1. Introduction
The accurate simulation of cloud properties is a longstanding challenge for modeling of regional climate and the global climate system (e.g., Randall et al. 2003; Boucher et al. 2013). Changes in cloud properties in response to surface temperature changes have profound impacts on Earth’s radiation budget and hydrological cycle, contributing important feedbacks to climate change (Ramanathan et al. 1989; Slingo 1990; Stephens 2005; Norris et al. 2016; Su et al. 2017; Webb et al. 2017). These responses and feedbacks occur across different time scales as a result of climate variability and external forcing (Andrews 2014; Zhou et al. 2015, 2016; Webb et al. 2017). Previous studies have demonstrated that the intermodel spread in climate sensitivity is largely due to differences in cloud feedback in response to long-term climate change (e.g., Stephens 2005; Bony et al. 2006; Vial et al. 2013; Webb et al. 2017). However, observational records, especially those of cloud properties, are deemed too short or too uncertain to monitor long-term cloud property trends, let alone provide direct constraints to long-term cloud feedback (Dessler 2010; Dessler and Loeb 2013; Zhou et al. 2015; Yue et al. 2017).
A few recent studies have established a strong correlation between the net global cloud feedback on short-term (interannual and decadal) and climate change time scales (Zhou et al. 2015; Colman and Hanson 2017, 2018). Although the strong correlations appear to partially result from consistent responses of low clouds to the changes of large-scale dynamics (Zhou et al. 2015, 2017), the physical reasons for these strong correlations still require further investigation. Moreover, there is mounting evidence from model simulations showing the dependence of cloud feedback on different geographic patterns of surface temperature changes (Stevens et al. 2016). This so-called pattern effect (Stevens et al. 2016) points out the additional state dependence of cloud feedback, which is especially important for interpreting results from observations and understanding the relationships between cloud feedbacks defined locally and globally. As observational estimates and further study on the mechanisms of short-term cloud feedbacks remains necessary (Klein and Hall 2015), it is important to investigate the temporal and spatial characteristics of these short-term cloud feedbacks revealed by observations, which will potentially provide additional observational constraints to the climate model cloud feedback.
The global-mean cloud feedback is defined as the linear response of cloud-induced radiative flux changes at the top of the atmosphere (TOA) to mean surface temperature change. Numerical modeling experiments have indicated that the pattern and magnitude of the global-mean cloud feedback change with different patterns of surface temperature change over varying time scales and in response to different climate variability and forcing (Andrews 2014; Rose et al. 2014; Zhou et al. 2016; Colman and Hanson 2017). Therefore, short-term cloud feedback estimates from observations and comparisons with model simulations are sensitive to the time scale. Low-cloud feedback is often calculated at the local scale by correlating the changes of local cloud properties (such as cloud cover, and optical as well as geometrical thickness) or cloud-induced TOA radiation with change of local thermodynamic and dynamic properties (Gordon and Klein 2014; Qu et al. 2015; Ceppi et al. 2016; Terai et al. 2016; Myers and Norris, 2015, 2016; Seethala et al. 2015; McCoy et al. 2017; Klein et al. 2017) to understand various physical mechanisms for low-cloud feedbacks (Bretherton 2015). Armour et al. (2013) formulated the local climate feedbacks as a linearization about the local surface temperature (e.g., Boer and Yu 2003a,b). Assumed to be time invariant, their results show that temporally varying climate feedbacks in the instantaneous CO2-doubling scenario primarily emerge from the scaling of time-invariant local feedbacks by different local temperature warming patterns, which are then summed globally to give the global climate feedback and hence climate sensitivity (Armour et al. 2013). Many studies have investigated the relationships between local and nonlocal feedbacks and their impact on climate sensitivity (e.g., Feldl and Roe 2013a,b; Roe et al. 2015). By perturbing sea surface temperature (SST) within individual spatial patches in model experiments, Zhou et al. (2017) show that both local and nonlocal SST anomalies are capable of causing significant changes on the local cloud feedback, which they define as the global-mean cloud-induced TOA radiation change in response to local surface temperature change. Zhou et al. (2017) further suggest that the local cloud feedbacks are unlikely to be time invariant, and therefore the global cloud changes in response to an SST anomaly pattern cannot be reconstructed by the scaling relationship between local feedback and the pattern of local surface temperature change. The temporal variability of cloud feedbacks, as well as the relationships between the local and nonlocal contributions, warrants further investigation.
The short-term cloud feedback arising from El Niño–Southern Oscillation (ENSO) variations has been estimated from satellite observations. Dessler (2010) and Dessler and Loeb (2013) used monthly averages of satellite and reanalysis TOA all-sky fluxes, and reanalysis clear-sky fluxes and meteorological fields to quantify total cloud feedback. Zhou et al. (2013) and Ceppi et al. (2016) combined cloud radiative kernels (CRKs; Zelinka et al. 2012) that are derived using the Fu–Liou radiative transfer model (Fu and Liou 1992) with NASA’s Terra Moderate Resolution Imaging Spectroradiometer (MODIS; Barnes et al. 1998) Collection 5 (C5) cloud observations, enabling the quantification of satellite-based cloud feedback as a function of cloud type. However, Yue et al. (2017) show that the cloud responses derived from C5 are adversely impacted by calibration degradation and retrieval artifacts, which are largely corrected in the Collection (C6) Aqua MODIS cloud product. Yue et al. (2016) propose an observation-based CRK (obs-CRK) method that avoids inconsistencies between the kernels and the cloud response term, which are especially important in cloud feedback calculations. Therefore, it is necessary to reexamine these previous estimates of short-term cloud feedback using the Aqua MODIS C6 cloud record and the obs-CRKs that correspond with the satellite-observed clouds. Cloud feedback is a transient feature on shorter time scales. As a result, calculations of short-term cloud feedbacks are sensitive to the temporal range of the observations.
