1. Introduction
Clouds have long been known to be a significant component of the climate system, not only because of their role in both the hydrological cycle and large-scale circulation, but also because of their impact in modulating incoming and outgoing radiation. Various cloud types with different altitudes (and thus cloud-top temperatures) and optical thicknesses have a spectrum of effects on radiation (Hartmann et al. 1992; Chen et al. 2000; Hartmann et al. 2001; Kubar et al. 2007). Top-of-atmosphere (TOA) cloud radiative forcing for individual cloud types has been calculated by considering 100% cloud cover with a particular cloud-top temperature and optical depth (τ), as in Kubar et al. (2007). The shortwave (SW) component, which induces TOA cooling, is primarily dependent on τ, while the longwave (LW) component is primarily a function of cloud-top temperature and induces TOA warming. High, thin clouds with cold temperatures relative to Earth’s surface have a net TOA warming effect due to their weak SW effect and strong LW effect, whereas moderately thick low-topped clouds have a net TOA cooling effect owing to their weak LW effect compared to their SW effect. Interestingly, although deep convective clouds have strong LW and SW effects that tend to nearly cancel each other (Hartmann et al. 2001; Ramanathan et al. 1989; Harrison et al. 1990), TOA radiative changes could still be important if their coverage, optical properties, or depth were to change.
The launch and operation of the millimeter-wavelength cloud-profiling radar (CPR) (Im et al. 2005) on CloudSat (Stephens et al. 2002) and the Cloud–Aerosol Lidar with Orthogonal Polarization (CALIOP) (Winker et al. 2007) on Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) occurred in late April 2006, and has offered unprecedented opportunities for global studies of hydrometeors and the vertical structure of cloud systems. CloudSat and CALIPSO are part of the A-Train satellite constellation with two equatorial passage times of approximately 0130 and 1330 local time. The CloudSat CPR is the first satellite-borne cloud radar (sun-synchronous), with an operational frequency of 94 GHz, for which backscatter from clouds can be measured. Its horizontal resolution of 2.5 km along track by 1.4 km across track, along with its effective vertical resolution of 240 m, provides a small spatial footprint and good vertical resolution, but only for a nadir curtain. CALIPSO combines an active lidar instrument with passive infrared and visible imagers to probe the vertical structure and properties of thin clouds and aerosols over the globe. The CALIOP of CALIPSO is sensitive to the presence of thin cloud layers, and its small footprint (as little as 333 m) is useful to identify small boundary layer clouds, especially in the absence of optically thick, high clouds.
Using the effective radar reflectivity factor Ze derived from the CPR, the inferred presence of precipitation, and ancillary data such as surface elevation and model predicted temperature profiles, Sassen and Wang (2008) developed an algorithm to classify clouds into eight types: high clouds (High), altostratus (As), altocumulus (Ac), stratus (St), stratocumulus (Sc), cumulus (Cu), nimbostratus (Ns), and deep convective (Dc). These high clouds include both cirrus and cirrostratus. The CloudSat cloud classification product based on this algorithm is called 2B-CLDCLASS and is available through the CloudSat website (http://www.cloudsat.cira.colostate.edu). Through a comparison of one year of zonally averaged 2B-CLDCLASS (version 5) cloud type frequency with long-term International Satellite Cloud Climatology Project (ISCCP) annual means and surface observer reports, Sassen and Wang (2008) showed overall consistency among the cloud classification records. However, as shown by them, the CloudSat algorithm identifies considerably fewer high clouds over land as well as fewer Cu and Ac clouds over land and ocean. Many reasons could contribute to the observed differences, including CPR surface return contamination for the first three or four bins (~1.0 km) above the surface as well as the known basic sensitivity limitation of CloudSat to physically detect optically thin clouds (e.g., Marchand et al. 2008).
We have used the 2B-CLDCLASS product to study the regional distribution of clouds over tropical oceans and the relations between various cloud types and SST. While a number of analyses have focused on relations between marine stratus, stratocumulus, or convective clouds and environmental factors such as SST, other cloud types have received much less attention. Using long-term surface observations, Norris and Leovy (1994) and Norris (1998a,b) examined interannual variations in marine stratiform cloudiness (MSC) and SST. They found extensive areas of negative correlations between anomalies in MSC amount and SST, mainly in northern midlatitude oceans, the eastern subtropical oceans, and the eastern tropical Pacific. They also found a negative correlation between SST and Ns and between SST and nonprecipitating midlevel clouds, particularly during summer and in the North Pacific and Atlantic Oceans. Other studies, such as Oreopoulos and Davies (1993), have focused on two primary subtropical stratus/stratocumulus regimes in the Southern Hemisphere, one off the west coast of Peru and the other off the west coast of Angola, and quantified negative correlations between low cloud albedo (and more weakly cloud fraction) and SST. In addition to SST, the effect of lower tropospheric stability (LTS) or other stability parameters on low-topped cloud fraction has been a subject of several studies (Klein and Hartmann 1993; Wood and Hartmann 2006; Clement et al. 2009; Kubar et al. 2011). These studies demonstrate that higher stability, which is linked to inversion frequency and strength, is well correlated with greater low-topped cloud coverage. Based on multidecadal observations by Clement et al. (2009) over the northeast Pacific, LTS and SST are inversely related, such that greater cloud coverage is geographically located over lower SST regions, with anticorrelations discerned on interannual time scales. The inverse relationship between SST and cloud fraction is most likely linked to the transition from largely homogeneous stratiform to more patchy and heterogeneous cumulus clouds as a result of decreased static stability and increased entrainment, such as via the “deepening–warming mechanism” as posited by Bretherton and Wyant (1997). In a recent study, Eitzen et al. (2011) analyzed Clouds and the Earth’s Radiant Energy System (CERES)–Terra and European Centre for Medium-Range Weather Forecasts (ECMWF) data to investigate the low-topped cloud physical and optical property changes with SST. Key findings include decreases in the low-cloud amount and the logarithm of low-cloud optical depth, and an increasingly less negative cloud radiative effect with increasing SST anomalies. Furthermore, the estimated inversion strength decreases with SST, explaining a significant percent of the measured cloud fraction and optical depth.
