1. Introduction
Clouds remain one of the largest sources of uncertainty in global climate projections (e.g., Bony and Dufresne 2005; Webb et al. 2013; Zelinka et al. 2020). Cloud processes are challenging to simulate directly in present-generation global climate models because they occur on spatial and temporal scales too small to be resolved in models and must therefore be parameterized. Efforts to improve the representation of clouds in models rely on a process-level understanding of the factors that govern various cloud types.
One common method to understand the processes responsible for global cloud cover and its variability is the cloud controlling factor framework (e.g., Stevens and Brenguier 2009; Klein et al. 2017). Cloud controlling factors are large-scale meteorological variables that have been linked to variability in global cloud cover and its radiative properties in the observed climate system, serving as proxies for the turbulent fluxes of heat, moisture, and momentum relevant for cloud formation (Scott et al. 2020). The advantage of the cloud controlling factor approach is that large-scale meteorological fields are more well represented in models than the smaller-scale cloud properties that depend upon them. Therefore, by understanding how observed cloud properties respond to perturbations in the cloud controlling factors, the response of model clouds to climate change can be inferred solely from the models’ projected future changes in large-scale meteorological factors, circumventing the need to rely on the models’ cloud parameterizations. This method has been used by a number of recent studies to provide observational constraints on global cloud feedbacks (Qu et al. 2015; Myers et al. 2021; Cesana and Del Genio 2021).
The cloud controlling factor framework is most commonly applied to understand marine low cloud variability, not only in the tropics and subtropics (e.g., Norris and Iacobellis 2005; Myers and Norris 2013, 2015; Qu et al. 2014; Seethala et al. 2015) but also in the extratropics (e.g., Wall et al. 2017; Zelinka et al. 2018; Kelleher and Grise 2019; Grise and Kelleher 2021; Blanco et al. 2023). Common cloud controlling factors linked to low cloud variability include vertical velocity, estimated inversion strength (EIS), near-surface temperature advection (TADV), sea surface temperature (SST), relative humidity (RH), and near-surface wind speed (see Scott et al. 2020 for a full review of these factors). Briefly, marine low clouds typically form in subsiding environments characterized by boundary layer temperature inversions over relatively cool SSTs. An increase in the boundary layer inversion strength (EIS) would further inhibit mixing between the dry free troposphere and the moist marine boundary layer, promoting an increase in low cloud fraction and thickness (Klein and Hartmann 1993; Wood and Bretherton 2006). Alternatively, if the subsidence becomes too large, it would reduce boundary layer height and thus inhibit low cloud height, thickness, and fraction (Myers and Norris 2013).
Observations show that warm SST anomalies reduce low cloud fraction in these environments (Qu et al. 2014; Myers and Norris 2015; Scott et al. 2020). One proposed mechanism for this relationship is that, if SSTs warm, there is more latent heat release and moisture in the boundary layer, increasing the effectiveness of mixing dry free-tropospheric air into the moist marine boundary layer and thus reducing low cloud fraction (Rieck et al. 2012; Frey and Kay 2018). Alternatively, if the free troposphere becomes more humid, this would limit the entrainment of dry air into the boundary layer and act to sustain the presence of low cloud (Eastman and Wood 2018; Scott et al. 2020). These low cloud sensitivities to SST and RH are weaker in midlatitudes, where the free troposphere is relatively moister compared to tropical low cloud environments (Kawai et al. 2017; Scott et al. 2020).
Finally, advection of relatively colder air over warmer ocean waters destabilizes the marine boundary layer and promotes an enhanced moisture source for low cloud formation through increased sensible and latent heat fluxes from the ocean surface into the atmosphere (Norris and Iacobellis 2005; Myers and Norris 2015; Seethala et al. 2015; Zelinka et al. 2018). In the tropics where there is less TADV, stronger surface winds promote enhanced evaporation and surface moisture fluxes, a deeper marine boundary layer, and enhanced low cloud amount (Nuijens and Stevens 2012; Kazil et al. 2016; Mieslinger et al. 2019; Scott et al. 2020). Surface wind speed may also be an important controlling factor for extratropical low clouds (Blanco et al. 2023).
The cloud controlling factor framework has been used to a lesser extent in understanding nonlow cloud environments. One disadvantage to the cloud controlling factor approach is that different cloud types may respond differently to the same large-scale meteorological variables, complicating the interpretation of the results. For example, ascending motion inhibits the development of low cloud but promotes enhanced cloud fraction in the mid-to-upper troposphere, such as that associated with tropical deep convection or in the warm conveyor belt of midlatitude cyclones (Lau and Crane 1995; Li et al. 2014; Wall et al. 2017; Grise and Kelleher 2021). Similarly, warm advection inhibits the development of low cloud but is associated with enhanced mid-to-upper-tropospheric cloud fraction at midlatitudes, consistent with the well-known relationship in the warm conveyor belt of midlatitude cyclones (Lau and Crane 1995; Grise and Kelleher 2021). Finally, warm SST anomalies are associated with enhanced deep convection in the tropics along the equator (Li et al. 2014; Grise and Kelleher 2021), and higher free-tropospheric RH also promotes more clouds in the free troposphere (Myers and Norris 2015).
Cloud controlling factors can be used to develop a conceptual model of observed global cloud cover (Datseris et al. 2022), but an alternative approach to studying global cloud cover is to classify global cloud structures using clustering algorithms applied to satellite data and then subsequently analyze meteorological variables to develop an understanding of the processes relevant for individual cloud types. One example of this approach is the weather states (such as deep convection, midlatitude storm clouds, shallow cumulus, and stratocumulus) identified from cloud optical depth–cloud top pressure histograms from International Satellite Cloud Climatology Project (ISCCP) satellite data (Jakob and Tselioudis 2003; Rossow et al. 2005; Tselioudis et al. 2013, 2021). A similar approach has also been applied to Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data (Oreopoulos et al. 2014). The advantage of this approach is that the climatology and variability of individual cloud types can be directly studied, but the disadvantage is that the clustering methodology inevitably involves somewhat arbitrary choices. This approach can also only be applied to climate models that have run the CFMIP Observation Simulator Package (COSP), limiting the application of this approach to only a small number of current climate models.
Studying individual cloud types is important, as they each have unique radiative and hydrological characteristics (Tselioudis et al. 2013; Oreopoulos et al. 2014). The purpose of this study is to document the relationship between cloud controlling factors and the ISCCP weather states in the observational record, identifying the large-scale meteorological factors associated with individual cloud types and the transitions among them. These results will serve as a benchmark for future studies seeking to apply the cloud controlling factor framework to study individual cloud types, allowing for alternative estimates of future changes in individual cloud types that do not rely on model cloud parameterizations. To date, several studies have classified cloud type based on climatological regimes of vertical velocity and lower-tropospheric stability, showing that annual-mean values of these two cloud controlling factors can be used to explain the climatological distribution of deep convective, midlatitude, trade cumulus, and stratocumulus clouds across global oceans (Medeiros and Stevens 2011; Scott et al. 2020; Myers et al. 2021). However, these studies did not examine the more detailed cloud types defined by the ISCCP weather states, nor did they address the relationship between temporal variations in individual cloud types and large-scale dynamical variables.
The study is organized as follows. Section 2 provides an overview of the data and methods. Section 3 reviews the climatology of the ISCCP weather states and cloud controlling factors and then assesses the relationship between the weather states and cloud controlling factors in the climatology. Section 4 subsequently assesses the relationship between day-to-day variability in the ISCCP weather states and the cloud controlling factors. Section 5 concludes with a summary and discussion.
