1. Introduction
As the temperature rises in response to increasing greenhouse gas (GHG) concentration, a variety of feedback processes arise, including those due to changes in the lapse rate, water vapor concentration, surface albedo, and cloud properties. Equilibrium climate sensitivity (ECS), the change of global mean surface air temperature needed to balance the radiative forcing of doubled CO2 concentration, depends on the magnitudes of the accumulated effect of these feedbacks. Among all the feedback processes, the cloud feedback contributes the most to intermodel spread in ECS (Bony and Dufresne 2005; Caldwell et al. 2016; Dufresne and Bony 2008; Vial et al. 2013). This makes understanding how clouds will change with global warming an abiding goal of climate science.
Observations provide clues of how clouds respond to external forcing in nature. However, variations in cloud cover and radiative fluxes in observations contain mixed signals from natural variability and external forcings. On interannual time scales, variations of clouds are primarily attributable to natural variability (Davis et al. 2013; Li et al. 2014; Li et al. 2017; Yang et al. 2016), while on multidecadal time scales, impacts of external forcing may become detectable for some cloud properties (Marvel et al. 2015; Norris et al. 2016). Dessler (2010) showed that the cloud feedback obtained from natural variability differs significantly from that occurring under CO2-induced global warming in eight CMIP3 GCMs, serving as a caution call for using short-term observational records to evaluate GCMs’ ECS. On the other hand, Zhou et al. (2015) reported that cloud feedbacks in response to natural variability are well correlated (r = 0.74) across CMIP5 models with those in response to greenhouse warming, offering the possibility of narrowing uncertainty in ECS via observationally constraining internal cloud variability. Isolating cloud changes due to natural variability and external forcings in satellite observations is an essential step for using observed data to evaluate GCMs and for assessing the relevance of observed cloud trends to cloud feedback and ECS.
Over the historical period, variations of cloud cover and cloud-mediated radiative fluxes are closely linked with variations in both the mean and pattern of SSTs (Gregory and Andrews 2016; Zhou et al. 2015, 2016). The climate feedback parameter shows substantial decadal variations in simulations performed with two atmospheric general circulation models (AGCMs) forced with prescribed observed historical SSTs (Gregory and Andrews 2016). These variations were shown to be driven by changing SST patterns. In similar experiments performed using CAM5 with prescribed historical SSTs, Zhou et al. (2016) showed that changing SST patterns can explain the cloud radiative effect (CRE) variations better than uniform warming over the historical period. They argue that changes in SST patterns lead to changes in the thermodynamic structure of the atmosphere that drive tropical marine low cloud anomalies, which are the primary contributors to CRE variations.
Several large-scale patterns of SST anomalies appear periodically due to internal dynamics of the ocean and atmosphere, and their interactions, recognized as climate modes (Deser et al. 2010). On decadal time scales, the Pacific decadal oscillation (PDO; Mantua et al. 1997) and the Atlantic multidecadal oscillation (AMO; Kerr 2000) represent the most important dynamical modes of global SST variation (Deser et al. 2010; Parker et al. 2007; Tung et al. 2019) and are reported to have important and near-global impacts on the climate system. The PDO has been shown to be associated with anomalous temperature and precipitation in North America (Barlow et al. 2001; Kurtzman and Scanlon 2007; McAfee 2014; Shukla et al. 2015; Wei et al. 2018), low cloud cover over the North American west coast (Schwartz et al. 2014), summer rainfall in East China (Chan and Zhou 2005; Yang et al. 2017; Yu et al. 2015; Zhu et al. 2015), and the tropical belt width (Allen et al. 2014). The AMO has been shown to impact the nearby regions (Enfield et al. 2001; Goldenberg et al. 2001; Knight et al. 2006; Lyu and Yu 2017; Sutton and Hodson 2005) and can also remotely impact the circulation in the Pacific (Kucharski et al. 2016a,b; Li et al. 2016; McGregor et al. 2014; Zhang and Delworth 2007). During the past four decades, both the PDO and the AMO indices exhibit a nonzero trend. Around the year 2000, the AMO switched from a negative to a positive phase, and the PDO switched from a positive to a negative phase. These strongly influenced the SST trends in recent decades and contributed to a decrease in the warming rate of global mean temperature in the early 2000s (Chen and Tung 2018; Dai et al. 2015; England et al. 2014; Kosaka and Xie 2013; Kucharski et al. 2016a; Li et al. 2016; Maher et al. 2014; McGregor et al. 2014; Meehl et al. 2013; Trenberth et al. 2014). However, the contribution of the phase shift of the PDO and the AMO to the observed cloud cover trends is not well established.
In this study, we investigate the PDO and the AMO impacts on cloud cover, and estimate the contribution of these decadal climate modes to the observed cloud cover trends during the satellite era. The data and methodology are described in section 2. GCM-derived estimates of the PDO- and AMO-induced cloud anomalies in the absence of forcing are described in section 3. The estimated contributions of the decadal variability to the observed cloud cover trends, along with the estimated contributions of GHG forcing, are described in section 4. In section 5 we summarize our findings.
2. Data and methodology
a. Data
The SST data used in this study were obtained from the Hadley Centre Global Sea Ice and Sea Surface Temperature dataset (HadISST; Rayner et al. 2003) and the Extended Reconstructed Sea Surface Temperature version 5 (ERSST v5; Huang et al. 2017) dataset. The HadISST dataset contains monthly global SST and sea ice concentration at 1° spatial resolution from 1870 to the present. The ERSST v5 dataset provides monthly mean global SST at 2° spatial resolution from 1854 to the present. In this study, SSTs from 1920 to 2018 are used. Monthly vertical motion at 500 hPa (ω500) was obtained from the NOAA-CIRES Twentieth Century Reanalysis V2c (Compo et al. 2011), which only assimilates the observed sea level pressure and provides a long record of data (1851–2014). The period of 1920–2014 is used. We also use ω500 from ERA-40 (Uppala et al. 2005), which assimilates satellite, surface, radiosonde and pilot observations. The period of 1957–2002 is used.
Cloud cover was obtained from the International Satellite Cloud Climatology Project (ISCCP; Schiffer and Rossow 1983) and the Pathfinder Atmospheres–Extended (PATMOS-x; Heidinger et al. 2014), two independent satellite cloud cover datasets with long periods of record (1983–2009 for ISCCP and 1982–present for PATMOS-x). The period of 1983–2009 is used. We make use of the ISCCP and PATMOS-x datasets that were adjusted for artifacts arising from changing satellite view angles, changing solar zenith angles, and other sources of spurious trends in the records (Norris and Evan 2015). During the procedure of removing spurious variability, global-mean cloud variability has also been subtracted, so the resulting dataset contains spatial anomalies relative to an unknown global mean value. The monthly total cloud cover, high cloud cover (above 440 hPa), and low cloud cover are used. The low and midlevel clouds in ISCCP are combined to yield a better estimation of low-level cloud cover (Seethala et al. 2015); as a result, the cloud cover below 440 hPa is considered as low cloud cover in ISCCP, whereas in PATMOS-x the low cloud cover is defined as the cloud cover below 680 hPa. These satellite data are provided on an evenly spaced 2.5° latitude–longitude grid. When calculating the spatial correlation between the trend patterns of cloud cover in satellite datasets with those obtained from GCM outputs, we average the satellite data into a 5° × 5° grid to make our results comparable with Norris et al. (2016). The patterns of cloud cover trends obtained from GCM are linearly interpolated into the 5° × 5° grid.
GCM simulations from CMIP5 are used to investigate the cloud responses to decadal variability and GHG forcing (see Table 1). The first 300 years of the preindustrial control (“piControl”) experiments are used to investigate cloud variations associated with the PDO and the AMO. In each model, the long-term trend is subtracted to remove the impacts of model drift. The removal of the long-term trend does not influence our results. The first 140 years of the “1pctCO2” (1% CO2) experiments, in which the CO2 concentration is increased by 1% each year, are used to investigate cloud responses to GHG forcing. A total of 29 models in CMIP5 provide cloud cover data for at least 300 years in piControl experiments and for the first 140 years of 1% CO2 experiments. The first ensemble member of each model is used.
