Ocean Salinity and Global Water Cycle (2015) PDF

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/274313550 Ocean Salinity and the Global Water Cycle Article in Oceanography · March 2015 DOI: 10.5670/oceanog.2015.03. CITATIONS...

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/274313550 Ocean Salinity and the Global Water Cycle Article in Oceanography · March 2015 DOI: 10.5670/oceanog.2015.03. CITATIONS READS 71 2,473 1 author: Paul J. Durack Lawrence Livermore National Laboratory 67 PUBLICATIONS 5,902 CITATIONS SEE PROFILE All content following this page was uploaded by Paul J. Durack on 06 July 2016. The user has requested enhancement of the downloaded file. Oceanography THE OFFICIAL MAGAZINE OF THE OCEANOGRAPHY SOCIETY CITATION Durack, P.J. 2015. Ocean salinity and the global water cycle. Oceanography 28(1):20–31, http://dx.doi.org/10.5670/oceanog.2015.03. DOI http://dx.doi.org/10.5670/oceanog.2015.03 COPYRIGHT This article has been published in Oceanography, Volume 28, Number 1, a quarterly journal of The Oceanography Society. Copyright 2015 by The Oceanography Society. All rights reserved. USAGE Permission is granted to copy this article for use in teaching and research. Republication, systematic reproduction, or collective redistribution of any portion of this article by photocopy machine, reposting, or other means is permitted only with the approval of The Oceanography Society. Send all correspondence to: [email protected] or The Oceanography Society, PO Box 1931, Rockville, MD 20849-1931, USA. DOWNLOADED FROM HTTP://WWW.TOS.ORG/OCEANOGRAPHY Salinity Processes in the Upper-ocean Regional Study Ocean Salinity and the Global Water Cycle By Paul J. Durack Atmosphere: 12.7 0.001% 15% 23% 77% 85% Evaporation Precipitation Precipitation Evaporation ~ 2.7 Sv ~ 3.5 Sv 12.2 ± 1.2 Sv 13.0 ± 1.3 Sv Rivers: 1.25 ± 0.1 Sv Ice (Land & Ocean): 25,540 Land: 15,662 2% Oceans: 1,335,040 1% 97% THE GLOBAL WATER CYCLE. The ocean contains the vast majority of Earth’s water reservoirs, and ~ 80% of surface water fluxes occur over the ocean. Reservoirs represented by solid boxes: 103 km3, fluxes represented by arrows: Sverdrups (106 m3 s–1). Sources: Baumgartner and Reichel (1975), Schmitt (1995), Trenberth et al. (2007), Schanze et al. (2010), Steffen et al. (2010) 20 Oceanography | Vol.28, No.1 ABSTRACT. Alterations to the global water cycle are of concern as Earth’s climate to the effects of coastal and other com- changes. Although policymakers are mainly interested in changes to terrestrial plex terrain and the icy, high albedo polar rainfall—where, when, and how much it’s going to rain—the largest component of the regions (>50°S/N), which limit the accu- global water cycle operates over the ocean where nearly all of Earth’s free water resides. racy of current algorithms. Computer Approximately 80% of Earth’s surface freshwater fluxes occur over the ocean; its surface algorithms have been available since salinity responds to changing evaporation and precipitation patterns by displaying salty 1979 to merge these independent data or fresh anomalies. The salinity field integrates sporadic surface fluxes over time, and streams and produce coarse-​ resolution after accounting for ocean circulation and mixing, salinity changes resulting from products (e.g., Global Precipitation long-term alterations to surface evaporation and precipitation are evident. Thus, ocean Climatology Project [GPCP]; Huffman salinity measurements can provide insights into water-cycle operation and its long- et al., 1997; Adler et al., 2003). term change. Although poor observational coverage and an incomplete view of the Beginning in 1997, high-resolution prod- interaction of all water-cycle components limits our understanding, climate models ucts became available, coinciding with the are beginning to provide insights that are complementing observations. This new launch of the Tropical Rainfall Measuring information suggests that the global water cycle is rapidly intensifying. Mission (TRMM, extended by the Global Precipitation Measurement [GPM] mis- INTRODUCTION operation, and by considering changes sion, which launched in February 2014). When we talk about planet Earth, we are to ocean properties linked to the water In situ gauge data and satellite retriev- really referring to planet “Ocean.” The cycle, insights to its long‐term change. als are often merged for climate studies global ocean covers 71% of Earth’s sur- For this reason, accurate assessments of (e.g., Huffman et al., 2007), keeping in face and contains 97% of all free water ocean salinity changes provide a novel mind that trends may be influenced by stored. If we consider total fluxes of fresh- method with which to investigate broad- inhomogeneities in both the satellite and water at Earth’s surface, the global ocean scale changes to Earth’s water cycle in the sparse in situ data available (Hegerl dominates totals: 85% of evaporation response to ongoing climate change. et al., 2014; Maidment et al., 2014). and 77% of precipitation occurs at the Regional evaporation (E) is more dif- ocean-atmosphere interface (Trenberth THE GLOBAL WATER CYCLE ficult to measure, as it depends on myr- et al., 2007; Schmitt, 2008; Schanze et al., When referring to the global water cycle, iad localized conditions, from wind speed 2010; see global water cycle image on regional precipitation (P) is generally the to the local vertical gradients of humid- p. 20). Additionally, when talking about key focus. The nature of P, with its spo- ity and temperature. Consequently, E is “global warming,” we really mean “ocean radic temporal and spatial patterns, along normally estimated from measurements warming.” Due to its great heat capac- with the sparse terrestrial rain gauge of meteorological variables using “bulk” ity, the ocean has been taking up 93% of observing network, makes it very diffi- formulas, or in a very few instances, from Earth’s energy increase since 1971 (Rhein cult to accurately observe, let alone esti- flux moorings that provide in situ mea- et al., 2013). With the ocean dominating mate, small changes over the observa- surements of all required meteorologi- global water cycle totals by every mea- tional record. cal conditions (see Josey et al., 2013, for a sure, and because it plays such a dom- Satellite-based sensors, in combination more complete description). inant role in changes to the planetary with in situ measurements, provide the Evaporation minus precipitation (E–P) energy budget, it is clearly an important only practical means of observing P over is a more accurate measure of the com- focus of climate research. the ocean and land combined. Thermal plete global water cycle than P or E alone, The importance of the ocean to cli- infrared-based estimates are available as both terms considered together (along mate cannot be overstated, and progress from geostationary satellites at high fre- with the smaller runoff term, R; Schanze requires continuing to work toward bet- quency, but they have only modest skill et al., 2010) provide the water sources ter understanding of ongoing ocean vari- at accurate instantaneous recording of and sinks that comprise the global ocean’s ability and change. Although our under- rainfall intensity (Kidd and Huffman, complete freshwater budget. Large global standing of global temperature changes 2011). Passive microwave data, avail- climatological mean discrepancies of is better quantified, the global water able since 1987, have improved the reli- order 25 W m–2 