2. Data and methods
For this study, we categorize cloud types using the ISCCP-H weather states of Tselioudis et al. (2021), which are freely available online (NASA 2021). To define these categories, Tselioudis et al. (2021) applied a K-means clustering algorithm to three-hourly cloud optical depth–cloud top pressure histograms (daylight periods only) from the most recent version of ISCCP (ISCCP-H; Young et al. 2018). Initially, ten different cloud clusters were found, but two similar cirrus and stratocumulus categories (with differences in optical depth) were each merged to yield a total of eight weather states (nine with clear sky). The weather states are available at three-hourly resolution on a 1° longitude × 1° latitude global grid, but in this study, we only present statistics on weather state frequency for daily or longer time scales. We focus our analysis on the 1984–2016 period, which are the full calendar years incorporated by the ISCCP-H weather state dataset.
As this study represents our first attempt at examining the relationship between the ISCCP weather states and cloud controlling factors, we choose to focus on cloud controlling factors that are well established in the literature (see Introduction for detailed review). The six cloud controlling factors chosen here are very similar to those chosen by Scott et al. (2020), with several key exceptions. First, we use vertical velocity at 500 hPa instead of 700 hPa, as we are examining mid- and upper-tropospheric cloud types in addition to low clouds in this study. Second, we do not assume a surface RH of 80% to calculate EIS; using the actual surface dewpoint to calculate EIS allows for a better discrimination between the shallow cumulus and stratocumulus weather states in our results. Finally, we follow the method of Seethala et al. (2015) and use the 925-hPa wind instead of the 10-m wind to calculate TADV.
We identify midlatitude cyclone centers using the feature tracking algorithm of Hodges (1994, 1995, 1999). As in Roebber et al. (2023), we apply the tracking algorithm to the ERA-5 reanalysis lower-tropospheric relative vorticity field and retain all cyclones that have a duration of 48 h or more, traverse at least 1000 km in distance, and achieve a minimum peak cyclonic vorticity of 3.0 × 10−5 s−1 (see Grise et al. 2013 for full methodological details). Here, we only examine cyclones at the time of peak vorticity to construct composites about the cyclone center.
3. Climatology
We begin by reviewing the annual-mean climatology of the six cloud controlling factors in Fig. 1, reproducing the results from Fig. 2 of Scott et al. (2020). Results are shown here only over global ocean regions, where the cloud controlling factor framework is most applicable (as additional factors would be relevant over land). The climatology of 500-hPa vertical velocity is characterized by ascending motion (ω < 0) over the western Pacific warm pool, ITCZ, SPCZ, and midlatitude storm track regions and descending motion (ω > 0) over eastern subtropical ocean basins (Fig. 1a). The climatology of 700-hPa RH largely mirrors that of the vertical velocity field, with higher RH in ascending regions (Fig. 1e). The climatology of EIS is characterized by negative values in the deep tropics and positive values in the eastern subtropical ocean basins and in the extratropics (Fig. 1b). Finally, SST maximizes in the tropics in the western Pacific warm pool (Fig. 1d), whereas TADV and surface wind speed maximize in the midlatitude storm tracks (Figs. 1c,f).
Annual-mean climatology of six cloud controlling factors over the period 1984–2016: (a) 500-hPa vertical velocity (ω < 0: ascending motion), (b) EIS, (c) TADV, (d) SST, (e) 700-hPa RH, and (f) surface (10-m) wind speed.
Citation: Journal of Climate 37, 20; 10.1175/JCLI-D-24-0011.1
We next review the climatology of the nine ISCCP-H weather states of Tselioudis et al. (2021). Figure 2 shows the average annual frequency of occurrence of the nine weather states, reproducing the results from Fig. 1 of Tselioudis et al. (2021). Weather state 1 (WS1; optically thick tropical deep convective and anvil clouds) and weather state 3 (WS3; optically thin cirrus clouds) dominate in tropical ascent regions and also occur to a lesser degree in the entrance regions of the midlatitude storm tracks (Figs. 2a,c). In contrast, weather state 2 (WS2; optically thick midlatitude storm clouds) and weather state 5 (WS5; optically thick and nearly overcast middle-top clouds) are most frequent in the midlatitude storm tracks (Figs. 2b,e). Weather state 7 (WS7; shallow cumulus clouds) and weather state 8 (WS8; stratocumulus clouds, which are optically thicker and have a higher cloud top than WS7) occur both in eastern subtropical ocean basins and in midlatitudes (Figs. 2g,h). Weather state 6 (WS6; fair-weather scattered thin cumulus and cirrus clouds) is the most frequent cloud type over global oceans, maximizing in occurrence in the descending regions in the central subtropical Pacific and Atlantic Oceans (Fig. 2f), whereas weather state 4 (WS4; polar clouds) and weather state 9 (WS9; clear sky) are the least frequent regimes (Figs. 2d,i).
Average annual frequency of occurrence (%) of the ISCCP-H weather states over the period 1984–2016.
Citation: Journal of Climate 37, 20; 10.1175/JCLI-D-24-0011.1
Scott et al. (2020) and Myers et al. (2021) suggest that the climatological distribution of global cloud types (as shown in Fig. 2) can be largely determined by the climatological values of vertical velocity and EIS. Consistent with these studies, Figs. 1 and 2 show that 1) deep convective (WS1) and cirrus (WS3) clouds are most frequent in tropical ascent regimes and 2) stratocumulus (WS8) clouds are most frequent in regimes with strong subsidence and large EIS, which occur in eastern subtropical ocean basins. However, the trade cumulus regime of Scott et al. (2020) and Myers et al. (2021), which is defined by weak subsidence and small values of EIS (e.g., in the central subtropical ocean basins), is dominated by fair weather (WS6), and shallow cumulus (WS7) is relatively infrequent in these regions (Fig. 2g). Furthermore, the midlatitude cloud regime of Scott et al. (2020) and Myers et al. (2021) includes not only midlatitude storm clouds (WS2) but also shallow cumulus and stratocumulus at the equatorward fringes of the storm tracks and middle-top clouds (WS5) at the poleward fringes of the storm tracks. Therefore, while the classification scheme of Scott et al. (2020) and Myers et al. (2021) is useful, it cannot explain the complete distribution of global cloud types shown in Fig. 2.
To investigate the large-scale meteorological factors associated with the cloud types shown in Fig. 2, Fig. 3 displays the most common ISCCP weather state as a function of daily values of each of the six cloud controlling factors. Many of the cloud controlling factors vary strongly in magnitude with latitude (Fig. 1), so the analysis in Fig. 3 is separated into 10° latitude bands from 70°S to 70°N to ensure that the results are not simply aliasing north–south variation. Since fair weather (WS6) is globally the most common weather state and dominates most dynamical regimes, we note it on the figure using a small light blue square when it is the most frequent and we use the color scale to denote the second most frequent weather state in that regime. To focus on the six cloud regimes that more strongly vary with large-scale dynamics, we similarly denote the prevalence of polar (WS4) and clear-sky (WS9) regimes with small purple and black squares. As shown in Fig. 3, WS6 dominates most dynamical regimes with the notable exception of ascending regimes at all latitudes (Fig. 3a), high (>75%) RH regimes at most latitudes (Fig. 3e), and strong positive or negative TADV regimes in the extratropics (Fig. 3c).
Colored shading shows most commonly occurring ISCCP-H weather state as a function of latitude and cloud controlling factor. Only WSs 1, 2, 3, 5, 7, and 8 are considered for this analysis. White color corresponds to latitude–cloud controlling factor combinations with insufficient sampling (≤50 daily observations). If WS 4, 6, or 9 is the most commonly occurring in a particular dynamical regime, we denote this using a small square of the corresponding color (purple, light blue, or black). Dynamical regimes with variable colors of shading indicate that there is not one prevailing weather state in that regime and that two or more weather states have similar frequencies of occurrence.
Citation: Journal of Climate 37, 20; 10.1175/JCLI-D-24-0011.1
In ascending regimes, a simple picture emerges with deep convective clouds (WS1) most frequent in the tropics and subtropics and midlatitude storm clouds (WS2) most frequent in middle and high latitudes (Fig. 3a). In descending regimes, the picture is more complicated. Excluding WS6, which dominates at all latitudes, shallow cumulus (WS7) is most frequent in the NH subtropics and midlatitudes and stratocumulus (WS8) is most frequent in the SH subtropics and midlatitudes. At higher latitudes in the SH, midlevel clouds (WS5) and cirrus (WS3) become more dominant.