In this study we will quantify the short-term global and local cloud feedbacks over a time series in response to interannual climate variability during the A-Train era following the conventional cloud feedback framework that describes the linear part of the cloud radiative responses to the surface temperature change (e.g., Knutti and Rugenstein 2015). The temporal and spatial signatures of the short-term cloud feedbacks will be examined in the near-simultaneous observations of atmospheric states, cloud properties, and TOA radiation from various sensors in the A-Train for a variety of cloud types. This study is the first step to improve our understanding on how the short-term cloud feedback derived from observations is relevant to climate models, what physical mechanisms affect short-term and long-term cloud feedbacks, and how satellite observations should be used to constrain cloud feedbacks in climate models. Partitioning cloud feedback by cloud type is an efficient approach to assessing the fidelity of cloud feedbacks estimated from both climate models and observations, as different cloud types have very distinct shortwave and longwave feedbacks (Hartmann et al. 1992; Williams and Webb 2009; Zelinka et al. 2012, 2016; Yue et al. 2016). Furthermore, different satellite sensors have observational strengths and limitations regarding the observation of certain cloud types (Pincus et al. 2012; Yue et al. 2011, 2013, 2017), which directly impacts the accuracy and uncertainty estimate of cloud feedback.
Section 2 describes the global and local cloud feedback definitions used in previous studies and their relationships. Section 3 describes the methodology and data used in this investigation including the obs-CRKs. Sections 4 and 5 provide results on the spatial and temporal variability of global and local short-term cloud feedbacks. Section 6 concludes with a discussion and summary of the main results.
2. Approaches to global and local cloud feedback calculations
The different definitions discussed above reveal different and unique aspects of cloud feedback. We therefore expect that these cloud feedback parameters will produce very different magnitudes and spatial patterns. A better quantification of cloud feedbacks on different time scales and spatial scales remains a high priority to understand the climate system (Knutti and Rugenstein 2015; Webb et al. 2017). This investigation aims at providing estimates on short-term cloud feedback from high-quality satellite observations and quantifying the different temporal and spatial variability of these local and global cloud feedback parameters within the A-Train era. Cloud cover responses (CR) are also calculated by replacing the TOA radiation anomaly with the cloud cover anomaly.
3. Data and methodology
The short-term cloud feedback calculations use a combination of nearly simultaneous observations from the Atmospheric Infrared Sounder (AIRS)/Advanced Microwave Sounding Unit (AMSU) (Chahine et al. 2006), MODIS, and the Clouds and the Earth’s Radiant Energy System (CERES; Wielicki et al. 1996) on NASA’s Aqua satellite (Parkinson 2003). Both the cloud responses and the obs-CRKs are directly derived from these observations in order to maintain consistency between the kernels and the response terms. The obs-CRKs are calculated following the methodology in Yue et al. (2016). The cloud responses as a function of cloud type are obtained from monthly joint CTP–τ cloud histograms using the C6 Aqua-MODIS Level 3 (L3) product (Yue et al. 2017). Note that these results are based on daytime observations at the Aqua overpass time only. This sampling impact on the observation-based cloud feedback estimate has not been quantified yet, although little discernible impact is found for longwave cloud feedbacks in climate models when only the sunlit portion of the diurnal cycle of cloudiness is sampled (Zelinka et al. 2012). For shortwave cloud feedback, the nighttime values are simply set to be zero following earlier studies (Zelinka et al. 2012; Zhou et al. 2013). No other diurnal adjustment is applied to either cloud or TOA flux measurements in order to avoid introducing additional artifacts to the observed CRK and cloud responses. As a result, the orbital sampling bias still exists due to inherent limitations of polar orbiting observations, causing uncertainty on short-term cloud feedbacks estimated from A-Train observations. The quantification of such uncertainty warrants further study using geostationary observations and/or numerical model simulations.
The global-mean surface temperature time series used in this study is from the combined land surface air and SST monthly anomalies from the NASA Goddard Institute for Space Studies (GISS) Surface Temperature Analysis (GISTEMP; GISTEMP Team 2016; Hansen et al. 2010). The local surface temperature is from the European Centre Medium-Range Weather Forecasting interim reanalysis product (ERA-Interim) at 1° × 1° resolution (Dee et al. 2011).
a. Observation-based cloud radiative kernels
The CRK is the TOA radiative sensitivity to changes of cloud in a given cloud regime and is calculated as the change of TOA cloud radiative forcing (CRF) due to a perturbation of one unit cloud fraction C for each cloud type (Zelinka et al. 2012). Following the obs-CRK method in Yue et al. (2016), CRKs are derived from collocated MODIS, AIRS, CERES, and reanalysis data within each 1° × 1° grid box, for each cloud type and month. At each grid box, the CRF for cloud type c is separately calculated by grid box only if the grid box has both cloudy (for cloud type c) and clear observations available within the same month. We use the cloud properties from the C6 Aqua MODIS L2 data and the TOA radiative fluxes from the Aqua CERES L2 footprint data product, specifically the Single Satellite Footprint (SSF) TOA/surface fluxes and clouds edition 4A (Loeb et al. 2016). Clear sky is identified using a threshold value of AIRS version 6 L2 effective cloud fraction (ECF) <0.02 (Kahn et al. 2014). One year (2009) of L2 swath data from different sensors, as well as the 3-hourly instantaneous Modern-Era Retrospective analysis for Research and Applications, version 2 (MERRA-2), atmospheric states and TOA clear-column radiation data (Gelaro et al. 2017), are collocated at the 45-km scale of the AIRS/AMSU field of regard (Lambrigtsen and Lee 2003). This approach ensures optimal correspondence among radiation, cloud, and atmospheric state variables, an essential ingredient for deriving obs-CRKs.
Forty-nine cloud types are specified by histograms partitioned with seven bins of CTP (edges at 50, 180, 310, 440, 560, 680, 800, and 1000 hPa) and seven bins of τ (edges at 0, 0.3, 1.3, 3.6, 9.4, 23, 60, and >60). Following the International Satellite Cloud Climatology Project (ISCCP) cloud histogram convention (Rossow and Schiffer 1999), high, middle, and low clouds are defined as clouds with CTP ≤ 440 hPa, 440 < CTP ≤ 680 hPa, and CTP > 680 hPa, respectively. The optically thin, medium, and thick clouds are defined as clouds with τ ≤ 3.6, 3.6 < τ ≤ 23, and τ > 23, respectively.
