Abstract

The modes and mechanisms of the annual water vapor variations over the twentieth century are investigated based on a newly developed twentieth-century atmospheric reanalysis product. It is found that the leading modes of global water vapor variations over the twentieth century are controlled by global warming, the Atlantic multidecadal oscillation (AMO), and ENSO. On the global scale, the variations in water vapor synchronize with the sea surface temperature, which can be explained by the simple thermal Clausius–Clapeyron theory under conditions of constant relative humidity. However, on regional scales, the spatial patterns of water vapor variations associated with global warming, AMO, and ENSO are largely attributed to the atmospheric circulation dynamics, particularly the planetary divergent circulation change induced by the sea surface temperature changes. In the middle and high latitudes, the transient eddy fluxes and thermodynamics also play significant roles.

1. Introduction

Water vapor is the most important gaseous source of infrared opacity in the atmosphere, which plays a decisive role in the transfer of radiation through the atmosphere. Meanwhile, water vapor, acting as one of the water phases, can regulate evaporation and transpiration processes and thus connects closely with both the global hydrological cycle and energy budget. It is therefore important to understand and monitor changes in atmospheric water vapor content, not only for detecting climate change but also for validating water vapor feedback in climate models.

Previous studies have primarily focused on the global mean water vapor in response to global warming, including surface humidity (Dai 2006; Willett et al. 2007; Philipona et al. 2004; Wang and Gaffen 2001), upper tropospheric moisture (Soden et al. 2005), and total column-integrated water vapor (Santer et al. 2006; Held and Soden 2006; Cubasch et al. 2001; Dai et al. 2001). Both observations and models indicate that the amount of water vapor is expected to increase under conditions of greenhouse gas–induced warming, leading to a significant positive feedback on the surface temperature (Hansen et al. 1984) and the associated climate changes (Wentz et al. 2007; Zhang et al. 2007; Allan and Soden 2008). Climate models further indicate that the increasing trend of water vapor is largely due to anthropogenic forcing rather than natural variability (Willett et al. 2007; Santer et al. 2006). Under the assumption that relative humidity remains approximately constant, for which there is considerable empirical support (Held and Soden 2006; Wentz and Schabel 2000; Soden et al. 2005), the increase in global mean water vapor is estimated to be about 6.5%–7% °C−1 warming of the lower troposphere according to the simple thermal Clausius–Clapeyron (C-C) equation. Water vapor feedback in global warming is an important positive feedback that increases the sensitivity of surface temperature to carbon dioxide by nearly a factor of 2, and possibly by as much as a factor of 3 or more when interactions with other feedback are considered (Hall and Manabe 1999; Held and Soden 2000).

Global mean water vapor in response to global warming is predominantly constrained by the thermodynamics. However, hemispheric mean water vapor response behaves quite differently. Khon et al. (2010) have pointed out that the greenhouse gas–induced Northern Hemispheric water vapor change is not only caused by increased evaporation, but is also attributable to the interhemispheric moisture exchange associated with the Hadley circulation (e.g., Oort and Yienger 1996; Clement et al. 2004). The significant role of interhemispheric moisture transport in the hemispheric mean water vapor response to global warming shares great similarities with the seasonal variability of water vapor. As demonstrated by Chen et al. (1996), water vapor is transported from the winter to summer hemisphere across the equator by the Hadley circulation.

So far the long-term variations of global water vapor in both temporal and spatial scales still remain poorly understood because of a lack of long-term observational data. This imposes an obstacle to detect changes of the greenhouse gas–induced water vapor feedback from natural climate variability. A newly developed twentieth-century atmospheric reanalysis product [Twentieth-Century Reanalysis, version 2 (20CRv2); Compo et al. 2011] provides a unique opportunity to examine the multiscale variations and mechanisms of global water vapor. In this paper, we investigate the variability and mechanisms of global water vapor based on the 20CRv2 reanalysis dataset. It is found that the leading modes of global water vapor variations over the twentieth century are controlled by global warming, the Atlantic multidecadal oscillation (AMO; Kerr 2000; Enfield and Mestas-Nuñez 1999), and El Niño–Southern Oscillation (ENSO), with the global mean controlled by thermodynamics and the regional pattern governed by atmospheric circulation dynamics.

The paper is organized as follows. Section 2 briefly presents the datasets we used. The leading modes of global water vapor variations are described in section 3. In section 4, mechanisms governing both the time scale and spatial pattern of different modes are investigated. The residence time of water vapor in the atmosphere is discussed in section 5. Discussion and summary are given in section 6.

2. Datasets

A new atmospheric reanalysis product for the twentieth-century (20CRv2) is used in this study, and it contains the synoptic observation–based estimate of global tropospheric variability spanning from 1871 to 2010 at 6-hourly temporal and 2° spatial resolutions (Compo et al. 2011). The data are derived by using observations of synoptic surface pressure and prescribing monthly SST and sea ice distributions as boundary conditions for the atmosphere. This dataset has been used in different studies (Compo et al. 2011).

The second dataset used in this study is the Hadley Centre Sea Ice and Sea Surface Temperature (HadISST) dataset (Rayner et al. 2003), with a spatial resolution of 1° × 1° and monthly temporal resolution. The data record spans 1870–2010.

To evaluate the water vapor variability in the reanalysis dataset, we compare with the observed water vapor variations from the Special Sensor Microwave Imager (SSM/I) (Alishouse et al. 1990). We use the sixth version of the precipitable water analyses synthesized from a series of eight intercalibrated satellites (F8, F10, F11, F13, F14, F15, F16, and F17) (Wentz 1997). Remote Sensing Systems generates the SSM/I products using a unified, physically based algorithm to simultaneously retrieve ocean wind speed (at 10 m), water vapor, cloud water, and rain rate. Finally, the twentieth-century multimodel simulations for the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) are also utilized to compare with results derived from the reanalysis products.

3. Modes of global water vapor

We first examine the variations of global mean water vapor. The global ocean mean column-integrated water vapor from 1871 to 2010, calculated from the annual data of the 20CRv2, is shown in Fig. 1. To validate the reanalysis data, we compare the observed water vapor variations from the SSM/I with the 20CRv2 in the last 23 yr (1988–2010) and find that the 20CRv2 reasonably captures the observed water vapor variability, albeit with a weak amplitude because of the coarse resolution of the atmosphere model. When the model resolution is low, it will miss some water vapor content that is generated by the small physical processes in the subgrid. The dominant change of global mean water vapor is an upward trend, suggesting an increase of moisture in response to global warming since the twentieth century (Santer et al. 2006). In the period from 1871 to 2010, the global mean water vapor in the reanalysis data increases by 0.0864 ± 0.01 kg m−2 decade−1. The upward trend in the last 20 yr becomes more substantial in both the SSM/I and 20CRv2, with 0.41 kg m−2 ± 0.21 kg m−2 decade−1, consistent with the estimation in Wentz et al. (2007).