Models used for the diagnostics of the preindustrial simulation and 1% CO2 simulation.
b. PDO and AMO signal
The PDO is defined by the leading EOF mode of the monthly anomalous SST in the North Pacific (poleward of 20°N) with global mean SST (GMSST) anomaly subtracted, following Mantua et al. (1997). The corresponding principal component (PC), after low-pass filtering and normalizing by its standard deviation, is defined as the PDO index. For the AMO, we make use of a revised AMO index proposed by Trenberth and Shea (2006), which is defined by the low-pass filtered time series of the area-weighted mean of the SST anomalies in the North Atlantic (0°–60°N, 0°–80°W), with near-global mean SST anomaly in 60°S–60°N subtracted to eliminate the impacts of background warming. In observational data, this revised AMO index yields a similar pattern of AMO-related SST anomalies compared to that of a traditional AMO index, which is defined by the average of North Atlantic SST anomalies of detrended data (Enfield et al. 2001). In GCM simulations, however, the revised AMO index better captures the observed remote impacts of the AMO outside of the North Atlantic, especially in the Pacific (Lyu and Yu 2017). As discussed in Lyu and Yu (2017), this may be associated with a bias of GCMs: the decadal variability of GMSST is more associated with the SST anomalies in the Atlantic than those in the Pacific in observation, while the reverse is true in GCMs. As a result, in GCM simulations, the traditional AMO index, which includes the signals of GMSST variation, would also include those signals of Pacific SST variation that are biasedly correlated with GMSST, leading to the biased SST responses in the Pacific. The removal of near-global mean temperature in the calculation of the revised AMO index minimizes the impacts of this bias, making it a more feasible index to apply in GCMs. Using the revised AMO index instead of the traditional AMO index may quantitatively affect our estimation of AMO’s contribution to observed cloud cover trend; however, the short observational records and the GCM biases limit us from estimating this potential uncertainty. The monthly anomalous fields of each variable, after low-pass filtering, are regressed onto the PDO index or the AMO index to obtain the PDO- or AMO-related anomalous fields. A Lanczos filter with a cutoff period of 13 years is used in all the filtering processes.
c. Attributing PDO- and AMO-related cloud cover changes to changes in large-scale meteorological parameters
Regression coefficient of SST and ω500 on the (left) PDO index and (right) AMO index: (a),(b) SST obtained from HadISST, (c),(d) ω500 from the Twentieth Century Reanalysis, and (e),(f) SST and (g),(h) ω500 obtained from CMIP5 model simulations. Dots in (e)–(h) indicate over 80% of models agree in sign.
Citation: Journal of Climate 32, 21; 10.1175/JCLI-D-18-0443.1
In addition to SST, previous studies find that the variation of low clouds is also related to changes in estimated inversion strength (EIS; Wood and Bretherton 2006) (Myers and Norris 2015; Qu et al. 2014, 2015b; Zhou et al. 2016). In particular, Zhou et al. (2016) reported a dominant role of EIS in driving the decadal variation of tropical marine low clouds over the historical period in CAM5 simulations. Employing multiple regression analysis with SST and EIS as independent variables to isolate their individual impacts on low clouds results in very little improvement to our statistical model, and SST anomalies explain the overall pattern of PDO- and the AMO-related low cloud changes better than EIS anomalies (Fig. S1 in the online supplemental material). The differences between our analysis and previous studies that report an important role of EIS in driving the variation of low clouds in the historical period may arise for two reasons. First, both external forcings and natural variability contribute to the variation of low cloud cover, SST, and EIS in the historical period. Certain regions can experience large EIS variations and hence large EIS-induced variations of low cloud cover in the historical period, while having few changes of EIS and small EIS-induced low cloud cover changes in the case of the PDO or the AMO. Second, in previous studies, multiple regression analysis is performed on data averaged over larger regions, whereas in the present study multiple regression analysis is applied on grid point data. Taking a regional mean before the multiple regression minimizes the nonlocal effects on low clouds from the upwind environment; also, it reduces the variability of SST, EIS, and low clouds to smaller ranges, in which the linear assumption may hold better and the nonlinear effects may be smaller. In regions with abundant stratiform clouds in CMIP5 models, typical magnitudes of correlations between monthly anomalous SST or EIS and low cloud cover are between 0.2 and 0.3 at the grid box scale. The magnitudes of these correlations increase to 0.5–0.7 if regionally averaged time series of SST, EIS, and low clouds are used (Fig. S2a), indicating that taking the regional mean improves the performance of the regression analysis. We have also examined multiple regression analysis on larger domain-averaged quantities (Fig. S2), similar to those methods used by previous studies (Qu et al. 2014, 2015b; Zhou et al. 2016). Off the coast of California, in the equatorial eastern Pacific, and in the tropical North Atlantic, the contributions of EIS to low cloud cover changes are larger than the contributions of SST. However, in the subtropical northeast Pacific and the midlatitudes of the northwest Pacific and the North Atlantic, the EIS contributions are much smaller. We choose to use grid point data for the statistical model, as they have the benefit of explaining the overall patterns of the PDO- and AMO-related low cloud changes, which are the particular interest of the present study. SST appears to be the variable that works the best for our grid point statistical model. We acknowledge that SST is not independent of many typical environmental conditions that are often used to understand low cloud changes such as inversion stability, subsidence, horizontal temperature advection, humidity, and surface wind speed. The influences of these cloud-controlling factors on low clouds are also implicitly included in our single-variable statistical model to the extent that they are correlated with SST.
d. Estimating the impacts of decadal variability and GHG forcing on observed cloud cover trends
3. PDO and AMO in climate model simulations
a. Characteristics of the PDO in model simulations
To set the stage for understanding the PDO-related changes of clouds, we first examine the changes of SST and circulation associated with the PDO in observational data and GCM simulations. The positive phase of PDO is characterized by negative SST anomalies in the midlatitude central Pacific straddled by positive horseshoe-shaped SST anomalies along the west coast of North America (Figs. 1a,e). As reported by previous studies, this pattern arises from several mechanisms, including the decadal variation of the Aleutian low caused by midlatitude stochastic forcing and remote influences from the tropics, the “reemergence” effect of the ocean due to oceanic thermal inertia, and the dynamical adjustments of the ocean gyre to basinwide anomalous wind forcing (Newman et al. 2016, and the citations therein). Outside of the North Pacific, the positive phase of the PDO is accompanied by the overall warming of the tropics with a maximum in the equatorial eastern Pacific. These features are consistent with the PDO-related SST anomalies in the ERSST v5 dataset (Fig. S3a) and are robust among the models.
The PDO-related ω500 changes in GCMs (Fig. 1g) show an enhancement of ascending motion in the tropical central Pacific and equatorial shift of the intertropical convergence zone (ITCZ) and the South Pacific convergence zone (SPCZ). The band of anomalous ascending motion extends south and east, to the subtropical South Pacific region. In the western North Pacific, the subtropical subsidence zone weakens and expands into midlatitudes, while in the eastern North Pacific there is anomalous ascent in the midlatitudes and anomalous descent around 15°N. These features are consistent across models, and are also roughly consistent with those in the ERA-40 and the Twentieth Century Reanalysis (Fig. 1c; see also Fig. S3c), except for the anomalous descent in the subtropical northeast Pacific that is not shown in the reanalysis datasets.
Figure 2a demonstrates the changes in total cloud cover associated with the PDO in CMIP5 models. The cloud cover increases in the central-to-western equatorial Pacific, and decreases over the Maritime Continent and the Indian Ocean. In the subtropical South Pacific, the cloud cover decreases on the western side and increases on the eastern side. In the subtropics and midlatitudes of the North Pacific, there is increasing cloud cover in the western side, and decreasing cloud cover in the regions along the west coast of North America and subtropical eastern Pacific.
Regression coefficient of cloud cover on the PDO index (shading) in CMIP5 models: (a) total cloud cover, (b) high cloud cover, and (c) low cloud cover; dots indicate over 80% of models agree in sign. The contours in (b) and (c) represent the omega-related high cloud changes and the SST-related low cloud changes, respectively, obtained by GCM simulations [see Eq. (2) for detailed description]. Black contours indicate positive values, and magenta contours indicate negative values, with a 0.15% per standard deviation contour interval and the zero contour omitted.