are apparent between cycle is the climate system element most ability of P retrievals and are particularly independent current-generation surface important to society. The ocean can pro- successful over the ocean, with land- heat flux (or, by definition, E) products, vide unique insights into water-cycle based retrievals more approximate due and suggest that similar issues exist with Oceanography | March 2015 21 current climatological mean E–P esti- the ocean is the ultimate source of all platforms, primarily on oceanographic mates (Josey et al., 2013). Better quan- terrestrial water (Gimeno et al., 2010; research vessels, and more recently from tification of the observed water cycle van der Ent and Savenije, 2013). The ter- automated profiling Argo floats (a more requires further work. restrial component is further complicated comprehensive history is detailed in It has long been noted that the clima- as a result of human interference due to Durack et al., 2013, and Gould et al., tological mean sea surface salinity (SSS) dams, diversions for irrigation, ground- 2013). As described in more detail in arti- and the surface E–P flux field are highly water usage, and land-use changes. cles in this special issue of Oceanography, correlated (Wüst, 1936). This strong cor- Additionally, water limitation (an issue more recently, surface salinity has also relation reflects the long-term balance not apparent with the ocean) and fresh- been measured using dedicated satellites. between ocean advection and mixing water cycling and recycling (van der Ent These data are providing exciting new insights into ocean processes. Historical measurements of ocean “ salinity are very sparse in space and time due to the difficulty in accessing the large The outlook is bright for a continuing portion of the ocean that is far away from improvement in our understanding of the the nearest port. Hydrographic (salinity, temperature, and select chemical tracers) oceanic water cycle due to the development of sampling first came to prominence in an a series of new salinity observing technologies early Atlantic Ocean survey undertaken ” from 1925–1927 on R/V Meteor. This and platforms.. expedition collected discrete (Nansen) bottle measurements obtained at ocean depths that were then analyzed for salin- ity and chemical properties, along with processes and E–P fluxes at the ocean et al., 2010) further obfuscate interpre- temperatures obtained using reversing surface that maintain local salinity gra- tation of long-term terrestrial changes thermometers. Since then, a number of dients. SSS provides an integrated and (e.g., Greve et al., 2014). These issues are coordinated expeditions have provided smoothed field from which the difficult-​ not relevant for the open ocean, and con- increasing global observational cover- to-​measure and poorly constrained E–P sequently do not affect assessments of age. Some key expeditions include the fields—which set the spatial pattern of salinity changes. International Geophysical Year (IGY, ocean SSS—can be inferred over the long 1956–1960), the Geochemical Ocean term. The spatial structure of the global OBSERVING OCEAN SALINITY Sections Study (GEOSECS, 1972–1978), ocean surface and subsurface salinity A key advantage of using the ocean salin- and the Transient Tracers in the Ocean field is maintained by ocean circulation ity field to study the global water cycle is study (TTO, 1981–1983). By the 1970s, and mixing, which are driven by ocean that over the period of observational cov- the significant variability associated density gradients and surface winds that erage, spatial patterns and climatolog- with the ocean mesoscale had been rec- act at the ocean-atmosphere interface. ical mean values are considered stable, ognized, and this led to the World This relationship between E–P and salin- with large deviations from present-day Ocean Circulation Experiment (WOCE) ity motivates the “ocean rain gauge” con- values having occurred during geologic Hydrographic Programme (WHP) that cept (e.g., Schmitt, 2008). The hope is (~100,000-year) rather than climate began in 1990. that improvements in salinity observa- (~100-year) history (e.g., Rubey, 1951; A new era of automated ocean observa- tions can lead to better estimates of the Holland, 1972). This long-term salinity tions began in 1999 with the development poorly observed E–P fluxes over the stability is useful for investigating small and implementation of the Argo float ocean and thus increase understanding of perturbations through time that result program (Gould et al., 2004). The goal of global water cycle operation, variability, from forced changes driven by increasing Argo is to continuously track the ocean’s and long-term change. Because ocean-​ anthropogenic CO2. broad-scale structure, addressing limita- atmosphere freshwater fluxes comprise The earliest measurements of ocean tions in our knowledge from discontinu- ~80% of global surface totals, a bet- salinity date back to 1772, when Captain ous hydrographic observations and poor ter understanding of the oceanic water James Cook embarked on his second voy- spatial and temporal observational cover- cycle should also aid our understanding age to circumnavigate the globe. Since that age. In particular, Argo was implemented of the smaller terrestrial component on time, ocean salinity has been measured to improve ocean coverage in the poorly which a large research focus is based—as using a number of different observing sampled Southern Hemisphere. Figure 1 22 Oceanography | Vol.28, No.1 shows the time and latitudinally distrib- clearer in recent assessments because of amplified water cycle (Allen and Ingram, uted historical evolution of salinity mea- Argo data coverage (Figure 2A,B,D,E; 2002; Held and Soden, 2006). surements. It is clear that while observa- blue denotes freshening regions and red These coherent changes to the global tions appear relatively continuous in time denotes regions with enhanced salinities). ocean salinity field are also apparent since ~1950, there is a strong bias toward In the broad scale, these changes suggest when considering subsurface waters, Northern Hemisphere observations prior that the existing climatological mean gra- with salinity change patterns also largely to Argo (CTD and bottle data). dients of SSS have intensified (Figure 2C enhancing the existing climatological and black contours in panels A, B, D, E), mean (Figure 3). As noted for the surface LONG-TERM CHANGES TO with increased values in the salinity max- salinity patterns, the interbasin contrast OCEAN SALINITY ima and decreased values in the salinity between the Atlantic (salty) and Pacific While limited historical observations minima zones. Taking a step back and just (fresh) also intensifies in the subsurface hinder accurate assessments, long-term considering the large basin-scale changes, (e.g., Boyer et al., 2005; Roemmich and SSS changes have been estimated based there is a marked contrast between the Gilson, 2009; von Schuckmann et al., on both trends fitted to ocean data freshening Pacific and the enhancing 2009; Durack and Wijffels, 2010; Skliris (e.g., Freeland et al., 1997; Wong et al., salinity in the Atlantic that is reinforcing et al., 2014). The penetration of salin- 1999; Curry et al., 2003; Boyer et al., 2005; the existing fresh (Pacific) versus salty ity anomalies into the upper 1,000 m Delcroix et al., 2007; Gordon and Giulivi, (Atlantic) climatological mean contrast suggests that persistent surface changes 