The obs-CRKs are derived empirically from readily available satellite observations and reanalysis data for cloud types determined by MODIS. Yue et al. (2016) showed that significant differences can exist between obs-CRKs and model-derived CRKs that arise mainly from cloud vertical structure. Yue et al. (2016) also show that the obs-CRKs are sensitive to the instrument characteristics and algorithm assumptions used in the retrieval of cloud properties. Therefore, to obtain observation-based estimates of cloud feedback, it is necessary to derive instrument-specific CRKs to maintain consistency between clouds that produce the kernels and clouds that produce the response term.
b. Cloud cover responses and cloud feedbacks to interannual climate variability
In section 2, we describe different approaches to quantify the global and local cloud responses and feedbacks. We will investigate both the temporal evolution and spatial structures using three different definitions for CR and λ by regressing the monthly anomaly time series of cloud cover and cloud-induced TOA radiation against the monthly surface temperature anomaly using a least squares fit:
The global cloud response CRGG and cloud feedback parameter λGG calculated following Eq. (2). Two different calculations of the global mean are tested: (i) by performing a global mean of gridbox–based regression coefficients and (ii) by regressing the global-mean anomaly
The local cloud response CRLL and cloud feedback λLL calculated using Eq. (3).
The local cloud response CRGL and cloud feedback λGL calculated using Eq. (5).
The regional and global-mean magnitudes for definitions 2 and 3 are obtained by taking a regional and global mean of the regression coefficients at each grid box. The magnitudes of the global-mean values and their spatial patterns for definitions 1–3 outlined above are calculated using the observations described earlier in section 3. We then investigate the temporal dependence of cloud feedbacks and their spatial patterns for different cloud types with respect to interannual variability within the A-Train era. Uncertainties are reported at 90% confidence levels using the Student’s t test.
4. Temporal variability of cloud response and cloud feedback parameter
The temporal variability of the global-mean magnitudes of different CR and λ is shown for high, middle, and low clouds in Fig. 1 calculated using the fixed-starting-point method. The net, shortwave, and longwave short-term cloud feedbacks are shown by green, blue, and red, respectively, and CR with black lines. The x axes in Fig. 1 correspond with the end points in the time series such that the leftmost point is obtained with 10 years of data (July 2002–June 2012) and the rightmost point with 15 years of data (July 2002–June 2017). To more directly compare the temporal dependence of different variables, 
Temporal variability of the global- (solid lines) and tropical- (dashed lines) mean cloud cover responses (CR; % K−1; black lines) and cloud feedback parameter λ (W m−2 K−1; colored lines) calculated using the three definitions given in section 2. Columns show (left) λGG, (center) λLL, and (right) λGL for (a)–(c) all cloud types, (d)–(f) high cloud, (g)–(i) middle cloud, and (j)–(l) low cloud. Net, shortwave, and longwave cloud feedbacks are indicated by green, blue, and red lines, respectively. For presentation purposes, λLL and CRLL magnitudes are reduced by half, and λGL and CRGL magnitudes are multiplied by 20. The starting point of the time series in the calculation is July 2006 (fixed-starting-point method). The x axes correspond with the end points in the time series such that the leftmost point is obtained with 10 years of data (July 2002–June 2012) and the rightmost point with 15 years of data (July 2002–June 2017). Gray lines are the zero lines.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
Correlation coefficients R of cloud feedback/response spatial patterns between different time periods with those obtained using data from July 2002 to June 2012. The x axis indicates the end points of data record used in the calculation. The definition of the line colors are as in Fig. 1.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
From the relationships among the three cloud feedback parameters given in Eqs. (4) and (6), these results suggests that for short-term cloud feedback estimated from satellite observations: 1) the local cloud feedback defined following Armour et al. (2013) λLL is nearly time-invariant during the A-Train era; 2) the short-term cloud feedback λGG exhibits significant temporal changes that arise from different spatial patterns of surface temperature changes during different time periods (e.g., Colman and Hanson 2017); and 3) the local cloud feedback parameter λGL as defined in Zhou et al. (2017) also exhibits significant temporal change which is intrinsically linked to its definition. The nonlocal effect is included in the calculation of λGL through ΔRG in the numerator of Eq. (5). Although middle clouds have a large temporal variation for λGG, this is mainly contributed from weak responses and feedbacks in high latitudes where satellite observations are more uncertain (see Figs. 7 and 10). Moreover, given the ambiguity in determining midlevel cloud with MODIS (Wang et al. 2016), caution is needed when interpreting the middle-cloud results.
Figure 3 depicts the uncertainty estimated at the 90% level of statistical significance for the magnitude of λGG. Since the uncertainty in the global-mean cloud feedback estimate cannot be obtained by averaging local uncertainty values, results in Fig. 3 are calculated by regressing the global-mean cloud-induced TOA anomaly against ΔTGS. The magnitude of the uncertainty decreases with time as more data increase statistical significance, and the magnitude of ΔTGS increases during the 2015/16 El Niño event. While the uncertainty for total cloud feedback (low, middle, and high cloud types combined) is significantly larger than the magnitude of the feedback parameter itself, the uncertainty is greatly reduced for some cloud types when they are treated individually. Furthermore, the magnitude of uncertainty is much smaller in terms of CR, mainly due to noisy signals for the shortwave cloud feedback (blue lines in Fig. 3). As will be shown regarding the spatial patterns (Figs. S1–S6), statistical significance of observations is dependent on region and cloud type, which includes most of the tropics and the low clouds over subtropical ocean. Despite the temporal shortness of the record of satellite observations, statistically robust signals are obtained for certain cloud types and regions, resulting from instrument calibration and algorithm improvements (Yue et al. 2017). This shows that efforts to improve instrument calibration, cloud detection, and retrieval algorithms have direct benefit on the quantification of cloud feedback despite the short temporal record of the A-Train.