Two different cloud transitions are apparent when examining the weather states as a function of EIS (Fig. 3b). First, at tropical and subtropical latitudes, cirrus (WS3) is prevalent at negative values of EIS, with a transition to stratocumulus (WS8) for positive values of EIS and midlevel clouds (WS5) for the largest positive values of EIS (∼15 K). Note, however, that the prominence of midlevel clouds for the largest values of EIS in the subtropics likely reflects ISCCP’s known bias in assigning low clouds as midlevel clouds in regions with strong inversions (Garay et al. 2008; see also Fig. 2e). Second, at middle and high latitudes, midlatitude storm clouds (WS2) dominate for most values of EIS, except at subpolar latitudes where cirrus (WS3) or polar (WS4) clouds dominate for large positive values of EIS. Similarly, cloud transitions are different between the tropics and extratropics when examining weather states as a function of TADV (Fig. 3c). At tropical and subtropical latitudes, the transition from cold to warm advection is characterized by a change from stratocumulus (WS8) to deep convective (WS1) clouds, while in the extratropics, it is characterized by a change from midlevel (WS5) to midlatitude storm (WS2) clouds.
In terms of SST, at tropical and subtropical latitudes, stratocumulus (WS8) dominates for relatively cool SSTs (<295 K), with cirrus (WS3) becoming more prominent in warmer SST regimes (Fig. 3d). At middle and high latitudes, WS2 is prevalent in nearly all SST regimes. In terms of 700-hPa RH, stratocumulus (WS8) is prevalent at nearly all latitudes in low RH regimes, except in the tropics where cirrus (WS3) is more dominant (Fig. 3e). As RH increases, WS8 transitions to WS2 in the extratropics when RH exceeds 30%–40% and WS3 transitions to WS1 in the tropics when RH exceeds 75%. Finally, in terms of surface wind speed, WS2 is prevalent at almost all values of surface wind speed at middle and high latitudes (Fig. 3f). At lower latitudes, cirrus (WS3) is more frequent at low wind speeds, stratocumulus (WS8) at moderate wind speeds, and deep convective (WS1) clouds at high wind speeds.
Further insight into the relationship between the cloud controlling factors and the ISCCP weather states can be gained by examining the most common ISCCP weather states as a function of the concurrent magnitudes of multiple cloud controlling factors. Selected examples from this analysis are shown in Fig. 4, with the results for the remaining combinations of cloud controlling factors provided for reference in the supplemental material (Figs. S1 and S2 in the online supplemental material). The results in Fig. 4 encompass the entire 70°S–70°N global ocean domain, but the plots are qualitatively similar when the analysis domain is confined to a subset of latitudes (not shown).
Colored shading shows most commonly occurring ISCCP-H weather state as a function of concurrent magnitudes of two cloud controlling factors over the 70°S–70°N domain. Only WSs 1, 2, 3, 5, 7, and 8 are considered for this analysis. White color corresponds to cloud controlling factor combinations with insufficient sampling (≤50 daily observations). If WS 4, 6, or 9 is the most commonly occurring in a particular dynamical regime, we denote this using a small square of the corresponding color (purple, light blue, or black). Dynamical regimes with variable colors of shading indicate that there is not one prevailing weather state in that regime and that two or more weather states have similar frequencies of occurrence.
Citation: Journal of Climate 37, 20; 10.1175/JCLI-D-24-0011.1
In ascending regimes, the low-to-high latitude transition from deep convective (WS1) to midlatitude storm (WS2) clouds found in Fig. 3a is associated with a transition from negative to positive EIS values. However, the meridional variation in the climatological values of the cloud controlling factors is aliased into the results, so the results in Fig. 4a may merely reflect the fact that WS1 is common in the tropics, where the climatological values of EIS are negative, and WS2 is common in the extratropics, where the climatological values of EIS are positive (Fig. 1b). Additionally, WS1 is most common in ascending regions characterized by weakly-to-moderately positive TADV, whereas WS2 occurs in ascending regimes with both warm and cold advection (Fig. 4c).
In descending regimes, excluding fair weather (WS6) occurrences, cirrus (WS3) dominates for negative EIS values, which are common in the tropics and subtropics, and very large positive EIS values, which are common in subpolar regions (Fig. 3b). For low-to-moderately positive EIS values, we see a systematic stratocumulus (WS8) to shallow cumulus (WS7) transition: stratocumulus occur in weaker descending regimes with larger values of EIS and shallow cumulus occur in stronger descending regimes with smaller values of EIS. At slightly higher values of EIS (but less than those associated with WS3), midlevel clouds (WS5) are also present in descending regimes. The presence of WS5 in descending regimes is further clarified when the effects of temperature advection are included. Figure 4c reveals that WS5 dominates for large values of cold advection (<−10 K day−1), whereas WS7 and WS8 are more prominent for smaller values of cold advection. However, WS5 can also dominate in regimes with weaker cold advection if the concurrent value of EIS is very large (∼15 K; Fig. 4d), such as occurring in the subtropics (Fig. 3b) where SSTs are relatively warm (Fig. 4b). Again though, note that ISCCP has a known bias in assigning subtropical low clouds as midlevel clouds in regions with strong inversions (Garay et al. 2008).
In summary, Figs. 3 and 4 provide an overall picture for climatological cloud type transitions in dynamical regimes defined by cloud controlling factors. Ascending regimes are dominated by the transition from deep convective (WS1) clouds in the tropics and subtropics to midlatitude storm (WS2) clouds at higher latitudes (Fig. 3a). For descending regimes, the picture is more complex. Shallow cumulus (WS7) is more frequent in regimes characterized by weakly positive values of EIS and strong subsidence (Fig. 4a). Stratocumulus (WS8) is more frequent in regimes with larger values of EIS, weaker subsidence, and relatively weak cold advection (Figs. 4a,c). Finally, midlevel clouds (WS5) are prominent in descending regimes with strong cold advection (Fig. 4c), such as occurring at midlatitudes (Fig. 3c).
The results in Figs. 3 and 4 suggest that climatological maps of the cloud controlling factors can be used to partition geographic regimes where a particular cloud type is most likely to occur. Future work should use these results to further refine the global cloud cover classification scheme of Scott et al. (2020) and Myers et al. (2021). A starting framework for such future work is explored in Fig. S3, which shows a global map of the most frequently occurring weather state and, then using Fig. 4a, a prediction of the most frequently occurring weather state based on the climatological values of vertical velocity and EIS at each global ocean location. The prediction based on Fig. 4a distinguishes more detail in subtropical weather state transitions than the classification scheme of Scott et al. (2020) and Myers et al. (2021) but still lacks detail in the extratropical weather state distributions. Figure S3 illustrates that cloud-type classifications based on vertical velocity and EIS alone are tailored for subtropical cloud transitions, and thus a global prediction of climatological weather states from cloud controlling factors likely requires the use of a larger set of parameters. However, it must be emphasized here that examining relationships between the cloud controlling factors and ISCCP weather states based on climatological values alone can be problematic. The magnitudes of the cloud controlling factors and the frequencies of the ISCCP weather states may covary geographically or seasonally, but such covariations may reflect coincidence rather than causality. Therefore, in the next section, we turn our attention to the high-frequency temporal covariability between the ISCCP weather states and cloud controlling factors.
4. Day-to-day variability
In this section, we investigate the relationship between temporal variability of each of the ISCCP-H weather states and the six cloud controlling factors. To conduct this analysis, we subtract the mean value for each calendar day from the daily time series of the cloud controlling factors and the ISCCP weather state occurrence frequencies at each global ocean grid point, such that the climatological seasonal cycle does not alias the results. Then, we perform a multiple linear regression analysis separately at each global ocean grid point, where the anomalous daily frequency of occurrence of a weather state is the predictand and the six cloud controlling factors are the predictors.