The relation between convective clouds and SST has also been studied fairly extensively over a wide range of interests. Many studies have concentrated on understanding mechanisms that regulate tropical temperature (e.g., Ramanathan and Collins 1991; Waliser and Graham 1993; Sud et al. 1999; Williams et al. 2009) or on SST thresholds with implications for tropical cyclones (Knutson et al. 2008). With respect to the upper limit of tropical SSTs, Waliser and Graham (1993) showed that maximum convective activity peaks over SSTs close to 302.5 K, and SSTs greater than 302.5 K are associated with reduced convection. This is consistent with findings of Xu et al. (2005, 2007), where they proposed a “cloud object” approach to analyze statistical properties of cloud systems from Earth Observing System (EOS) satellites in order to more rigorously validate model simulations. Waliser and Graham (1993) also found that at SSTs greater than 302 K intense deep convection is associated with a cooling effect of about 0.1°C month−1, whereas suppression of deep convection in convectively active regions results in a similar amount of sea surface warming. As noted by Lin et al. (2006), the observed decrease of clouds in the warmest SSTs (SST > ~304 K) is linked to weaker convergence in those areas compared to their slightly cooler SST counterparts. Kubar et al. (2011) recently showed that in a cross section from low SSTs near the California coast to high SSTs over the central equatorial regime, cloud frequencies at all vertical modes from CALIPSO and CloudSat reach a maximum within a relatively narrow surface temperature window between 2-m temperatures of 297 and 300 K, coinciding with deep ascending motion. Free-tropospheric mean subsidence occurs over the highest surface temperatures, thus demonstrating the suppression of deep convection by the large-scale circulation over the highest SSTs.
The SST threshold for the onset of deep convection usually occurs between 299 and 301 K, although SST is not the sole factor that controls deep convection; for instance, regions of mean subsidence and surface wind divergence have a paucity of convection (e.g., Graham and Barnett 1987). Furthermore, deep convective regions with varying upward vertical velocity profiles and structure may have not only different convection characteristics, but also a range of SSTs at which the onset of deep convection occurs (e.g., Lau et al. 1997). According to a recent study by Johnson and Xie (2010), where convection is dynamically probable, both the convective threshold and tropical mean SST have shown a parallel upward trend of approximately 0.1°C decade−1 over the past 30 years. Based on general circulation model simulations and observational data, Johnson and Xie (2010) also concluded that as a consequence of approximate moist-adiabatic lapse rate adjustment (Stone and Carlson 1979), the SST threshold for convection will continue to rise together with the tropical mean SST. This suggests that the tropical SST distribution observed in today’s climate will likely not be locked in a warmer climate by the upper SST limit seen today.
In this paper, we expand the study of local relations between tropical SST and clouds by analyzing such relations among various cloud types and by considering regional differences. The advent of CloudSat in particular enables a profile-by-profile examination of various low-topped, middle, and high-topped cloud types as they occur in concert with underlying SSTs, allowing for a careful assessment of convective and nonconvective high cloud occurrence with local temperature.
2. Study area, dataset, and methodology
We focus on tropical oceans between 30°N and 30°S, where a strong connection among the large-scale circulation, SSTs, and cloud types exists. The selected zone to a large extent avoids the complicating influence of higher-latitude frontal systems driven largely by strong horizontal temperature gradients and baroclinic instability, although north of about 22°N baroclinicity occurs perhaps nearly 10% or more of the time during boreal winter (please see the appendix). This tropical latitude range also largely bounds our analysis to ascending and subsiding branches of the large-scale Hadley circulation. We use two years (2007/08) of data from the latest version of the CloudSat 2B-CLDCLASS product to identify cloud types. As described in Sassen and Wang (2008), the 2B-CLDCLASS product is based on two main steps: 1) using a cloud mask algorithm to identify clouds in each CPR column, followed by cloud clustering, and 2) classifying cloud clusters using extracted features such as vertical and horizontal extent, maximum radar reflectivity, corresponding temperature at the maximum reflectivity level, and presence or absence of precipitation reaching the surface. The cloud classification process is further refined by identifying high-, middle-, and low-level clouds together with cloud-cluster features, including mean cloud height, temperature, and liquid water path amount as well as cloud height variability at top and bottom. For example, a Dc cloud is a precipitating thick cloud with a base lower than 3 km, a horizontal extent of about 10 km, and a liquid water path greater than zero. Detailed information about the 2B-CLDCLASS product and all variables calculated for each cloud cluster can be obtained from Sassen and Wang (see their “Level 2 cloud scenario classification product process description and interface control document,” version 5.0, 2007; available online at http://www.cloudsat.cira.colostate.edu/ICD/2B-CLDCLASS/2B-CLDCLASS_PDICD_5.0.pdf).
With respect to the 2B-CLDCLASS product two issues need to be considered:
Because of high surface contamination at three or four of the lowest vertical CPR bins (from the surface to ~1 km above the surface), the product may not provide an accurate observation of low-level clouds such as St and Sc. Therefore, following Sassen and Wang (2008) the two low-level cloud classes (St and Sc) are combined and are hereafter referred to as StSc.
So far, 2B-CLDCLASS does not incorporate CALIPSO lidar information, and therefore thin cloud layers might be missed in the cloud classification product. For a more thorough collection of cloud occurrence observations, the version 3 2B-GEOPROF lidar product is used, which is also available through the CloudSat website, and is the latest product available at the time of this writing. The combined product uses radar (CPR) aboard CloudSat and lidar (CALIOP) aboard CALIPSO to probe both optically thick large-particle layers as well as to sense optically thin layers and tenuous cloud tops. The two instruments fly in close coordination with one another. CloudSat orbits in front of CALIPSO with an approximately 12.5-s average delay between the two instruments. The SST data are obtained from ECMWF-AUX (http://www.cloudsat.cira.colostate.edu/dataSpecs.php?prodid=6), which is a quality-controlled intermediate product that contains a set of ancillary variables from the ECMWF data interpolated to each CloudSat profile. To map the regional variability of the studied variables, our study area is first divided into 20 × 120 = 2400 grid boxes, with each grid box 3° × 3° latitude/longitude, so as to collect a sufficient number of samples for each one, while preserving high spatial resolution to study regional variations.