Citation: Journal of Climate 32, 21; 10.1175/JCLI-D-18-0443.1
The anomalous cloud cover pattern depends on altitude. The shadings in Figs. 2b and 2c show the changes of high clouds and low clouds associated with the PDO, respectively. A positive PDO index is associated with enhanced high cloud cover in the equatorial tropical Pacific and extending south and east, along with decreased high cloud cover over the Maritime Continent and the Indian Ocean. In the North Pacific, the changes of high clouds are marked by a quadrupole structure. High clouds increase in the western side and decrease in the eastern side of the subtropical North Pacific, and vice versa at midlatitudes. Associated with a positive PDO, low cloud cover increases in the north central Pacific and decreases along the west coast of North America and in the tropical eastern Pacific.
The relative contribution of anomalous low clouds and high clouds to the changes of total cloud cover depends on the climatology of the vertical distribution of clouds. Over the Maritime Continent and the tropical and subtropical western Pacific, where convective clouds are prevalent, the changes of total cloud cover are mostly due to changes in high clouds. Conversely, in the midlatitude North Pacific and the stratocumulus zone of the northeast Pacific, where the stratiform clouds are prevalent, the anomalous total cloud cover is dominated by low cloud changes. It is worth noting that the signs of total cloud changes and high cloud changes are robust across models in nearly the whole Pacific and the tropical Indian Ocean, and the sign of low cloud changes is also robust in the regions where the changes of low clouds dominate (Fig. 2).
b. Characteristics of the AMO in model simulations
The AMO represents the coherent decadal variation of the SST anomalies in the North Atlantic, which is suggested to be driven by the variation in the Atlantic meridional overturning circulation (AMOC) and the associated heat transport (Ba et al. 2014; Delworth and Mann 2000; Knight et al. 2005; Latif et al. 2004; Wang and Zhang 2013; Zhang et al. 2016). The positive phase of the AMO is marked by anomalously warm SSTs over the North Atlantic Ocean, which are also in horseshoe-shaped bands (Figs. 1b,f), and anomalously cool SSTs over the South Atlantic. In the Pacific, the patterns are similar to those of PDO albeit with opposite signs. These features are shared in both observation and GCM simulations.
The AMO-related ω500 changes in GCMs (Fig. 1h) featured a dipole structure in the tropical Atlantic, with anomalous ascent in the North Atlantic and anomalous descent in the South Atlantic. The dipole structure is accompanied by anomalous descent in the subtropical eastern Pacific of both hemispheres and the equatorial western Pacific. These features of ω500 changes in the Pacific are consistent with previous studies that showed how warming in the tropical Atlantic can remotely influence the eastern Pacific through Rossby waves and force anomalous descent in the central-to-western equatorial Pacific through equatorial Kelvin waves (Kucharski et al. 2016b; Li et al. 2016; McGregor et al. 2014). ERA-40 and the Twentieth Century Reanalysis also exhibit consistent large-scale features of the AMO-related ω500 anomalies, but the exact locations of anomalous descent in the equatorial western Pacific are different.
The AMO-related changes in cloud cover are demonstrated in Fig. 3. Similar to the changes of ω500, the high cloud cover shows a dipole structure in the tropical Atlantic with increasing high clouds in the North Atlantic and decreasing high clouds in the South Atlantic, coinciding with decreasing high cloud cover in the subtropical eastern Pacific of both hemispheres and tropical central Pacific (Fig. 3b). The anomalously negative low cloud cover in the North Atlantic has a horseshoe shape that is similar to the SST pattern, and the anomalously positive low cloud cover is most significant in subtropical northeast Pacific (Fig. 3c). These characteristics are robust across models (Fig. 3c). However, the changes in high and low cloud cover compensate in most regions, leaving the extratropical North Atlantic and equatorial central Pacific as the only two regions that exhibit consistent total cloud cover changes among models (Fig. 3a).
Regression coefficient of cloud cover on AMO index (shading) in CMIP5 models: (a) total cloud cover, (b) high cloud cover, and (c) low cloud cover; dots indicate over 80% of models agree in sign. The contours in (b) and (c) represent the omega-related high cloud changes and the SST-related low cloud changes, respectively, obtained by GCM simulations [see Eq. (2) for detailed description]. Black contours indicate positive values, and magenta contours indicate negative values, with a 1% K−1 contour interval and the zero contour omitted.
Citation: Journal of Climate 32, 21; 10.1175/JCLI-D-18-0443.1
c. Linking the PDO- and AMO-related cloud changes to meteorological factors
To understand the physics behind the robust PDO- and AMO-related cloud changes in GCM simulations and examine the reliability of GCMs, we estimate the cloud cover changes that are related to changes of SST or circulation associated with the PDO and AMO via regression analysis [Eq. (2)]. The contours in Figs. 2b and 3b show the high cloud changes related to ω500 in CMIP5 models. The pattern and magnitudes are similar to the anomalous high cloud cover (shading in Figs. 2b and 3b), suggesting that the PDO- and AMO-related high cloud changes are mostly due to changes in circulation. The area-weighted spatial correlations between the PDO-related high cloud changes and high cloud changes related to vertical motion are above 0.5 across all models. In the case of the AMO, the spatial correlations are higher than 0.5 in 26 out of 29 models (Table 2), indicating that the relation between changes of high clouds and circulation holds in most models.
The area-weighted spatial correlation between high cloud changes associated with PDO/AMO (shading of Figs. 2b and 3b) and vertical motion–related high cloud changes (contours of Figs. 2b and 3b), and the area-weighted spatial correlation between low cloud changes associated with PDO/AMO (shading of Figs. 2c and 3c) and SST-related low cloud changes (contours of Figs. 2c and 3c). Values higher than 0.5 are shown in bold.
The contours in Figs. 2c and 3c show the SST-related low cloud changes. The pattern and magnitudes are also similar to the low cloud changes associated with the PDO and the AMO (shading in Figs. 2c and 3c), suggesting that the low cloud changes associated with the PDO and the AMO are related to the SST changes. The area-weighted spatial correlations between the PDO-related low cloud changes and low cloud changes related to SST are above 0.5 in 27 out of 29 models, and 23 out of 29 models in the case of the AMO (Table 2), showing that the linkage of the PDO- and AMO-related low cloud changes and SST changes hold in most of the models. This result is consistent with previous studies that found significant correlation between low clouds and SST in observational data (Clement et al. 2009; Eastman et al. 2011; McCoy et al. 2017; Myers and Norris 2015) and GCMs (Myers and Norris 2015; Qu et al. 2014, 2015b). The mechanisms of how anomalous SSTs alter the profiles of temperature and moisture and thereby influence low clouds have been investigated in previous studies. For the impacts on temperature profile, a warmer local SST leads to a decrease of EIS, which favors smaller low cloud coverage (Myers and Norris 2015; Qu et al. 2014, 2015b; Wood and Bretherton 2006). It is worth noting that EIS changes are also influenced by temperature changes in the free atmosphere, which can be caused by nonlocal SST changes (Ceppi and Gregory 2017; Qu et al. 2015a; Zhou et al. 2016; 2017). For the impacts on moisture profile, previous studies suggest that a warmer SST increases the moisture flux from the surface to the boundary layer, and hence increases the moisture gradient between the boundary layer and the free troposphere (Bretherton 2015; Klein et al. 2017; Qu et al. 2015b). The increasing moisture gradient leads to more efficient entrainment drying, favoring reduced low cloud coverage (Bretherton and Blossey 2014; Qu et al. 2015b). As a result, warmer SSTs correspond to fewer low clouds and vice versa.
In summary, the cloud changes associated with the PDO and the AMO can be separated into changes of high clouds and low clouds, with the former being strongly tied to changes in atmospheric circulations and the latter being strongly tied to changes in local SST. In the case of a positive PDO, the warming in the eastern and central equatorial Pacific leads to an eastward shift of the tropical Pacific convective zone and an equatorial shift of the SPCZ, and hence influences the high clouds in the tropical Pacific and the Indian Ocean. On the other hand, the warm SST anomalies in the northeast Pacific and tropics lead to the decreasing low cloud cover in these regions, and the cool SST anomalies in the midlatitudes of the northwest Pacific lead to the increasing low cloud cover. In the case of a positive AMO, the hemispherically asymmetric SST anomalies in the tropical Atlantic lead to the asymmetric changes of vertical motion in the northern and southern Atlantic, which influences high cloud cover. Anomalous ascent in the tropical North Atlantic induces remote descent in the tropical central Pacific and the subtropics and midlatitudes of the eastern Pacific of both hemispheres, suppressing high cloud cover in these regions. Meanwhile, warming in the North Atlantic leads to decreasing low cloud cover. In the subtropical northeast Pacific, the anomalous cool SST and anomalous descent both contribute to the increase of EIS, resulting in increasing low cloud cover. We find that although the magnitude can vary, the patterns of cloud changes associated with the PDO are consistent and robust across models in nearly the whole Pacific and the tropical Indian Ocean. When separated into changes of high clouds and low clouds, the AMO impacts are also robust in the North Atlantic and the eastern Pacific, although the compensation between anomalous low cloud cover and high cloud cover may lead to diverse patterns of anomalous total cloud cover.