2008; Cravatte et al., 2009; Durack and (Figure 2). These results are particularly from past decades have propagated into Wijffels, 2010; Chen et al., 2012; Skliris prominent in the well-sampled Northern the ocean’s interior, increasing salinity et al., 2014) and comparisons of Argo-era Hemisphere Pacific and Atlantic Oceans in the high-salinity subtropical waters (2003–present) data to historical ocean (Figures 1B and 2), and are in broad- and freshening those in the high lati- climatologies (e.g., Johnson and Lyman, scale agreement with the regional stud- tudes. The salty subtropical gyre “bowls” 2007; Hosoda et al., 2009; Roemmich and ies of Curry et al. (2003) and Cravatte in each basin, particularly those of the Gilson, 2009; von Schuckmann et al., 2009; et al. (2009) for the Atlantic and Pacific, Southern Hemisphere Atlantic and Helm et al., 2010). Although these studies respectively. These SSS changes demon- Indian Oceans (not shown), show posi- suggest that long-term, broad-scale and strate that P-dominated wet regions get tive anomalies in the salinity maxima at coherent changes have occurred, it must fresher and E-dominated dry regions salt- depths shallower than 300 m. The pat- be noted that because of the long time ier, following the expected response of an terns found in the Northern Hemisphere scales of ocean variability (e.g., Bullister et al., 2013; Durack et al., 2013; Kirtman et al., 2013; Latif, 2013), elucidating what 140 Profiles (per year) in 1000’s Argo 120 CTDs is forced change (in response to increas- Bottles 100 ing atmospheric CO2) and what is longer 80 time scale natural variability remains the 60 focus of ongoing research. 40 A number of studies have consid- 20 A ered salinity changes on a global scale 0 1950 1960 1970 1980 1990 2000 2010 (Boyer et al., 2005; Hosoda et al., 2009; Year Roemmich and Gilson, 2009; Durack and Profiles (2.5° bins) in 1000’s Wijffels, 2010; Helm et al., 2010; Skliris 240 Argo CTDs et al., 2014). They show that surface pat- 200 Bottles terns of multidecadal salinity change are 160 remarkably similar to the mean salin- 120 ity field (and also the mean E–P; see 80 subsequent section). Rainfall (P) domi- 40 B nated regions such as the western Pacific 0 90°S 70°S 50°S 30°S 10°S 0 10°N 30°N 50°N 70°N 90°N warm (and fresh) pool, for example, have Latitude undergone a long-term freshening, and FIGURE 1. Ocean profile data from salinity observing platforms that comprise the World Ocean subtropical regions in the evaporation Database 2013, updated to 2014 (Boyer et al., 2013). Platform type for (A) per year (1950–2014) and (B) in 2.5° zonal (latitude) bins for the period 1950–2014. The global nature of Argo program obser- (E) dominated “desert latitudes” have vations is evident in the even distribution of profiles across hemispheres. There is a clear Northern generally increased in salinity in each Hemisphere bias in the historical archive comprised of data derived from samples collected in basin. These observations have become Nansen bottles and by CTDs (instruments that measure conductivity, temperature, and depth). Oceanography | March 2015 23 are less homogeneous across basins, of an isopycnal (density) outcrop, which compelling evidence of long-term change which is not surprising considering the is generally poleward due to the broad- to the ocean’s salinity field and agree with North Atlantic and North Pacific have scale warming, can lead to a migration theories of surface-forced ocean changes very different water mass geographies through ocean surface climate zones in response to ongoing climate change, (e.g., Talley, 2008). (e.g., from a low salinity P-dominated the limited number of observations war- Although these changes would appear region to a high salinity E-dominated rants continued consideration of coin- consistent with theories of surface-forced region) and yield a subsurface salinity cident salinity variability (e.g., Dickson salinity changes, coincident ocean warm- anomaly that is unrelated to surface E–P et al., 1988; Belkin et al., 1998; Delcroix ing must also be considered. Durack and changes. Such changes are particularly et al., 2007; Reverdin, 2010; Durack et al., Wijffels (2010) show that salinity changes relevant in regions of very strong meridi- 2013) and its influence on long-term esti- expressed in an ocean density framework onal salinity gradients. mates of change. Such an assessment will are driven by broad-scale surface warm- While these near-surface and sub- be possible in the future as our observa- ing. A temperature-driven lateral shift surface changes appear to provide tional understanding improves with the 70°N 50°N DW10 B05 30°N A B 10°N 10°S 30°S 50°S 70°S 0° 60°E 120°E 180° 120°W 60°W 0° 0° 60°E 120°E 180° 120°W 60°W 0° 70°N DW10 − Mean 33 35 34 50°N 33 C 36 34 30°N 35 37 10°N 36 34 35 35 36 34 10°S 36 35 30°S 35 36 35 34 35 34 50°S 34 34 34 34 34 70°S 0° 60°E 120°E 180° 120°W 60°W 0° 70°N 50°N H09 EN4 30°N D E 10°N 10°S 30°S 50°S 70°S 0° 60°E 120°E 180° 120°W 60°W 0° 0° 60°E 120°E 180° 120°W 60°W 0° −0.20 −0.15 −0.10 −0.05 0.00 0.05 0.10 0.15 0.20 FIGURE 2. Four long-term estimates of global sea surface salinity (SSS) change and the 1950–2000 climatological mean after (A, C) Durack and Wijffels (2010; analysis period 1950–2008), (B) Boyer et al. (2005; analysis period 1955–1998), (D) Hosoda et al. (2009; analysis period 1975–2005), and (E) Good et al. (2014; analysis period 1950–2012) all scaled to represent equivalent magnitude changes over a 50-year period (Practical Salinity Scale 1978 [PSS-78], 50 yr–1). Black contours show the associated climatological mean SSS for the analysis period. Broad-scale similarities exist between each independent analysis of long-term change, and suggest an increase in spatial gradients of salinity have occurred over the period of analysis. However, regional scale differences are due to differing data sources (the B05 analysis did not benefit from the availability of Argo data), temporal periods of anal- ysis, and analysis methodologies. Blue denotes freshening regions and red denotes regions with enhanced salinities. 24 Oceanography | Vol.28, No.1  availability of new observing platforms atmosphere, ocean, and land-surface compares observed climatological mean that provide more comprehensive mea- modules. Recent advances include Earth estimates of SSS (panel A) and E–P surement coverage. System Models (ESMs) that addition- (panel C) to equivalent fields from a rep- ally incorporate modules for simulat- resentative AOGCM from the Coupled LINKING OCEAN SALINITY ing interactions with biogeochemical cli- Model Intercomparison Project Phase 5 CHANGES TO THE GLOBAL mate components. As model complexity (CMIP5; Taylor et al., 2012), ACCESS-1.0 WATER CYCLE and horizontal resolution have increased, (panels B and D). It is clear in Figure 4 Measurable changes to the ocean’s salinity the fidelity of model simulations when that these fields are not identical; how- field have been recorded over the period compared to observation-based estimates ever, unforced climate variability can of observational coverage, but how do has improved (e.g., Bellenger et al., 2013; drive the marked regional changes that changing salinity patterns relate to long- Polade et al., 2013). However, it is worth result in the largest discrepancies between term changes in the global water cycle? noting that significant biases, such as the these panels. When a more broad-scale This question is difficult to answer con- well-described Pacific Ocean “double view is considered, the realism of the sidering the poor state of E–P observa- ITCZ” (Intertropical Convergence Zone; model is impressive: for SSS, the model tions, sparse salinity observations, and e.g., Lin, 2007; Li and Xie, 2014) and captures the broad-scale structure of the coincident ocean warming. What other Atlantic zonal sea surface temperature ocean’s salinity field, with P-dominated tools can oceanographers use to investi- bias (e.g., Richter and Xie, 2008), along fresh regions such as the western Pacific gate relationships between ocean salinity with others, still exist (e.g., Flato et al., warm pool and the high-latitude North and water cycle (E–P) changes? 