3. Results
We consider the occurrence of clouds within each radar footprint and, given the large number of samples used in this study, we consider the total frequency of clouds within a given region or SST bin to be a cloud fraction, and henceforth use this terminology. Figure 1a shows cloud fractions observed by CloudSat radar and the combined CALIPSO–CloudSat product (hereafter referred to as the lidar–radar) as a function of SST in 1-K bins for the global tropical oceans between 30°N and 30°S. The number of observations in each SST bin is also plotted in Fig. 1b. The reported cloud fractions are based on the presence of at least one layer of cloud although multilayered clouds may exist. The cloud fractions derived from radar only and from lidar–radar are significantly different, with lidar–radar indicating consistently higher cloud fractions than radar across all SSTs with a difference ranging from 0.3 to 0.4. This is due to the high sensitivity of the lidar to detect optically thin clouds, which are not necessarily captured by radar with its sensitivity limited to approximately −31 dBZ. Additionally, CloudSat misses many shallow clouds with tops below ~1 km (e.g., Marchand et al. 2008; Kubar et al. 2011), and such clouds are abundant over lower SSTs in subtropical subsidence regions, primarily off the west coasts of continents. While the two products yield different absolute cloud fractions within each SST bin, they demonstrate fairly similar patterns of cloud fraction dependence on SST. The radar shows two distinct cloud fraction maxima at SSTs of about 292 and 303 K, and a minimum cloud fraction near 299 K. Lidar–radar also shows two distinct maxima at SSTs of 291 and 303 K, and a minimum cloud fraction at an SST of 299 K. Both radar and lidar–radar observations indicate that the cloud fraction decreases as SSTs increase from 292 to 299 K, and then sharply increases between SSTs of 299 and 303 K. Above 303 K, both products demonstrate a very rapid decrease in cloud fraction. Also noteworthy is the slight displacement between the location of the radar maximum cloud fraction at 292 K versus at 291 K from lidar–radar. Since low-topped clouds grow vertically as SST increases (e.g., Wyant et al. 1997; Lin et al. 2009; Kubar et al. 2011), an even larger proportion of shallow low-topped clouds at the coldest SSTs are missed by CloudSat.
(a) Fraction of clouds observed by radar (solid line) and lidar–radar (dashed line) products plotted as a function of SST for the global tropical oceans. (b) Number of observations within each 1-K SST bin.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
(a) Fraction of clouds observed by radar (solid line) and lidar–radar (dashed line) products plotted as a function of SST for the global tropical oceans. (b) Number of observations within each 1-K SST bin.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
(a) Fraction of clouds observed by radar (solid line) and lidar–radar (dashed line) products plotted as a function of SST for the global tropical oceans. (b) Number of observations within each 1-K SST bin.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Within our large domain the maximum SST population occurs near 301 K (Fig. 1b), which is identical to that reported by Graham and Barnett (1987) and Waliser and Graham (1993). The latter study demonstrated that the reduction in surface solar radiation by clouds associated with organized convection could explain why the observed “equilibrium” SST appears at 301 K. With respect to the upper limit of SST population curves, few SSTs greater than 305 K are observed. This is also consistent with previous studies suggesting that SSTs rarely exceed 305 K because of an apparent “natural” limit that is placed on SST (Waliser and Graham 1993; Newell 1979, among others). Lin et al. (2006) described that the observed decrease of clouds in the warmest SSTs (SST > ~304 K) is linked to the weaker convergence there compared to slightly cooler SST areas (~302 K), suggesting a preferred SST envelope in today’s climate for deep convection. It is important to note that subsequent studies have indicated that while a particular SST threshold of convection and an upper limit of SSTs may exist in today’s climate, these thresholds and upper limits may change in a warmer climate, whereby the mean population of SST may be shifted upward (e.g., Waliser and Graham 1993; Williams et al. 2009; Johnson and Xie 2010). The explanation of the upper bound of SSTs in today’s climate and under future warming conditions continues to be an area of active research.
Figure 2 shows a regional distribution of cloud fractions based on radar-only (Fig. 2a) and lidar–radar (Fig. 2b) products and their differences (Fig. 2c) calculated as lidar–radar minus radar cloud fraction in each 3° × 3° box. Knowing that CALIPSO more thoroughly detects clouds because of its higher sensitivity, a large fraction of clouds is missed by the radar compared to the lidar, particularly between the equator and 10°–15°S, including the Atlantic and the east and central Pacific Ocean, as shown in Fig. 2c. The difference in cloud fractions reaches nearly 0.6 in some areas. Both radar and lidar–radar exhibit a large triangular-shaped local minimum of cloud fraction in the central Pacific between 0° and 20°S. This is an area generally under large-scale subsidence yet with warm enough SSTs such that low patchy shallow cumuliform clouds may exist, which occupy much less space than the more homogeneous stratiform clouds over lower SSTs toward the east. Close to the coasts of South America and Africa, many low clouds are quite shallow and thus may be missed by CloudSat as the radar has reduced sensitivity below 1.2 km due to near-surface contamination, accounting for larger differences between CALIPSO and CloudSat in those regions. Additionally, tropopause transition layer (TTL) thin cirrus clouds may be quite pervasive in the lower latitudes either as outflow from primary convective areas as the intertropical convergence zone (ITCZ) or South Pacific convergence zone (SPCZ) or from in situ formation (McFarquhar et al. 2000; Winker and Trepte 1998), and such thin high clouds would be readily detected by the lidar but not the radar.
Regional distribution of cloud fraction based on (a) radar-only, (b) lidar and radar, and (c) their differences calculated by subtracting radar cloud fraction from lidar–radar cloud fraction at each individual grid for the entire period of study.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Regional distribution of cloud fraction based on (a) radar-only, (b) lidar and radar, and (c) their differences calculated by subtracting radar cloud fraction from lidar–radar cloud fraction at each individual grid for the entire period of study.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Regional distribution of cloud fraction based on (a) radar-only, (b) lidar and radar, and (c) their differences calculated by subtracting radar cloud fraction from lidar–radar cloud fraction at each individual grid for the entire period of study.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
In Fig. 3, the distribution of cloud fraction versus SST is decomposed for various cloud types identified by the CloudSat 2B-CLDCLASS product. Figure 3 is constructed by calculating cloud fractions within 27 SST bins, each containing an equal number of observations (including both cloudy and clear sky observations). Therefore, as shown in Fig. 3, the bins are not uniformly distributed with respect to SST but rather with respect to the number of observations in each bin. As a result, smaller bin sizes are located in regions of denser SST observations, so that each point carries as much statistical weight as any other. Hereafter we use this binning scheme in our analysis. In Fig. 3, the cloud fractions for a specific type of cloud are calculated based on the presence of that type of cloud whether alone or in combination with other types. Figure 3b is obtained by accumulating all cloud fractions shown in Fig. 3a. Based on Fig. 3 we highlight the following:
For SSTs less than approximately 301 K, StSc clouds are the dominant type of clouds (cloud fraction between 0.15 and 0.37) followed by high, As, and Ac clouds. For SSTs greater than around 301 K, high clouds are dominant (cloud fraction between 0.18 and 0.28).