Uncertainty at the 90% statistical significance interval for global- (solid lines) and tropical- (dashed lines) mean cloud cover responses CRGG (% K−1; black lines) and cloud feedback parameter λGG (W m−2 K−1; colored lines). Results are shown for (top to bottom) all cloud types, high clouds, middle clouds, and low clouds. Net, shortwave, and longwave cloud feedbacks are indicated by green, blue, and red lines, respectively.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
Figures 4–6 show the feedback parameters λGG, λLL, and λGL, respectively, for each bin in the 3 × 3 CTP–τ histograms. The cloud types exhibit a wide range of temporal variability of cloud cover responses and short-term feedbacks calculated with definitions like λGG (Fig. 4) and λGL (Fig. 6), with less temporal variability found for optically thin clouds in comparison to optically medium and thick cloud. The nearly time-invariant λLL type of responses and feedbacks (Fig. 5) holds for all cloud types, except slightly more variability is observed for medium optical thickness clouds (center column in Fig. 5). It is notable, however, that the magnitude of CRLL and λLL is substantially different with respect to cloud type (Fig. 5).
Global- (solid lines) and tropical- (dashed lines) mean cloud cover responses CRGG (% K−1; black lines) and cloud feedback parameter λGG (W m−2 K−1; colored lines) for each cloud type in the 3 × 3 CTP–τ histograms following the ISCCP convention: (a)–(c) high, (d)–(f) middle, and (g)–(i) low clouds are defined as clouds with CTP ≤ 440 hPa, 440 < CTP ≤ 680 hPa, and CTP > 680 hPa, respectively. The optically (left) thin, (center) medium, and (right) thick clouds are defined as cloud with τ ≤ 3.6, 3.6 < τ ≤ 23, τ > 23, respectively. Line styles and colors are as in Fig. 1. Gray lines are the zero lines.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
As in Fig. 4, but for cloud cover responses CRLL (% K−1; black lines) and cloud feedback parameter λLL (W m−2 K−1; colored lines) defined in Eq. (3).
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
As in Fig. 4, but for cloud cover responses CRGL (% K−1; black lines) and cloud feedback parameter λGL (W m−2 K−1; colored lines) defined in Eq. (5). Note that the magnitudes of CRGL and λGL are multiplied by 20 for presentation purposes.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
The negative to positive transition of λGG for global total net cloud feedback (Fig. 1a) is primarily due to reductions in magnitude of negative shortwave feedback of low cloud with medium τ (Fig. 4h) and increases in the magnitude of positive shortwave feedback of optically thick low clouds (Fig. 4i). The latter corresponds with a transition from near-zero to negative cloud cover response for optically thick low clouds (black line in Fig. 4i) and suggests that a transition to a less negative total low-cloud shortwave feedback shown in Fig. 1j is related to a decreasing optically thick low cloud cover anomaly with increasing global-mean surface temperature anomaly ΔTGS. The increasing positive feedback from total high cloud (Fig. 1d) also contributes to the transition to a positive net total global feedback in Fig. 1a, in particular from increased positive shortwave feedback by optically thick high clouds (more negative cloud response during 2015/16 El Niño event for this cloud type; see Fig. 4c). The high clouds with medium τ show an increased positive cloud response; however, their longwave and shortwave effects seem to compensate each other (Fig. 4b). The magnitude of λGG for optically thin clouds is much smaller than for other cloud types, which is related to the small TOA radiative sensitivity (kernel values) and the limited sensitivity of MODIS to cloud with τ < 0.4 (Holz et al. 2008; Zhou et al. 2014). As shown in Fig. 5, the magnitudes of the shortwave and longwave components of λLL for the optically thin high and low clouds are much larger than those for λGG and λGL, which results from the large local cloud responses by these clouds (CRLL; black lines in Fig. 5). For local feedback λGL (Fig. 1, right column, and Fig. 6), the time variations before 2015 are fairly small with more dramatic changes that are associated with the 2015/16 El Niño event.
The dashed lines in Figs. 1–6 are for tropical means (30°N–30°S) but with magnitudes reduced by half for presentation purposes. Although obtaining observational estimates of cloud feedbacks in the tropics neglects the effect of horizontal energy transport on the energy balance (Rose et al. 2014; Zhou et al. 2017), the results presented here show that similar conclusions regarding time dependence can be drawn for both tropical and global cloud feedbacks, while the magnitudes are generally 2–3 times larger in the tropics with a smaller statistical uncertainty for each cloud type (Fig. 3). This is likely because ENSO variability is the dominant mode of interannual climate variability with the strongest signals over the tropical ocean.
5. Spatial pattern of short-term CR and λ
The three definitions given in section 2 reveal different aspects of cloud feedback and thus different spatial patterns and magnitudes are expected. In this section, the characteristics of the spatial structures of CR (Fig. 7–9) and net cloud feedback parameters (Figs. 10–12) are shown for four different time periods: July 2002–June 2012 (first columns), July 2002–June 2015 (second columns), July 2002–June 2016 (third columns), and July 2002–June 2017 (fourth columns). These time periods are chosen because they correspond with a gradual change of λGG from negative, to zero, and then to positive values, and also with significant changes in high cloud during the 2015/16 El Niño event. (The regions that reach the 90% statistical significance level are shown in the supplemental material Figs. S3–S8.)
The spatial patterns of cloud cover responses CRGG (% K−1) for four different time periods: (left to right) July 2002–June 2012, July 2002–June 2015, July 2002–June 2016, and July 2002–June 2017. Results for (a)–(d) all cloud and (e)–(h) high, (i)–(l) middle, and (m)–(p) low clouds are shown. Results with only regions reaching a 90% statistically significant regression slope are shown in Fig. S3.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
As in Fig. 7, but for CRLL. Results with only regions reaching a 90% statistically significant regression slope are shown in Fig. S4.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
As in Fig. 7, but for CRGL. Results with only regions reaching a 90% statistically significant regression slope are shown in Fig. S5.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
As in Fig. 7, but for net cloud feedback λGG (W m−2 K−1). Definition in Eq. (2) is used before taking global mean. Results with only regions reaching a 90% statistically significant regression slope are shown in Fig. S6.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
As in Fig. 10, but for net cloud feedback λLL. Results with only regions reaching a 90% statistically significant regression slope are shown in Fig. S7.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
As in Fig. 10, but for net cloud feedback λGL. Definition in Eq. (5) is used. Results with only regions reaching a 90% statistically significant regression slope are shown in Fig. S8.