2013). Also, by necessity, models are Pacific and Southern Oceans clearly evi- Since the late 1950s, numerical tools simplified versions of reality and con- dent. Similarly, the E-dominated sub- called General Circulation Models sequently do not include all the compo- tropical gyres are well defined, and so (GCMs) have been used in attempts nents that may strongly influence regional too are their maxima at the approximate to simulate the global climate system. salinity features such as terrestrial ice latitude for each basin independently. Thanks to exponential growth in com- sheets and glaciers. However, models The salty Atlantic and fresh Pacific con- puting power, these models have become have enabled new insights into the rela- trast is also evident. For the E–P field, more complex over time and their realism tionship between ocean salinity and the the model captures regional features (when compared to observed data) has water cycle, which has strongly enhanced such as the P-dominated ITCZ, which consistently improved. The more modern the understanding attainable from the is a principal driver of tropical atmo- model versions, known as Atmosphere- limited observational data alone. spheric circulation. The model also cap- Ocean General Circulation Models So, how do models compare with tures the E-dominated subtropical cells (AOGCMs), include fully coupled our observed view of Earth? Figure 4 and the regions of strong E–P gradients. 0 35 34 35 35.5 34 34 35 35 35 5 35.5 34.5 35.5 35.5 34 334 4. 34.5 34.5 35.5 34.5 35 34.5 34 35 35.5 100 34.5 34.5 35.5 34.5 35 35 35 200 35 35 Pressure/Depth (dbar/m) 35 35 34.5 35 300 34.5 35 35 34.5 35 400 35 34.5 35 34.5 34.5 34.5 34.5 35 34.5 34.5 35 34.5 34.5 4.5 500 34 34.5 35.5 35 34.5 35 750 1000 34.5 34.5 1250 34.5 1500 1750 2000 A DW10 B05 B EN4 C 70°S 50°S 30°S 10°S 10°N 30°N 50°N 70°N 70°S 50°S 30°S 10°S 10°N 30°N 50°N 70°N 70°S 50°S 30°S 10°S 10°N 30°N 50°N 70°N Latitude Latitude Latitude −0.20 −0.15 −0.10 −0.05 0.00 0.05 0.10 0.15 0.20 FIGURE 3. Three long-term estimates of global zonal mean subsurface salinity changes after (A) Durack and Wijffels (2010; analysis period 1950–2008), (B) Boyer et al. (2005; analysis period 1955–1998), and (C) Good et al. (2014; analysis period 1950–2012), all scaled to represent equivalent magnitude changes over a 50-year period (PSS-78, 50 yr–1). Black contours show the associated climatological mean subsurface salinity for the analysis period from each corresponding climatological assessment. Broad-scale similarities also exist in the subsurface salinity changes, which suggests decreasing salinity in ocean waters fresher than the global average, and increasing salinity in waters saltier than the global average. However, regional differences, particularly in the high-latitude regions are apparent and are due to limited data sources, temporal periods of analysis, and analysis methodologies. Oceanography | March 2015 25 Collectively, the model realism for the studies may be useful proxies for provid- changes, a linear regression is undertaken climatological mean at broad scales is ing insight into poorly observed long- on basin zonal mean quantities, which impressive and warrants further consid- term historical changes. express the local anomaly from the global eration of simulated forced changes. SSS mean salinity (x-axis; fresh regions Using models and idealized experi- CHANGES TO OCEAN SURFACE are negative, salty positive) and the corre- ments, how do the SSS and E–P fields SALINITY—A MARKER OF THE sponding temporal trend (y-axis; freshen- interact and change? Williams et al. (2006) CHANGING GLOBAL WATER CYCLE ing regions are negative, enhanced salin- have shown the modeled dependence of To quantify salinity changes over the ity positive). The resulting slope of this the three-dimensional ocean salinity observed record, Durack et al. (2012) relationship was defined as the PA and structure on E–P forcing, which decays developed a technique to quantify and the corresponding correlation coeffi- (gradients dissipate) over a remarkably compare the strength of broad-scale salin- cient (R) as the pattern correlation (PC). short time scale of 10–20 years if the E–P ity pattern amplification (PA) between Using this technique, the salinity change forcing is artificially removed. Further observed estimates and models. They results from both observations and mod- studies by Williams et al. (2007, 2010) developed ocean basin (Atlantic, Pacific, els can be quantitatively compared to one show consistent amplification in sur- and Indian) zonal averages for both the another in a change per degree context— face and subsurface salinity gradients 50-year (nominally 1950–2000) climato- the natural thermodynamic framework if the E–P field is doubled from present logical mean SSS and its corresponding of water cycle amplification (Figure 5A). observed values. Using idealized ocean- change (see Figure 2, panels C and A). Using a per-degree framework allows a only model simulations, Durack et al. The key advantage of this methodology much larger suite of model simulations to (2012, 2013) also investigate the role of is averaging across regional unforced be investigated, augmenting those from E–P changes as drivers of surface and climate variability, which is apparent the CMIP5 "historical" experiment that subsurface salinity changes. They find in both models and observations, and correspond with the observed period, that the three-dimensional salinity gra- also averaging across the patchy spa- with projected simulations for the future. dients are enhanced if E–P surface fluxes tial patterns in observations that are a Figure 5 presents the results of this anal- are amplified at the surface, yielding com- result of the sparse observational cov- ysis, which show a strong linear depen- parable broad-scale simulated changes to erage through time. Additionally, the dence of salinity PA with temperature the observed estimates discussed ear- 50-year trends (the upper limit of obser- (Figure 5A). Symbol colors in the figure lier. Collectively, these model stud- vational coverage) provide change esti- indicate the strength of the correlation ies support the E–P to SSS relationship, mates that are believed to be mostly unaf- (PC) between the climatological mean which underpins the “ocean rain gauge” fected by unforced modes of decadal and salinity and its change; red (a feature of concept and suggests that modeling multidecadal variability. To quantify SSS the strongly forced future simulations 70°N 37.0 Argo ACCESS1−0 50°N A 2004−2013 B 2004−2013 36.5 30°N 36.0 35.5 10°N 35.0 10°S 34.5 30°S 34.0 50°S 33.5 70°S 33.0 70°N 2.5 OAFlux/GPCP EP ACCESS1−0 50°N C 1987−2013 D 1987−2013 2.0 1.5 30°N 1.0 10°N 0.5 0.0 10°S −0.5 30°S −1.0 −1.5 50°S −2.0 70°S −2.5 0° 60°E 120°E 180° 120°W 60°W 0° 0° 60°E 120°E 180° 120°W 60°W 0° FIGURE 4. Climatological mean 2004–2013 near-surface salinity (SSS; PSS-78; A, B) and 1987–2013 evaporation minus precipitation (E–P; m yr–1; C, D) in modern observations (A, C) and a representative model (ACCESS-1.0; B, D) from the CMIP5 suite. 