Fig. 1.

Time series of column-integrated water vapor (kg m−2 or mm) over the global ocean derived from the 20CRv2 (red area) and SSM/I (black line). Blue line depicts the water vapor fluctuations due to temperature changes under conditions of constant relative humidity. The magnitudes of water vapor are not directly comparable between the observation and reanalysis data because of the relatively coarse resolution of reanalysis data. Thus, the observational water vapor (black line) is reduced by 2 kg m−2 to compare with 20CRv2.

Fig. 1.

Time series of column-integrated water vapor (kg m−2 or mm) over the global ocean derived from the 20CRv2 (red area) and SSM/I (black line). Blue line depicts the water vapor fluctuations due to temperature changes under conditions of constant relative humidity. The magnitudes of water vapor are not directly comparable between the observation and reanalysis data because of the relatively coarse resolution of reanalysis data. Thus, the observational water vapor (black line) is reduced by 2 kg m−2 to compare with 20CRv2.

In addition to the secular trend, the global water vapor also displays significant interannual and multidecadal variations (Fig. 1). A visual examination finds that interannual variability may be affected by ENSO. For example, the global water vapor content in 1997–99 is higher probably because of the strong El Niño in 1997/98, which increased the global mean column-integrated water vapor due to the warming in the equatorial Pacific. Furthermore, the time series also displays three to four cycles of rises and falls, implying a superposition of some multidecadal fluctuations on the long-term trend (will be discussed later).

To document the global mean water vapor variability on various time scales, a singular spectrum analysis (SSA) (Broomhead and King 1986; Allen et al. 1992) is performed on the water vapor time series. Figure 2a displays the normalized eigenvalues λk together with the 95% confidence interval δλk = 2λk(2/N)1/2 (Fraedrich 1986; Vautard and Ghil 1989), where N is the length of the time series in years. The leading two modes account for 63% and 13% of the total variance, respectively, and both of their distances above the “noise floor” indicate that these modes are deterministic rather than stochastic. Figure 2b displays eigenvectors that represent a trend and a quasi-periodic oscillation. The reconstructed time series of the leading two modes are shown in Fig. 2c. It can be seen that the first mode is characterized by an increasing trend, although it is different from the linear trend. The second mode displays an oscillatory pattern with a period of about 65 yr, showing a reduction of water vapor in 1900–30 and 1970–95 and an increase prior to 1900, in 1929–68 and after 1995. The residual time series, calculated by subtracting the first two modes from the original water vapor time series, displays a distinct interannual variability (Fig. 2d).

Fig. 2.

SSA for the column-integrated water vapor anomaly derived from 20CRv2. (a) Variance contribution associated with the 95% confidence interval. (b) The first (dashed black line) and the second (solid black line) eigenvectors. (c) Reconstructed water vapor time series of mode 1 (black line) and mode 2 (gray line). (d) Residuals after modes 1 and 2 removed from the original water vapor time series. We chose a nesting spatial dimension of M = 30. Similar results are obtained for 30 ≤ M ≤ 50.

Fig. 2.

SSA for the column-integrated water vapor anomaly derived from 20CRv2. (a) Variance contribution associated with the 95% confidence interval. (b) The first (dashed black line) and the second (solid black line) eigenvectors. (c) Reconstructed water vapor time series of mode 1 (black line) and mode 2 (gray line). (d) Residuals after modes 1 and 2 removed from the original water vapor time series. We chose a nesting spatial dimension of M = 30. Similar results are obtained for 30 ≤ M ≤ 50.

A further inspection finds that the second mode is in phase with the AMO with a high correlation coefficient of 0.89, a mode characterized by an interhemispheric SST seesaw in the Atlantic basin with global expressions. A high correlation coefficient up to 0.68 between the residual time series and ENSO index, defined as the SST anomalies over the Niño-3 region (5°S–5°N, 150°–270°E), indicates that the residual time series is associated with ENSO variability.

The variations of global mean column-integrated water vapor may be simply controlled by the C-C thermodynamic relationship. To verify that, we use the gridded observational SST dataset HadISST (Rayner et al. 2003) to reconstruct the water vapor based on the C-C equation with an assumption of constant relative humidity. It is found that most of the water vapor variability can be explained by the SST changes (Fig. 1). The SSA decomposition of the reconstructed water vapor time series also displays three distinct modes: a secular trend, a multidecadal mode, and an ENSO mode. That implies, to the first order, that the global mean column-integrated water vapor variations are controlled by the SST changes.

To further examine the potential relationship between the water vapor and SST fluctuations, we regress the annual SST anomalies against the decomposed water vapor time series. As shown in Fig. 3a, an increasing moisture trend is associated with warming over the global ocean, with the patterns resembling the linear trend of SST over the same period (not shown). In particular, strong warming occurs in the western boundary currents and their extensions (Wu et al. 2012). The multidecadal variation of global water vapor is largely associated with the AMO (Fig. 3b), featuring a bipolar seesaw pattern over the Atlantic Ocean, with a warming in the north and a cooling in the south. The regression also shows warming over the North and South Pacific and cooling over the Pacific cold tongue. A close inspection finds that the maximum SST anomaly is located over the North Atlantic deep convection region, supporting the hypothesis that the driving mechanism of the AMO involves fluctuations of the Atlantic meridional overturning circulation (Delworth and Mann 2000; Knight et al. 2005; Dijkstra et al. 2006). As expected, the interannual variation of global water vapor is primarily associated with ENSO (Fig. 3c), with strong anomalies in the eastern equatorial Pacific and global teleconnections (e.g., Enfield and Mayer 1997; Klein et al. 1999; Giannini et al. 2000).

Fig. 3.

Regression of SST on the (a) trend, (b) multidecadal oscillation, and (c) residual time series of global ocean mean water vapor [°C (kg m−2)−1] from Fig. 2.

Fig. 3.

Regression of SST on the (a) trend, (b) multidecadal oscillation, and (c) residual time series of global ocean mean water vapor [°C (kg m−2)−1] from Fig. 2.

The spatial patterns of the water vapor variations are obtained by regressing the column-integrated water vapor against the decomposed time series of global mean water vapor (Fig. 2). It can be found that the water vapor associated with global warming tends to increase nearly over the global oceans (Fig. 4a). The largest increase of moisture occurs over the Indo-Pacific warm pool region. The spatial distribution of the water vapor variability associated with the AMO shares great similarities with that associated with global warming (Fig. 4b), but some discrepancies still exist. Over the Atlantic Ocean, moisture increase in response to global warming is much larger in the Southern Hemisphere than in the Northern Hemisphere. The opposite is true for the AMO-modulated water vapor distribution.

Fig. 4.

Regression of column-integrated water vapor on the (a) trend, (b) multidecadal oscillation, and (c) residual time series of global ocean mean water vapor (Fig. 2) on each grid.

