15 AUGUST 2006 S TARK ET AL. 4075 Reevaluating the Causes of Observed Changes in Indian Ocean Water Masses SHEILA STARK,RICHARD A. WOOD, AND HELENE T. BANKS Hadley Centre for Climate Prediction and Research, Met Office, Exeter, Devon, United Kingdom (Manuscript received 16 February 2005, in final form 8 December 2005) ABSTRACT The consistency between observed changes in Subantarctic Mode Water (SAMW) properties at 32°S in the Indian Ocean and model simulations is explored using the Third Hadley Centre Coupled Ocean– Atmosphere GCM (HadCM3). Hydrographic data collected in 2002 show that the water mass is warmer and saltier on isopycnals than in 1987, in contrast to the isopycnal freshening observed between 1962 and 1987. The response of HadCM3 under a range of forcing scenarios is explored and the observed freshening is only seen in experiments that include greenhouse gas forcing; however, there is no subsequent return to more saline conditions in 2002. The response of the model to greenhouse gas forcing is dominated by a persistent freshening trend, the simulated water mass variability agrees well with that suggested by the limited observations. Comparing model isopycnal changes from the forced experiments with a control run shows that the changes from the 1960s to 2002 are best explained by internal variability. This is in contrast to earlier work, which attributed the observed isopycnal freshening to anthropogenic forcing. Although the model shows that at present an anthropogenic climate change signal is not detectable in SAMW, the model water mass freshens on isopycnals during the twenty-first century under increased greenhouse gas forcing. This is consistent with recent heat content observations, which suggest that the salting is unlikely to persist. In HadCM3, this freshening is due to an increasing surface heat flux and Ekman heat and freshwater flux into the water mass formation region. This paper emphasizes the importance of higher-frequency obser- vations of SAMW if detection and attribution statements are to be made. 1. Introduction to global climate through the large-scale transport of heat and freshwater in the oceans. Despite the impor- The hydrological cycle is a fundamental component tant role of the ocean in the global hydrological cycle, of the planetary energy budget yet it remains one of the direct measurements of oceanic precipitation and least understood elements of the climate system. Over evaporation are too sparse to detect any patterns of recent years there has been a growing body of evidence change or variability in these fluxes. Interior water for a global-scale shift in the distribution of freshwater, masses, which are directly ventilated at the ocean sur- which may be linked to global warming and a possible face, are an attractive way to gauge changes in surface strengthening in the hydrological cycle [e.g., Curry et al. fluxes as they act to integrate highly variable surface (2003) in the Atlantic and Wong et al. (1999) in the changes in heat and freshwater improving the signal to Indo-Pacific]. Most ocean model simulations, under in- noise ratio. Despite this potential, the sparsity of oce- creasing greenhouse gas emissions, predict a weakening anic measurements makes it difficult to determine North Atlantic thermohaline circulation due to fresh- whether changes in water mass properties reflect inter- ening and warming in the subpolar seas (Houghton et nal climate variability and are “normal,” or reflect, for al. 2001), so an intensified hydrological cycle character- example, anthropogenic climate change, and are “un- ized by increased high-latitude precipitation could have usual.” For this reason, general circulation models pro- significant climatic impacts. vide a unique opportunity to understand the past and to The study of ocean water masses is intimately linked predict future ocean climatic changes. Subantarctic Mode Water (SAMW) is a globally im- portant water mass formed in large quantities in the Corresponding author address: Sheila Stark, Hadley Centre for Climate Prediction and Research, Met Office, Fitzroy Road, Ex- Southern Ocean (McCartney 1977). Along with Ant- eter, Devon EX1 3PB, United Kingdom. arctic Intermediate Water (AAIW), SAMW exported E-mail: [email protected] from the Southern Ocean forms an upper limb of the Unauthenticated | Downloaded 09/28/21 06:38 PM UTC JCLI3845 4076 JOURNAL OF CLIMATE VOLUME 19 Ϫ FIG. 1. Zonal mean isopycnal salinity in the range of the modeled mode waters (25.5–26.5 kg m 3) for 32°Sin the Indian Ocean for the 400 yr of the CTL run over which the ensemble members were run (a) before and (b) after the linear trend was removed from each isopycnal surface. global overturning circulation (Sloyan and Rintoul during the twentieth century, focusing primarily on the 2001). An observed cooling and freshening of SAMW last 50 yr. HadCM3 