4. Estimating the contribution of decadal variability and GHG forcing to the observed cloud trends
In this section, we investigate the role of decadal variability in driving the long-term trend of the observed cloud cover and further compare with the estimated contribution of GHG forcing. Figure 4a shows the trend of total cloud cover in ISCCP observations. During the satellite era, regions with the more significant increase of total cloud cover include the tropical western Pacific, poleward and westward of the SPCZ, and the subtropical eastern Pacific of both hemispheres. Regions with decreasing total cloud cover include the equatorial central Pacific and midlatitudes of both hemispheres. In the Atlantic, the total cloud cover shows a hemispherically asymmetric structure, with decreasing cloud cover in the Northern Hemisphere and increasing cloud cover in the Southern Hemisphere. The trends of cloud cover in the ISCCP and the PATMOS-x datasets have also been discussed in Norris et al. (2016), who have shown that the increasing cloud cover in the tropical western Pacific, the poleward shift of the SPCZ, and the decreasing cloud cover in the midlatitudes of both hemispheres are consistent with trends in albedo and liquid water path in satellite observations.
Trend of cloud cover (percentage amount) in ISCCP observations (1983–2009). The plus signs indicate the sign of the trend of PATMOS-x observations in the same period is consistent with ISCCP.
Citation: Journal of Climate 32, 21; 10.1175/JCLI-D-18-0443.1
The trends of high clouds and low clouds are very different, being opposite to each other in many regions. Figures 4b and 4c show the trends of high cloud cover and low cloud cover, respectively. In the tropical western Pacific, there are reducing high clouds in the equatorial region and a poleward shift of high clouds in the region around the SPCZ. The changes of high clouds are offset by changes in low clouds over the equatorial western Pacific. In the subtropical Pacific of both hemispheres, the low clouds increase in the eastern side and decrease in the western side, while the reverse is true for high clouds. In the midlatitudes of the South Pacific, high clouds increase and low clouds decrease. In the tropical Atlantic, high clouds and low clouds both exhibit a dipole structure. The positive high cloud cover trends in the Northern Hemisphere and the negative high cloud cover trends in the Southern Hemisphere are compensated by the opposite low cloud changes. It should be noted that to some extent, the compensation between high and low cloud cover may be attributable to the changes in obscuration, that increasing high clouds mask low clouds from the satellite and decreasing high clouds reveal more previously masked low clouds.
Figure 5 shows the estimated contribution of decadal variability to the observed cloud cover trends [Eq. (3), with the terms ∂C/∂PDO and ∂C/∂AMO obtained from GCM outputs, and the terms dPDO/dt and dAMO/dt estimated from observational data]. During the satellite era, the PDO transitions from positive to negative phases, whereas the phase of the AMO switches from negative to positive. These gradual phase switches of the PDO and the AMO are accompanied by unique signatures of cloud cover changes that exhibit zonal variations in the Pacific, hemispheric asymmetry in the Atlantic, and a relatively more uniform increase in the tropical Indian Ocean (Figs. 5a,c,e). Many of these cloud pattern changes are consistent with the linear cloud cover trends observed over the satellite data (Figs. 5b,d,f). The estimated contributions of decadal variability to observed high cloud changes are characterized by a poleward and westward shift of the SPCZ, increasing high clouds in the Indian Ocean and the western North Pacific, and decreasing high clouds in the eastern North Pacific. In the tropical Atlantic, high cloud cover increases in the Northern Hemisphere and decreases in the Southern Hemisphere. The estimated contributions of decadal variability to observed low cloud changes are characterized by increasing low clouds in the northeast Pacific and decreasing low clouds in the northwest Pacific and the North Atlantic. In the Pacific, the estimated trends of total, high, and low cloud cover associated with decadal variability are dominated by the PDO component (Fig. 6). Exceptions are in the subtropics of the eastern Pacific and the tropical central Pacific, where the cloud cover trends associated with the AMO have magnitudes similar to those associated with the PDO (Fig. 7). In the Atlantic, the PDO component vanishes and the AMO component dominates the trends of total, high, and low cloud cover.
(a),(c),(e) The estimated contribution of decadal variability (PDO + AMO) to observed cloud cover trends [Eq. (3), with the terms ∂C/∂PDO and ∂C/∂AMO obtained from GCM outputs, and the terms dPDO/dt and dAMO/dt estimated from observational data]. (b),(d),(f) The pattern consistency between the contribution of decadal variability in (a), (c), and (e) and the observed cloud cover trend in ISCCP (Figs. 4a–c, respectively). Orange indicates the signs are both positive, and blue indicates the signs are both negative. Dots indicate over 80% of models agree in sign. The 60°S–60°N mean trend has been removed.
Citation: Journal of Climate 32, 21; 10.1175/JCLI-D-18-0443.1
(a),(c),(e) The estimated contribution of the PDO to observed cloud cover trends [Eq. (3), with the term ∂C/∂PDO obtained from GCM outputs, and the term dPDO/dt estimated from observational data]. (b),(d),(f) The pattern consistency between the PDO contribution in (a), (c), and (e) and the observed cloud cover trend in ISCCP (Figs. 4a–c, respectively). Orange indicates the signs are both positive, and blue indicates the signs are both negative. Dots indicate over 80% of models agree in sign. The 60°S–60°N mean trend has been removed.
Citation: Journal of Climate 32, 21; 10.1175/JCLI-D-18-0443.1
(a),(c),(e) The estimated contribution of the AMO to observed cloud cover trends [Eq. (3), with the term ∂C/∂AMO obtained from GCM outputs, and the term dAMO/dt estimated from observational data]. (b),(d),(f) The pattern consistency between the PDO contribution in (a), (c), and (e) and the observed cloud cover trend in ISCCP (Figs. 4a–c, respectively). Orange indicates the signs are both positive, and blue indicates the signs are both negative. Dots indicate over 80% of models agree in sign. The 60°S–60°N mean trend has been removed.
Citation: Journal of Climate 32, 21; 10.1175/JCLI-D-18-0443.1
To account for potential biases in GCMs, we applied the method described in section 2d [Eq. (4)], in which the estimated contribution of decadal variability to the observed low cloud cover trends is estimated entirely using observational data (Fig. 8). In the Pacific and the Atlantic, the estimated contribution of decadal variability to low cloud cover captures the features of the spatial pattern of the observed low cloud cover trends reasonably well (spatial correlation R = 0.48), although the magnitudes are underestimated. It is worth noting that, although the AMO-related low cloud changes in the South Atlantic are very weak and not robust in CMIP5 models (Fig. 7e), observational data suggest that the hemispherically asymmetric structure of the AMO impacts low clouds in the Atlantic, consistent with the observed low cloud cover trends (Figs. 8e,f). The AMO component dominates the estimated low cloud cover trends in the Atlantic, and also contributes to about half of the estimated trends of low clouds in the subtropical northeast Pacific. In other regions of the Pacific, the estimated low cloud cover trends are dominated by the PDO component with negligible contribution from the AMO component.
(a),(c),(e) The estimated contribution of decadal variability (PDO +AMO), the PDO, and the AMO to observed low cloud cover trends, respectively [Eq. (4), with all the terms estimated from observational data]. (b),(d),(f) The pattern consistency between the contribution of decadal variability in (a), (c), and (e) and the observed cloud cover trend in ISCCP (Figs. 4a–c, respectively). Orange indicates the signs are both positive, and blue indicates the signs are both negative. The 60°S–60°N mean trend has been removed.