Citation: Journal of Climate 32, 6; 10.1175/JCLI-D-18-0335.1
Both high clouds and low clouds produce changes in the spatial patterns of CRGG (Fig. 7) and λGG (Fig. 10). High clouds (Figs. 10e–h) exhibit an increased magnitude of negative feedback in the central and eastern Pacific, and an increased positive feedback in the warm pool region due to the strong 2015/16 El Niño event. The warm pool response plays a dominant role such that it causes the global net cloud feedback λGG for high cloud to become increasingly positive around 2016 (Figs. 10g and 1d). For low clouds, the net cloud feedback in the southeastern Pacific flips from negative (Fig. 10m) to positive (Fig. 10p) with time, which, in turn, contributes to a less negative global-mean low-cloud feedback (Fig. 1j). Significant changes over the subtropical Pacific occur with respect to cloud type, including an increased magnitude of positive feedback for high clouds (Fig. 10g) and a decreased magnitude of negative low-cloud feedback in the central subtropical Pacific (Fig. 10o), and an increased magnitude of positive low-cloud feedback off the West Coast of United States (Fig. 10o). Middle-cloud (Figs. 7i–l and 10i–l) contributions are much smaller than high and low clouds, and are primarily located in the high latitudes where the uncertainties of satellite cloud observations are large and noise in the shortwave cloud feedback increases (not shown).
As previously pointed out, the spatial patterns remain nearly unchanged for CRLL (Fig. 8) and λLL (Fig. 11) for each cloud type. Large and statistically significant signals are present over the tropical and subtropical ocean (see the supplemental figures) with signals reaching the 90% significance level regionally. This primarily results from a negative net feedback of high clouds (Figs. 11e–h) and a positive low-cloud feedback (Figs. 11m–p) with respect to the local surface temperature change. The correlation coefficients between spatial patterns from different time periods are above 0.9 except for middle clouds (Fig. 3), for which the CRLL and λLL are close to zero most everywhere.
Figures 9 and 12 show CRGL and λGL, respectively. These local feedbacks are calculated by regressing the global cloud cover and cloud radiative forcing anomalies against the local surface temperature change. The spatial patterns for all cloud types combined (top rows in Figs. 9 and 12) shown in this study are similar to those in Zhou et al. (2017), which are derived from idealized model simulations by changing SSTs within different spatial patches using a Green function approach and CERES observations using the radiative kernel method (Shell et al. 2008). The temporally varying spatial patterns from satellite data shown here indicate that the local feedbacks defined in this fashion may be also driven by different local warming patterns, similar to CRGG and λGG, supported by their relationship in Eq. (6). The magnitudes of CRGL and λGL are much smaller than parameters with subscripts GG and LL, and smaller than those in Zhou et al. (2017). Figure 12 shows that high, middle, and low cloud types all significantly contribute to the net all-cloud feedback λGL.
6. Conclusions
Observations from multiple sensors on the NASA Aqua satellite are used to estimate the temporal and spatial variability of the short-term (interannual) cloud responses (CR) and cloud feedback parameters λ for different cloud types using the interannual variability within the A-Train era (July 2002–June 2017). The calculations use a synergy of instantaneous observations from the Atmospheric Infrared Sounder (AIRS)/Advanced Microwave Sounding Unit (AMSU), Moderate Resolution Imaging Spectroradiometer (MODIS), and the Clouds and the Earth’s Radiant Energy System (CERES), as well as the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). The observation-based cloud radiative kernel (obs-CRK) method (Yue et al. 2016) is applied to ensure consistency between the radiative kernels and climate response terms in the cloud feedback calculations. The Collection 6 (C6) Aqua MODIS cloud product is used to minimize impacts from calibration degradation and retrieval artifacts as demonstrated by Yue et al. (2017). The global and local cloud feedbacks are estimated from satellite observations using three different definitions in the literature: 1) the conventionally defined global-mean cloud feedback parameter λGG from regressing the global-mean cloud-induced radiation anomaly ΔRG at the top of atmosphere (TOA) with the global-mean surface temperature change ΔTGS; 2) the local feedback λLL from regressing the local cloud induced TOA radiative flux change ΔR(c, ϕ) with the local surface temperature change ΔTS(c, ϕ) (e.g., Armour et al. 2013); and 3) the local feedback λGL from regressing ΔRG with ΔTS(c, ϕ) (Zhou et al. 2017).
With 15 years of observations, more significant temporal variability is found for λGG and λGL, while a nearly time-invariant λLL is found for all cloud types. The magnitude of λGG gradually changes from a negative to positive global net cloud feedback between 2002 and 2017. This is primarily due to a less negative shortwave feedback for low clouds, and partially due to a rapid increase of positive high-cloud feedback in response to the strong 2015/16 El Niño event. With a (nearly) time-invariant local feedback λLL, the conventionally defined cloud feedback parameter λGG can be obtained by scaling λLL with different local surface warming patterns [cf. Eq. (4); Colman and Hanson 2017, 2018], which indicates the potential of using relatively short satellite observations to constrain the long-term cloud feedback in climate models. Such a scaling relationship does not hold between λGL and λGG, and the nonlocal effect is included in the λGL type of feedbacks by definition [cf. Eq. (6)]. The relationships between the local and global cloud feedbacks are closely related to the atmospheric and oceanic energy transport of the climate system, and affect the uncertainty in regional climate predictability (Webb et al. 2006; Hwang and Frierson 2010; Feldl et al. 2014; Roe et al. 2015).
The spatial patterns remain nearly unchanged for CRLL and λLL for each cloud type. However, significant temporal variations occurred for the patterns of CRGG and λGG, as well as of CRGL and λGL, within the A-Train era. High clouds generally show a negative net λLL and low clouds generally show a positive λLL in response to local temperature change when combining all optical depth ranges. (Note that optically thin high clouds show a positive net λLL as shown in Fig. 5a) These signals are large and statistically significant over the tropical and subtropical oceans. For the more conventionally defined short-term cloud feedback parameter λGG, the increasing of positive feedback from high clouds in the warm pool region dominates over the increased magnitude of negative feedback in the central and eastern Pacific in association with the strong 2015/16 El Niño event. This change, together with the negative to positive change in low-cloud net feedback in the southeastern Pacific Ocean, causes the global net cloud feedback λGG to gradually switch from negative to positive within the A-Train era. Although the feedback parameter λGL is calculated with respect to local temperature change, in its definition the nonlocal effect is included through the global-mean TOA radiation changes. The magnitude of λGL is much smaller than λGG and λLL, and also smaller than those derived using the radiative kernel method (Shell et al. 2008) by Zhou et al. (2017). Unlike λGG and λLL where the contributions from middle clouds are negligible, all cloud types (high, middle, and low) contribute significantly to the total cloud feedback for λGL.