Fig. 4.

Regression of column-integrated water vapor on the (a) trend, (b) multidecadal oscillation, and (c) residual time series of global ocean mean water vapor (Fig. 2) on each grid.

It should be noted that the spatial patterns of the water vapor variations in response to global warming and the AMO do not follow the SST fluctuations (Fig. 3 versus Fig. 4). That implies that simple thermodynamics (C-C theory) cannot explain the regional distribution of water vapor. However, at interannual time scales, the spatial pattern of the water vapor response follows ENSO SST variability (Fig. 4c). In the following section, we will examine the column-integrated water vapor budget to assess roles of atmospheric circulation in determining the spatial distribution of water vapor variability.

4. Mechanisms controlling the spatial patterns

a. Column-integrated water vapor budget

The water vapor budget analysis below follows Seager et al. (2010), but for different time scales. The column-integrated water vapor budget can be written as (Trenberth and Guillemot 1995)

 
formula

where is the column-integrated water vapor of the atmosphere and q, U, ps, E, and P are specific humidity, horizontal velocity, surface pressure, evaporation, and precipitation, respectively. The integral on the right-hand side describes the divergence of water vapor horizontal flux, which should be zero if integrated over the globe. Apparently, any variations in the global mean water vapor result from an imbalance between global mean evaporation and precipitation. The divergence of the water vapor flux can be divided into the mean and transient components, and Eq. (1) can be expressed in the form of

 
formula

in which overbars indicate monthly mean and primes indicate departures from the monthly mean. The last term (which has not been broken into monthly-mean and transient components) involves surface quantities. We found that in the 20CRv2 this term was reasonably approximated when evaluated with monthly mean, as suggested by previous studies (Seager et al. 2010; Wang et al. 2013; Zhang and Wang 2012). The water vapor flux divergence can be further broken up into contributions that depend mostly on the mass divergence in the lower atmosphere and horizontal advection by the divergent wind. Thus, Eq. (2) can be further decomposed as

 
formula

denoting

 
formula

where indicates each term of Eq. (3) at every month and indicates the long-term annual mean value, so Eq. (3) can be approximated as

 
formula

According to Seager et al. (2010), terms in Eq. (5) involving changes in q but no changes in U are referred to as thermodynamic contributors, and terms involving changes in U but no changes in q are referred to as dynamic contributors. We leave the transient eddy moisture convergence as is since there is no straightforward way to divide the covariance into contributions from changes in eddy humidity and eddy flow. Note that the nonlinear term , which is the product of changes in both time-mean specific humidity and flow, is neglected because of its small magnitude. Briefly, thermodynamic and dynamic contributions can be expressed in the form of

 
formula

and

 
formula

respectively.

Meanwhile, we can further decompose the thermodynamic and dynamic contributions into terms due to the advection of moisture (subscript A) and the convergence or divergence of moisture (subscript D) as

 
formula
 
formula
 
formula
 
formula
 
formula
 
formula

All terms in these equations are obtained with the original daily or monthly data and then are averaged to annual mean time series to focus on various time scale variations. Given that the mathematical formulas of the water vapor budget are very complicated, we add a table to show symbols for the primary terms of the equations (Table 1).

Table 1.

Moisture budget terms analyzed in this study and their source equations, physical processes, time scales, and symbols.

Moisture budget terms analyzed in this study and their source equations, physical processes, time scales, and symbols.
Moisture budget terms analyzed in this study and their source equations, physical processes, time scales, and symbols.

b. Contributions to water vapor secular trend

To diagnose contributions of different physical processes to the water vapor secular trend, we regress main terms of the water vapor budget against the first mode of water vapor SSA (Fig. 2c). Figure 5 shows changes associated with the secular trend in moisture flux divergence (DivQ) (Fig. 5a), evaporation minus precipitation (EP) (Fig. 5b), horizontal moisture advection by the wind (DivQA) (Fig. 5c), mass divergence of moisture (DivQD) (Fig. 5d), transient eddies (Fig. 5e), and surface quantities (Fig. 5f). It can be seen that the water vapor secular trend can be largely accounted for by the water vapor flux divergence (DivQ), including moistening in much of the warm pool regions and the southern flank of the equatorial Pacific and Atlantic and drying in the northern flank of the equatorial Pacific and Atlantic (Fig. 5a), while EP plays a damping role (Fig. 5b). A further decomposition finds that the secular trend of DivQ is primarily due to changes in its divergent part DivQD (Fig. 5d), particularly in the tropics. In the extratropics, its advective part DivQA (Fig. 5c) and transient eddies (Fig. 5e) also determine the moisture flux divergence change.

Fig. 5.

Regression of primary terms in water vapor budget against the normalized water vapor secular trend (mm day−1). (a) Moisture flux divergence DivQ. (b) Evaporation minus precipitation EP. DivQ is further decomposed into its (c) advective part DivQA, (d) divergent part DivQD, (e) transient eddies, and (f) surface quantities, all of which come from Eq. (3). See Table 1 for equation terms.

Fig. 5.

Regression of primary terms in water vapor budget against the normalized water vapor secular trend (mm day−1). (a) Moisture flux divergence DivQ. (b) Evaporation minus precipitation EP. DivQ is further decomposed into its (c) advective part DivQA, (d) divergent part DivQD, (e) transient eddies, and (f) surface quantities, all of which come from Eq. (3). See Table 1 for equation terms.

It is seen that DivQD (Fig. 5d) exhibits a complicated structure. The Indo-Pacific warm pool region is characterized by an increased moisture convergence. However, in the eastern equatorial Pacific and Atlantic Ocean, there is a tendency for a decrease of moisture in the north and an increase of moisture in the south. In comparison, the DivQA (Fig. 5c) change in response to global warming is relatively simple, with a strengthened magnitude but spatial pattern similar to climatology (see Fig. 1 in Seager et al. 2010), indicating a phenomenon of deficit getting deficit and surplus getting surplus. That is, more moisture advection occurs in the midlatitude westerly regions and less moisture advection occurs in the equatorward easterly regions. The transient eddy (Fig. 5e) change also provides a simple pattern in both hemispheres, with moistening in the higher latitudes and drying at the poleward flank of the subtropics, as a consequence of strengthening of the transient eddy moisture transport in response to global warming (Wu et al. 2011). Additionally, in both hemispheres there is an indication of a poleward shift of the pattern of transient eddy moisture flux. Quite clearly the increased transient eddy moisture flux convergence contributes significantly to the increased moisture in the midlatitude and subpolar regions.

As expected, the surface quantities (Fig. 5f) are much smaller than other terms, particularly in the interior ocean, and thus we can neglect them. Until now, although we can conclude that DivQ plays a significant role in water vapor secular trend, particularly DivQD, it is still not clear which component, q or U change, plays the dominant role.