Citation: Journal of Climate 32, 21; 10.1175/JCLI-D-18-0443.1
We also compare the estimated contribution of decadal variability to the observed cloud cover trends with the estimated contribution of GHG forcing (Fig. 9). The GHG-induced total cloud cover trends, in contrast to those associated with the PDO and the AMO, are more zonally uniform in the Pacific and are more hemispherically symmetric in the Atlantic. For high cloud cover, there are positive trends in the entire equatorial tropical Pacific region, negative trends in the tropical Atlantic in both hemispheres, and zonally uniform positive trends in poleward of 50° in both hemispheres. For low cloud cover, a westward and poleward expansion of the stratocumulus zones occurs in the subtropical southeast Pacific and the subtropics of the Atlantic in both hemispheres, while in the midlatitude North Pacific the low cloud cover decreases. These GHG-induced cloud cover trends may partially explain the observed cloud cover trends in specific regions, including positive low cloud cover trends in the subtropical southeast Pacific, negative total cloud cover trends in the midlatitudes, and positive high cloud cover trends poleward of 50°N/S.
(a),(c),(e) The estimated contribution of GHG forcing to observed cloud cover trends [Eq. (5), with the term dC/dGMT obtained from GCM outputs, and the term dGMT/dt estimated from observational data]. (b),(d),(f) The pattern consistency between GHG contribution in (a), (c), and (e) and the observed cloud cover trend in ISCCP (Figs. 4a–c, respectively). Orange indicates the signs are both positive, and blue indicates the signs are both negative. Dots indicate over 80% of models agree in sign. The 60°S–60°N mean trend has been removed.
Citation: Journal of Climate 32, 21; 10.1175/JCLI-D-18-0443.1
Comparing Fig. 5 and Fig. 9, the multimodel mean of the estimated decadal variability contributions to the observed cloud cover trends are larger than those due to GHG-induced warming. This is also the case in most of the models when assessing them individually. We examine the absolute value of the ratio of the estimated decadal variability contribution [left-hand side of Eq. (4)] to the GHG contribution [left-hand side of Eq. (5)] at each grid point in each model. A ratio larger than 1 indicates that at the grid point, the joint impacts of the PDO and the AMO variation on cloud cover trend are larger than the impacts of GHG-induced warming during 1983–2009; on the other hand, a ratio smaller than 1 indicates that the GHG forcing is more important. In nearly all the regions over the globe, in over 50% of the models the ratios fall within the range of 1–10 (Fig. 10a), indicating that the contribution of decadal variability to observed total cloud cover trend is larger than GHG contribution, while the two are still of the same order of magnitude and the GHG contribution is not negligible. About 30% of the models show a ratio larger than 10 in the regions of the North Pacific, the southwest Pacific, the North Atlantic, and the Indian Ocean (Fig. 10b), suggesting a dominating decadal variability contribution and a negligible GHG contribution in these regions. In the subtropical southeast Pacific, about 20% of the models show a ratio smaller than 1, suggesting that the impacts of GHG forcing may be important in that region. Overall, in the majority of the models, the contribution of decadal variability to observed cloud cover trends is more important, while the GHG contribution may be smaller but not negligible. As the observational records become longer, the contribution of decadal variability to the observed linear trend is expected to decrease while the relative importance of GHG forcing is expected to increase.
Percentage of models in which absolute values of the ratio of decadal variability contribution (PDO + AMO) to GHG contribution are (a) within the range of 1–10, (b) larger than 10, and (c) smaller than 1.
Citation: Journal of Climate 32, 21; 10.1175/JCLI-D-18-0443.1
The area-weighted spatial correlations between the observed cloud cover trends and contributions from the PDO and the AMO, the joint impacts of the PDO and the AMO, and GHG forcing are summarized in Table 3. The pattern of decadal variability-related total cloud cover trends is significantly correlated with those observed by the ISCCP (R = 0.37) and the combined ISCCP/PATMOS-x dataset (R = 0.32). However, the correlations between the patterns of decadal variability-related cloud cover trends and those observed by the PATMOS-x are weak and not significant, reflecting the large uncertainty in observed clouds trends. For the high cloud cover trends, the spatial correlations between the pattern of decadal variability-related high cloud cover trends and those in the ISCCP and the PATMOS-x observation are not significant, which may be caused by the small number of spatial degrees of freedom of the high cloud data (see Table S1 in the online supplemental material). The pattern of low cloud cover trends associated with decadal variability is significantly correlated with those observed by the ISCCP (R = 0.38). Both the PDO and the AMO components show similar significance as in the case of the joint impacts of decadal variability, suggesting the correlations are contributed by both climate modes rather than dominated by either mode. The estimated GHG contribution shows no significant spatial correlation in any case, suggesting a relatively minor role for GHG-induced cloud changes on observed cloud trends.
The area-weighted spatial correlation between the estimated decadal variability/GHG contribution (Figs. 5–7 and 9) and the observed cloud trend (Fig. 4). Parentheses show the p values obtained from a one-tailed Student’s t test; values with p value < 0.05 are highlighted in bold. The effective numbers of spatial degrees of freedom used in the statistical test are listed in Table S1.
Although we highlight in this study the importance of decadal variability to the cloud cover trends in the satellite era, the impacts of external forcings may not be ignored. Norris et al. (2016) found a spatial correlation of 0.39 between the cloud cover trends in the period of 1983–2009 in the combined ISCCP/PATMOS-x dataset and that in the coincident CMIP5 historical simulations that include GHG, aerosol, ozone, and volcano forcings. The correlation of 0.39 in the case of historical simulations is higher than the correlation in the case of GHG single-forcing simulations (R = 0.21), suggesting that the external forcings other than GHG, which are not discussed in the present study, may also contribute to the observed cloud cover trend. Norris et al. (2016) also note that a correlation as high as 0.39 is extremely rare when considering all possible 27-yr-long chunks of 15 000 years of preindustrial simulations, suggesting that the observed trends are at least partly forced and were extremely unlikely to have occurred due solely to internal climate variability. Consistently, in our analysis, the spatial correlation between the cloud cover trends associated with decadal variability and the observed cloud cover trends in the combined ISCCP/PATMOS-x dataset (R = 0.32) is slightly lower than the spatial correlation between cloud cover trends in CMIP5 historical simulations and observation found in Norris et al. (2016) (R = 0.39), suggesting that the joint impacts of external forcings may also be important to the pattern of observed cloud cover trends. Nevertheless, the spatial correlation between the observed cloud cover trends and decadal variability contribution (R = 0.32) is larger than the correlation between the observed cloud cover trends and the contribution of any single external forcing [comparing with Table 1 in Norris et al. (2016)]. By considering the specific changing rates of the PDO and the AMO indices, the present study suggests that the observed cloud cover trends are more closely linked with decadal variability than GHG forcing. Some previous studies suggest that external forcings may contribute to part of the patterns and strength of the PDO and the AMO through their impacts on regional SST (Booth et al. 2012; McGregor et al. 2014; Takahashi and Watanabe 2016; Zhang et al. 2013). While we cannot exclude the possibility of external forcings contributing to the PDO- and AMO-like patterns of SST anomalies and hence influencing the trend of the PDO and the AMO indices in observation, our results highlight the contributions of the internally generated decadal variability to the observed cloud cover trends in the majority of the Pacific and North Atlantic via investigating unforced preindustrial simulations.
In summary, the estimated contribution of decadal variability to the cloud cover trends shows a similar pattern and order of magnitude with the observed cloud cover trends. The spatial patterns of the contribution of decadal variability to total cloud cover trends are significantly correlated with the cloud cover trends in the ISCCP and the combined ISCCP/PATMOS-x datasets. In the Pacific, the PDO component dominates the zonal variation cloud cover trends associated with decadal variability, except in the subtropical eastern Pacific and in the tropical central Pacific where the AMO can also contribute to the high cloud cover trends. In the Atlantic, the AMO component dominates the hemispherically asymmetric cloud cover trends associated with decadal variability. The estimated GHG contribution to total cloud cover trends shows a more zonally uniform pattern in the Pacific as well as a more hemispheric symmetric pattern in the Atlantic. This pattern differs markedly from the cloud cover trends associated with the PDO and the AMO, and shows no significant correlation with the pattern of total cloud cover trends in observation. When compared with the contribution of GHG forcing, the contribution of decadal variability shows larger magnitudes in nearly all the regions over the globe. It has been discussed in Seethala et al. (2015) that the trends of low cloud cover in the five main stratocumulus zones during the period of 1983–2009 may be associated with natural variability rather than GHG-induced warming, as the low cloud cover trends and their relative contributions of meteorological factors are more diverse across regions compared to those in the future warming experiments simulated by climate models (Qu et al. 2014). Consistent with their results, the present study further suggests that the phase shift of the PDO and the AMO during the period of 1983–2009 has significantly contributed to the long-term trends of the observed cloud cover.