Using the conventional linearized climate feedback framework, short-term cloud feedbacks by cloud type are estimated from A-Train observations with respect to local and global interannual climate variability. The statistical uncertainty of the satellite-based global-mean cloud feedback is large, more so for the shortwave component than the longwave component. However, this investigation shows that despite the short data record, statistically robust signals are derived from the A-Train to quantify the cloud response and feedback parameter for tropical high clouds and subtropical oceanic low clouds, which shows the capability of the synergy of A-Train observations in reducing the high- and low-cloud feedback uncertainties, a key objective identified by the 2017 decadal survey (National Academies of Sciences, Engineering, and Medicine 2018). Moreover, the temporal and spatial structures of short-term cloud feedback revealed by the satellite observations provide additional observational constraints to the climate model cloud feedback. These new estimates of local and global cloud feedback parameters serve as a first step to understand the connections between cloud feedback, energy and moisture transport, and local energy budget. The capability of high-quality long-term observations shown in this study also highlights the importance of the continuous efforts to improve instrument calibration, cloud detection, and retrieval algorithms.
Acknowledgments
The authors thank Chen Zhou from the Lawrence Livermore National Laboratory (now at Nanjing University, China) for the helpful discussions. The research described in this paper was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. QY, EJF, SW, and BHK were supported by NASA’s Making Earth Science Data Records for Use in Research Environments (MEaSUREs) program (NNH17ZDA001N). QY and XLH were supported by the NASA CloudSat and CALIPSO Science Team Recompete NNH15ZDA001N-CCST grant. QY and SW were supported by NASA’s Modeling Analysis and Prediction grant (NNH16ZDA001N-MAP). QY, EJF, SW, MS, and BHK acknowledge the support of the AIRS Project at JPL. XLH was supported by NASA Grant NNX15AJ50G awarded to the University of Michigan. MODIS data were obtained through the Level-1 Atmosphere Archive and Distribution System (LAADS; http://ladsweb.nascom.nasa.gov/). CERES data were obtained from the CERES SSF-Level2 ordering page (http://ceres.larc.nasa.gov/products-info.php?product=SSF-Level2). The ECMWF Interim Reanalysis fields and GISTEMP data are publicly available from http://www.ecmwf.int/ and the GISTEMP website. The data used to generate the figures and tables in this study can be obtained by contacting the corresponding author.
REFERENCES
Andrews, T., 2014: Using an AGCM to diagnose historical effective radiative forcing and mechanisms of recent decadal climate change. J. Climate, 27, 1193–1209, https://doi.org/10.1175/JCLI-D-13-00336.1.
Armour, K. C., C. M. Bitz, and G. H. Roe, 2013: Time-varying climate sensitivity from regional feedbacks. J. Climate, 26, 4518–4534, https://doi.org/10.1175/JCLI-D-12-00544.1.
Barnes, W. L., T. S. Pagano, and V. V. Salomonson, 1998: Prelaunch characteristics of the Moderate Resolution Imaging Spectroradiometer (MODIS) on EOS-AM1. IEEE Trans. Geosci. Remote Sens., 36, 1088–1100, https://doi.org/10.1109/36.700993.
Boer, G. J., and B. Yu, 2003a: Climate sensitivity and climate state. Climate Dyn., 21, 167–176, https://doi.org/10.1007/s00382-003-0323-7.
Boer, G. J., and B. Yu, 2003b: Climate sensitivity and response. Climate Dyn., 20, 415–429, https://doi.org/10.1007/s00382-002-0283-3.
Bony, S., and Coauthors, 2006: How well do we understand and evaluate climate change feedback processes? J. Climate, 19, 3445–3482, https://doi.org/10.1175/JCLI3819.1.
Boucher, O., and Coauthors, 2013: Clouds and aerosols. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 571–657.
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.
Ceppi, P., D. T. McCoy, and D. L. Hartmann, 2016: Observational evidence for a negative shortwave cloud feedback in middle to high latitudes. Geophys. Res. Lett., 43, 1331–1339, https://doi.org/10.1002/2015GL067499.
Chahine, M. T., and Coauthors, 2006: AIRS: Improving weather forecasting and providing new data on greenhouse gases improving weather forecasting and providing new data on greenhouse gases. Bull. Amer. Meteor. Soc., 87, 911–926, https://doi.org/10.1175/BAMS-87-7-911.
Colman, R., and L. Hanson, 2017: On the relative strength of radiative feedbacks under climate variability and change. Climate Dyn., 49, 2115–2129, https://doi.org/10.1007/s00382-016-3441-8.
Colman, R., and L. Hanson, 2018: Correction to: On the relative strength of radiative feedbacks under climate variability and change. Climate Dyn., 50, 4783–4785, https://doi.org/10.1007/s00382-017-4048-4.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
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.
Dessler, A. E., and N. G. Loeb, 2013: Impact of dataset choice on calculations of the short-term cloud feedback. J. Geophys. Res., 118, 2821–2826, https://doi.org/10.1002/jgrd.50199.
Feldl, N., and G. H. Roe, 2013a: The nonlinear and nonlocal nature of climate feedbacks. J. Climate, 26, 8289–8304, https://doi.org/10.1175/JCLI-D-12-00631.1.
Feldl, N., and G. H. Roe, 2013b: Four perspectives on climate feedbacks. Geophys. Res. Lett., 40, 4007–4011, https://doi.org/10.1002/grl.50711.
Feldl, N., D. M. W. Frierson, and G. H. Roe, 2014: The influence of regional feedbacks on circulation sensitivity. Geophys. Res. Lett., 41, 2212–2220, https://doi.org/10.1002/2014GL059336.