is a fully coupled ocean–atmo- on isopycnals between the 1960s and 1990s has been sphere model, without flux adjustments, which has been well documented (Bindoff and McDougall 2000; described in detail elsewhere (Gordon et al. 2000). All Johnson and Orsi 1997; Wong et al. 1999) and attrib- of the forced model runs are compared to a control uted to changes in surface dynamic forcing and the experiment (CTL), which was initialized with the hy- warming or freshening of surface waters. A coupled drography of the ocean given by Levitus and Boyer climate model study using the Third Hadley Centre (1994) and run with a fixed atmospheric composition ϭ Coupled Ocean–Atmosphere GCM (HadCM3) by representative of 1860 (pCO2 290 ppm). The first Banks et al. (2000) found that with anthropogenic forc- member of each forced ensemble is started from year ing SAMW became both cooler and fresher on isopy- 370 of CTL, by which time both the heat and freshwater cnals, and Banks and Bindoff (2003) classify the cooling budgets have come into balance (Pardaens et al. 2003). and freshening of midlatitude isopycnals in the Indo- We analyze ensembles forced with natural forcing Pacific as a possible fingerprint of anthropogenic cli- (NAT) from historical records of volcanic emissions mate change in the ocean. The most recent occupation and solar irradiance, with anthropogenic forcing (ANT) of the 32°S section in the Indian Ocean, however, has consisting of imposed historical changes in greenhouse shown the properties of SAMW return to near-1960s gases, ozone and sulfur, and with a combination of the conditions (Bryden et al. 2003; McDonagh et al. 2005), two (ALL). For each ensemble, four simulations are a reversal not seen in the HadCM3 simulations. In this carried out with the same forcing applied but different paper, the work of Banks et al. (2000) is continued, with initial conditions, generated by starting 100 yr apart in a more detailed analysis of SAMW in HadCM3 using a the control integration. Each ensemble covers different larger number of experiments than was available for time periods and we examine in detail the NAT en- the earlier study, and taking into account the recent semble from 1900 to 1996, the ANT ensemble from Bryden et al. (2003) results. The consistency between 1900 to 1999, and ALL from 1900 to 2002. the HadCM3 simulations and the observations is exam- ined using a range of forcing scenarios with a view to b. Control drift addressing what is driving the water mass changes and Examination of zonal mean isopycnal salinity at 32°S whether the observed isopycnal changes reflect anthro- in CTL at mode water densities reveals a freshening pogenic forcing. trend on all levels, as illustrated by Fig. 1. It is found that the drift on each isopycnal surface is linear (R2 Ͼ 2. Methods 0.95 for each level) and a unique trend is removed from each isopycnal individually. On the 25.7 kg mϪ3 isopy- a. Model runs cnal, the core of the model SAMW in the western A suite of ensemble runs using HadCM3 are used to mode, the drift has a magnitude of 0.0005 psu yrϪ1; the look at the changing properties of SAMW along 32°S trend in the eastern mode is of a similar magnitude. Unauthenticated | Downloaded 09/28/21 06:38 PM UTC 15 AUGUST 2006 S TARK ET AL. 4077 FIG. 2. (a) Potential vorticity along 32°S in the HadCM3 control run. Note that the plot shows the magnitude of PV, with the minimum values being those closest to 0, though for stable stratification PV is negative in the Southern Hemisphere. SAMW is clearly visible as a band of low PV between 25.5 and 26.0 kg mϪ3. (b) The temperature and salinity of the mid-depth PV minimum. From Fig. 1, it is evident that this procedure success- in HadCM3 at densities ranging from 25.5 to 26.0 kg fully removes the freshening drift on the mode water mϪ3. The modeled SAMW is both fresher and lighter isopycnals. To ensure that any patterns and trends in than observations because of the initial drift, as the each of the forced ensemble members represents water masses adjust to balance the freshwater budgets, changes due to the applied forcing the linear trend in during the model spinup (see Pardaens et al. 2003 for a CTL on each isopycnal is removed prior to all analyses. full discussion). Examination of temperature and salin- ity on the PV minimum (Fig. 2b) reveals that the model c. Differences from Banks et al. (2000) thermocline along this section contains modes whose The work presented here is complimentary to the characteristics fall into two broad categories. The west- earlier analysis described in Banks et al. (2000) where ern half of the section, to approximately 75°E, is occu- the observed isopycnal freshening of SAMW between pied by a water mass with an average temperature of 1962 and 1987 was compared with HadCM3 simula- 13.8°C while the mode water in the east of the section tions.