5. Summary
The relationships between clouds and the PDO and the AMO are investigated in GCM simulations. The patterns of PDO-related total cloud cover changes feature an eastward and equatorial shift of clouds in the tropical western Pacific and the Indian Ocean, along with zonally asymmetric total cloud cover changes in the North Pacific. On the other hand, patterns of AMO-related cloud cover changes are characterized by hemispherically asymmetric anomalies of high and low cloud cover in the Atlantic, as well as decreasing high cloud cover in the subtropical eastern Pacific and the tropical central Pacific. These PDO- and AMO-related changes of cloud cover can be linked to the anomalous SST and circulation. Regions with anomalous ascent experience anomalously greater high cloud fraction and vice versa; regions with anomalously cool SST tend to experience increases in low cloud fraction and vice versa. Although the method of linearly estimating the SST- or vertical motion-related cloud changes [Eq. (2)] is highly simplified, and may miss some nonlocal impacts, it still captures the pattern and magnitude of anomalous high and low cloud coverage associated with the PDO and the AMO.
When compared with the observed cloud cover trends during the satellite era, the estimated contribution of decadal variability shows a similar pattern and order of magnitude to the observed cloud cover trends. The spatial pattern of the decadal variability–related total and low cloud cover trends are significantly correlated with the cloud cover trends in ISCCP, while for the PATMOS-x dataset the correlation is weak and insignificant. Nevertheless, both datasets share some features that are consistent with the signature of decadal variability impacts, including the poleward shift of the SPCZ, decreasing clouds in the equatorial central Pacific, the zonally asymmetric structure in the North Pacific, and the hemispherically asymmetric anomalous low cloud cover in the Atlantic. In the majority of the models, the estimated contribution of decadal variability to cloud cover trends is larger than that from GHG forcing in nearly all regions over the globe. These results suggest that the phase shifts of the PDO and the AMO around the year 2000 have significantly contributed to the cloud cover trends in the satellite record.
Our results show that the impacts of the PDO, the AMO, and GHG forcing on clouds are regionally dependent and have different signatures. The distinct patterns may provide hints for attributing the observed cloud cover trend. To the extent that the observed cloud cover trends have a large component that does not arise from GHG-induced global warming, our results also suggest that the cloud feedback obtained from natural variability may differ from that obtained from global warming scenarios, since the associated cloud cover patterns are very different. Caution should be taken when comparing the cloud feedback obtained from observational data, in which forced and internally generated responses are comparable, with the cloud feedback obtained from warming experiments in GCM, in which the forced signal dominates over the noise of natural variability by design (Hawkins and Sutton 2009, 2011). Further attribution work that isolates the forced cloud responses from the impacts of internal variability may be needed to yield an observational constraint for climate models.
Acknowledgments
Y -J. Chen and Y.-T. Hwang were supported by the Ministry of Science and Technology of Taiwan (105-2628-M-002-009-MY4, 107-2636-M-002-001-, and 108-2636-M-002-007-). MDZ’s work was supported by DOE’s Regional and Global Model Analysis Program under the project “Identifying Robust Cloud Feedbacks in Observations and Models” and was performed under the auspices of the United States Department of Energy by Lawrence Livermore National Laboratory under contract DE-AC52-07NA27344. CZ’s work was supported by NSFC (41875095). We acknowledge the World Climate Research Programme’s Working Group on Coupled Modelling, which is responsible for CMIP, and we thank the climate modeling groups (listed in Table 1) for producing and making available their model output. For CMIP, the U.S. DOE’s Program for Climate Model Diagnosis and Intercomparison provided coordinating support and led development of software infrastructure in partnership with the Global Organization for Earth System Science Portals. We thank Joel Norris for providing artifact-corrected ISCCP and PATMOS-x data. NOAA_ERSST_V5 data and 20th Century Reanalysis V2c data provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado, USA, from their Web site at https://www.esrl.noaa.gov/psd/. We thank three anonymous reviewers for their constructive comments.
REFERENCES
Allen, R. J., J. R. Norris, and M. Kovilakam, 2014: Influence of anthropogenic aerosols and the Pacific Decadal Oscillation on tropical belt width. Nat. Geosci., 7, 270–274, https://doi.org/10.1038/ngeo2091.
Ba, J., and Coauthors, 2014: A multi-model comparison of Atlantic multidecadal variability. Climate Dyn., 43, 2333–2348, https://doi.org/10.1007/s00382-014-2056-1.
Barlow, M., S. Nigam, and E. H. Berbery, 2001: ENSO, Pacific decadal variability, and U.S. summertime precipitation, drought, and stream flow. J. Climate, 14, 2105–2128, https://doi.org/10.1175/1520-0442(2001)014<2105:EPDVAU>2.0.CO;2.
Bodas-Salcedo, A., and Coauthors, 2011: COSP: Satellite simulation software for model assessment. Bull. Amer. Meteor. Soc., 92, 1023–1044, https://doi.org/10.1175/2011BAMS2856.1.
Bony, S., and J. L. Dufresne, 2005: Marine boundary layer clouds at the heart of tropical cloud feedback uncertainties in climate models. Geophys. Res. Lett., 32, L20806, https://doi.org/10.1029/2005GL023851.
Booth, B. B. B., N. J. Dunstone, P. R. Halloran, T. Andrews, and N. Bellouin, 2012: Aerosols implicated as a prime driver of twentieth-century North Atlantic climate variability. Nature, 484, 228–232, https://doi.org/10.1038/nature10946.
Bretherton, C. S., 2015: Insights into low-latitude cloud feedbacks from high-resolution models. Philos. Trans. Roy. Soc., 373A, 20140415, https://doi.org/10.1098/rsta.2014.0415.
Bretherton, C. S., and P. N. Blossey, 2014: Low cloud reduction in a greenhouse-warmed climate: Results from Lagrangian LES of a subtropical marine cloudiness transition. J. Adv. Model. Earth Syst., 6, 91–114, https://doi.org/10.1002/2013MS000250.
Caldwell, P. M., M. D. Zelinka, K. E. Taylor, and K. Marvel, 2016: Quantifying the sources of intermodel spread in equilibrium climate sensitivity. J. Climate, 29, 513–524, https://doi.org/10.1175/JCLI-D-15-0352.1.
Ceppi, P., and J. M. Gregory, 2017: Relationship of tropospheric stability to climate sensitivity and Earth’s observed radiation budget. Proc. Natl. Acad. Sci. USA, 114, 13 126–13 131, https://doi.org/10.1073/PNAS.1714308114.
Chan, J. C. L., and W. Zhou, 2005: PDO, ENSO and the early summer monsoon rainfall over south China. Geophys. Res. Lett., 32, L08810, https://doi.org/10.1029/2004GL022015.
Chen, X., and K.-K. Tung, 2018: Global-mean surface temperature variability: Space–time perspective from rotated EOFs. Climate Dyn., 51, 1719–1732, https://doi.org/10.1007/s00382-017-3979-0.
Clement, A. C., R. Burgman, and J. R. Norris, 2009: Observational and model evidence for positive low-level cloud feedback. Science, 325, 460–465, https://doi.org/10.1126/science.1171255.
Clemesha, R. E. S., A. Gershunov, S. F. Iacobellis, and D. R. Cayan, 2017: Daily variability of California coastal low cloudiness: A balancing act between stability and subsidence. Geophys. Res. Lett., 44, 3330–3338, https://doi.org/10.1002/2017GL073075.
Compo, G. P., and Coauthors, 2011: The Twentieth Century Reanalysis Project. Quart. J. Roy. Meteor. Soc., 137, 1–28, https://doi.org/10.1002/qj.776.