Forster, P. M., and Coauthors, 2016: Recommendations for diagnosing effective radiative forcing from climate models for CMIP6. J. Geophys. Res., 121, 12 460–12 475, https://doi.org/10.1002/2016JD025320.
Fu, Q., and K. N. Liou, 1992: On the correlated k-distribution method for radiative transfer in nonhomogeneous atmosphere. J. Atmos. Sci., 49, 2139–2156, https://doi.org/10.1175/1520-0469(1992)049<2139:OTCDMF>2.0.CO;2.
Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 5419–5454, https://doi.org/10.1175/JCLI-D-16-0758.1.
GISTEMP Team, 2016: GISS Surface Temperature Analysis (GISTEMP). NASA Goddard Institute for Space Studies, accessed November 2017, http://data.giss.nasa.gov/gistemp/.
Gordon, N. D., and S. A. Klein, 2014: Low-cloud optical depth feedback in climate models. J. Geophys. Res., 119, 6052–6065, https://doi.org/10.1002/2013JD021052.
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.
Hansen, J., R. Ruedy, M. Sato, and K. Lo, 2010: Global surface temperature change. Rev. Geophys., 48, RG4004, https://doi.org/10.1029/2010RG000345.
Hartmann, D. L., M. E. Ockert-Bell, and M. L. Michelsen, 1992: The effect of cloud type on Earth’s energy balance: Global analysis. J. Climate, 5, 1281–1304, https://doi.org/10.1175/1520-0442(1992)005<1281:TEOCTO>2.0.CO;2.
Holz, R. E., S. A. Ackerman, F. W. Nagle, R. Frey, S. Dutcher, R. E. Kuehn, M. A. Vaughan, and B. Baum, 2008: Global Moderate Resolution Imaging Spectroradiometer (MODIS) cloud detection and height evaluation using CALIOP. J. Geophys. Res., 113, D00A19, https://doi.org/10.1029/2008JD009837.
Hwang, Y.-T., and D. M. W. Frierson, 2010: Increasing atmospheric poleward energy transport with global warming. Geophys. Res. Lett., 37, L24807, https://doi.org/10.1029/2010GL045440.
Kahn, B. H., and Coauthors, 2014: The Atmospheric Infrared Sounder version 6 cloud products. Atmos. Chem. Phys., 14, 399–426, https://doi.org/10.5194/acp-14-399-2014.
Klein, S. A., and A. Hall, 2015: Emergent constraints for cloud feedbacks. Curr. Clim. Change Rep., 1, 276–287, https://doi.org/10.1007/s40641-015-0027-1.
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.
Knutti, R., and M. A. A. Rugenstein, 2015: Feedbacks, climate sensitivity and the limits of linear models. Philos. Trans. Roy. Soc., 373A, 20150146, https://doi.org/10.1098/rsta.2015.0146.
Lambrigtsen, B. H., and S.-Y. Lee, 2003: Co-alignment and synchronization of the AIRS instrument suite. IEEE Trans. Geosci. Remote Sens., 41, 343–351, https://doi.org/10.1109/TGRS.2002.808246.
Loeb, N. G., N. Manalo-Smith, W. Su, M. Shankar, and S. Thomas, 2016: CERES top-of-atmosphere Earth radiation budget climate data record: Accounting for in-orbit changes in instrument calibration. Remote Sens., 8, 182, https://doi.org/10.3390/rs8030182.
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 ECMWF-Interim reanalysis. J. Climate, 30, 3609–3620, https://doi.org/10.1175/JCLI-D-15-0734.1.
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.
Myers, T. A., and J. R. Norris, 2016: Reducing the uncertainty in subtropical cloud feedback, Geophys. Res. Lett., 43, 2144–2148, https://doi.org/10.1002/2015GL067416.
National Academies of Sciences, Engineering, and Medicine, 2018: Thriving on Our Changing Planet: A Decadal Strategy for Earth Observation from Space. The National Academies Press, 716 pp., https://doi.org/10.17226/24938.
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 data record. Nature, 536, 72–75, https://doi.org/10.1038/nature18273.
Parkinson, C. L., 2003: Aqua: An Earth-observing satellite mission to examine water and other climate variables. IEEE Trans. Geosci. Remote Sens., 41, 173–183, https://doi.org/10.1109/TGRS.2002.808319.
Pincus, R., S. Platnick, S. A. Ackerman, R. S. Hemler, and R. J. P. Hofmann, 2012: Reconciling simulated and observed views of clouds: MODIS, ISCCP, and the limits of instrument simulators. J. Climate, 25, 4699–4720, https://doi.org/10.1175/JCLI-D-11-00267.1.
Qu, X., A. Hall, S. A. Klein, and A. DeAngelis, 2015: Positive tropical marine low-cloud cover feedback inferred from cloud-controlling factors. Geophys. Res. Lett., 42, 7767–7775, https://doi.org/10.1002/2015GL065627.
Ramanathan, V., R. D. Cess, E. F. Harrison, P. Minnis, B. R. Barkstrom, E. Ahmad, and D. Hartmann, 1989: Cloud-radiative forcing and climate: Results from the Earth Radiation Budget Experiment. Science, 243, 57–63, https://doi.org/10.1126/science.243.4887.57.
Randall, D., M. Khairoutdinov, A. Arakawa, and W. Grabowski, 2003: Breaking the cloud parameterization deadlock. Bull. Amer. Meteor. Soc., 84, 1547–1564, https://doi.org/10.1175/BAMS-84-11-1547.
Roe, G. H., N. Feldl, K. C. Armour, Y.-T. Hwang, and D. M. W. Frierson, 2015: The remote impacts of climate feedbacks on regional climate predictability. Nat. Geosci., 8, 135–139, https://doi.org/10.1038/ngeo2346.
Rose, B. E. J., K. Armour, D. S. Battisti, N. Feldl, and D. Koll, 2014: The dependence of transient climate sensitivity and radiative feedbacks on the spatial pattern of ocean heat uptake. Geophys. Res. Lett., 41, 1071–1078, https://doi.org/10.1002/2013GL058955.
Rossow, W. B., and R. A. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc., 80, 2261–2287, https://doi.org/10.1175/1520-0477(1999)080<2261:AIUCFI>2.0.CO;2.