To separate this ambiguity, we next further separate the DivQ trend into the thermodynamics contribution δTH and the contribution from the mean circulation dynamics δMCD. As shown in Fig. 6a, δTH can partly explain water vapor secular trend over the warm pool and subpolar regions. It is found that δTH moistens in much of the intertropical convergence zone (ITCZ) and subpolar regions and dries in the trade winds region. This change follows simply from an increase in specific humidity in a warmer atmosphere and has the spatial pattern of the climatological mean low-level divergence (drying) and convergence (moistening). This is further confirmed by the changes in the divergent part of thermodynamics δTHD (Fig. 6d). There is an exception over the eastern equatorial Pacific Ocean. where moisture has a tendency to decrease rather than increase over the ITCZ region. This is largely because SST response over the tropical Pacific is characterized by a La Niña–like pattern in response to global warming in the 20CRv2 (Fig. 3a). The δTH change in response to global warming is also associated with its advective part δTHA, particularly over the eastern equatorial Pacific Ocean (Fig. 6c).

Fig. 6.

Regression of primary terms in water vapor budget against the normalized water vapor secular trend (mm day−1). (a) Thermodynamics contribution δTH. (b) Mean circulation dynamics δMCD. The δTH equation is further decomposed into (c) the advection of moisture δTHA and (d) the convergence or divergence of moisture δTHD. The δMCD equation is further decomposed into (e) the advection part δMCDA and (f) the divergence part δMCDD. See Table 1 for equation terms.

Fig. 6.

Regression of primary terms in water vapor budget against the normalized water vapor secular trend (mm day−1). (a) Thermodynamics contribution δTH. (b) Mean circulation dynamics δMCD. The δTH equation is further decomposed into (c) the advection of moisture δTHA and (d) the convergence or divergence of moisture δTHD. The δMCD equation is further decomposed into (e) the advection part δMCDA and (f) the divergence part δMCDD. See Table 1 for equation terms.

The δMCD (Fig. 6b) shares great similarities with the DivQ changes (Fig. 5a), suggesting it is the dominant contribution of δMCD to the water vapor secular trend. In the tropical Pacific and Atlantic Ocean, it is clearly seen that δMCD tends to increase moisture over the warm pool regions and the southern flank of the equator, while the opposite occurs in the northern flank of the equator. A close inspection finds that δMCD change in the tropics primarily arises from its divergent part δMCDD (Fig. 6f), which is closely linked to the low-level divergent circulation change (Fig. 7a). The strengthening of divergent circulation is seen over the western part of the Pacific Ocean, with an anomalous convergence over the Indo-Pacific warm pool region and an anomalous divergence over the western subtropics (Fig. 7a), implying a strengthened Hadley cell in the western Pacific Ocean. In addition to the Hadley cell change, the Walker circulation is strengthened because of the La Niña–like SST response to global warming in the tropical Pacific, as presented in Fig. 3a. Similarly, there is an anomalous convergence circulation over the Caribbean Sea and the Gulf of Mexico, and the Atlantic ITCZ also shifts to the south (Fig. 7a). The tropical planetary divergent circulation changes in response to global warming indicate that the water vapor is trapped in the warm pool regions and south of the equator over the Pacific and Atlantic sectors, consistent with the maximum moisture increase in Fig. 4a. The divergent circulation also exhibits some changes over the middle and high latitudes. On the poleward flank of the Hadley cell in the subtropical Atlantic, there is a tendency of moisture reduction because of a poleward expansion of the Hadley cell. In the Pacific and Indian Oceans, the expansion of the Hadley cell is not obvious. Over the extratropics, δMCD also partly depends on its advective part δMCDA, which mainly reflects changes in low-level winds (Fig. 7b). The strengthened and poleward-shifted westerlies are clearly seen in both hemispheres, consistent with previous studies (Lu et al. 2007).

Fig. 7.

(a) Climatological divergence (106 s−1) at the surface (contours) and the regression of surface divergence on the normalized water vapor secular trend (colors). (b) Regression of surface wind (m s−1) against the normalized water vapor secular trend.

Fig. 7.

(a) Climatological divergence (106 s−1) at the surface (contours) and the regression of surface divergence on the normalized water vapor secular trend (colors). (b) Regression of surface wind (m s−1) against the normalized water vapor secular trend.

c. Contributions to water vapor multidecadal variability

Figure 8 shows the relative contributions of DivQ, EP, transient eddy, and thermodynamical and mean circulation dynamical processes to water vapor multidecadal fluctuations. It appears that a large portion of water vapor variations associated with the AMO can be explained by DivQ (Fig. 8a), whereas EP plays an opposite role (Fig. 8b). A further inspection finds that DivQ change associated with the AMO is largely due to the mass divergence of moisture DivQD (Fig. 8c), particularly in the tropics. In the middle and high latitudes, DivQA (not shown) and transient eddy (Fig. 8d) also significantly contribute to DivQ change. Note that the transient eddy change associated with the AMO exhibits an opposite response between the two hemispheres, with a weakened and equatorward-shifted transient eddy in the Northern Hemisphere and a strengthened and poleward-shifted transient eddy in the Southern Hemisphere (Fig. 8d), which covaries with the mean circulation change that displays a weakened westerly in the Northern Hemisphere and a strengthened westerly in the Southern Hemisphere (Fig. 9b).

Fig. 8.

Regression of primary terms in water vapor budget against the normalized water vapor multidecadal variability (mm day−1). (a) Moisture flux divergence DivQ, (b) EP, (c) mass divergence of moisture DivQD, (d) transient eddies, and (e) thermodynamics contribution δTH. (f) Mean circulation dynamics δMCD; The δMCD equation is further decomposed into (g) the advection part δMCDA and (h) the divergence part δMCDD. See Table 1 for equation terms.

Fig. 8.

Regression of primary terms in water vapor budget against the normalized water vapor multidecadal variability (mm day−1). (a) Moisture flux divergence DivQ, (b) EP, (c) mass divergence of moisture DivQD, (d) transient eddies, and (e) thermodynamics contribution δTH. (f) Mean circulation dynamics δMCD; The δMCD equation is further decomposed into (g) the advection part δMCDA and (h) the divergence part δMCDD. See Table 1 for equation terms.

Fig. 9.

As in Fig. 7, but against the water vapor multidecadal variability.

Fig. 9.

As in Fig. 7, but against the water vapor multidecadal variability.