The results of this analysis are displayed in Figs. 5 and 6, which show the anomalous daily frequency of occurrence of each of the weather states associated with a one standard deviation increase in each of the six cloud controlling factors. Note that results are not shown for WS4, as this weather state occurs very infrequently at nonpolar latitudes and shows little temporal covariability with the cloud controlling factors. Overall, the patterns shown in Figs. 5 and 6 are relatively insensitive to different formulations of the regression model, such as using fewer predictors or using ridge regression (not shown). The most notable exception is the sensitivity of the weather states to EIS in regions of the tropics where deep convection is dominant. In strongly ascending regimes, EIS, which was designed for low cloud environments, is not a reliable predictor, and the regression coefficient for EIS should be viewed with caution in these conditions.
Anomalous daily frequency of occurrence (%) in ISCCP-H WSs 1, 2, 3, and 5 with a one standard deviation increase in each cloud controlling factor. Daily frequency of occurrence is defined as the percentage of the available daily three-hourly periods occupied by a given ISCCP weather state. Results are based on a multiple linear regression analysis with six cloud controlling factors, conducted separately at each grid point. Stippling indicates regions that are 95% statistically significant via the Student’s t test.
Citation: Journal of Climate 37, 20; 10.1175/JCLI-D-24-0011.1
We first discuss the two weather states that are associated with conditions of anomalous ascent (WS1 and WS2). In agreement with Fig. 3, the left column of Fig. 5 reveals that deep convective (WS1) clouds occur more frequently in the tropics with a one standard deviation increase in 700-hPa RH and surface wind speed and with a one standard deviation decrease in 500-hPa ω (i.e., anomalous ascent). WS1 is also more common with positive SST anomalies over the eastern tropical Pacific Ocean, consistent with an El Niño event. Outside of the tropics, WS1 can occur in midlatitude regions with anomalous ascent and warm advection, such as in the warm sector of midlatitude cyclones.
Midlatitude storm (WS2) clouds occur more frequently in the extratropics under conditions of anomalous ascent, positive EIS, warm advection, and high RH (Fig. 5, second column). To better understand these relationships, Fig. 7 shows a composite midlatitude cyclone at 50°N at its time of maximum intensity; qualitatively similar results are found for a midlatitude cyclone at 50°S (not shown). WS2 primarily occurs in the warm conveyor belt region of the cyclone, characterized by anomalous ascent, warm advection, and high RH. The similarly signed cloud controlling factor anomalies associated with WS1 and WS2 in extratropical regions are consistent with the fact that deep convective elements are often embedded within midlatitude storm clouds in the warm sector of midlatitude cyclones (see Fig. 3 of Tselioudis et al. 2021). However, WS2 can also occur under conditions of anomalous ascent, cold advection, and high RH in the comma head where the cold conveyor belt wraps around the low-pressure center (see also Fig. 4c). We note that EIS anomalies are weak and are of variable sign in the frontal region of the cyclone (Fig. 7c), suggesting that the positive relationship between EIS and WS2 in Fig. 5f is not associated with midlatitude cyclones at their time of maximum intensity.
Composites around NH midlatitude cyclone centers near 50°N (48°–52°N) at time of maximum intensity: (a) Most common ISCCP weather state, (b) 500-hPa vertical velocity anomaly, (c) EIS anomaly, (d) TADV anomaly, and (e) 700-hPa RH anomaly. In (a), WS6 is most common in all regions, where WS2 is not most common. In these regions, we display the next most common weather state (other than WS2 and WS6).
Citation: Journal of Climate 37, 20; 10.1175/JCLI-D-24-0011.1
We next discuss the five weather states that are generally associated with conditions of anomalous descent (WS3, WS5, WS6, WS7, and WS8). Cirrus (WS3) clouds occur more frequently in the tropics under conditions of anomalous descent, high RH, and weak surface wind speed (Fig. 5, third column; see also Fig. 3). On the equatorward side of the extratropics, WS3 is more common under conditions of anomalous warm advection, consistent with its presence on the equatorward side of the warm conveyor belt of midlatitude cyclones (Fig. 7a).
Mid-level (WS5) clouds occur more frequently in the subtropics and midlatitudes under conditions of anomalously positive EIS, cold advection, cold SST, high RH, and strong surface wind speed (Fig. 5, fourth column). As shown in Fig. 7, WS5 is common in the cold-air sectors of midlatitude cyclones, particularly where cold advection is strong and EIS is anomalously positive (or only weakly anomalously negative) (see also Fig. 4d). While WS5 is associated with anomalous descent over most regions, it is associated with anomalous ascent over eastern subtropical ocean basins (Fig. 5d). It is important to note that the cloud controlling factor anomalies associated with WS5 may be combining the dynamical conditions for multiple cloud morphologies. WS5 can indicate not only clouds with tops in the midtroposphere but also cirrus clouds overlying low-level clouds, as both conditions produce a midtropospheric infrared signature (Jin and Rossow 1997). Furthermore, as discussed above, ISCCP has a known bias in categorizing subtropical low-level clouds as midlevel clouds (e.g., note similar cloud controlling factor anomalies associated with WS5 and WS8 in the subtropics in Figs. 5 and 6). However, this bias is observed primarily in subtropical stratocumulus regions (Garay et al. 2008), while Fig. 2e shows that WS5 occurs predominantly in the midlatitude storm track areas. In those areas, studies using active CloudSat–CALIPSO retrievals have documented the presence of middle-top cloud decks in the cold-air sector of extratropical cyclones (e.g., Posselt et al. 2008; Govekar et al. 2011; Naud et al. 2015).
Fair weather (WS6) occurs more frequently under conditions of anomalous descent, negative EIS, low RH, and weak surface wind speed (Fig. 6, left column; see also Fig. 3). WS6 is also associated with anomalously positive SST anomalies over eastern subtropical ocean basins. The TADV anomalies associated with WS6 are variable and are of inconsistent sign globally. We note that, in contrast to WS6, clear-sky occurrences (WS9) are not associated with strong anomalies in vertical velocity. WS9 is only associated with anomalously low RH and surface wind speed in the tropics (Fig. 6, fourth column).
Of all weather states, the relationship between shallow cumulus (WS7) and the cloud controlling factors appears to be the most regionally dependent (Fig. 6, second column), reflecting the different mechanisms by which these clouds form globally. First, WS7 is associated with anomalous descent globally and low RH in nearly all regions. For all other cloud controlling factors, there is a contrast between tropical/subtropical regions and the midlatitudes, particularly in the SH. In the tropics and subtropics, WS7 is associated with anomalous negative EIS, cold advection, cool SST, low RH, and strong surface wind speed. The weakening of the boundary layer temperature inversion and reduced RH would promote increased dry air entrainment into the moist marine boundary layer, acting to break up stratocumulus clouds into shallow cumulus. However, at midlatitudes, shallow cumulus formation can also be driven by surface latent heating and boundary layer instability (Tselioudis et al. 2021), suggesting a different combination of cloud controlling factors may be in play. At NH midlatitudes, WS7 is associated with anomalous negative EIS, cold advection, and warm SST, but at SH midlatitudes, WS7 is associated with anomalous positive EIS, warm advection, warm SST, and weak surface wind speed. The differences between NH and SH midlatitudes in Fig. 6 (second column) likely reflect competing processes in different regions of a midlatitude cyclone. WS7 is common both in the dry conveyor belt region near the cyclone center characterized by strongly negative EIS anomalies and cold advection, indicative of a breakup of thicker midlevel and stratocumulus clouds (Tselioudis et al. 2021), and in the warm sector of the cyclone outside of the region of strong ascent where there is warm advection and positive EIS anomalies (Fig. 7).