In contrast to other cloud types, StSc cloud fraction consistently decreases with SST (with a very strong correlation coefficient of −0.99). The observation is consistent with several previous studies suggesting that more marine stratocumulus clouds are found at lower SSTs. Other studies have also shown that stratus and stratocumulus clouds are anticorrelated with SSTs, especially on longer time scales from several weeks to seasonal (e.g., Oreopoulos and Davies 1993; Kubar et al. 2012) to interannual (e.g., Oreopoulos and Davies 1993; Norris and Leovy 1994; Clement et al. 2009; Eastman et al. 2011); our results show the geographic correlation over a 2-yr climatology.
High, Dc, As, and Ac cloud fractions reach a maximum at a SST of near 303 K, while Cu has a broad cloud fraction peak at an SST close to 301 K. Few Ns clouds are observed over the global tropical oceans with no apparent peak. Ns clouds are mainly located at higher latitudes beyond 50° (Sassen and Wang 2008), and also seasonally in the midlatitudes in association with baroclinic frontal systems.
For High, Dc, As, and Ac clouds, as the SST exceeds 303 K (corresponding to the highest cloud fraction), cloud fractions are reduced substantially. This is consistent with Waliser and Graham (1993), who demonstrated that for SSTs exceeding 302.5 K, deep convection is reduced.
The SST onsets (the SST at which the cloud fraction amount starts to rise sharply as SST increases) for High, Dc, Ac, and As clouds fall between 299 K (High clouds) and 300.5 K (As clouds). Note that the observed onset of Dc clouds (about 300 K from Fig. 3a) is consistent with several previous studies showing that Dc clouds occur more frequently as SSTs exceed 300–301 K (Graham and Barnett 1987; Waliser and Graham 1993; Xu et al. 2005, 2007; Lin et al. 2007). Studies suggest that within this range of SST the vertical stability of the tropical troposphere is sufficiently reduced to cause the onset of large-scale moist convection (Lau and Shen 1988; Betts and Ridgway 1989). More recently, Kubar et al. (2011) showed that deep ascent in the central tropical Pacific occurs within a 2-m surface temperature SST range between 297 and 300 K, coinciding with where clouds at all vertical modes are observed, thus indicating such an onset point and window for deep convection.
By inclusion of all types of cloud, whether in a single layer or in multiple layers, the minimum and maximum cloud fractions occur at SSTs of 299 and ~303 K, respectively (Fig. 3b). These two extremes are consistent with those shown in Fig. 1.
(a) Distribution of cloud fraction vs SST for various cloud types identified by CloudSat 2B-CLDCLASS product, (b) Fraction of all clouds observed by radar using SST bins that are uniformly distributed with respect to the number of observations in each bin. In (b), the cloud fraction at each bin is the aggregate of all cloud types shown in (a).
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
(a) Distribution of cloud fraction vs SST for various cloud types identified by CloudSat 2B-CLDCLASS product, (b) Fraction of all clouds observed by radar using SST bins that are uniformly distributed with respect to the number of observations in each bin. In (b), the cloud fraction at each bin is the aggregate of all cloud types shown in (a).
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
(a) Distribution of cloud fraction vs SST for various cloud types identified by CloudSat 2B-CLDCLASS product, (b) Fraction of all clouds observed by radar using SST bins that are uniformly distributed with respect to the number of observations in each bin. In (b), the cloud fraction at each bin is the aggregate of all cloud types shown in (a).
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Distribution of various cloud types over the tropical oceans of (a) StSc, (b) Cu, (c) Ac, (d) As, (e) High, and (f) Dc clouds. Note that the color bar scales are different for each panel.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Distribution of various cloud types over the tropical oceans of (a) StSc, (b) Cu, (c) Ac, (d) As, (e) High, and (f) Dc clouds. Note that the color bar scales are different for each panel.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Distribution of various cloud types over the tropical oceans of (a) StSc, (b) Cu, (c) Ac, (d) As, (e) High, and (f) Dc clouds. Note that the color bar scales are different for each panel.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Distribution of mean SSTs corresponding to (a) CloudSat observations, including cloudy and clear sky scenes, (b) scenes with only StSc clouds, and (c) scenes with only Dc clouds.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Distribution of mean SSTs corresponding to (a) CloudSat observations, including cloudy and clear sky scenes, (b) scenes with only StSc clouds, and (c) scenes with only Dc clouds.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Distribution of mean SSTs corresponding to (a) CloudSat observations, including cloudy and clear sky scenes, (b) scenes with only StSc clouds, and (c) scenes with only Dc clouds.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Cumulus clouds are more frequent around the central Pacific ITCZ and southern Indian Ocean where SSTs are close to 301 K, as seen in Fig. 4b. Many of these clouds, which have low bases (between the surface and 3 km) and shallow to moderate thicknesses, likely represent the shallow latent heating mode as analyzed by Takayabu et al. (2010) using the Tropical Rainfall Measuring Mission Precipitation Radar (TRMM PR) spectral latent heating algorithm. Furthermore, our 2B-CLDCLASS cumulus horizontal distribution is qualitatively similar to the 2-km peak heating mode distribution from Takayabu et al. (2010) and to a lesser extent to the midlevel cumulus distribution in Lebsock et al. (2010), although that study shows a larger amount of cumulus in some areas where CloudSat indicates StSc, particularly near the Californian and Peruvian regimes. Other cloud types, including Ac (Fig. 4c), As (Fig. 4d), High (Fig. 4e), and Dc (Fig. 4f) exhibit high fractions over the ITCZ and a vast area over the Indian and west Pacific Oceans, where the mean SST exceeds 300 K. Figure 5c shows a map of mean SSTs in the presence of Dc clouds, suggesting that the presence of Dc clouds is strongly tied to the existence of SSTs greater than 300 K. Indeed, CloudSat profiles with active deep convection have considerably higher SSTs than the mean state, as visually seen by considering the differences between Figs. 5c and 5a. This could be reflecting the annual cycle of convection, in which the seasonal migration of high SST and Dc are in phase with each other.