Dai, A., J. C. Fyfe, S. P. Xie, and X. Dai, 2015: Decadal modulation of global surface temperature by internal climate variability. Nat. Climate Change, 5, 555–559, https://doi.org/10.1038/nclimate2605.
Davis, S. M., C. K. Liang, and K. H. Rosenlof, 2013: Interannual variability of tropical tropopause layer clouds. Geophys. Res. Lett., 40, 2862–2866, https://doi.org/10.1002/grl.50512.
Delworth, T. L., and M. E. Mann, 2000: Observed and simulated multidecadal variability in the Northern Hemisphere. Climate Dyn., 16, 661–676, https://doi.org/10.1007/s003820000075.
Deser, C., M. A. Alexander, S. Xie, and A. S. Phillips, 2010: Sea surface temperature variability: Patterns and mechanisms. Annu. Rev. Mar. Sci., 2, 115–143, https://doi.org/10.1146/annurev-marine-120408-151453.
Dessler, A. E., 2010: A determination of the cloud feedback from climate variations over the past decade. Science, 330, 1523–1527, https://doi.org/10.1126/science.1192546.
Dufresne, J. L., and S. Bony, 2008: An assessment of the primary sources of spread of global warming estimates from coupled atmosphere–ocean models. J. Climate, 21, 5135–5144, https://doi.org/10.1175/2008JCLI2239.1.
Eastman, R., S. G. Warren, and C. J. Hahn, 2011: Variations in cloud cover and cloud types over the ocean from surface observations, 1954–2008. J. Climate, 24, 5914–5934, https://doi.org/10.1175/2011JCLI3972.1.
Enfield, D. B., A. M. Mestas-Nuñez, and P. J. Trimble, 2001: The Atlantic multidecadal oscillation and its relation to rainfall and river flows in the continental U.S. Geophys. Res. Lett., 28, 2077–2080, https://doi.org/10.1029/2000GL012745.
England, M. H., and Coauthors, 2014: Recent intensification of wind-driven circulation in the Pacific and the ongoing warming hiatus. Nat. Climate Change, 4, 222–227, https://doi.org/10.1038/nclimate2106.
Goldenberg, S. B., C. W. Landsea, A. M. Mestas-Nuñez, and W. M. Gray, 2001: The recent increase in Atlantic hurricane activity: Causes and implications. Science, 293, 474–480, https://doi.org/10.1126/science.1060040.
Gregory, J. M., and T. Andrews, 2016: Variation in climate sensitivity and feedback parameters during the historical period. Geophys. Res. Lett., 43, 3911–3920, https://doi.org/10.1002/2016GL068406.
Ham, Y., J. Kug, and J. Park, 2013a: Two distinct roles of Atlantic SSTs in ENSO variability: North tropical Atlantic SST and Atlantic Niño. Geophys. Res. Lett., 40, 4012–4017, https://doi.org/10.1002/grl.50729.
Ham, Y., J. Kug, J. Park, and F. Jin, 2013b: Sea surface temperature in the north tropical Atlantic as a trigger for El Niño/Southern Oscillation events. Nat. Geosci., 6, 112–116, https://doi.org/10.1038/ngeo1686.
Hawkins, E., and R. Sutton, 2009: The potential to narrow uncertainty in regional climate predictions. Bull. Amer. Meteor. Soc., 90, 1095–1107, https://doi.org/10.1175/2009BAMS2607.1.
Hawkins, E., and R. Sutton, 2011: The potential to narrow uncertainty in projections of regional precipitation change. Climate Dyn., 37, 407–418, https://doi.org/10.1007/s00382-010-0810-6.
Heidinger, A. K., M. J. Foster, A. Walther, and X. Zhao, 2014: The Pathfinder Atmospheres-Extended AVHRR climate dataset. Bull. Amer. Meteor. Soc., 95, 909–922, https://doi.org/10.1175/BAMS-D-12-00246.1.
Huang, B., and Coauthors, 2017: Extended Reconstructed Sea Surface Temperature, version 5 (ERSSTv5): Upgrades, validations, and intercomparisons. J. Climate, 30, 8179–8205, https://doi.org/10.1175/JCLI-D-16-0836.1.
Kerr, R. A., 2000: A North Atlantic climate pacemaker for the centuries. Science, 288, 1984–1985, https://doi.org/10.1126/science.288.5473.1984.
Klein, S. A., D. L. Hartmann, and J. R. Norris, 1995: On the relationships among low-cloud structure, sea surface temperature, and atmospheric circulation in the summertime northeast Pacific. J. Climate, 8, 1140–1155, https://doi.org/10.1175/1520-0442(1995)008<1140:OTRALC>2.0.CO;2.
Klein, S. A., A. Hall, J. R. Norris, and R. Pincus, 2017: Low-cloud feedbacks from cloud-controlling factors: A review. Surv. Geophys., 38, 1307–1329, https://doi.org/10.1007/S10712-017-9433-3.
Knight, J. R., R. J. Allan, C. K. Folland, M. Vellinga, and M. E. Mann, 2005: A signature of persistent natural thermohaline circulation cycles in observed climate. Geophys. Res. Lett., 32, L20708, https://doi.org/10.1029/2005GL024233.
Knight, J. R., C. K. Folland, and A. A. Scaife, 2006: Climate impacts of the Atlantic Multidecadal Oscillation. Geophys. Res. Lett., 33, L17706, https://doi.org/10.1029/2006GL026242.
Kosaka, Y., and S.-P. Xie, 2013: Recent global-warming hiatus tied to equatorial Pacific surface cooling. Nature, 501, 403–407, https://doi.org/10.1038/nature12534.
Kucharski, F., and Coauthors, 2016a: Atlantic forcing of Pacific decadal variability. Climate Dyn., 46, 2337–2351, https://doi.org/10.1007/s00382-015-2705-z.
Kucharski, F., and Coauthors, 2016b: The teleconnection of the tropical Atlantic to Indo-Pacific sea surface temperatures on inter-annual to centennial time scales: A review of recent findings. Atmosphere, 7, 29, https://doi.org/10.3390/ATMOS7020029.
Kurtzman, D., and B. R. Scanlon, 2007: El Niño–Southern Oscillation and Pacific Decadal Oscillation impacts on precipitation in the southern and central United States: Evaluation of spatial distribution and predictions. Water Resour. Res., 43, W10427, https://doi.org/10.1029/2007WR005863.
Latif, M., E. Roeckner, M. Botzet, M. Esch, H. Haak, S. Hagemann, and Coauthors, 2004: Reconstructing, monitoring, and predicting multidecadal-scale changes in the North Atlantic thermohaline circulation with sea surface temperature. J. Climate, 17, 1605–1614, https://doi.org/10.1175/1520-0442(2004)017<1605:RMAPMC>2.0.CO;2.
Li, K.-F., H. Su, S.-N. Mak, T. M. Chang, J. H. Jiang, J. R. Norris, and Y. L. Yung, 2017: An analysis of high cloud variability: Imprints from the El Niño–Southern Oscillation. Climate Dyn., 48, 447–457, https://doi.org/10.1007/s00382-016-3086-7.
Li, X., S. Xie, S. T. Gille, and C. Yoo, 2016: Atlantic-induced pan-tropical climate change over the past three decades. Nat. Climate Change, 6, 275–279, https://doi.org/10.1038/nclimate2840.
Li, Y., D. W. J. Thompson, Y. Huang, and M. Zhang, 2014: Observed linkages between the northern annular mode/North Atlantic Oscillation, cloud incidence, and cloud radiative forcing. Geophys. Res. Lett., 41, 1681–1688, https://doi.org/10.1002/2013GL059113.
Lyu, K., and J.-Y. Yu, 2017: Climate impacts of the Atlantic Multidecadal Oscillation simulated in the CMIP5 models: A re-evaluation based on a revised index. Geophys. Res. Lett., 44, 3867–3876, https://doi.org/10.1002/2017GL072681.
Maher, N., A. Sen Gupta, and M. H. England, 2014: Drivers of decadal hiatus periods in the 20th and 21st centuries. Geophys. Res. Lett., 41, 5978–5986, https://doi.org/10.1002/2014GL060527.
Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78, 1069–1080, https://doi.org/10.1175/1520-0477(1997)078<1069:APICOW>2.0.CO;2.
