1. Introduction
Marine boundary layer clouds are an influential lever in the climate system. Stratocumuli cover approximately 20% of Earth’s surface, the majority of this over the ocean (Wood 2012), and exert a stronger cooling tendency on the planet than any other cloud type (Chen et al. 2000). Changes on the order of just 4% in the global low cloud cover or 4%–7% in the globally averaged low cloud liquid water path could result in a radiative forcing similar in magnitude to a doubling of the atmospheric CO2 concentration (Randall et al. 1984; Slingo 1990).
High concentrations of stratocumulus occur in the eastern subtropical ocean basins, where the surface water is cold and subsidence is relatively strong, creating a shallow boundary layer with a strong inversion. As air follows the trade winds toward the equator and away from the coast, the boundary layer (BL) grows deeper, the surface temperature grows warmer, and the stratocumulus gradually decouples from the surface and breaks up into trade cumulus clouds (Wyant et al. 1997) with a lower cloud fraction and lower cumulative albedo. Their leverage in the climate system has made the accurate parameterization of stratocumuli a key problem in climate modeling (Stephens 2005). Our ability to understand and model the marine stratocumulus-to-cumulus transition (SCT) in particular is a prerequisite to confident prediction of the cloud response to global warming (Teixeira et al. 2011).
Various strategies have been used in efforts to understand stratocumuli and the SCT in particular. Satellite and maritime observations, in conjunction with reanalysis data, can be used in large-scale statistical studies, which often search for correlations between stratocumuli and controlling meteorological factors (Klein and Hartmann 1993; Kawai and Teixeira 2010). Large-eddy simulation (LES) has filled a critical role in developing our understanding on smaller scales, providing a proxy for experiments (see, e.g., Wyant et al. 1997; Sandu and Stevens 2011; Zhang et al. 2013) and providing a foundation for the development of single-column models that can be used to parameterize subgrid-scale processes in the global models (Duynkerke et al. 2004; Zhu et al. 2005). In situ campaigns such as the Atlantic Stratocumulus Transition Experiment (ASTEX; Albrecht et al. 1995) and the Second Dynamics and Chemistry of Marine Stratocumulus field study (DYCOMS-II; Stevens et al. 2003) have gathered data on marine stratocumulus clouds and the SCT, which can then be used to evaluate models (Bretherton et al. 1999; Stevens et al. 2005). Mixed-layer models can be used to study decoupling along the SCT (Bretherton and Wyant 1997). Still others have attempted to construct simple physical models of the transition based on data or LES results (Chung and Teixeira 2012; Chung et al. 2012; Karlsson and Teixeira 2014).
Many of these studies attempt to make sense of the complexity in the marine stratocumulus system and the SCT by focusing attention on one or a few features of the transition. By contrast, an approach that considers the flows of energy and water through a well-defined closed atmospheric system can yield a fundamental and holistic check on our observational and modeling capabilities and complement our understanding by providing integral constraints on our understanding of the small-scale turbulent processes that govern the flows of water and energy through the system. For example, Trenberth et al. (2007, 2009) and Stephens et al. (2012) have demonstrated the usefulness of the holistic budget approach on the global scale. Regional budgets are perhaps more difficult to close than global budgets because they include advective terms which vanish in the global mean (Wong et al. 2011), but they may be useful in isolating and evaluating key atmospheric processes such as the SCT.
In this paper we report on the climatological mean energy and water budgets in the subtropical marine BL, focusing on the northeastern Pacific basin in summer. We limit our analysis to a portion of the Global Energy and Water Cycle Experiment Cloud System Study/Working Group on Numerical Experimentation (GCSS/WGNE) Pacific Cross-Section Intercomparison (GPCI) transect (Teixeira et al. 2011) from 35° to 15°N, which includes the BL transition from stratocumulus to trade cumulus. (We will refer to this transect simply as “the transect” throughout this work.) Our goals are to advance the fundamental understanding of the marine stratocumulus and trade cumulus BL system and to provide a holistic observational reference for global and regional models, LES, BL parameterization schemes, and simple physical models.
Figure 1 shows the sea surface temperature (SST), lower-tropospheric stability (LTS; the difference between potential temperature at 700 hPa and the SST), and surface winds in the northeastern Pacific region, obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) Interim Re-Analysis (ERA-Interim; Dee et al. 2011). The SCT evolves as air moves along the northern portion of the GPCI transect, shown in the figure as a white line. Figure 2 shows the corresponding transition in terms of cloud cover and BL height, as the stratocumulus sheet breaks up into cumulus, in the climatological mean.