Consistent with prior literature on the controlling factors on stratocumulus clouds (see Introduction), WS8 occurs more frequently under conditions of positive EIS, cold advection, and cool SST (Fig. 6, third column). In the tropics, WS8 is associated with anomalously high RH and surface wind speed. In the extratropics, WS8 is associated with anomalously low RH and cold advection, as it occurs in the cold conveyor belt region of midlatitude cyclones (Fig. 7). Across most of the world’s oceans, WS8 occurs more frequently under conditions of anomalous descent (Fig. 6c), as stratocumulus clouds require mean subsidence relative to the ascending regimes associated with deep convection or midlatitude storm clouds. However, in most eastern subtropical ocean basins where there is already strong mean subsidence (Fig. 1a), WS8 is associated with anomalous ascent, consistent with the expectation from prior literature that very strong subsidence inhibits stratocumulus (Myers and Norris 2013; see also Fig. 4a).
To summarize the results for the five weather states associated with anomalous descent, cirrus (WS3), fair weather (WS6), and shallow cumulus (WS7) are associated with negative EIS anomalies (excluding WS7 in the warm sector of midlatitude cyclones). WS3 is the only one of these three associated with anomalously high RH in the tropics, and WS7 is the only one of these three associated with anomalously strong surface wind speeds in the tropics. The connection between high surface wind speed and WS7 in the tropics is consistent with Mieslinger et al. (2019), who found that high surface wind speeds produce larger tropical shallow cumulus clouds with greater cloud amount and cloud top height (see also Naud et al. 2023). For the weather states associated with positive EIS anomalies [midlevel clouds (WS5), stratocumulus (WS8), and shallow cumulus (WS7) in the warm sector of midlatitude cyclones], only WS7 is associated with anomalous warm advection. In extratropical regions, WS5 is associated with positive RH anomalies and WS8 is associated with negative RH anomalies. In tropical and subtropical regions, WS8 is more strongly associated with anomalous descent.
One caveat to the results shown in Figs. 5 and 6 is that they were derived using daily time-scale data. A key assumption of the cloud controlling factor framework is that relationships between cloud properties and large-scale meteorological controlling factors are approximately time-scale invariant (see Klein et al. 2017 for further discussion). In other words, if this assumption is true, short-term covariability in cloud properties with the cloud controlling factors can be used to understand longer-term cloud behavior, including that associated with climate change. Previous studies have generally found that the sign of the relationship between cloud properties and large-scale meteorology does not change across time scale (Klein 1997; Tselioudis et al. 1998; de Szoeke et al. 2016; McCoy et al. 2017; Kelleher and Grise 2019), provided that the time scale is long enough for clouds to adjust to their environment, which can be up to several days for tropical low clouds (Schubert et al. 1979; Bretherton 1993). In the extratropics, there is strong evidence that cloud controlling factors can be relevant in describing subdaily or daily cloud behavior (Norris and Iacobellis 2005; Wall et al. 2017; Kelleher and Grise 2019).
In Fig. 8, we investigate whether the relationships documented in Figs. 5 and 6 are time-scale invariant. To do this, we have repeated the multiple linear regression analysis from Figs. 5 and 6, except now with the ISCCP weather state frequencies bandpass filtered into four temporal bands: 1–3 days, 3–10 days, 10–30 days, and 30+ days. Results are shown for WS1, WS3, and WS9 for the tropical western Pacific warm pool region (top row); WS6, WS7, and WS8 for the SH subtropics (middle row); and WS2, WS5, and WS7 for the SH midlatitudes (bottom row). Note that, in all cases, the summation of the regression coefficients from all four temporal bands is approximately equal to the results shown in Figs. 5 and 6.
Anomalous daily frequency of occurrence (%) in the indicated ISCCP-H weather states with a one standard deviation increase in each cloud controlling factor (as in Figs. 5 and 6). Results are shown for averages over three regions: (top) warm pool (25°S–25°N, 90°E–180°), (middle) SH subtropics (45°–25°S), and (bottom) SH midlatitudes (60°–45°S). For each cloud controlling factor, four colored bars are shown, reflecting the magnitude of regression coefficients that have been calculated separately for the time series separated into four time scales using a bandpass filter: 1–3 days, 3–10 days, 10–30 days, and 30+ days.
Citation: Journal of Climate 37, 20; 10.1175/JCLI-D-24-0011.1
Overall, the results in Fig. 8 reveal that the relationships identified in Figs. 5 and 6 persist across all four temporal bands. In general, the cloud controlling factors have the most impact on cloud frequency on the 3–10-day and 10–30-day time scales, with shorter time scales dominating more in the extratropics (consistent with the time scale of midlatitude synoptic-scale weather systems). The 30+ day time-scale relationships are generally weaker than those at submonthly time scales, except for SST, which unlike the atmospheric controlling factors, has most of its variance at lower frequency time scales (de Szoeke et al. 2016). Several interesting exceptions occur when the sign of the relationship notably changes with time scale. First, in the tropics, cirrus (WS3) is associated with negative SST anomalies on submonthly time scales and positive SST anomalies on 30+ day time scales (Fig. 8b). Second, in the SH subtropics, fair weather (WS6) is associated with anomalous cold advection on time scales of less than 10 days and anomalous warm advection on longer time scales (Fig. 8d). In both of these cases, the sign of the regression coefficient in Figs. 5 and 6 was unclear in these regions, which masks the large compensation between the oppositely signed relationships occurring on different temporal scales. Finally, in the SH subtropics, stratocumulus (WS8) appears to be associated with anomalous warm SST on 3–10-day time scales and anomalous cold SST on 30+ day time scales (Fig. 8f). However, in this case, the sign of the regression coefficients in Fig. 8f reflects competing signs within the chosen region (45°–25°S): a positive relationship with SST at the equatorward fringe of the midlatitude storm tracks (which occurs primarily on synoptic time scales) and a negative relationship with SST over most ocean basins equatorward of 30° latitude (which occurs primarily on 30+ day time scales) (see also Fig. 6o).
In summary, in this section, we have identified distinct relationships between the ISCCP weather states and the six cloud controlling factors using temporal covariability. With only a few exceptions, the relationships maintain the same sign across a range of time scales (Fig. 8) and generally agree with the climatological relationships identified in section 3. Consequently, our results suggest that cloud controlling factors may be able to be used to isolate dynamical processes responsible for transitions between individual cloud types over tropical and extratropical oceans, in association with both short-term climate variability and longer-term climate change.
To illustrate the effectiveness of the cloud controlling factor approach in isolating dynamical transitions among cloud types, we conclude with two examples. First, in Fig. 9, we display the daily probability of occurrence of each weather state within the western Pacific warm pool of the tropics, as a function of the concurrent values of 700-hPa RH and surface wind speed. As discussed above, these two cloud controlling factors are particularly helpful in discriminating among cloud types in the tropics (see also Scott et al. 2020). The tropical weather states with the highest probability (∼60%–80%) are deep convection (WS1) on high RH/high wind speed days and fair weather (WS6) on low RH (<50%) days, particularly with wind speeds between 5 and 10 m s−1. Cirrus (WS3) increases in likelihood with moderate-to-high RH and low wind speed conditions (Figs. 3e,f; Fig. 5, third column), such that WS1 and WS3 are equally likely on high RH/low wind speed days. At very high values of RH, midlatitude storm (WS2) and midlevel (WS5) clouds become more prevalent than WS1, likely reflecting cumulus congestus (WS5) or mesoscale convection (WS2) with larger grid point values of RH compared to smaller-scale tropical deep convective systems. On low RH days, shallow cumulus (WS7) and stratocumulus (WS8) increase in likelihood with higher values of surface wind speed (Figs. 6v,w) and occur on a comparable frequency to WS6 on low RH/high wind speed days, with WS8 occurring more commonly than WS7 at slightly higher values of RH (see also Figs. 6r,s). On days with very low values of RH and surface wind speed, clear-sky occurrences (WS9) slightly compete with WS6.
Daily probability of occurrence for each ISCCP-H weather state as a function of 700-hPa RH and surface wind speed, within the western Pacific warm pool (25°S–25°N, 90°E–180°) region. Cloud controlling factor combinations with insufficient sampling (≤50 daily observations) are omitted.