To gain more understanding of the mechanisms of water vapor multidecadal variability, we further decompose the water vapor change into terms due to the thermodynamic and dynamic contributions. Figures 8e and 8f show the thermodynamical contribution δTH and the contribution from changes in the mean circulation δMCD, respectively. It can be seen that δTH provides a large amount of water vapor in the warm pool regions, particularly over the Indo-Pacific warm pool. Further examination finds that δTH primarily arises from δTHD (not shown), which has a moisture increase tendency in the regions of low-level convergence (the ITCZ and subpolar region) and a reduction tendency in the regions of low-level divergence (the subtropics). Similar to the water vapor secular trend, δMCD is the dominant factor to control water vapor multidecadal variability (Fig. 8f). A further investigation reveals that the major contribution to δMCD arises from δMCDD (Fig. 8h), which shows a strengthened convergence over the Indo-Pacific warm pool region and a strengthened divergence over the eastern Pacific cold tongue and south subtropics, indicating that the Walker circulation and Southern Hemispheric Hadley cell in the Pacific Ocean are strengthened (Fig. 9a). The anomalous planetary divergent circulation is primarily due to SST changes (Fig. 3b). Moreover, the dipole-like SST response associated with the AMO (Fig. 3b) over the Atlantic Ocean induces a northward shift of the Atlantic ITCZ, which can be revealed from the divergent circulation change in which there is a decreased convergence over the Atlantic ITCZ region and an increased convergence in its northern flank (Fig. 9a). The divergent circulation change in turn leads to a moisture convergence over the northern tropical Atlantic and a divergence over the southern tropical Atlantic. In addition to δMCDD, δMCDA, which reflects low-level wind change, partly accounts for the water vapor multidecadal variability in the middle and high latitudes (Fig. 8g).

d. Contributions to water vapor interannual variability

Similar to water vapor variability associated with the global warming and AMO, water vapor interannual fluctuations are also dominated by DivQ (Fig. 10a), whereas EP plays a damping role (Fig. 10b). DivQ changes mainly arise from DivQD (Fig. 10c). The transient eddy (Fig. 10d) contributes to the water vapor changes almost negatively in the middle and high latitudes, which is opposite to the variability associated with global warming and the AMO.

Fig. 10.

As in Fig. 8, but against the water vapor interannual variability.

Fig. 10.

As in Fig. 8, but against the water vapor interannual variability.

Thermodynamic and dynamic decompositions further point out that the water vapor changes associated with ENSO also can be explained by δMCD (Fig. 10f). The other component δTH is of secondary importance (Fig. 10e). As displayed in Figs. 10g and 10h, the moisture increase over the tropical oceans is predominantly accounted for by δMCD, while elsewhere the contribution of δMCDA to water vapor is comparable to δMCDD. As expected, δMCDD is largely associated with divergent circulation changes, with a weakened Walker circulation in the tropical Pacific Ocean and a strengthened Hadley cell in the eastern Pacific (Fig. 11a). Outside of the tropical Pacific Ocean, there is a significant convergence anomaly in the tropical North Atlantic (Fig. 11a), implying a weakened North Atlantic subtropical high (Klein et al. 1999). As the subtropical high becomes weak, winds in the southern branch change accordingly (Fig. 11b) and thus induce anomalous poleward winds to cause moisture increase over the tropical North Atlantic (Fig. 10g). Similarly, the North Pacific response to ENSO is characterized by a strengthened Aleutian low, leading to a strengthened convergence over the subpolar region (Fig. 11a). This induces a corresponding moisture increase (Fig. 10h).

Fig. 11.

As in Fig. 7, but against the water vapor interannual variability.

Fig. 11.

As in Fig. 7, but against the water vapor interannual variability.

e. Water vapor flux potential

Section 3 points out that global mean water vapor variability is primarily determined by the thermodynamics, which can be simply explained by the thermal C-C theory. However, on regional scales, water vapor fluctuations are dominated by the mean circulation dynamics, particularly the divergent circulation change, while the thermodynamics plays a secondary role. The significant role of divergent circulation change can be further seen from the water vapor flux potential , where should fulfill .

Water vapor flux potential and the associated planetary divergent water vapor transport are shown in Fig. 12. Climatological water vapor flux potential reveals that the water vapor flux has a convergent tendency in the warm pool regions, favoring an accumulation of water vapor (Fig. 12a). Over the cold tongue and subtropics, there is a tendency to diverge water vapor (Fig. 12a). This climatological distribution is largely associated with low-level transport of the Walker circulation and the Hadley cell. Additionally, it can be seen that the climatological water vapor convergence center in the Indo-Pacific region is located north of the equator, indicating a climatological northward water vapor transport across the equator, which is consistent with Chen et al. (1996).

Fig. 12.

(a) Climatological water vapor potential (contour interval is 1 × 1012 kg day−1) and associated planetary divergent water vapor transport (kg day−1 m−1). (b) Trend of water vapor potential (contour interval is 1 × 1010 kg day−1 yr−1) and the associated planetary divergent water vapor transport (kg day−1 m−1 yr−1). (c) Water vapor potential (contour interval is 1 × 1011 kg day−1) and the associated planetary divergent water vapor transport difference (kg day−1 m−1) between the AMO positive phase and the AMO negative phase. (d) As in (c), but for the ENSO composites.

Fig. 12.

(a) Climatological water vapor potential (contour interval is 1 × 1012 kg day−1) and associated planetary divergent water vapor transport (kg day−1 m−1). (b) Trend of water vapor potential (contour interval is 1 × 1010 kg day−1 yr−1) and the associated planetary divergent water vapor transport (kg day−1 m−1 yr−1). (c) Water vapor potential (contour interval is 1 × 1011 kg day−1) and the associated planetary divergent water vapor transport difference (kg day−1 m−1) between the AMO positive phase and the AMO negative phase. (d) As in (c), but for the ENSO composites.

Water vapor flux potential change in response to global warming is characterized by an increased water vapor flux convergence over the warm pool regions and a tendency to increase water vapor flux divergence over the cold tongue, as revealed in Fig. 12b. This west–east dipole of the water vapor flux potential is largely associated with the strengthened Walker circulation (Fig. 7a) as a result of La Niña–like SST response to global warming (Fig. 3a), which in turn induces moisture surplus over the warm pool and moisture deficit over the cold tongue, consistent with Fig. 4a. Additionally, water vapor flux convergence center over the warm pool exhibits a northwestward shift, suggesting that there is a net water vapor transport from the Southern Hemisphere to the Northern Hemisphere, as proposed by Khon et al. (2010). The interhemispheric water vapor transport can be also inferred from the Hadley cell change. Figure 13a presents the climatological Hadley cell and its trend. It is seen that the Hadley cell in the Southern Hemisphere is strengthened, whereas the Northern Hemispheric branch is weakened. Meanwhile, the southern Hadley cell tends to shift toward the north, favoring more northward water vapor transport across the equator. In general, the water vapor flux potential associated with global warming (Fig. 12b) shares great similarities with planetary divergent circulation change in the low level (Fig. 7a) and, thus, the water vapor spatial distribution (Fig. 4a), with water vapor flux convergence (divergence) coinciding with planetary circulation convergence (divergence) and water vapor surplus (deficit).

Fig. 13.

Climatological Hadley cell defined as the zonal mean streamfunction (black contours, interval is 20 × 109 kg s−1). (a) Hadley cell trend (20 × 107 kg s−1 decade−1) and Hadley cell composites during the (b) AMO and (c) ENSO different phases (20 × 109 kg s−1).