The CloudSat classification product also makes it possible to analyze and quantify single and multiple cloud types within a CPR column. Figure 6a shows that as SST increases, the fraction of CPR columns containing only one type of cloud decreases until the SST reaches 303 K, followed by an increase of one cloud type for SSTs greater than 303 K. At the same time, as SST increases, the fraction of CPR columns containing two or more distinct cloud types increases up to an SST of 303 K, and then decreases for SSTs greater than 303 K (Figs. 6b,c). For various SST bins, Table 1 provides detailed information about the fraction of one and two distinct cloud type combinations within CPR columns. For the case of two distinct cloud types, Table 1 provides the seven most dominant combinations. Table 1 suggests that for SSTs between 288 and 290 K, approximately 75% of total clouds are StSc clouds with no other coincident cloud type. Within this SST bin Dc clouds are the least prevalent (0.28%), whereas between 302 and 304 K High (~33%) and Dc (~13%) clouds are the most dominant types. At this temperature bin about 4% of clouds consist of high clouds overlying Ac clouds. At the highest SST bin, High clouds are the most frequent (~42%) cloud type that exists in isolation.
The relationship between SST and fraction of cloudy scenes containing (a) a single cloud type, (b) two different cloud types, and (c) three or more cloud types in vertical columns of radar observations.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
The relationship between SST and fraction of cloudy scenes containing (a) a single cloud type, (b) two different cloud types, and (c) three or more cloud types in vertical columns of radar observations.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
The relationship between SST and fraction of cloudy scenes containing (a) a single cloud type, (b) two different cloud types, and (c) three or more cloud types in vertical columns of radar observations.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Percent of cloudy profiles that contain single and double cloud classifications as observed by CloudSat.
While the full diurnal cycle cannot be captured by either CloudSat or CALIPSO, twice-daily overpasses allow for a quantitative assessment of cloud property differences between the early morning and afternoon. Using the two CloudSat equatorial overpass times of 0130 and 1330 local time, we calculate cloud fraction difference (CFD) maps between the two overpasses for individual cloud types (Fig. 7), defined by subtracting the afternoon from morning cloud fraction. Figure 7a shows a strong positive CFD, mostly in dominant StSc regions. In contrast, CFDs are quite small for Cu and Ac clouds (Figs. 7b,c), whereas As and especially High clouds are much more abundant in the afternoon (Figs. 7d,e), which is consistent with Tian et al. (2004), who indicate that high clouds, some of which are directly detrained from convective systems, tend to lag the peak of deep convection near local sunrise by ~6–9 h. In fact, Luo and Rossow (2004) used trajectory analysis not only to quantify the formation of detrained high cloud from deep convective clouds (with high cirrus clouds persisting for ~19–30 h), but also to estimate that 44% of cirrus is formed directly from convection, and 56% is formed in situ. In contrast to the high clouds, Dc (Fig. 7f) shows a wide area of positive CFD over the ITCZ, SPCZ, and Indian Ocean. Although deep convection peaks several hours later than the early morning CloudSat overpass, a recent study by Sato et al. (2009) shows that TRMM tropical (15°S–15°N) precipitation, a good proxy for deep convection (Aumann et al. 2011), is also higher during the early morning versus early afternoon. Our CloudSat findings are thus consistent with this.
Maps of morning cloud fraction minus afternoon cloud fraction for various cloud types: (a) StSc, (b) Cu, (c) Ac, (d) As, (e) High, and (f) Dc clouds.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Maps of morning cloud fraction minus afternoon cloud fraction for various cloud types: (a) StSc, (b) Cu, (c) Ac, (d) As, (e) High, and (f) Dc clouds.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Maps of morning cloud fraction minus afternoon cloud fraction for various cloud types: (a) StSc, (b) Cu, (c) Ac, (d) As, (e) High, and (f) Dc clouds.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
For each cloud type we also calculate a mean percent CFD which for StSc, Cu, Ac, As, High, and Dc are 4.26, 0.30, 0.43, −0.43, −1.47, and 1.64, respectively. Overall there are 10% more clouds during the early morning as observed by CloudSat, expressed as the total CFD divided by total mean CF. Although these values represent tropics-wide differences of cloud amount between early morning and afternoon, it is instructive to examine the geographic distribution of CFD, which is presented in Fig. 8 for the radar (top panel) and lidar–radar (bottom panel). While the radar overall shows a greater difference between early morning and afternoon cloud amounts than the lidar–radar, the overall pattern (the areas of positive and negative CFDs) obtained from both is similar. The lidar–radar product overall indicates 4% more early morning versus afternoon clouds, smaller than CloudSat alone but still of the same sign.
Maps of morning minus afternoon cloud fraction for all cloud types observed by (a) radar and (b) lidar–radar.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Maps of morning minus afternoon cloud fraction for all cloud types observed by (a) radar and (b) lidar–radar.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Maps of morning minus afternoon cloud fraction for all cloud types observed by (a) radar and (b) lidar–radar.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
For low-topped clouds our observations are consistent with previous studies, such as Minnis and Harrison (1984) and Rozendaal et al. (1995), who observed a considerable diurnal modulation of low cloud amounts. The latter study also found that the diurnal cycle is larger downwind of the center of the subtropical stratocumulus regimes, primarily because those clouds are more susceptible to solar absorption where the boundary layer is deeper and low-topped clouds may be more broken climatologically. A large diurnal cycle of liquid water path has also been observed in various studies (e.g., Zuidema and Hartmann 1995; Weng and Grody 1994; Greenwald and Christopher 1999). Using the TRMM Microwave Imager (TMI), Wood et al. (2002) observed an early morning peak of low clouds throughout most of the subtropics, consistent with a diurnal cycle driven largely by the diurnal solar cycle. Cloud-top longwave cooling is most effective at night, important for driving and maintaining high stability and a strong temperature inversion as well as for generation of in-cloud instability and overturning circulation. This couples the cloud layer to the surface and thus provides a consistent moisture source from the surface (Wood 2012). During the day, cloud solar absorption counters longwave cloud-top cooling, and thus overall the aforementioned circulation is also weaker. In our analysis, the difference between morning and afternoon StSc clouds is largest where cloud amounts are also largest, although it is important to note again that CloudSat by definition preferentially senses deeper low-topped clouds because of the radar’s reduced sensitivity near the surface. This means that StSc cloud maxima are likely downwind of the real maxima, such that the observation of the largest morning and afternoon cloud differences that CloudSat shows is consistent with Rozendaal et al. (1995).