Citation: Journal of Climate 37, 20; 10.1175/JCLI-D-24-0011.1
The second example is shown in Fig. 10, which displays analogous results to Fig. 9 but for the SH subtropics and midlatitudes (25°–60°S). Qualitatively similar results are found for the NH subtropics and midlatitudes (not shown). Here, the two cloud controlling factors that are particularly helpful in discriminating among cloud types are EIS and TADV (see also Scott et al. 2020). The results in Fig. 10 are derived only from daily grid points with descending motion. Daily grid points with ascending motion are dominated by midlatitude storm clouds (WS2), except when EIS is negative; when EIS is negative, deep convection (WS1) dominates in regions of strong warm advection and fair weather (WS6) dominates elsewhere (see also Figs. 4c,d).
Daily probability of occurrence for each ISCCP-H weather state as a function of EIS and TADV, within the SH subtropics and midlatitudes (25°–60°S). Only daily grid points with descending vertical velocity (500-hPa ω > 0) are considered for this analysis. Cloud controlling factor combinations with insufficient sampling (≤50 daily observations) are omitted.
Citation: Journal of Climate 37, 20; 10.1175/JCLI-D-24-0011.1
Figure 10 shows that, in descending regions of the SH subtropics and midlatitudes, the two weather states with the highest probability are fair weather (WS6) on negative EIS days and midlevel clouds (WS5) on positive EIS days with strong cold advection. Midlevel (WS5) and stratocumulus (WS8) clouds both occur primarily on days with cold advection and positive EIS, with WS5 more favored for larger values of cold advection (Fig. 10d; see also Fig. 4d). On positive EIS/warm advection days, midlatitude storm clouds (WS2) are the most likely, with shallow cumulus (WS7) playing a secondary role. This sensitivity of WS7 to positive EIS and warm advection is consistent with the SH midlatitude shallow cumulus behavior shown in Figs. 6f and 6j, which likely arises from the warm sector of midlatitude cyclones (Fig. 7). The secondary peak in likelihood of WS7 on days with weakly positive EIS and weak cold advection reflects the different controlling factors on shallow cumulus in the SH subtropics (Figs. 6f,j) and in the dry conveyor belt region of midlatitude cyclones (Fig. 7). On negative EIS days, deep convective (WS1) and cirrus (WS3) clouds slightly compete with fair weather (WS6), with this behavior arising from tropical clouds on the equatorward margin of the SH extratropics. Cirrus (WS3) also occurs on days with weak temperature advection and very large values of EIS (>15 K), which is associated with subpolar clouds (see also Fig. 3b).
5. Summary and discussion
A detailed process-based understanding of global cloud cover is essential to the accurate modeling of clouds and their response to climate change. One approach to understand global cloud processes is through the identification of cloud controlling factors, which are large-scale meteorological variables that covary with cloud properties and serve as proxies for the smaller-scale processes responsible for cloud formation (e.g., Klein et al. 2017). By understanding observed cloud dependence on the cloud controlling factors, estimates of changes in model cloud behavior can be made by relying only on models’ large-scale meteorological fields instead of highly uncertain cloud parameterizations (Qu et al. 2015; Myers et al. 2021; Cesana and Del Genio 2021). However, the cloud controlling factor approach is generally only applied to marine low clouds, and only recent studies have begun to use it to try to discriminate between marine low cloud types (Myers et al. 2021; Cesana and Del Genio 2021).
An alternative approach to understand global cloud processes is through the clustering of global cloud optical depth and height data from satellites into distinct categories, such as the ISCCP-H weather states of Tselioudis et al. (2021). The weather state approach is useful in that it allows a direct study of individual cloud types and the meteorological phenomena associated with them, but this approach can only be applied to a small number of present-day climate models with satellite simulator cloud output. In this study, we attempt to reconcile the cloud controlling factor and weather state approaches. If robust relationships between the cloud controlling factors and the ISCCP weather states exist, this would circumvent the need for clustering algorithms and satellite simulator output in models and provide a pathway forward toward the estimation of future changes in individual weather states in models using the cloud controlling factor approach.
Our results show that, over global oceans, most of the ISCCP weather states have distinct relationships with the six cloud controlling factors examined in this study. We establish these relationships by examining both the long-term climatology (Figs. 3 and 4) and high-frequency temporal variability (Figs. 5–7) and have shown that they occur across a range of time scales (Fig. 8):
-
WS1 (deep convection) most commonly occurs in tropical regions with ascending motion, high values of relative humidity, and large surface wind speeds (Fig. 3). It also occurs to a lesser extent in midlatitude regions with warm advection (Fig. 5i), such as the warm sector of midlatitude cyclones where deep convective elements are embedded in frontal clouds.
-
WS2 (midlatitude storm clouds) occurs in extratropical regions and is associated with anomalous ascent, high values of EIS, warm advection, and high relative humidity (Fig. 5, second column). WS2 not only dominates in the warm conveyor belt region of midlatitude cyclones but also occurs in the comma head where the cold conveyor belt wraps around the low pressure center (Fig. 7a; Tselioudis et al. 2021).
-
WS3 (cirrus) most commonly occurs in tropical regions characterized by anomalous descent, moderate values of relative humidity, and weak surface wind speeds (Fig. 3; Fig. 5, third column; Fig. 9c).
-
WS4 (polar clouds) is mostly confined to the polar regions and does not have a close relationship with the cloud controlling factors.
-
WS5 (middle-top clouds) occurs in descending regions of both the subtropics and midlatitudes under conditions of anomalous positive EIS, cold advection, cold SST, high relative humidity, and high surface wind speed (Fig. 5, fourth column). In the midlatitudes, WS5 dominates in regions characterized by descent and strong cold advection (Fig. 4c), such as occurring in the cold air outbreak region of midlatitude cyclones (Fig. 7a; Tselioudis et al. 2021).
-
WS6 (fair weather) is the most frequent of all the weather states and dominates over global oceans under quiescent conditions (descent, negative EIS, low relative humidity, and low-to-moderate surface wind speed) (Fig. 3; Fig. 6, first column). Positive SST anomalies in subtropical ocean basins also favor WS6 (Fig. 6m), as they inhibit stratocumulus formation in these regions.
-
WS7 (shallow cumulus) reflects two different modes of low cloud behavior, which are both associated with anomalous descent and low RH (Fig. 6, second column). In the subtropics, WS7 is associated with relatively weak EIS, cold advection, cool SST anomalies, and stronger surface wind speed. The stronger surface wind speeds help to organize the clouds more than WS6 (Mieslinger et al. 2019; Naud et al. 2023), but the weaker EIS, lower RH, and stronger descent inhibit the formation of thicker stratocumulus clouds (Fig. 4a). In midlatitudes, WS7 similarly occurs in regions of weak EIS and cold advection (such as the dry conveyor belt region of midlatitude cyclones), but it can also be associated with stronger EIS, warm advection, warm SST anomalies, and weaker surface wind speeds, likely reflecting clouds forced by surface convection in the warm sector of midlatitude cyclones (Fig. 7a).
-
WS8 (stratocumulus) occurs in descending regions with strongly positive EIS, cold advection, cold SST anomalies in the tropics, and strong surface wind speeds (Fig. 6, third column). It is associated with higher RH in the tropics and lower RH in the extratropics.
-
WS9 (clear sky) occurs in tropical regions with low RH and surface wind speed (Fig. 6, fourth column; Fig. 9h).