Fig. 13.

Climatological Hadley cell defined as the zonal mean streamfunction (black contours, interval is 20 × 109 kg s−1). (a) Hadley cell trend (20 × 107 kg s−1 decade−1) and Hadley cell composites during the (b) AMO and (c) ENSO different phases (20 × 109 kg s−1).

The water vapor flux potential associated with the AMO positive phase shares great similarities with global warming, with anomalous water vapor flux convergence over the Pacific warm pool regions and anomalous divergence over the Pacific cold tongue, as shown in Fig. 12c, which is consistent with the anomalous low-level divergent circulation change (Fig. 9a) due to the La Niña–like SST response over the Pacific Ocean (Fig. 3b). However, there are some differences over the Atlantic Ocean (Fig. 12b versus Fig. 12c). The water vapor flux potential exhibits a distinct north–south dipole associated with the AMO over the Atlantic Ocean (Fig. 12c) rather than a west–east dipole, as seen in global warming scenario (Fig. 12b). Dipole SST associated with the AMO warm phase is accompanied by a northward shift of the Atlantic ITCZ (Zhang and Delworth 2005), which results in an anomalous atmospheric overturning circulation with rising motion north of the equator and descending motion south of the equator (Fig. 9a) and hence leads to a moisture accumulated in the northern tropics (Fig. 12c). Figure 12c further points out that the water vapor flux convergence center has shifted to the north, particularly in the Atlantic. As expected, the southern Hadley cell in response to the AMO warm phase is strengthened and shifts northward (Fig. 13b), favoring more moisture transport from the Southern Hemisphere to the Northern Hemisphere.

The water vapor flux potential associated with the ENSO (Fig. 12d) is characterized by an anomalous convergence in the Pacific cold tongue region and a divergence over the Pacific warm pool regions, consistent with the SST-induced (Fig. 3c) divergent circulation change (Fig. 11a). The covariability of SST, planetary divergent circulation, and water vapor flux potential further suggests the dominant role of divergent circulation in regional water vapor distribution. Given that the divergent circulation change associated with the ENSO is symmetric about the equator, it is expected that interhemispheric water vapor transport under the ENSO conditions should be weakened. As shown in Fig. 13c, there is an anticyclonic Hadley cell anomaly north of the equator and a cyclonic Hadley cell anomaly south of the equator, representing that water vapor fluxes tend to converge at the equator, which in turn decreases the climatological northward water vapor transport across the equator.

5. Residence time of water vapor in atmosphere

In the 20CRv2 dataset, the global mean moisture increases with the surface temperature at a rate of 6.4% K−1, whereas the precipitation exhibits a slower increasing rate at 1% K−1. This different rate of global mean moisture and precipitation change has one immediate consequence on the changes in global mean water vapor residence time, defined as the ratio of global mean column-integrated water vapor and precipitation W/P (e.g., Trenberth 1998; Schneider et al. 2010). The rate of change in global mean residence time of water vapor with respect to surface temperature can be estimated as the difference between the corresponding rates of water vapor and precipitation:

 
formula

where W, P, and T are the column-integrated water vapor, precipitation, and surface temperature, respectively. Both 20CRv2 and SSM/I have their own water vapor and precipitation data, so it is easy for us to calculate their corresponding residence time.

The time series of global mean water vapor residence time are shown in Fig. 14a. In comparison with the SSM/I, 20CRv2 reasonably captures the residence time change in the last 20 yr, albeit with a weak amplitude. Additionally, it is found that the water vapor residence time increases since the twentieth century, which may be interpreted as a weakening of the atmospheric circulation in response to global warming, particularly in the tropics, where most of the water vapor is concentrated and maximum precipitation is located (e.g., Betts and Ridgway 1989; Betts 1998; Held and Soden 2006; Vecchi et al. 2006; Vecchi and Soden 2007). A further investigation finds an increase of the global mean residence time of water vapor at a rate of about 5.4% K−1, consistent with the prediction from moisture and precipitation changes (6.4% K−1 − 1% K−1 = 5.4% K−1). Meanwhile, it is worth noting that the global mean residence time displays a distinct multidecadal variability. SSA analysis shows that the first two low-frequency modes of the residence time are characterized by a trend and an AMO-like oscillation (Fig. 14b).

Fig. 14.

(a) Time series of global mean residence time (days) of atmospheric moisture in the 20CRv2 and SSM/I datasets. The residence time in SSM/I is reduced by two to better compare with the 20CRv2. (b) The first two modes (black line for trend, gray line for multidecadal oscillation) of SSA for the moisture residence time.

Fig. 14.

(a) Time series of global mean residence time (days) of atmospheric moisture in the 20CRv2 and SSM/I datasets. The residence time in SSM/I is reduced by two to better compare with the 20CRv2. (b) The first two modes (black line for trend, gray line for multidecadal oscillation) of SSA for the moisture residence time.

6. Discussion and summary

The global water vapor variations over the twentieth century are identified to display three major variability modes: secular trend and multidecadal and interannual variability modes (Fig. 2). The long-term trend is suggested to be associated with global warming (Santer et al. 2006) (Fig. 3a), in which the water vapor acts as a positive feedback to amplify SST (Hall and Manabe 1999; Held and Soden 2000). At interannual and multidecadal time scales, water vapor fluctuates with the ENSO and AMO, respectively (Figs. 3b,c). The synchronization of the water vapor and SST on the global scale can be explained by the simple thermodynamic C-C theory under conditions of constant relative humidity. However, thermodynamics cannot explain the spatial patterns of water vapor variations. The column-integrated water vapor budget points out that the water vapor spatial patterns associated with global warming, AMO, and ENSO are largely attributable to the atmospheric mean circulation dynamics, particularly the planetary divergent circulation as a result of SST change. In the middle and high latitudes, the transient eddy fluxes and thermodynamics also play significant roles.

Under global warming, the divergent circulation change in the tropics is characterized by an anomalous convergence over the warm pool regions and the southern flank of the tropics and an anomalous divergence over the northern flank of the tropics (Fig. 6f). In the extratropics, transient eddy response to global warming exerts a poleward shift and strengthening (Fig. 5e). Thermodynamic response contributes to moistening in the high latitudes and drying in the subtropics as a result of moisture increase by the mean divergent circulation (Fig. 6a). Divergent circulation changes associated with the AMO in the tropics exhibit an increased convergence in the Pacific warm pool and an increased divergence in the Pacific cold tongue (Fig. 8h), which is indicative of the strengthened Walker circulation. Over the tropical Atlantic, the ITCZ shifts to the north, which in turn generates a dipole-like divergent circulation response, with a convergence in the north and a divergence in the south (Fig. 8h). Thermodynamics associated with the AMO share great similarities with global warming (Fig. 8e); however, transient eddy response exerts large differences, which displays a weakened (strengthened) and an equatorward (a poleward) shift of transient eddy in the Northern (Southern) Hemisphere (Fig. 8d). Divergent circulation change associated with the ENSO in the tropical Pacific is opposite to that in the AMO warm phase, with a convergence anomaly in the cold tongue and a divergence anomaly in the warm pool regions (Fig. 10h), which predominantly causes water vapor distribution in the tropical Pacific. However, in the tropical North Atlantic, North Pacific, and South Pacific, water vapor distribution is associated not only with divergent circulation change but also with the moisture advection by the anomalous wind (Fig. 10g). Different from global warming and the AMO, both the thermodynamics and transient eddy play a negative role in the water vapor distribution associated with the ENSO (Figs. 10d,e).