In general, although a sun-synchronous instrument such as CloudSat does not permit observations of the full diurnal cycle, it can nonetheless provide insight into particular tropical cloud types that do exhibit a significant diurnal cycle, such that appreciable differences between early morning and early afternoon can be seen, as we have demonstrated. Better quantification of the diurnal cycle of cloud cover, cloud water, and even cloud type is important in modulating the diurnal radiation budget and reducing radiative flux calculation errors (Bergman and Salby 1997).
4. Summary and conclusions
The CloudSat 2B-CLDCLASS product intrinsically defines clouds based on the horizontal and vertical features achievable with an active instrument, and thus is consistent with the morphological definition of cloud types, themselves governed by different large-scale dynamics. Indeed, an assessment of cloud morphology from observations is fundamental when comparisons are made to general circulation models. We have analyzed the distribution of several cloud types over tropical oceans (30°N–30°S) by collecting two years (2007/08) of data from CloudSat 2B-CLDCLASS and 2B-GEOPROF lidar products, and have also examined cloud type and SST relationships. We have furthermore quantified that the influence of clouds associated with baroclinic midlatitude frontal systems is insignificant except during winter, when the baroclinicity fraction may exceed 10% over regions poleward of 22°. This suggests that our analysis primarily captures tropical and subtropical cloud systems associated with the Hadley and Walker circulations. Across all SSTs, the 2B-GEOPROF lidar shows considerably larger cloud fractions than the 2B-CLDCLASS product, an anticipated result as the lidar has a much greater sensitivity compared to the radar and thus senses more optically thin clouds invisible to the radar. Many of these clouds may be TTL cirrus clouds, which have small optical depths and are readily detected by the lidar but mostly missed by the radar. Furthermore, many very shallow low-topped clouds are missed by CloudSat because of near-surface contamination below one kilometer, as demonstrated in Marchand et al. (2008) and Kubar et al. (2011). In the latter study, fewer than half of uniform low-topped clouds observed by the Moderate Resolution Imaging Spectroradiometer (MODIS) or CALIPSO are observed by CloudSat over the tropical and subtropical northeast Pacific, a region where subsidence-forced shallow clouds are abundant. In our study, both tropics-wide CloudSat and the merged CALIPSO and CloudSat products indicate cloud fraction peaks at SSTs of 303 and 292 K, representing population maxima of high-topped and low-topped clouds, respectively, which likely correspond to ascending and descending branches of the Hadley and Walker circulations.
Stratiform and stratocumulus (StSc) clouds are the dominant cloud type (cloud fraction between 0.15 and 0.37) over SSTs less than 301 K, and their fraction is strongly inversely related to SST, with a correlation coefficient of −0.99. This is physically reasonable as both static stability and large-scale subsidence scale well with decreasing SST. At SSTs greater than 301 K, high clouds (with cloud fractions ranging from 0.18 to 0.28) are the most abundant cloud type. All cloud types, except nimbostratus and stratocumulus, become rapidly more abundant at SSTs greater than a window between 299 and 300.5 K, depending on cloud type. The fractions of high, deep convective, altostratus, and altocumulus clouds peak at an SST close to 303 K, while cumulus clouds have a broad cloud fraction peak centered near 301 K. The greater abundance of cumulus clouds over lower SSTs may be suggestive of an evolution from shallow to deeper convection with increasing SST. Geographically, we show that these cumulus clouds are more widespread in the east Pacific ITCZ region, consistent with findings from Kubar et al. (2007) in which MODIS midlevel clouds in the east Pacific are more abundant where mean SSTs are lower than in the west Pacific. The presence of more cumulus clouds in the east Pacific is also dynamically consistent with Back and Bretherton (2006), who show that upward vertical velocity profiles in the east Pacific are more “bottom-heavy” versus the more “top-heavy” profiles in the west Pacific, where ascending motion peaks in the mid to upper troposphere. Physically, the east Pacific is characterized by stronger SST gradients, which have been shown to be an important forcing for low-level convergence (Lindzen and Nigam 1987), reflected in the cloud distribution as a greater abundance of cumulus clouds there.
Despite some previous arguments of monotonic convective increases with SST (e.g., Ramanathan and Collins 1991), all cloud types in our study, including deep convective clouds, decrease strongly at the highest SSTs, consistent with previous studies (e.g., Waliser and Graham 1993; Waliser et al. 1993) showing that diminished convection occurs at the highest SSTs. Indeed, our analysis indicates a reduction in overall cloudiness at SSTs greater than 303 K. Ocean surface cooling occurs in conjunction with deep convection, and ocean surface warming in nonconvective areas of very high SSTs (Waliser and Graham 1993), where free-tropospheric subsidence is also observed (Kubar et al. 2011). Furthermore, the maximum SST observed in the presence of fewer clouds does not necessarily imply a tight limit on maximum SST in climate regimes different from today’s, as the SST distribution in a warmer climate would likely shift toward higher values (Waliser and Graham 1993; Williams et al. 2009). Indeed, the SST threshold for convection has been shown to have increased over the past several decades in concert with increases in surface and tropospheric tropical temperatures (Johnson and Xie 2010), and climate simulations project that the fractional area of active convection would change little in a warmer climate, consistent with a population shift in SST and corresponding convective characteristics.
Finally, although the full diurnal cycle cannot be captured by either CloudSat or CALIPSO, twice-daily overpasses allow for a quantitative assessment of cloud property differences in the early morning and afternoon, from which considerable differences are observed . The cloud fraction differences are not uniform and vary significantly from one region to another and among different cloud types. We observe that the amount of stratocumulus, high, and deep convective clouds exhibit the largest differences, suggesting that these clouds have a relatively strong diurnal cycle. Among all cloud types, CloudSat radar indicates that early morning clouds are about 10% more frequent than afternoon clouds and joint lidar and radar suggests that this frequency is about 4%, likely arising from the combination of marine boundary layer cloud fraction and oceanic deep convection peaks before sunrise.
Acknowledgments
The authors would like to express their appreciation to two anonymous reviewers for their helpful comments and suggestions to improve the manuscript. 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.
APPENDIX
Influence of Extratropical Clouds
Baroclinicity fraction as an indicator for frequency of intrusion of extratropical baroclinic systems into the studied region. The fractions are calculated for various σBI thresholds during winter and summer and for each (a),(b) 3° latitude zone and (c),(d) 1-K SST bin.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Baroclinicity fraction as an indicator for frequency of intrusion of extratropical baroclinic systems into the studied region. The fractions are calculated for various σBI thresholds during winter and summer and for each (a),(b) 3° latitude zone and (c),(d) 1-K SST bin.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
Baroclinicity fraction as an indicator for frequency of intrusion of extratropical baroclinic systems into the studied region. The fractions are calculated for various σBI thresholds during winter and summer and for each (a),(b) 3° latitude zone and (c),(d) 1-K SST bin.