Marvel, K., and Coauthors, 2015: External influences on modeled and observed cloud trends. J. Climate, 28, 4820–4840, https://doi.org/10.1175/JCLI-D-14-00734.1.
Mauger, G. S., and J. R. Norris, 2010: Assessing the impact of meteorological history on subtropical cloud fraction. J. Climate, 23, 2926–2940, https://doi.org/10.1175/2010JCLI3272.1.
McAfee, S. A., 2014: Consistency and the lack thereof in Pacific decadal oscillation impacts on North American winter climate. J. Climate, 27, 7410–7431, https://doi.org/10.1175/JCLI-D-14-00143.1.
McCoy, D. T., R. Eastman, D. L. Hartmann, and R. Wood, 2017: The change in low cloud cover in a warmed climate inferred from AIRS, MODIS, and ERA-Interim. J. Climate, 30, 3609–3620, https://doi.org/10.1175/JCLI-D-15-0734.1.
McGregor, S., A. Timmermann, M. F. Stuecker, M. H. England, M. Merrifield, F.-F. Jin, and Y. Chikamoto, 2014: Recent Walker circulation strengthening and Pacific cooling amplified by Atlantic warming. Nat. Climate Change, 4, 888–892, https://doi.org/10.1038/NCLIMATE2330.
Meehl, G. A., A. Hu, J. M. Arblaster, J. Fasullo, and K. E. Trenberth, 2013: Externally forced and internally generated decadal climate variability associated with the interdecadal Pacific oscillation. J. Climate, 26, 7298–7310, https://doi.org/10.1175/JCLI-D-12-00548.1.
Myers, T. A., and J. R. Norris, 2015: On the relationships between subtropical clouds and meteorology in observations and CMIP3 and CMIP5 models. J. Climate, 28, 2945–2967, https://doi.org/10.1175/JCLI-D-14-00475.1.
Newman, M., and Coauthors, 2016: The Pacific Decadal Oscillation, revisited. J. Climate, 29, 4399–4427, https://doi.org/10.1175/JCLI-D-15-0508.1.
Norris, J. R., and A. T. Evan, 2015: Empirical removal of artifacts from the ISCCP and PATMOS-x satellite cloud records. J. Atmos. Oceanic Technol., 32, 691–702, https://doi.org/10.1175/JTECH-D-14-00058.1.
Norris, J. R., R. J. Allen, A. T. Evan, M. D. Zelinka, C. W. O’Dell, and S. A. Klein, 2016: Evidence for climate change in the satellite cloud record. Nature, 536, 72–75, https://doi.org/10.1038/nature18273.
Parker, D., C. Folland, A. Scaife, J. Knight, A. Colman, P. Baines, and B. Dong, 2007: Decadal to multidecadal variability and the climate change background. J. Geophys. Res., 112, D18115, https://doi.org/10.1029/2007JD008411.
Qu, X., A. Hall, S. A. Klein, and P. M. Caldwell, 2014: On the spread of changes in marine low cloud cover in climate model simulations of the 21st century. Climate Dyn., 42, 2603–2626, https://doi.org/10.1007/s00382-013-1945-z.
Qu, X., A. Hall, S. A. Klein, and P. M. Caldwell, 2015a: The strength of the tropical inversion and its response to climate change in 18 CMIP5 models. Climate Dyn., 45, 375–396, https://doi.org/10.1007/s00382-014-2441-9.
Qu, X., A. Hall, S. A. Klein, and A. M. DeAngelis, 2015b: Positive tropical marine low-cloud cover feedback inferred from cloud-controlling factors. Geophys. Res. Lett., 42, 7767–7775, https://doi.org/10.1002/2015GL065627.
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, https://doi.org/10.1029/2002JD002670.
Schiffer, R. A., and W. B. Rossow, 1983: The International Satellite Cloud Climatology Project (ISCCP)—The first project of the World Climate Research Programme. Bull. Amer. Meteor. Soc., 64, 779–784, https://doi.org/10.1175/1520-0477-64.7.779.
Schwartz, R. E., A. Gershunov, S. F. Iacobellis, and D. R. Cayan, 2014: North American west coast summer low cloudiness: Broadscale variability associated with sea surface temperature. Geophys. Res. Lett., 41, 3307–3314, https://doi.org/10.1002/2014GL059825.
Seethala, C., J. R. Norris, and T. A. Myers, 2015: How has subtropical stratocumulus and associated meteorology changed since the 1980s? J. Climate, 28, 8396–8410, https://doi.org/10.1175/JCLI-D-15-0120.1.
Shukla, S., A. Steinemann, S. F. Iacobellis, and D. R. Cayan, 2015: Annual drought in California: Association with monthly precipitation and climate phases. J. Appl. Meteor. Climatol, 54, 2273–2281, https://doi.org/10.1175/JAMC-D-15-0167.1.
Sutton, R. T., and D. L. R. Hodson, 2005: Atlantic Ocean forcing of North American and European summer climate. Science, 309, 115–118, https://doi.org/10.1126/science.1109496.
Takahashi, C., and M. Watanabe, 2016: Pacific trade winds accelerated by aerosol forcing over the past two decades. Nat. Climate Change, 6, 768–772, https://doi.org/10.1038/nclimate2996.
Trenberth, K. E., and D. J. Shea, 2006: Atlantic hurricanes and natural variability in 2005. Geophys. Res. Lett., 33, L12704, https://doi.org/10.1029/2006GL026894.
Trenberth, K. E., J. T. Fasullo, G. Branstator, and A. S. Phillips, 2014: Seasonal aspects of the recent pause in surface warming. Nat. Climate Change, 4, 911–916, https://doi.org/10.1038/nclimate2341.
Tung, K.-K., X. Chen, J. Zhou, and K. Li, 2019: Interdecadal variability in pan-Pacific and global SST, revisited. Climate Dyn., 52, 2145–2157, https://doi.org/10.1007/s00382-018-4240-1.
Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 2961–3012, https://doi.org/10.1256/qj.04.176.
Vial, J., J. L. Dufresne, and S. Bony, 2013: On the interpretation of inter-model spread in CMIP5 climate sensitivity estimates. Climate Dyn., 41, 3339–3362, https://doi.org/10.1007/s00382-013-1725-9.
Wang, C., and L. Zhang, 2013: Multidecadal ocean temperature and salinity variability in the tropical North Atlantic: Linking with the AMO, AMOC, and subtropical cell. J. Climate, 26, 6137–6162, https://doi.org/10.1175/JCLI-D-12-00721.1.
Wei, Y., H. Yu, J. Huang, Y. He, B. Yang, X. Guan, and X. Liu, 2018: Comparison of the Pacific decadal oscillation in climate model simulations and observations. Int. J. Climatol., 38, e99–e118, https://doi.org/10.1002/joc.5355.
Wood, R., and C. S. Bretherton, 2006: On the relationship between stratiform low cloud cover and lower-tropospheric stability. J. Climate, 19, 6425–6432, https://doi.org/10.1175/JCLI3988.1.
Yang, Q., Z. Ma, X. Fan, Z.-L. Yang, Z. Xu, and P. Wu, 2017: Decadal modulation of precipitation patterns over eastern China by sea surface temperature anomalies. J. Climate, 30, 7017–7033, https://doi.org/10.1175/JCLI-D-16-0793.1.
Yang, Y., and Coauthors, 2016: Impacts of ENSO events on cloud radiative effects in preindustrial conditions: Changes in cloud fraction and their dependence on interactive aerosol emissions and concentrations. J. Geophys. Res. Atmos., 121, 6321–6335, https://doi.org/10.1002/2015JD024503.
Yu, L., T. Furevik, O. H. Ottera, and Y. Gao, 2015: Modulation of the Pacific decadal oscillation on the summer precipitation over East China: A comparison of observations to 600-years control run of Bergen Climate Model. Climate Dyn., 44, 475–494, https://doi.org/10.1007/s00382-014-2141-5.
Zhang, R., and T. L. Delworth, 2007: Impact of the Atlantic Multidecadal Oscillation on North Pacific climate variability. Geophys. Res. Lett., 34, L23708, https://doi.org/10.1029/2007GL031601.
Zhang, R., and Coauthors, 2013: Have aerosols caused the observed Atlantic multidecadal variability? J. Atmos. Sci., 70, 1135–1144, https://doi.org/10.1175/JAS-D-12-0331.1.