The twentieth-century atmospheric reanalysis product provides a unique opportunity to assess climate model simulations of the twentieth century. Not all the climate models simulate water vapor variability in the twentieth century reasonably, especially water vapor multidecadal oscillation. Figure 15a displays the time series of global mean water vapor from the selected models in which the water vapor multidecadal variability is relatively well captured. The first visual impression from Fig. 15a is the significant moisture increase in response to global warming. Consistent with the 20CRv2, the second mode extracted based on the SSA method is associated with the AMO, with an increase of moisture in the AMO warm phase and a decrease of moisture in the AMO cold phase (Fig. 15b). Those models, which cannot simulate water vapor multidecadal oscillation, are deficient in capturing the AMO phase. Here, we only show simple results for the models, which will be focused on in the next step.

Fig. 15.

(a) Global ocean mean column-integrated water vapor time series (kg m−2) for the selected Coupled Model Intercomparison Project, phase 3 (CMIP3), models. (b) Multidecadal oscillation of ensemble mean water vapor time series decomposed by SSA method.

Fig. 15.

(a) Global ocean mean column-integrated water vapor time series (kg m−2) for the selected Coupled Model Intercomparison Project, phase 3 (CMIP3), models. (b) Multidecadal oscillation of ensemble mean water vapor time series decomposed by SSA method.

Acknowledgments

This work is supported by the China National Natural Science Foundation Key Project (41130859) and the National Creative Research Group Project (40921004).