As shown in Figs. 9 and 10, the cloud controlling factors can be used to partition dynamical regimes where individual ISCCP weather states are more or less likely. These relationships provide an important framework to understand the physical processes responsible for observed transitions among individual cloud types. However, we caution that it would be difficult to definitively predict the occurrence of one particular ISCCP weather state on a given day based only on the values of two cloud controlling factors, as two or more weather states often occur with similar likelihoods given the same values of pairs of cloud controlling factors (Figs. 9 and 10). Conducting a similar analysis with three or more cloud controlling factors may improve the accuracy of such quantitative predictions, but this is not the objective of the present study. Here, we have established clear relationships between the weather states and the cloud controlling factors over global oceans, and we view these relationships as tools for better process understanding rather than for quantitative predictability. Future work is needed to understand the processes governing the weather states over land, where additional controlling factors, such as surface roughness, vegetation, and soil moisture, are likely relevant (e.g., Scott et al. 2020).
In conclusion, cloud controlling factors appear to provide an effective framework to pinpoint regions prone to the development of certain cloud types and to isolate processes associated with transitions among individual cloud types. While this study has served to provide an observational benchmark for the relationships between the ISCCP weather states and cloud controlling factors, future work is necessary to determine the usefulness of this approach in climate prediction. Combined with previous cloud controlling factor studies, the relationships established here have the potential to be used to predict the changes in prevalence of different cloud types with climate variability and change based on predicted changes in the cloud controlling factors from models, which could ultimately help to improve our confidence in the simulation of model cloud climate feedbacks.
Acknowledgments.
We thank Joaquín Blanco and two anonymous reviewers for helpful comments. K. Grise acknowledges support from the National Science Foundation under Grant AGS-1752900. G. Tselioudis acknowledges support from the NASA Modeling, Analysis, and Prediction (MAP) and Making Earth System Data Records for Use in Research Environments (MEaSUREs) programs. We thank K. I. Hodges for making his tracking algorithm available to us.
Data availability statement.
All data used in this study are freely and publicly available from NASA (2021), NOAA (2021), and ECMWF (Hersbach et al. 2023). Online locations for these data products are provided in the citations to these datasets in the references section below.
REFERENCES
Blanco, J. E., R. Caballero, G. Datseris, B. Stevens, S. Bony, O. Hadas, and Y. Kaspi, 2023: A cloud-controlling factor perspective on the hemispheric asymmetry of extratropical cloud albedo. J. Climate, 36, 1793–1804, https://doi.org/10.1175/JCLI-D-22-0410.1.
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.
Bretherton, C. S., 1993: Understanding Albrecht’s model of trade cumulus cloud fields. J. Atmos. Sci., 50, 2264–2283, https://doi.org/10.1175/1520-0469(1993)050<2264:UAMOTC>2.0.CO;2.
Cesana, G. V., and A. D. Del Genio, 2021: Observational constraint on cloud feedbacks suggests moderate climate sensitivity. Nat. Climate Change, 11, 213–218, https://doi.org/10.1038/s41558-020-00970-y.
Datseris, G., J. Blanco, O. Hadas, S. Bony, R. Caballero, Y. Kaspi, and B. Stevens, 2022: Minimal recipes for global cloudiness. Geophys. Res. Lett., 49, e2022GL099678, https://doi.org/10.1029/2022GL099678.
de Szoeke, S. P., K. L. Verlinden, S. E. Yuter, and D. B. Mechem, 2016: The time scales of variability of marine low clouds. J. Climate, 29, 6463–6481, https://doi.org/10.1175/JCLI-D-15-0460.1.
Eastman, R., and R. Wood, 2018: The competing effects of stability and humidity on subtropical stratocumulus entrainment and cloud evolution from a Lagrangian perspective. J. Atmos. Sci., 75, 2563–2578, https://doi.org/10.1175/JAS-D-18-0030.1.
Frey, W. R., and J. E. Kay, 2018: The influence of extratropical cloud phase and amount feedbacks on climate sensitivity. Climate Dyn., 50, 3097–3116, https://doi.org/10.1007/s00382-017-3796-5.
Garay, M. J., S. P. de Szoeke, and C. M. Moroney, 2008: Comparison of marine stratocumulus cloud top heights in the southeastern Pacific retrieved from satellites with coincident ship-based observations. J. Geophys. Res., 113, D18204, https://doi.org/10.1029/2008JD009975.
Georgakakos, K. P., and R. L. Bras, 1984: A hydrologically useful station precipitation model: 1. Formulation. Water Resour. Res., 20, 1585–1596, https://doi.org/10.1029/WR020i011p01585.
Govekar, P. D., C. Jakob, M. J. Reeder, and J. Haynes, 2011: The three-dimensional distribution of clouds around Southern Hemisphere extratropical cyclones. Geophys. Res. Lett., 38, L21805, https://doi.org/10.1029/2011GL049091.
Grise, K. M., and M. K. Kelleher, 2021: Midlatitude cloud radiative effect sensitivity to cloud controlling factors in observations and models: Relationship with Southern Hemisphere jet shifts and climate sensitivity. J. Climate, 34, 5869–5886, https://doi.org/10.1175/JCLI-D-20-0986.1.
Grise, K. M., S.-W. Son, and J. R. Gyakum, 2013: Intraseasonal and interannual variability in North American storm tracks and its relationship to equatorial Pacific variability. Mon. Wea. Rev., 141, 3610–3625, https://doi.org/10.1175/MWR-D-12-00322.1.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Hersbach, H., and Coauthors, 2023: ERA5 hourly data on pressure levels from 1940 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS), accessed 16 January 2023, https://doi.org/10.24381/cds.bd0915c6.
Hodges, K. I., 1994: A general method for tracking analysis and its application to meteorological data. Mon. Wea. Rev., 122, 2573–2586, https://doi.org/10.1175/1520-0493(1994)122<2573:AGMFTA>2.0.CO;2.
Hodges, K. I., 1995: Feature tracking on the unit sphere. Mon. Wea. Rev., 123, 3458–3465, https://doi.org/10.1175/1520-0493(1995)123<3458:FTOTUS>2.0.CO;2.
Hodges, K. I., 1999: Adaptive constraints for feature tracking. Mon. Wea. Rev., 127, 1362–1373, https://doi.org/10.1175/1520-0493(1999)127<1362:ACFFT>2.0.CO;2.
Huang, B., C. Liu, V. Banzon, E. Freeman, G. Graham, B. Hankins, T. Smith, and H.-M. Zhang, 2021: Improvements of the Daily Optimum Interpolation Sea Surface Temperature (DOISST) version 2.1. J. Climate, 34, 2923–2939, https://doi.org/10.1175/JCLI-D-20-0166.1.
Jakob, C., and G. Tselioudis, 2003: Objective identification of cloud regimes in the tropical western Pacific. Geophys. Res. Lett., 30, 2082, https://doi.org/10.1029/2003GL018367.
Jin, Y., and W. B. Rossow, 1997: Detection of cirrus overlapping low-level clouds. J. Geophys. Res., 102, 1727–1737, https://doi.org/10.1029/96JD02996.
Kawai, H., T. Koshiro, and M. J. Webb, 2017: Interpretation of factors controlling low cloud cover and low cloud feedback using a unified predictive index. J. Climate, 30, 9119–9131, https://doi.org/10.1175/JCLI-D-16-0825.1.
Kazil, J., G. Feingold, and T. Yamaguchi, 2016: Wind speed response of marine non-precipitating stratocumulus clouds over a diurnal cycle in cloud-system resolving simulations. Atmos. Chem. Phys., 16, 5811–5839, https://doi.org/10.5194/acp-16-5811-2016.
Kelleher, M. K., and K. M. Grise, 2019: Examining Southern Ocean cloud controlling factors on daily time scales and their connections to midlatitude weather systems. J. Climate, 32, 5145–5160, https://doi.org/10.1175/JCLI-D-18-0840.1.
Klein, S. A., 1997: Synoptic variability of low-cloud properties and meteorological parameters in the subtropical trade wind boundary layer. J. Climate, 10, 2018–2039, https://doi.org/10.1175/1520-0442(1997)010<2018:SVOLCP>2.0.CO;2.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6, 1587–1606, https://doi.org/10.1175/1520-0442(1993)006<1587:TSCOLS>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.