26 Oceanography | Vol.28, No.1 RCP85, RCP60, RCP45) is suggestive of observational coverage (Figure 3). Are salinity concentrations. These steric (den- a coherent PA, where the change field models also capturing these changes? sity) components are termed the thermo- reports fresh getting fresher and salty get- Data sparsity is an even larger steric (heat) and halosteric (salt) proper- ting saltier. The strongest results are found issue in the deeper ocean (>300 m). ties of sea­water. When integrated over the in the CMIP5 future simulations, with Consequently, estimates of subsurface full ocean depth, steric changes impact these relationships shown for the 2050– salinity changes are poorly constrained. local and global sea levels. 2100 period compared to 1950–2000 for Such observed changes to the depth- While there have been numerous observations and historical simulations, integrated ocean can provide a more rig- investigations into observed changes respectively. These model results suggest orous test for models, as surface and to globally integrated thermosteric sea that an 8% PA for E–P change occurs mixed-layer variability is not apparent level (e.g., Church et al., 2013, and ref- per degree of global surface temperature in the deeper ocean. As a result, deeper erences therein), there have been rel- change, a result that agrees with the simple changes are likely more representative of atively few that focused on halosteric thermodynamic relationship described persistent long-term changes, reflecting changes (e.g., Antonov et al., 2002; by Clausius-Clapeyron (6–7% increase anomalies subducted into the ocean inte- Levitus et al., 2005; Ishii et al., 2006; in atmospheric water vapor per degree rior from their surface source regions on Durack et al., 2014b), particularly at the temperature change), and is consistent decadal times scales. If coherent and per- larger basin scales. Comparing halosteric with results obtained from the previous sistent subsurface salinity changes can be changes in both observed and simu- CMIP3 model suite (Durack et al., 2012). substantiated by additional salinity mea- lated estimates is a more difficult chal- As coupled models simulate both surements, they will provide further evi- lenge for climate models because the ocean and atmospheric changes together, dence supporting the hypothesis of an time scales for such anomalies to propa- it is possible to investigate the coinci- enhanced water cycle. gate into the deeper ocean are much lon- dent changes to E–P alongside the SSS Seawater density is a fundamen- ger than those of surface or mixed-layer changes that are strongly related to sur- tal physical property that decreases changes. Additionally, the processes that face temperature change (Figure 5A). (increases) when heat is added drive modeled halosteric changes are Figure 5B shows the relationship between (removed). Similarly, salinity changes less well understood than their thermo- the E–P PA and the SSS PA. Like the also drive density changes, with a den- steric counterparts, and inherent vari- SSS-temperature relationship, there also sity decrease (expansion) occurring in ability in the model simulations leads to appears to be a relationship between these response to a decreasing salinity concen- more divergent results and a correspond- variables, with increased SSS PA associ- tration, and a density increase (contrac- ing lower signal-to-noise ratio. ated with increased E–P PA. However, tion) occurring in response to increasing The results presented in Figure 6 unlike the strong linear relationship between SSS PA and temperature change, there is a larger spread, which suggests 14 that variability and dynamics play a larger 24 22 A CMIP5 12 B 200% 100% 50% role in E–P changes. This is indicative of 24 16 12 Water Flux PA % 50 yr –1 20 Salinity PA % 50 yr –1 OCEAN OBS 18 8 10 the enhanced unforced variability in E–P 16 (of which P is the most variable) when 14 6 8 compared to SSS change. As noted in the 12 25% 10 6 introduction, the SSS field acts to integrate 8 Hist RCP85 4 the higher frequency and more sporadic 6 4 RCP60 10% E–P fluxes (in concert with ocean advec- 4 2 2 RCP45 2 tion and mixing) and so provides an effec- 0 n = 370 0 n = 427 tive measure of persistent and ongoing 0.0 0.5 1.0 1.5 2.0 2.5 0 5 10 15 20 25 30 Global Surface Warming °C 50 yr –1 Salinity PA % 50 yr –1 forced change with a better signal-to- noise than E–P (Figure 5) or P changes 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 alone (Durack et al., 2012, their Figure 2). FIGURE 5. (A) 50-year near-surface salinity (SSS) pattern amplification (PA) plotted against coin- cident surface warming. (B) 50-year E–P PA plotted against coincident SSS. Small circles present DEPTH-INTEGRATED CHANGES TO results from the CMIP5 historical analysis experiment (for the period 1950–2000), and squares, THE OCEAN’S SALINITY FIELD large circles, and diamonds present results from the RCP85, RCP60, and RCP45 (analysis period 2050–2099) CMIP5 experiments, respectively. These simulations represent the most to least Along with changes to near-surface salin- strongly forced future projections undertaken by the CMIP5 model suite. The gray shaded area ity, the ocean’s subsurface salinity field bounds the pattern correlation (PC)-weighted linear best fit to the model ensemble for a line inter- has been changing over the period of secting 0 in black. Oceanography | March 2015 27 capture the depth-integrated halosteric around the Southern Hemisphere sub- historical simulations), whereas the cor- anomaly from the surface to 2,000 dbar tropics, regions that are ventilated in the responding halosteric (and thermo- for a 50-year period, notionally high latitudes and show a near-surface steric, not shown) basin-scale agree- 1950–2000. These results indicate regions freshening trend (Figure 2). The larger ment with observed estimates is absent where the depth-integrated water column Southern Hemisphere halosteric magni- in simulations that exclude CO2 forcing has contracted (blue) because of a salin- tudes, compared to those in the Northern (CMIP5 historical simulations with nat- ity increase, or expanded (orange) due Hemisphere, are not a feature of other ural forcing only; Durack et al., 2014b). to a freshening. Although there are some halosteric estimates of change (e.g., Ishii Although there appears to be a com- regional differences, on basin scales there and Kimoto, 2009; Levitus et al., 2005) pelling link between surface E–P forced is a strong correspondence between the and may be symbolic of long-term change salinity changes and subsurface salinity observed and modeled results (Figure 6). underestimates for the poorly sampled patterns in observations and in idealized These results feature a coherent Southern Hemisphere (e.g., Gregory model simulations (Durack et al., 2012), halosteric contraction in the Atlantic, et al., 2004; Gouretski and Koltermann, the effect of coincident warming (Durack with the largest magnitudes in the North 2007; Gille, 2008; Rhein et al., 2013; and Wijffels, 2010) and the influence of Atlantic, and a tendency for halosteric Durack et al., 2014a). It is worth not- ocean dynamical changes (Bouttes et al., expansion (depth-integrated freshen- ing