REFERENCES

REFERENCES
Alishouse
,
J. C.
,
S. A.
Snyder
,
J.
Vongsathorn
, and
R. R.
Ferraro
,
1990
:
Determination of oceanic total precipitable water from the SSM/I
.
IEEE Trans. Geosci. Remote Sens.
,
28
,
811
816
.
Allan
,
R. P.
, and
B. J.
Soden
,
2008
:
Atmospheric warming and the amplification of precipitation extremes
.
Science
,
321
,
1481
1484
.
Allen
,
M. R.
,
P. L.
Read
, and
L. A.
Smith
,
1992
:
Temperature time series
.
Nature
,
333
,
686
.
Betts
,
A. K.
,
1998
:
Climate-convection feedbacks: Some further issues
.
Climatic Change
,
39
,
35
38
.
Betts
,
A. K.
, and
W.
Ridgway
,
1989
:
Climatic equilibrium of the atmospheric convective boundary layer over a tropical ocean
.
J. Atmos. Sci.
,
46
,
2621
2641
.
Broomhead
,
D. S.
, and
G. P.
King
,
1986
:
Extracting qualitative dynamics from experimental data
.
Physica D
,
20
,
217
236
.
Chen
,
T.
,
M.
Yan
,
J.
Pfaendtner
, and
Y.
Sud
,
1996
:
Annual variation of the global precipitable water and its maintenance: Observation and climate simulation
.
Tellus
,
48
,
1
16
.
Clement
,
A. C.
,
A.
Hall
, and
A. J.
Broccoli
,
2004
:
The importance of processional signals in the tropical climate
.
Climate Dyn.
,
22
,
327
341
.
Compo
,
G. P.
, and
Coauthors
,
2011
:
The Twentieth Century Reanalysis Project
.
Quart. J. Roy. Meteor. Soc.
,
137
,
1
28
.
Cubasch
,
U.
, and
Coauthors
,
2001
: Projections of future climate change. Climate Change 2001: The Scientific Basis, J. T. Houghton et al., Eds., Cambridge University Press, 525–582.
Dai
,
A.
,
2006
:
Recent climatology, variability, and trends in global surface humidity
.
J. Climate
,
19
,
3589
3606
.
Dai
,
A.
,
G. A.
Meehl
,
W. M.
Washington
,
T. M. L.
Wigley
, and
J. M.
Arblaster
,
2001
:
Ensemble simulation of twenty-first century climate changes: Business-as-usual versus CO2 stabilization
.
Bull. Amer. Meteor. Soc.
,
82
,
2377
2388
.
Delworth
,
T. L.
, and
M. E.
Mann
,
2000
:
Observed and simulated multidecadal variability in the Northern Hemisphere
.
Climate Dyn.
,
16
,
661
676
.
Dijkstra
,
H. A.
,
L.
Raa
,
M.
Schmeits
, and
J.
Gerrits
,
2006
:
On the physics of the Atlantic multidecadal oscillation
.
Ocean Dyn.
,
56
,
36
50
.
Enfield
,
D. B.
, and
D. A.
Mayer
,
1997
:
Tropical Atlantic sea surface temperature variability and its relation to El Niño-Southern Oscillation
.
J. Geophys. Res.
,
102
,
929
945
.
Enfield
,
D. B.
, and
A. M.
Mestas-Nuñez
,
1999
:
Multiscale variability in global sea surface temperatures and their relationships with tropospheric climate patterns
.
J. Climate
,
12
,
2719
2733
.
Fraedrich
,
K.
,
1986
:
Estimating the dimension of weather and climate attractors
.
J. Atmos. Sci.
,
43
,
419
432
.
Giannini
,
A.
,
Y.
Kushnir
, and
M. A.
Cane
,
2000
:
Interannual variability of Caribbean rainfall, ENSO, and the Atlantic Ocean
.
J. Climate
,
13
,
297
311
.
Hall
,
A.
, and
S.
Manabe
,
1999
:
The role of water vapor feedback in unperturbed climate variability and global warming
.
J. Climate
,
12
,
2327
2346
.
Hansen
,
J.
,
A.
Lacis
,
D.
Rind
,
G.
Russell
,
P.
Stone
,
I.
Fung
,
R.
Ruedy
, and
J.
Lerner
,
1984
:
Climate sensitivity: Analysis of feedback mechanisms
. Climate Processes and Climate Sensitivity, Geophys. Monogr., Vol. 29, Amer. Geophys. Union, 130–163.
Held
,
I. M.
, and
B. J.
Soden
,
2000
:
Water vapor feedback and global warming
.
Annu. Rev. Energy Environ.
,
25
,
441
475
.
Held
,
I. M.
, and
B. J.
Soden
,
2006
:
Robust response of the hydrological cycle to global warming
.
J. Climate
,
19
,
5686
5699
.
Kerr
,
R. A.
,
2000
:
A North Atlantic pacemaker for the centuries
.
Science
,
288
,
1984
1986
.
Khon
,
V. C.
,
W.
Park
,
M.
Latif
,
I. I.
Mokhov
, and
B.
Schneider
,
2010
:
Response of the hydrological cycle to orbital and greenhouse gas forcing
.
Geophys. Res. Lett.
,
37
,
L19705
,
doi:10.1029/2010GL044377
.
Klein
,
S. A.
,
B. J.
Soden
, and
N. C.
Lau
,
1999
:
Remote sea surface temperature variations during ENSO: Evidence for a tropical atmospheric bridge
.
J. Climate
,
12
,
917
932
.
Knight
,
J. R.
, and
Coauthors
,
2005
:
A signature of persistent natural thermohaline circulation cycles in observed climate
.
Geophys. Res. Lett.
,
32
,
L20708
,
doi:10.1029/2005GL024233
.
Lu
,
J.
,
G. A.
Vecchi
, and
T.
Reichler
,
2007
:
Expansion of the Hadley cell under global warming
.
Geophys. Res. Lett.
,
34
,
L06805
,
doi:10.1029/2006GL028443
.
Oort
,
A. H.
, and
J. J.
Yienger
,
1996
:
Observed interannual variability in the Hadley circulation and its connection to ENSO
.
J. Climate
,
9
,
2751
2767
.
Philipona
,
R.
,
B.
Durr
,
C.
Marty
,
A.
Ohmura
, and
M.
Wild
,
2004
:
Radiative forcing—measured at Earth's surface—corroborate the increasing greenhouse effect
.
Geophys. Res. Lett.
,
31
,
L03202
,
doi:10.1029/2003GL018765
.
Rayner
,
N. A.
,
D. E.
Parker
,
E. B.
Horton
,
C. K.
Folland
,
L. V.
Alexander
,
D. P.
Rowell
,
E. C.
Kent
, and
A.
Kaplan
,
2003
:
Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century
.
J. Geophys. Res.
,
108
,
4407
,
doi:10.1029/2002JD002670
.
Santer
,
B. D.
, and
Coauthors
,
2006
:
Forced and unforced ocean temperature changes in Atlantic and Pacific tropical cyclogenesis regions
.
Proc. Natl. Acad. Sci. USA
,
103
,
13 905
13 910
.
Schneider
,
T.
,
P. O.
Gorman
, and
X.
Levine
,
2010
:
Water vapor and the dynamics of climate changes
.
Rev. Geophys.
,
48
,
RG3001
,
doi:10.1029/2009RG000302
.
Seager
,
R.
,
N.
Naik
, and
G. A.
Vecchi
,
2010
:
Thermodynamic and dynamic mechanisms for large-scale changes in the hydrological cycle in response to global warming
.
J. Climate
,
23
,
4651
4668
.
Soden
,
B. J.
,
D. L.
Jackson
,
V.
Ramaswamy
,
D.
Schwarzkopf
, and
X.
Huang
,
2005
:
The radiative signature upper tropospheric moistening
.
Science
,
310
,
841
844
.
Trenberth
,
K. E.
,
1998
:
Atmospheric moisture residence times and cycling: Implications for rainfall rates and climate change
.
Climatic Change
,
39
,
667
694
.
Trenberth
,
K. E.
, and
C. J.
Guillemot
,
1995
:
Evaluation of the global atmospheric moisture budget as seen from analyses
.
J. Climate
,
8
,
2255
2272
.
Vautard
,
R.
, and
M.
Ghil
,
1989
:
Singular spectrum analysis in non-linear dynamics with applications to paleoclimatic time series
.
Physica D
,
35
,
395
424
.
Vecchi
,
G. A.
, and
B. J.
Soden
,
2007
:
Global warming and the weakening of the tropical circulation
.
J. Climate
,
20
,
4316
4340
.
Vecchi
,
G. A.
,
A. T.
Wittenberg
,
I. M.
Held
,
A.
Leetmaa
, and
M. J.
Harrison
,
2006
:
Weakening of tropical Pacific atmospheric circulation due to anthropogenic forcing
.
Nature
,
441
,
73
76
.
Wang
,
C.
,
L.
Zhang
, and
S.-K.
Lee
,
2013
:
Response of freshwater flux and sea surface salinity to variability of the Atlantic warm pool
.
J. Climate
,
26
,
1249
1267
.
Wang
,
J. X. L.
, and
D. J.
Gaffen
,
2001
:
Late-twentieth-century climatology and trends of surface humidity and temperature in China
.
J. Climate
,
14
,
2833
2845
.
Wentz
,
F. J.
,
1997
:
A well-calibrated ocean algorithm for special sensor microwave/imager
.
J. Geophys. Res.
,
102
,
8703
8718
,
doi:10.1029/96JC01751
.
Wentz
,
F. J.
, and
M.
Schabel
,
2000
:
Precise climate monitoring using complementary satellite data sets
.
Nature
,
403
,
414
416
.
Wentz
,
F. J.
,
L.
Ricciardulli
,
K.
Hilburn
, and
C.
Mears
,
2007
:
How much more rain will global warming bring?
Science
,
317
,
233
235
.
Willett
,
K. M.
,
N. P.
Gillett
,
P. D.
Jones
, and
P. W.
Thorne
,
2007
:
Attribution of observed surface humidity changes to human influence
.
Nature
,
449
,
710
712
.
Wu
,
L.
, and
Coauthors
,
2012
:
Enhanced warming over the global subtropical western boundary currents
.
Nat. Climate Change
,
2
,
161
166
.
Wu
,
Y.
,
M.
Ting
,
R.
Seager
,
H.
Huang
, and
M.
Cane
,
2011
:
Changes in storm tracks and energy transports in a warmer climate simulated by the GFDL CM2.1 model
.
Climate Dyn.
,
37
,
53
72
,
doi:10.1007/s00382-010-0776-4
.
Zhang
,
L.
, and
C.
Wang
,
2012
:
Remote influences on freshwater flux variability in the Atlantic warm pool region
.
Geophys. Res. Lett.
,
39
,
L19714
,
doi:10.1029/2012GL053530
.
Zhang
,
R.
, and
T.
Delworth
,
2005
:
Simulated tropical response to a substantial weakening of the Atlantic thermohaline circulation
.
J. Climate
,
18
,
1853
1860
.
Zhang
,
X.
,
F. W.
Zwiers
,
G. C.
Hegerl
,
F. H.
Lambert
,
N. P.
Gillett
,
S.
Solomon
,
P. A.
Stott
, and
T.
Nozawa
,
2007
:
Detection of human influence on twentieth-century precipitation trends
.
Nature
,
448
,
461
465
.