Lau, N.-G., and M. W. Crane, 1995: A satellite view of the synoptic-scale organization of cloud properties in midlatitude and tropical circulation systems. Mon. Wea. Rev., 123, 1984–2006, https://doi.org/10.1175/1520-0493(1995)123<1984:ASVOTS>2.0.CO;2.
Li, Y., D. W. J. Thompson, G. L. Stephens, and S. Bony, 2014: A global survey of the instantaneous linkages between cloud vertical structure and large-scale climate. J. Geophys. Res. Atmos., 119, 3770–3792, https://doi.org/10.1002/2013JD020669.
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.
Medeiros, B., and B. Stevens, 2011: Revealing differences in GCM representations of low clouds. Climate Dyn., 36, 385–399, https://doi.org/10.1007/s00382-009-0694-5.
Mieslinger, T., Á. Horváth, S. A. Buehler, and M. Sakradzija, 2019: The dependence of shallow cumulus macrophysical properties on large-scale meteorology as observed in ASTER imagery. J. Geophys. Res. Atmos., 124, 11 477–11 505, https://doi.org/10.1029/2019JD030768.
Myers, T. A., and J. R. Norris, 2013: Observational evidence that enhanced subsidence reduces subtropical marine boundary layer cloudiness. J. Climate, 26, 7507–7524, https://doi.org/10.1175/JCLI-D-12-00736.1.
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., R. C. Scott, M. D. Zelinka, S. A. Klein, J. R. Norris, and P. M. Caldwell, 2021: Observational constraints on low cloud feedback reduce uncertainty of climate sensitivity. Nat. Climate Change, 11, 501–507, https://doi.org/10.1038/s41558-021-01039-0.
NASA, 2021: Global weather states from ISCCP HGG Cloud Dataset. NASA Goddard Institute for Space Studies, accessed 31 March 2022, https://isccp.giss.nasa.gov/outgoing/HGGWS/.
Naud, C. M., D. J. Posselt, and S. C. van den Heever, 2015: A CloudSat–CALIPSO view of cloud and precipitation properties across cold fronts over the global oceans. J. Climate, 28, 6743–6762, https://doi.org/10.1175/JCLI-D-15-0052.1.
Naud, C. M., G. S. Elsaesser, and J. F. Booth, 2023: Dominant cloud controlling factors for low-level cloud fraction: Subtropical versus extratropical oceans. Geophys. Res. Lett., 50, e2023GL104496, https://doi.org/10.1029/2023GL104496.
NOAA, 2021: Daily Optimum Interpolation Sea Surface Temperature (DOISST) version 2.1. NOAA/Physical Sciences Library, accessed 3 January 2023, https://www.psl.noaa.gov/data/gridded/data.noaa.oisst.v2.highres.html.
Norris, J. R., and S. F. Iacobellis, 2005: North Pacific cloud feedbacks inferred from synoptic-scale dynamic and thermodynamic relationships. J. Climate, 18, 4862–4878, https://doi.org/10.1175/JCLI3558.1.
Nuijens, L., and B. Stevens, 2012: The influence of wind speed on shallow marine cumulus convection. J. Atmos. Sci., 69, 168–184, https://doi.org/10.1175/JAS-D-11-02.1.
Oreopoulos, L., N. Cho, D. Lee, S. Kato, and G. J. Huffman, 2014: An examination of the nature of global MODIS cloud regimes. J. Geophys. Res. Atmos., 119, 8362–8383, https://doi.org/10.1002/2013JD021409.
Posselt, D. J., G. L. Stephens, and M. Miller, 2008: CLOUDSAT: Adding a new dimension to a classical view of extratropical cyclones. Bull. Amer. Meteor. Soc., 89, 599–610, https://doi.org/10.1175/BAMS-89-5-599.
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 A. M. 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.
Rieck, M., L. Nuijens, and B. Stevens, 2012: Marine boundary layer cloud feedbacks in a constant relative humidity atmosphere. J. Atmos. Sci., 69, 2538–2550, https://doi.org/10.1175/JAS-D-11-0203.1.
Roebber, P. J., K. M. Grise, and J. R. Gyakum, 2023: The histories of well-documented maritime cyclones as portrayed by an automated tracking method. Mon. Wea. Rev., 151, 2905–2924, https://doi.org/10.1175/MWR-D-22-0287.1.
Rossow, W. B., G. Tselioudis, A. Polak, and C. Jakob, 2005: Tropical climate described as a distribution of weather states indicated by distinct mesoscale cloud property mixtures. Geophys. Res. Lett., 32, L21812, https://doi.org/10.1029/2005GL024584.
Schubert, W. H., J. S. Wakefield, E. J. Steiner, and S. K. Cox, 1979: Marine stratocumulus convection. Part II: Horizontally inhomogeneous solutions. J. Atmos. Sci., 36, 1308–1324, https://doi.org/10.1175/1520-0469(1979)036<1308:MSCPIH>2.0.CO;2.
Scott, R. C., T. A. Myers, J. R. Norris, M. D. Zelinka, S. A. Klein, M. Sun, and D. R. Doelling, 2020: Observed sensitivity of low-cloud radiative effects to meteorological perturbations over the global oceans. J. Climate, 33, 7717–7734, https://doi.org/10.1175/JCLI-D-19-1028.1.
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.
Stevens, B., and J.-L. Brenguier, 2009: Cloud-controlling factors: Low clouds. Clouds in the Perturbed Climate System: Their Relationship to Energy Balance, Atmospheric Dynamics, and Precipitation, J. Heintzenberg and R. J. Charlson, Eds., MIT Press, 173–196.
Tselioudis, G., A. D. DelGenio, W. Kovari Jr., and M.-S. Yao, 1998: Temperature dependence of low cloud optical thickness in the GISS GCM: Contributing mechanisms and climate implications. J. Climate, 11, 3268–3281, https://doi.org/10.1175/1520-0442(1998)011<3268:TDOLCO>2.0.CO;2.
Tselioudis, G., W. Rossow, Y. Zhang, and D. Konsta, 2013: Global weather states and their properties from passive and active satellite cloud retrievals. J. Climate, 26, 7734–7746, https://doi.org/10.1175/JCLI-D-13-00024.1.
Tselioudis, G., W. B. Rossow, C. Jakob, J. Remillard, D. Tropf, and Y. Zhang, 2021: Evaluation of clouds, radiation, and precipitation in CMIP6 models using global weather states derived from ISCCP-H cloud property data. J. Climate, 34, 7311–7324, https://doi.org/10.1175/JCLI-D-21-0076.1.
Wall, C. J., D. L. Hartmann, and P.-L. Ma, 2017: Instantaneous linkages between clouds and large-scale meteorology over the Southern Ocean in observations and a climate model. J. Climate, 30, 9455–9474, https://doi.org/10.1175/JCLI-D-17-0156.1.
Webb, M. J., F. H. Lambert, and J. M. Gregory, 2013: Origins of differences in climate sensitivity, forcing and feedback in climate models. Climate Dyn., 40, 677–707, https://doi.org/10.1007/s00382-012-1336-x.
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.
Young, A. H., K. R. Knapp, A. Inamdar, W. Hankins, and W. B. Rossow, 2018: The international satellite cloud climatology project H-series climate data record product. Earth Syst. Sci. Data, 10, 583–593, https://doi.org/10.5194/essd-10-583-2018.
Zelinka, M. D., K. M. Grise, S. A. Klein, C. Zhou, A. M. DeAngelis, and M. W. Christensen, 2018: Drivers of the low-cloud response to poleward jet shifts in the North Pacific in observations and models. J. Climate, 31, 7925–7947, https://doi.org/10.1175/JCLI-D-18-0114.1.
Zelinka, M. D., T. A. Myers, D. T. McCoy, S. Po‐Chedley, P. M. Caldwell, P. Ceppi, S. A. Klein, and K. E. Taylor, 2020: Causes of higher climate sensitivity in CMIP6 models. Geophys. Res. Lett., 47, e2019GL085782, https://doi.org/10.1029/2019GL085782.