that the basin-scale contrasts (Pacific 2012, 2013) cannot yet be ruled out as the ing) in the Pacific. Focusing on obser- halosteric expansion, Atlantic halosteric primary drivers of resolved changes. vation in the Pacific (Figure 6A), mod- contraction) are only reproduced in els show the largest magnitudes of change models that include CO2 forcing (CMIP5 CONCLUSION The outlook is bright for a continuing improvement in our understanding of the oceanic water cycle due to the develop- 70°N ment of a series of new salinity observing 50°N DW10 technologies and platforms. Along with 30°N A the existing Argo float array, two new sat- ellite platforms have begun to provide 10°N near-global-ocean surface salinity obser- 10°S vations. The Soil Moisture and Ocean 30°S Salinity (SMOS; Berger et al., 2002) plat- form was launched in November 2009, 50°S and the dedicated surface salinity satel- 70°S lite Aquarius (Lagerloef et al., 2008) was 70°N launched in June 2011. Together, they are 50°N CMIP5 Hist. providing reprocessed data sets of three- 30°N B daily to seasonal coverage and horizontal resolution of better than 150 km for the 10°N surface salinity field. Such new data are increasing our understanding of surface 10°S salinity variability and are resolving salin- 30°S ity at temporal and spatial resolutions never before seen. Preliminary stud- 50°S ies are providing new insights into high- 70°S 0° 60°E 120°E 180° 120°W 60°W 0° frequency variability in the tropical Pacific (Lee et al., 2012), salinity variability due −4 −3 −2 −1 0 1 2 3 4 to river outflows from the Amazon and FIGURE 6. Long-term trends in the 0–2,000 dbar halosteric anomaly presented as 50-year trends. Orinoco Rivers (Grodsky et al., 2012), Units are mm yr–1. The observed trend (A) is from Durack and Wijffels (2010; analysis period and more refined and sharper seasonal 1950–2008), and the modeled trend (B) is from the CMIP5 historical simulations multimodel mean gradients around the ITCZ in the Pacific (MMM; analysis period 1950–2004). Stippling is used to mark regions where the observational and Atlantic ocean basins (Lagerloef result does not agree in sign with one other observed halosteric trend estimate (Ishii and Kimoto, 2009; analysis period 1950–2008) (A) and where less than 66% of the contributing models do not et al., 2012). These satellites are mea- agree in sign with the averaged (multimodel mean, or MMM) map obtained from the CMIP5 his- suring salinity responses and their vari- torical ensemble (B). ability in regions of currently ungauged 28 Oceanography | Vol.28, No.1  FIGURE 7. Maps of 50-year salinity trends for the near-surface ocean. (A) The 1950–2000 observational change and (B) the correspond- ing 1950–2000 climatological mean of Durack and Wijffels (2010; analysis period 1950–2008). (C) Modeled changes for the 1950–2000 period from the CMIP5 historical experiment MMM (analysis period 1950–1999) and (D) 2050–2099 future projected changes for the most strongly forced RCP85 experiment MMM (analysis period 2050–2099). Black contours bound the climatological mean salinity associated with each map, and white contours bound the salinity trend in increments of 0.25 PSS-78. river and groundwater outflows and in processes that govern regions of high and ACKNOWLEDGEMENTS. This paper is a con- tribution to the US Department of Energy, Office of areas of high seasonal precipitation. They low surface salinity. Science, Climate and Environmental Sciences Division, are also providing observational cover- What’s in store for future research Regional and Global Climate Modeling Program under contract DE-AC52-07NA27344. The author thanks age in ocean regions that the Argo pro- into the oceanic water cycle and ongoing J. Durack of the University of California, San Francisco, gram has been unable to access, such as changes to ocean salinity? Considering and the three guest editors (R. Schmitt, F. Bryan, and E. Lindstrom) for helpful comments with early drafts the Indonesian seas and heavily pirated that models appear to capture most of of this manuscript. The sources of observed data regions such as the tropical Indian Ocean. the broad-scale features of both clima- used in this study are acknowledged: T. Boyer (B05), S. Hosoda (H09), S. Good (EN4), and the International Though an unintended capability of tological mean E–P and SSS (Figure 4), Argo Program and the national programs that contrib- Aquarius but the primary mission objec- further work is needed to quantita- ute to it. The efforts of the World Climate Research Programme’s Working Group on Coupled Modelling, tive of SMOS, both satellites are also pro- tively examine modeled variability esti- which is responsible for CMIP, is acknowledged, viding new insights into soil moisture for mates and compare the results to the new and considerable thanks goes to the climate mod- eling groups for producing and making available the top 5 cm of the global land surface—a observations available from Argo floats, their model output. For CMIP, the US Department of bonus that further improves our global the SMOS and Aquarius satellite mis- Energy’s Program for Climate Model Diagnosis and Intercomparison provides coordinating support and water cycle understanding. sions, and process-based studies such as led development of software infrastructure in part- These new satellite data are comple- SPURS. If modeled variability estimates nership with the Global Organization for Earth System Science Portals. The DW10 data presented in this mented by the Salinity Processes in the stand up well against these new observa- study can be downloaded from the CSIRO Ocean Upper-ocean Regional Study (SPURS) tional insights, then more confidence can Change website at www.cmar.csiro.au/oceanchange. LLNL Release # LLNL-JRNL-665262. field experiment in the North Atlantic be placed in the changes models suggest (SPURS-1), which set out to assess the will occur to the global water cycle as a REFERENCES salinity budget of this ocean region’s result of ongoing CO2 forcing. Adler, R.F., G.J. Huffman, A. Chang, R. Ferraro, P. Xie, J. Janowiak, B. Rudolf, U. Schneider, S. Curtis, salinity maximum. This special issue of CMIP5 models project significant D. Bolvin, and others. 2003. The Version-2 Oceanography highlights many new and changes will continue to occur to the Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–pres- exciting results from the SPURS-1 exper- ocean’s salinity field in response to cli- ent). Journal of Hydrometeorology 4:1,147–1,167, iment, and further insights are expected mate change, with future changes in http://dx.doi.org/10.1175/1525-7541(2003)004​ 2.0.CO;2. from SPURS-2 (Farrar et al., 2014), broad agreement with resolved observed Allen, M.R., and W.J. Ingram. 2002. Constraints on which will focus on a high rainfall and changes (Figure 7). As the ocean consti- future changes in climate and the hydrologic cycle. Nature 419:224–232, http://dx.doi.org/10.1038/ salinity minima region in the tropical tutes Earth’s largest possible rain gauge, nature01092. Pacific. By examining both the Atlantic it is time that we paid more attention to and Pacific, the two SPURS experiments what it is trying to tell us. promise to greatly illuminate the oceanic Oceanography | March 2015 29 Antonov, J.I., S. Levitus, and T.P. Boyer. 2002. Curry, R., B. Dickson, and I. Yashayaev. 2003. uncertainty estimates. Journal of Geophysical Steric sea level variations during 1957–1994: A change in the freshwater balance of the Research 118:6,704–6,716, http://dx.doi.org/​ Importance of salinity. Journal of Geophysical Atlantic Ocean over the past four decades. 