming of the World OceanSydney Levitus,* John I. Antonov, Timothy P. Boyer, Cathy StephensWe quantify the interannual-to-decadal variability of the heat content (meantemperature) of the world ocean from the surface through 3000-meter depthfor the period 1948 to 1998. The heat content of the world ocean increased by⬃2 ⫻ 1023joules between the mid-1950s and mid-1990s, representing avolume mean warming of 0.06°C. This corresponds to a warming rate of 0.3watt per meter squared (per unit area of Earth’s surface). Substantial changesin heat content occurred in the 300- to 1000-meter layers of each ocean andin depths greater than 1000 meters of the North Atlantic. The global volumemean temperature increase for the 0- to 300-meter layer was 0.31°C, corre-sponding to an increase in heat content for this layer of ⬃1023joules betweenthe mid-1950s and mid-1990s. The Atlantic and Pacific Oceans have undergonea net warming since the 1950s and the Indian Ocean has warmed since themid-1960s, although the warming is not monotonic.The Intergovernmental Program on ClimateChange (1), the World Climate Research Pro-gram CLIVAR (2), and the U.S. NationalResearch Council (3) have identified the roleof the ocean as being critical to understandingthe variability of Earth’s climate system.Physically we expect this to be so because ofthe high density and specific heat of seawater.Water can store and transport large amountsof heat.Simpson (4) conducted the first study ofEarth’s heat balance which concluded that theEarth system is not in local radiative balance,and therefore transport of heat from the trop-ics to the poles is required for the Earthsystem to be in global radiative balance. Iden-tifying the mechanisms by which heat istransported from the tropics to the poles isone of the central problems of climate re-search. In addition, Rossby (5) drew attentionto the fact that because of its large specificheat capacity and mass, the world oceancould store large amounts of heat and removethis heat from direct contact with the atmo-sphere for long periods of time. The results ofthese studies are the subject of this researcharticle.Until recently, little work has been donein systematically identifying ocean subsur-face temperature variability on basin andglobal scales, in large part due to the lack ofdata [recent studies include (6–8)]. The firststep in examining the role of the ocean inclimate change is to construct the appropriatedatabases and analysis fields that can be usedto describe ocean variability. About 25 yearsago, ship-of-opportunity programs were ini-tiated to provide measurements of subsurfaceupper ocean temperature. Before the initia-tion of these programs, subsurface oceano-graphic data were not reported in real time, asis the case with much meteorological data.During the past 10 years, projects have beeninitiated (9) that have resulted in a largeincrease in the amount of historical upperocean thermal data available to examine theinterannual variability of the upper ocean.Using these data, yearly, objectively ana-lyzed, gridded analyses of the existing datawere prepared and distributed (7) for individ-ual years for the period 1960 to 1990. Wehave used the recently published WorldOcean Database 1998 (10–13) to prepareyearly and year-season objectively analyzedtemperature anomaly fields. Detailed infor-mation about the temperature data used inthis study can be found in this series. Com-putation of the anomaly fields was similar toour earlier work (7), but some procedureswere changed (7).To estimate changes in heat content atdepths greater than 300 m, we prepared ob-jective analyses of running 5-year compositesof all historical oceanographic observationsof temperature for the period 1948 to 1996 atstandard depth levels from the surfacethrough 3000-m depth using the proceduresdescribed above. Constructing composites ofdeep-ocean data by multiyear periods is nec-essary due to the lack of deep-ocean obser-vations. Most of the data from the deep oceanare from research expeditions. The amount ofdata at intermediate and deep depths decreas-es as we go back further in time.Temporal Variability of Upper OceanHeat ContentFigure 1 shows the variability of yearly heatcontent anomalies in the upper 300 m for1948 to 1998 for individual ocean basinsdefined using the Equator as a boundary.Each yearly estimate includes the standarderror of the mean anomaly value for eachyear plotted as a vertical bar. The anomalyfields for the Atlantic and Indian oceans, forboth the entire basins and Northern andSouthern Hemisphere basins of each ocean,show a positive correlation. In each basinbefore the mid-1970s, temperatures werenearly all relatively cool, whereas after themid-1970s these oceans are in a warm state.The year of largest yearly mean temperatureand heat content for the North Atlantic is1998. In 1998 heat content reaches a value of⬃4 ⫻ 1022J, equivalent to a volume meantemperature anomaly of 0.37°C. [Expandedversions of Figs. 1 and 4 with volume meantemperature scales as well as heat contentscale and similar time series for heat contentintegrated through 1000-m depth can beviewed at Science Online (14) as Web figures1to3.]Both Pacific Ocean basins exhibit quasi-bidecadal changes in upper ocean heat content,with the two basins positively correlated. Dur-ing 1997 the Pacific achieved its maximum heatcontent. A decadal-scale oscillation in NorthPacific sea surface temperature (Pacific Dec-adal Oscillation) has been identified (15, 16 ),but it is not clear if the variability we observe inPacific Ocean heat content is correlated withthis phenomenon or whether there are addition-al phenomena that contribute to the observedheat content variability.In order to place our results in perspective,we compare the range of upper ocean heatcontent with the range of the climatologicalannual cycle of heat content for the NorthernHemisphere and world ocean computed asdescribed by (8) but using a more completeoceanographic database (10–13). There isrelatively little contribution to the climatolog-ical range of heat content from depths below300 m. Our results indicate that the decadalvariability of the upper ocean heat content ineach basin is a significant percentage of therange of the annual cycle for each basin. Forexample, the climatological range of heatcontent for the North Atlantic is about 5.6 ⫻1022J, and the interdecadal range of heatcontent is about 3.8 ⫻ 1022J.Changes in Temperature at