Citation: Monthly Weather Review 140, 10; 10.1175/MWR-D-11-00247.1
REFERENCES
Aumann, H. H., S. G. DeSouza-Machado, and A. Behrangi, 2011: Deep convective clouds at the tropopause. Atmos. Chem. Phys., 11, 1167–1176.
Back, L. E., and C. S. Bretherton, 2006: Geographic variability in the export of moist static energy and vertical motion profiles in the tropical Pacific. Geophys. Res. Lett., 33, L17810, doi:10.1029/2006GL026672.
Bergman, J. W., and M. L. Salby, 1997: The role of cloud diurnal variations in the time-mean energy budget. J. Climate, 10, 1114–1124.
Betts, A. K., and W. Ridgway, 1989: Climatic equilibrium of the atmospheric convective boundary layer over a tropical ocean. J. Atmos. Sci., 46, 2621–2641.
Bony, S., K. M. Lau, and Y. C. Sud, 1997: Sea surface temperature and large-scale circulation influences on tropical greenhouse effect and cloud radiative forcing. J. Climate, 10, 2055–2077.
Bretherton, C. S., and M. C. Wyant, 1997: Moisture transport, lower-tropospheric stability, and decoupling of cloud-topped boundary layer. J. Atmos. Sci., 54, 148–167.
Chen, T., W. B. Rossow, and Y. C. Zhang, 2000: Radiative effects of cloud-type variations. J. Climate, 13, 264–286.
Clement, A. C., R. Burgman, and J. R. Norris, 2009: Observational and model evidence for positive low-level cloud feedback. Science, 325, 460–464.
Eastman, R., S. G. Warren, and C. J. Hahn, 2011: Variations in cloud cover and cloud types over the ocean from surface observations, 1954–2008. J. Climate, 24, 5914–5934.
Eitzen, Z. A., K.-M. Xu, and T. Wong, 2011: An estimate of low cloud feedbacks from variations of cloud radiative and physical properties with sea surface temperature on interannual time scales. J. Climate, 24, 1106–1121.
Emanuel, K. A., M. Fantini, and A. Thorpe, 1987: Baroclinic instability in an environment of small stability to slantwise moist convection. Part I: Two-dimensional models. J. Atmos. Sci., 44, 1559–1573.
Graham, N. E., and T. P. Barnett, 1987: Sea surface temperature, surface wind divergence, and convection over tropical oceans. Science, 238, 657–659.
Greenwald, T. J., and S. A. Christopher, 1999: Daytime variation of marine stratocumulus microphysical properties as observed from geostationary satellite. Geophys. Res. Lett., 26, 1723–1726.
Harrison, E. F., P. Minnis, B. R. Barkstrom, V. Ramanathan, R. D. Cess, and G. G. Gibson, 1990: Seasonal variation of cloud radiative forcing derived from the Earth Radiation Budget Experiment. J. Geophys. Res., 95, 18 687–18 703.
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.
Hartmann, D. L., L. A. Moy, and Q. Fu, 2001: Tropical convection and the energy balance at the top of the atmosphere. J. Climate, 14, 4495–4511.
Im, E., C. Wu, and S. L. Durden, 2005: Cloud profiling radar for the CloudSat mission. IEEE Aerosp. Electron. Syst. Mag., 20, 15–18.
Johnson, N. C., and S.-P. Xie, 2010: Changes in the sea surface temperature threshold for tropical convection. Nat. Geosci., 3, 842–845.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6, 1587–1606.
Knutson, T. R., J. J. Sirutis, S. T. Garner, G. A. Vecchi, and I. M. Held, 2008: Simulated reduction in Atlantic hurricane frequency under twenty-first-century warming conditions. Nat. Geosci., 1, 359–364.
Kubar, T. L., D. L. Hartmann, and R. Wood, 2007: Radiative and convective driving of tropical high clouds. J. Climate, 20, 5510–5526.
Kubar, T. L., D. E. Waliser, and J.-L. Li, 2011: Boundary layer and cloud structure controls on tropical low cloud cover using A-Train satellite data and ECMWF analyses. J. Climate, 24, 194–215.
Kubar, T. L., D. E. Waliser, J.-L. Li, and X. Jiang, 2012: On the annual cycle, variability, and correlations of oceanic low-topped clouds with large-scale circulation using Aqua MODIS and ECMWF-Interim. J. Climate, 25, 6152–6174.
Lau, K.-M., and S. Shen, 1988: On the dynamics of intraseasonal oscillations and ENSO. J. Atmos. Sci., 45, 1781–1797.
Lau, K.-M., H.-T. Wu, and S. Bony, 1997: The role of large-scale atmospheric circulation in the relationship between tropical convection and sea surface temperature. J. Climate, 10, 381–392.
Lebsock, M. D., C. Kummerow, and G. L. Stephens, 2010: An observed tropical oceanic radiative–convective cloud feedback. J. Climate, 23, 2065–2078.
Lin, B., B. A. Wielicki, P. Minnis, L. H. Chamber, K.-M. Xu, Y. Hu, and A. Fan, 2006: The effect of environmental conditions on tropical deep convective systems observed from the TRMM satellite. J. Climate, 19, 5745–5761.
Lin, B., K.-M. Xu, P. Minnis, B. A. Wielicki, Y.-X. Hu, L. Chambers, T.-F. Fan, and W. Sun, 2007: Coincident occurrences of tropical individual cirrus clouds and deep convective systems derived from TRMM observations. Geophys. Res. Lett., 34, L14804, doi:10.1029/2007GL029768.
Lin, W., M. Zhang, and N. Loeb, 2009: Seasonal variation of the physical properties of marine boundary layer clouds off the California coast. J. Climate, 22, 2624–2638.
Lindzen, R. S., and B. F. Farrell, 1980: A simple approximate result for maximum growth rate of baroclinic instabilities. J. Atmos. Sci., 37, 1648–1654.
Lindzen, R. S., and S. Nigam, 1987: On the role of sea surface temperature gradients in forcing low-level winds and convergence in the tropics. J. Atmos. Sci., 44, 2418–2436.