10.1002/2013JC009067. Research 107(C12), 8013, http://dx.doi.org/​ Nature 426:826–829, http://dx.doi.org/10.1038/ Gordon, A.L., and C.F. Giulivi. 2008. Sea surface 10.1029/2001JC000964. nature02206. salinity trends over fifty years within the subtrop- Baumgartner, A., and E. Reichel. 1975. The World Delcroix, T., S. Cravatte, and M.J. McPhaden. 2007. ical North Atlantic. Oceanography 21(1):20–29, Water Balance: Mean Annual Global, Continental Decadal variations and trends in tropical Pacific sea http://dx.doi.org/10.5670/oceanog.2008.64. and Maritime Precipitation, Evaporation and surface salinity since 1970. Journal of Geophysical Gould, J., D. Roemmich, S. Wijffels, H. Freeland, Runoff. Elsevier Science Ltd. Amsterdam, 179 pp. Research 112, C03012, http://dx.doi.org/​10.1029/​ N. Ignaszewsky, X. Jianping, S. Pouliquen, Belkin, I.M., S. Levitus, J. Antonov, and 2006JC003801. Y. Desaubies, U. Send, K. Radhakrishnan, and oth- S.-A. Malmberg. 1998. “Great Salinity Dickson, R.R., J. Meincke, S.-A. Malmberg, and ers. 2004. Argo profiling floats bring new era of Anomalies” in the North Atlantic. Progress in A.J. Lee. 1988. The “great salinity anomaly” in in situ ocean observations. Eos, Transactions Oceanography 41:1–68, http://dx.doi.org/10.1016/ the Northern North Atlantic 1968–1982. Progress American Geophysical Union 85(19):185, S0079-6611(98)00015-9. in Oceanography 20:103–151, http://dx.doi.org/​ http://dx.doi.org/10.1029/2004EO190002. Bellenger, H., E. Guilyardi, J. Leloup, M. Lengaigne, 10.1016/0079-6611(88)90049-3. Gould, J., B. Sloyan, and M. Visbeck. 2013. In and J. Vialard. 2013. ENSO representation in cli- Durack, P.J., P.J. Gleckler, F.W. Landerer, and situ ocean observations: A brief history, pres- mate models: From CMIP3 to CMIP5. Climate K.E. Taylor. 2014a. Quantifying underestimates of ent status, and future directions. Pp. 59–81 Dynamics 42:1,999–2,018, http://dx.doi.org/10.1007/ long-term upper-ocean warming. Nature Climate in Ocean Circulation and Climate: A 21st s00382-013-1783-z. Change 4:999–1,005, http://dx.doi.org/10.1038/ Century Perspective. G. Siedler, S.M. Griffies, Berger, M., A. Camps, J. Font, Y. Berr, J. Miller, nclimate2389. J. Gould, and J.A. Church, eds, International J. Johannessen, J. Boutin, M.R. Drinkwater, Durack, P.J., and S.E. Wijffels. 2010. Fifty-year Geophysics, vol. 103, Academic Press, N. Skou, N. Floury, and others. 2002. Measuring trends in global ocean salinities and their rela- Elsevier, Oxford, UK, http://dx.doi.org/​10.1016/ ocean salinity with ESA’s SMOS Mission. ESA tionship to broad-scale warming. Journal of B978-0-12-391851-2.00003-9. Bulletin 111:113–121, http://esamultimedia.esa.int/ Climate 23:4,342–4,362, http://dx.doi.org/​ Gouretski, V., and K.P. Koltermann. 2007. How docs/Cryosat/ESABulletin111-SMOSOSactivities.pdf. 10.1175/2010JCLI3377.1. much is the ocean really warming? Geophysical Bouttes, N., J.M. Gregory, T. Kuhlbrodt, and R.S. Smith. Durack, P.J., S.E. Wijffels, and T.P. Boyer. 2013. Research Letters 34, L01610, http://dx.doi.org/​ 2013. The drivers of projected North Atlantic sea Long-term salinity changes and implica- 10.1029/2006GL027834 level change. Climate Dynamics 43:1,531–1,544, tions for the global water cycle. Pp. 727–757 Gregory, J.M., H.T. Banks, P.A. Stott, J.A. Lowe, http://dx.doi.org/10.1007/s00382-013-1973-8. in Ocean Circulation and Climate: A 21st and M.D. Palmer. 2004. Simulated and Bouttes, N., J.M. Gregory, T. Kuhlbrodt, and T. Suzuki. Century Perspective. G. Siedler, S.M. Griffies, observed decadal variability in ocean heat con- 2012. The effect of windstress change on J. Gould, and J.A. Church, eds, International tent. Geophysical Research Letters 31, L15312, future sea level change in the Southern Ocean. Geophysics, vol. 103, Academic Press, http://dx.doi.org/10.1029/2004GL020258. Geophysical Research Letters 39, L23602, Elsevier, Oxford, UK, http://dx.doi.org/​10.1016/ Greve, P., B. Orlowsky, B. Mueller, J. Sheffield, http://dx.doi.org/10.1029/2012GL054207. B978-0-12-391851-2.00028-3. M. Recihstein, and S.I. Seneviratne. 2014. Boyer, T.P., J.I. Antonov, O.K. Baranova, C. Coleman, Durack, P.J., S.E. Wijffels, and P.J. Gleckler. 2014b. Global assessment of trends in wetting and dry- H.E. Garcia, A. Grodsky, D.R. Johnson, Long-term sea-level change revisited: The role of ing over land. Nature Geoscience 7:716–721, R.A. Locarnini, A.V. Mishonov, T.D. O’Brien, and oth- salinity. Environmental Research Letters 9, 114017, http://dx.doi.org/​10.1038/ngeo2247. ers. 2013. World Ocean Database 2013. S. Levitus, http://dx.doi.org/10.1088/1748-9326/9/11/114017. Grodsky, S.A., N. Reul, G. Lagerloef, G. Reverdin, ed., A. Mishonov, technical ed., NOAA Atlas Durack, P.J., S.E. Wijffels, and R.J. Matear. 2012. J.A. Carton, B. Chapron, Y. Quilfen, NESDIS 72, 209 pp., http://data.nodc.noaa.gov/ Ocean salinities reveal strong global water V.N. Kudryavtsev, and H.-Y. Kao. 2012. Haline woa/WOD13/DOC/wod13_intro.pdf. cycle intensification during 1950 to 2000. hurricane wake in the Amazon/Orinco plume: Boyer, T.P., S. Levitus, J.I. Antonov, R.A. Locarnini, Science 336:455–458, http://dx.doi.org/10.1126/ AQUARIUS/SACD and SMOS observations. and H.E. Garcia. 2005. Linear trends in salin- science.1212222. Geophysical Research Letters 39, L20603, ity for the world ocean, 1955–1998. Geophysical Farrar, J.T., R. Schmitt, L. Rainville, W. Asher, http://dx.doi.org/10.1029/2012GL053335. Research Letters 32, L01604, http://dx.doi.org/​ B. Hodges, A. Jessup, F. Bingham, A. Shcherbina, Hegerl, G.C., E. Black, R.P. Allan, W.J. Ingram, 10.1029/2004GL021791. W.S. Kessler, L. Centurioni, and others. 2014. D. Polson, K.E. Trenberth, R.S. Chadwick, P.A. Arkin, Bullister, J.L., M. Rhein, and C. Mauritzen. SPURS-2: Diagnosing the Physics of a Rainfall- B.B. Sarojini, A. Becker, and others. 2014. 2013. Deepwater formation. Pp. 227–253 Dominated Salinity Minimum. Report of a Challenges in quantifying changes in the global in Ocean Circulation and Climate: A 21st Workshop in Pasadena, April 16–18 2014. water cycle. Bulletin of the Meteorological Society, Century Perspective. G. Siedler, S.M. Griffies, 17 pp., http://spurs.jpl.nasa.gov/pdf/SPURS-2_ http://dx.doi.org/10.1175/BAMS-D-13-00212.1. J. Gould, and J.A. Church, eds, International WhitePaper_v12rev.pdf. Held, I.M., and B.J. Soden. 2006. Robust responses of Geophysics, vol. 103, Academic Press, Flato, G., J. Marotzke, B. Abiodun, P. Braconnot, the hydrological cycle to global warming. Journal Elsevier, Oxford, UK, http://dx.doi.org/​10.1016/ S.C. Chou, W. Collins, P. Cox, F. Driouech, S. Emori, of Climate 19:5,686–5,699, http://dx.doi.org/10.1175/ B978-0-12-391851-2.00010-6. V. Eyring, and others. 2013. Evaluation of cli- JCLI3990.1. Chen, J., R. Zhang, H. Wang, Y. An, P. Peng, and mate models. Pp. 741–866 in Climate Change Helm, K.P., N.L. Bindoff, and J.A. Church. 2010. W. Zhang. 