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.
Shell, K. M., J. T. Kiehl, and C. A. Shields, 2008: Using the radiative kernel technique to calculate climate feedbacks in NCAR’s Community Atmospheric Model. J. Climate, 21, 2269–2282, https://doi.org/10.1175/2007JCLI2044.1.
Slingo, A., 1990: Sensitivity of the Earth’s radiation budget to changes in low clouds. Nature, 343, 49–51, https://doi.org/10.1038/343049a0.
Stephens, G. L., 2005: Cloud feedbacks in the climate system: A critical review. J. Climate, 18, 237–273, https://doi.org/10.1175/JCLI-3243.1.
Stevens, B., S. C. Sherwood, S. Bony, and M. J. Webb, 2016: Prospects for narrowing bounds on Earth’s equilibrium climate sensitivity. Earth’s Future, 4, 512–522, https://doi.org/10.1002/2016EF000376.
Su, H., and Coauthors, 2017: Tightening of Hadley ascent and tropical high cloud region key to precipitation change in a warmer climate. Nat. Commun., 8, 15771, https://doi.org/10.1038/ncomms15771.
Terai, C. R., M. Zelinka, and S. A. Klein, 2016: Constraining the low-cloud optical depth feedback at middle and high latitudes using satellite observations. J. Geophys. Res., 121, 9696–9716, https://doi.org/10.1002/2016JD025233.
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, T., E. J. Fetzer, S. Wong, B. H. Kahn, and Q. Yue, 2016: Validation of MODIS cloud mask and multilayer flag using CloudSat-CALIPSO cloud profiles and a cross-reference of their cloud classifications. J. Geophys. Res., 121, 11 620–11 635, https://doi.org/10.1002/2016JD025239.
Webb, M. J., and Coauthors, 2006: On the contribution of local feedback mechanisms to the range of climate sensitivity in two GCM ensembles. Climate Dyn., 27, 17–38, https://doi.org/10.1007/s00382-006-0111-2.
Webb, M. J., and Coauthors, 2017: The Cloud Feedback Model Intercomparison Project (CFMIP) contribution to CMIP6. Geosci. Model Dev., 10, 359–384, https://doi.org/10.5194/gmd-10-359-2017.
Wielicki, B. A., B. R. Barkstrom, E. F. Harrison, R. B. Lee III, G. L. Smith, and J. E. Cooper, 1996: Clouds and the Earth’s Radiant Energy System (CERES): An Earth observing system experiment. Bull. Amer. Meteor. Soc., 77, 853–868, https://doi.org/10.1175/1520-0477(1996)077<0853:CATERE>2.0.CO;2.
Williams, K. D., and M. J. Webb, 2009: A quantitative performance assessment of cloud regimes in climate models. Climate Dyn., 33, 141–157, https://doi.org/10.1007/s00382-008-0443-1.
Yue, Q., B. H. Kahn, E. J. Fetzer, and J. Teixeira, 2011: Relationship between oceanic boundary layer clouds and lower tropospheric stability observed by AIRS, CloudSat and CALIOP. J. Geophys. Res., 116, D18212, https://doi.org/10.1029/2011JD016136.
Yue, Q., E. J. Fetzer, B. H. Kahn, S. Wong, G. Manipon, A. Guilaume, and B. Wilson, 2013: Cloud-state-dependent sampling in AIRS observations based on CloudSat cloud classification. J. Climate, 26, 8357–8377, https://doi.org/10.1175/JCLI-D-13-00065.1.
Yue, Q., B. H. Kahn, E. J. Fetzer, M. Schreier, S. Wong, X. Chen, and X. Huang, 2016: Observation-based longwave cloud radiative kernels derived from the A-Train. J. Climate, 29, 2023–2040, https://doi.org/10.1175/JCLI-D-15-0257.1.
Yue, Q., B. H. Kahn, E. J. Fetzer, S. Wong, R. Frey, and K. G. Meyer, 2017: On the response of MODIS cloud coverage to global-mean surface air temperature. J. Geophys. Res. Atmos., 122, 966–979, https://doi.org/10.1002/2016JD025174.
Zelinka, M. D., S. A. Klein, and D. L. Hartmann, 2012: Computing and partitioning cloud feedbacks using cloud property histograms. Part I: Cloud radiative kernels. J. Climate, 25, 3715–3735, https://doi.org/10.1175/JCLI-D-11-00248.1.
Zelinka, M. D., C. Zhou, and S. A. Klein, 2016: Insights from a refined decomposition of cloud feedbacks. Geophys. Res. Lett., 43, 9259–9269, https://doi.org/10.1002/2016GL069917.
Zhang, M. H., J. J. Hack, J. T. Kiehl, and R. D. Cess, 1994: Diagnostic study of climate feedback processes in atmospheric general circulation models. J. Geophys. Res., 99, 5525–5537, https://doi.org/10.1029/93JD03523.
Zhou, C., M. D. Zelinka, A. E. Dessler, and P. Yang, 2013: An analysis of the short-term cloud feedback using MODIS data. J. Climate, 26, 4803–4815, https://doi.org/10.1175/JCLI-D-12-00547.1.
Zhou, C., A. E. Dessler, M. D. Zelinka, P. Yang, and T. Wang, 2014: Cirrus feedback on interannual climate fluctuations. Geophys. Res. Lett., 41, 9166–9173, https://doi.org/10.1002/2014GL062095.
Zhou, C., M. D. Zelinka, A. E. Dessler, and S. Klein, 2015: The relationship between interannual and long-term cloud feedback. Geophys. Res. Lett., 42, 10 463–10 469, https://doi.org/10.1002/2015GL066698.
Zhou, C., M. D. Zelinka, and S. A. Klein, 2016: Impact of decadal cloud variations on the Earth’s energy budget. Nat. Geosci., 9, 871–874, https://doi.org/10.1038/ngeo2828.
Zhou, C., M. D. Zelinka, and S. A. Klein, 2017: Analyzing the dependence of global cloud feedback on the spatial pattern of sea surface temperature change with a Green’s function approach. J. Adv. Model. Earth Syst., 9, 2174–2189, https://doi.org/10.1002/2017MS001096.