Luo, Z., and W. B. Rossow, 2004: Characterizing tropical cirrus life cycle, evolution, and interaction with upper-tropospheric water vapor using Lagrangian trajectory analysis of satellite observations. J. Climate, 17, 4541–4563.
Marchand, R., G. G. Mace, T. Ackerman, and G. Stephens, 2008: Hydrometeor detection using CloudSat—An Earth-orbiting 94-GHz cloud radar. J. Atmos. Oceanic Technol., 25, 519–533.
McFarquhar, G. M., A. J. Heymsfield, J. Spinhirne, and B. Hart, 2000: Thin and subvisual tropopause tropical cirrus: Observations and radiative impacts. J. Atmos. Sci., 57, 1841–1853.
Minnis, P., and E. F. Harrison, 1984: Diurnal variability of regional cloud and clear-sky radiative parameters derived from GOES data. Part II: November 1978 cloud distributions. J. Climate Appl. Meteor., 23, 1012–1031.
Newell, R. E., 1979: Climate and the ocean. Amer. Sci., 67, 405–416.
Norris, J. R., 1998a: Low cloud type over the ocean from surface observations. Part I: Relationship to surface meteorology and the vertical distribution of temperature and moisture. J. Climate, 11, 369–382.
Norris, J. R., 1998b: Low cloud type over the ocean from surface observations. Part II: Geographical and seasonal variations. J. Climate, 11, 383–403.
Norris, J. R., and C. B. Leovy, 1994: Interannual variability in stratiform cloudiness and sea surface temperature. J. Climate, 7, 1915–1925.
Oreopoulos, L., and R. Davies, 1993: Statistical dependence of albedo and cloud cover on sea surface temperature for two tropical marine stratocumulus regions. J. Climate, 6, 2434–2447.
Ramanathan, V., and W. Collins, 1991: Thermodynamics regulation of ocean warming by cirrus clouds deduced from observations of the 1987 El Niño. Nature, 351, 27–32.
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.
Rozendaal, M. A., C. B. Leovy, and S. A. Klein, 1995: An observational study of diurnal variations of marine stratiform cloud. J. Climate, 8, 1795–1809.
Sassen, K., and Z. Wang, 2008: Classifying clouds around the globe with the CloudSat radar: 1 year of results. Geophys. Res. Lett., 35, L04805, doi:10.1029/2007GL032591.
Sato, T., H. Miura, M. Satoh, Y. N. Takayabu, and Y. Q. Wang, 2009: Diurnal cycle of precipitation in the tropics simulated in a global cloud-resolving model. J. Climate, 22, 4809–4826.
Stephens, G. L., and Coauthors, 2002: The CloudSat mission and the A-Train—A new dimension of space-based observations of clouds and precipitation. Bull. Amer. Meteor. Soc., 83, 1771–1790.
Stone, P. H., and J. H. Carlson, 1979: Atmospheric lapse rate regimes and their parameterization. J. Atmos. Sci., 36, 415–423.
Sud, Y. C., G. K. Walker, and K.-M. Lau, 1999: Mechanisms regulating sea-surface temperatures and deep convection in the tropics. Geophys. Res. Lett., 26, 1019–1022.
Takayabu, Y. N., S. Shige, W.-K. Tao, and N. Hirota, 2010: Shallow and deep latent heating modes over tropical oceans observed with TRMM PR spectral latent heating data. J. Climate, 23, 2030–2046.
Tian, B., B. J. Soden, and X. Wu, 2004: Diurnal cycle of convection, clouds, and water vapor in the tropical upper troposphere: Satellites versus a general circulation model. J. Geophys. Res., 109, D10101, doi:10.1029/2003JD004117.
Waliser, D. E., and N. E. Graham, 1993: Convective cloud systems and warm-pool sea surface temperatures: Coupled interactions and self-regulation. J. Geophys. Res., 98, 12 881–12 893.
Waliser, D. E., N. E. Graham, and C. Gautier, 1993: Comparison of the highly reflective cloud and outgoing longwave radiation datasets for use in estimating tropical deep convection. J. Climate, 6, 331–353.
Weng, F. Z., and N. C. Grody, 1994: Retrieval of cloud liquid water using the Special Sensor Microwave Imager (SSM/I). J. Geophys. Res., 99 (D12), 25 535–25 551.
Williams, I. N., R. T. Pierrehumbert, and M. Huber, 2009: Global warming, convective threshold and false thermostats. Geophys. Res. Lett., 36, L21805, doi:10.1029/2009GL039849.
Winker, D. M., and C. R. Trepte, 1998: Laminar cirrus observed near the tropical tropopause by LITE. Geophys. Res. Lett., 25, 3351–3354.
Winker, D. M., W. H. Hunt, and M. J. McGill, 2007: Initial performance assessment of CALIOP. Geophys. Res. Lett., 34, L19803, doi:10.1029/2007GL030135.
Wood, R., 2012: Stratocumulus clouds. Mon. Wea. Rev., 140, 2373–2423.
Wood, R., and D. L. Hartmann, 2006: Spatial variability of liquid water path in marine low cloud: The importance of mesoscale cellular convection. J. Climate, 19, 1748–1764.
Wood, R., C. S. Bretherton, and D. L. Hartmann, 2002: Diurnal cycle of liquid water path over the subtropical and tropical oceans. Geophys. Res. Lett., 29, 2092, doi:10.1029/2002GL015371.
Wyant, M. C., C. S. Bretherton, H. A. Rand, and D. E. Stevens, 1997: Numerical simulations and a conceptual model of the stratocumulus to trade cumulus transition. J. Atmos. Sci., 54, 168–192.
Xu, K.-M., T. Wong, B. A. Wielicki, L. Parker, and Z. A. Eitzen, 2005: Statistical analyses of satellite cloud object data from CERES. Part I: Methodology and preliminary results of 1998 El Niño/2000 La Niña. J. Climate, 18, 2497–2514.
Xu, K.-M., T. Wong, B. A. Wielicki, L. Parker, B. Lin, Z. A. Eitzen, and M. Branson, 2007: Statistical analyses of satellite cloud object data from CERES. Part II: Tropical convective cloud objects during 1998 El Niño and evidence for supporting the fixed anvil temperature hypothesis. J. Climate, 20, 819–842.
Zuidema, P., and D. L. Hartmann, 1995: Satellite determination of stratus cloud microphysical properties. J. Climate, 8, 1638–1657.