2012. Isolation of sea surface salinity 2013: The Physical Science Basis. Contribution of Changes in the global hydrological-cycle inferred maps on various timescales in the tropical Pacific Working Group I to the Fifth Assessment Report of from ocean salinity. Geophysical Research Ocean. Journal of Oceanography 68(5):687–701, the Intergovernmental Panel on Climate Change. Letters 37, L18701, http://dx.doi.org/10.1029/​ http://dx.doi.org/10.1007/s10872-012-0126-8. T.F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, 2010GL044222. Church, J.A., P.U. Clark, A. Cazenave, J.M. Gregory, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, Holland, H.D. 1972. The geologic history of S. Jevrejeva, A. Levermann, M.A. Merrifield, and P.M. Midgley, eds, Cambridge University sea water: An attempt to solve the problem. G.A. Milne, R.S. Nerem, P.D. Nunn, and oth- Press, Cambridge, UK, and New York, NY, Geochimica et Cosmochimica Acta 36(6):637–651, ers. 2013. Sea level change. Pp. 1,137–1,216 in http://dx.doi.org/​10.1017/CBO9781107415324.020. http://dx.doi.org/​10.1016/0016-7037(72)90108-1. Climate Change 2013: The Physical Science Freeland, H., K. Denman, C.S. Wong, F. Whitney, and Hosoda, S., T. Suga, N. Shikama, and K. Mizuno. Basis. Contribution of Working Group I to the Fifth R. Jacques. 1997. Evidence of change in the win- 2009. Global surface layer salinity change Assessment Report of the Intergovernmental ter mixed layer in the Northeast Pacific Ocean. detected by Argo and its implication for Panel on Climate Change. T.F. Stocker, D. Qin, Deep-Sea Research Part I 44(12):2,117–2,129, hydrological cycle intensification. Journal of G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, http://dx.doi.org/​10.1016/S0967-0637(97)00083-6. Oceanography 65:579–596, http://dx.doi.org/​ A. Nauels, Y. Xia, V. Bex, and P.M. Midgley, eds, Gille, S.T. 2008. Decadal-scale temperature trends 10.1007/s10872-009-0049-1. Cambridge University Press, Cambridge, UK, in the Southern Hemisphere ocean. Journal Huffman, G.J., R.F. Adler, P. Arkin, A. Chang, and New York, NY, http://dx.doi.org/10.1017/ of Climate 21:4,749–4,765, http://dx.doi.org/​ R. Ferraro, A. Gruber, J. Janowiak, A. McNab, CBO9781107415324.026. 10.1175/2008JCLI2131.1. B. Rudolf, and U. Schneider. 1997. The Global Cravatte, S., T. Delcroix, D. Zhang, M. McPhaden, and Gimeno, L., A. Drumon, R. Nieto, R.M. Trigo, and Precipitation Climatology Project (GPCP) com- J. LeLoup. 2009. Observed freshening and warm- A. Stohl. 2010. On the origin of continental precip- bined precipitation dataset. Bulletin of the ing of the western Pacific Warm Pool. Climate itation. Geophysical Research Letters 37, L13804, American Meteorology Society 78:5–20, Dynamics 33:565–589, http://dx.doi.org/10.1007/ http://dx.doi.org/10.1029/2010GL043712. http://dx.doi.org/​10.1175/​1520-0477(1997)078​ s00382-009-0526-7. Good, S.A., M.J. Martin, and N.A. Rayner. 2014. EN4: 2.0.CO;2. Quality controlled ocean temperature and salin- ity profiles and monthly objective analyses with 30 Oceanography | Vol.28, No.1  Huffman, G.J., D.T. Bolvin, E.J. Nelkin, D.B. Wolff, Li, G., and S.-P. Xie. 2014. Tropical biases in CMIP5 Talley, L.D. 2008. Freshwater transport estimates R.F. Adler, G. Gu, Y. Hong, K.P. Bowman, and multimodel ensemble: The excessive equatorial and the global overturning circulation: Shallow, E.F. Stocker. 2007. The TRMM Multisatellite Pacific cold tongue and double ITCZ problems. deep and throughflow components. Progress in Precipitation Analysis (TMPA): Quasi-global, multi- Journal of Climate 27:1,765–1,780, http://dx.doi.org/​ Oceanography 78:257–303, http://dx.doi.org/​ year, combined-sensor precipitation estimates at 10.1175/JCLI-D-13-00337.1. 10.1016/j.pocean.2008.05.001. fine scales. Journal of Hydrometeorology 8:38–55, Lin, J.-L. 2007. The double-ITCZ problem in IPCC Taylor, K.E., R.J. Stouffer, and G.A. Meehl. 2012. http://dx.doi.org/10.1175/JHM560.1. AR4 coupled GCMs: Ocean-atmosphere feed- An overview of CMIP5 and the experiment Ishii, M., and M. Kimoto. 2009. Reevaluation of histori- back analysis. Journal of Climate 20:4,497–4,525, design. Bulletin of the American Meteorological cal ocean heat content variations with time-varying http://dx.doi.org/10.1175/JCLI4272.1. Society 93:485–498, http://dx.doi.org/10.1175/ XBT and MBT depth bias corrections. Journal of Maidment, R.I., D. Grimes, R.P. Allan, E. Tarnavsky, BAMS-D-11-00094.1. Oceanography 65:287–299, http://dx.doi.org/​ M. Stringer, T. Hewison, R. Roebeling, and Trenberth, K.E., L. Smith, T. Qian, A. Dai, and 10.1007/s10872-009-0027-7. E. Black. 2014. The 30 year TAMSAT African J. Fasullo. 2007. Estimates of the global Ishii, M., M. Kimoto, K. Sakamoto, and S.-I. Iwasaki. Rainfall Climatology and Time series (TARCAT) water budget and its annual cycle using 2006. Steric sea level changes estimated from his- data set. Journal of Geophysical Research: observational and model data. Journal of torical ocean subsurface temperature and salin- Atmospheres 119:10,619–10,644, http://dx.doi.org/​ Hydrometeorology 8:758–769, http://dx.doi.org/​ ity analyses. Journal of Oceanography 62:155–170, 10.1002/2014JD021927. 10.1175/JHM600.1. http://dx.doi.org/10.1007/s10872-006-0041-y. Polade, S.D., A. Gershunov, D.R. Cayan, van der Ent, R.J., and H.H. Savenije. 2013. Oceanic Johnson, G.C., and J.M. Lyman. 2007. Global M.D. Dettinger, and D.W. Pierce. 2013. Natural cli- sources of continental precipitation and the oceans: Sea surface salinity. Supplement mate variability and teleconnections to precipi- correlation with sea surface temperature. to State of the Climate in 2006. A. Arguez, tation over the Pacific-North American region in Water Resources Research 49:3,993–4,004, ed. Bulletin of the American Meteorological CMIP3 and CMIP5 models. Geophysical Research http://dx.doi.org/​10.1002/wrcr.20296. Society 88:s34–s35, http://dx.doi.org/10.1175/ Letters 40:2,296–2,301, http://dx.doi.org/10.1002/ van der Ent, R.J., H.H.G. Savenije, B. Schaefli, and BAMS-88-6-StateoftheClimate. grl.50491. S.C. Steele-Dunne. 2010. Origin and fate of atmo- Josey, S.A., S. Gulev, and L. Yu. 2013. Exchanges Reverdin, G. 2010. North Atlantic subpolar gyre spheric moisture over continents. Water Resources through the ocean surface. Pp 115–140 surface variability (1895-2009). Journal of Research 46(9), W09525, http://dx.doi.org/​ in Ocean Circulation and Climate: A 21st Climate 23:4,571–4,584, http://dx.doi.org/​ 10.1029/2010WR009127. Century Perspective. G. Siedler, S.M. Griffies, 10.1175/2010JCLI3493.1. von Schuckmann, K., F. Gaillard, and P.-Y. Le Traon. J. Gould, and J.A. Church, eds, International Rhein, M., S.R. Rintoul, S. Aoki, E. Campos, 2009. Global hydrographic variability patterns Geophysics, vol. 103, Academic Press, D. Chambers, R.A. Feely, S. Gulev, G.C. Johnson, during 2003–2008. Journal of Geophysical Elsevier, Oxford, UK, http://dx.doi.org/​10.1016/ S.A. Josey, A. Kostianoy, and others. 2013. Research 114(C9), C09007, http://dx.doi.org/​ B978-0-12-391851-2.00005-2. Observations: Ocean. Pp. 255–315 in Climate 10.1029/2008JC005237. Kidd, C., and G. Huffman. 2011. Global pre- Change 2013: The Physical Science Basis. Williams, P.D., E. Guilyardi, G. Madec, S. Gualdi, and cipitation measurement. Meteorological Contribution of Working Group I to the Fifth E. Scoccimarro. 2010. The role of mean ocean Applications 18:34–353. Assessment Report of the Intergovernmental salinity in climate. Dynamics of Atmospheres and Kirtman, B.P., T. Stockdale, and R. Burgman. Panel on Climate Change. T.F. Stocker, D. Qin, Oceans 49:108–123, http://dx.doi.org/10.1016/​ 2013. The ocean’s role in modeling and pre- G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, j.dynatmoce.2009.02.001. dicting seasonal-to-interannual climate varia- A. Nauels, Y. Xia, V. Bex, and P.M. Midgley, eds, Williams, P.D.E., Guilyardi, R.T. Sutton, J.M. Gregory, tions. Pp. 625–643 in Ocean Circulation and Cambridge University Press, Cambridge, UK, and G. Madec. 2006. On the climate response Climate: A 21st Century Perspective.