JJA mean SST (colored contours; K), LTS (black contours; K), and surface (10 m) winds (white arrows) in the northeastern Pacific region from ERA-Interim. The single white track shows the GPCI transect; the multiple white tracks show the JJA ground tracks for the MAGIC campaign. For a sense of scale, the mean wind speed along the GPCI transect is 7 m s−1.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
CloudSat/CALIPSO fractional cloud cover up to 680 hPa (colored contours) and COSMIC BL height (black contours; km). All data are shown in the JJA mean.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
We compare our results to energy and water budgets derived at a single point in the southeastern subtropical Pacific by Caldwell et al. (2005). Our budget estimates build on the work of Caldwell et al. (2005) by considering a larger spatiotemporal domain that includes the SCT, demonstrating the use of spaceborne observations, and more completely characterizing uncertainty in budget terms. In the course of calculating the budgets, we also derive the mean cloud-top entrainment rate and compare our estimates to those derived by Wood and Bretherton (2004), who, like us, give estimates for both the stratocumulus and cumulus regimes.
Section 2 defines the energy and water budgets. Section 3 introduces each observational ingredient in the budget equations and assembles the terms from the data. Section 4 puts the observational ingredients together to derive the entrainment rate and to close the budgets along the transect. Section 5 concludes the paper.
2. Equations for the energy and water budgets
We are interested in the bulk column-integrated energy and water budgets for the stratocumulus-topped boundary layer. Figure 3 diagrams the energy and water budget terms.
Diagram of the terms in the boundary layer energy and water budgets. Our system is bounded by the ocean surface at the bottom and by the BL top (i.e., the cloud tops in this diagram). Filled arrows show positive budget terms (i.e., energy or water added to the boundary layer), while hollow arrows show negative terms. Arrow widths approximate the relative importance of the respective term in the budget.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1










3. Data
We present the energy and water budget terms from monthly mean data, in the June–August (JJA) mean. Sandu et al. (2010) suggested that this climatology is equivalent to the mean Lagrangian evolution of individual air masses. We limit the domain of our budgets to the atmospheric boundary layer between 35° and 15°N along the GPCI transect.
To construct the budgets, we use data from satellite and surface observations and from reanalyses. To determine radiation, precipitation, and cloud fraction (CF), we use satellite data from instruments on three A-Train satellites, CloudSat, Aqua, and Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO), including the first global-scale observations of light precipitation and improved radiation budget products from active sensors (Mace et al. 2009; Lebsock and L’Ecuyer 2011; Henderson et al. 2013). To determine the BL height we use radio occultation data from Constellation Observing System for Meteorology, Ionosphere, and Climate (COSMIC) microsatellites, which intercept signals from global positioning system satellites (Ao et al. 2012). To estimate horizontal advection, surface fluxes, and subsidence we use ECMWF reanalysis data. We neglect the time derivative or “storage” terms in the budget equations, which should be zero in the steady state.
We also rely on radiosonde data obtained by the shipborne Marine Atmospheric Radiation Measurement Program (ARM) GPCI Investigation of Clouds (MAGIC) campaign.1 MAGIC consisted of a U.S. Department of Energy ARM Mobile Facility installed on a container ship traveling back and forth between Los Angeles and Honolulu every 2 weeks; ground tracks by the ship are shown in Fig. 1. In this analysis, we use data from 212 JJA radiosonde launches.
To estimate uncertainties in the individual budget components and in the resulting budgets, we rely on several secondary data sources, approximating uncertainty as the RMS error between datasets. In general, we report uncertainties at the one standard deviation level; while the RMS error equals the standard deviation for an unbiased estimator, the errors we estimate typically also include a bias.
a. Cloud fraction
We obtain cloud data from version 4 of the CloudSat level 2B radar–lidar geometric profile product (2B-GEOPROF-lidar; Mace et al. 2009), which is merged from a CloudSat radar cloud mask that can penetrate optically thick clouds and a CALIPSO lidar cloud mask capable of detecting optically thin clouds. Cloud fraction is a three-dimensional field in 2B-GEOPROF-lidar giving the fraction of lidar volumes in a radar resolution volume containing hydrometeors. Cloud cover is a two-dimensional product, giving the percentage of the pixel that is cloudy; we construct it from the cloud layers field in 2B-GEOPROF-lidar. We average the data within 4° longitude by 2° latitude cells, with 250-m vertical bins.
Figure 2 shows the JJA mean of low cloud cover in the northeastern Pacific region from 2007 to 2010, while Fig. 4 shows low cloud cover and total cloud cover along the transect. We set our threshold for low clouds at 680 hPa (approximately 3 km) to include low clouds along the entire transect. Figure 5 shows the transect slice of cloud fraction, giving the vertical profile of cloud density.
2B-GEOPROF-lidar cloud cover along the transect, from the surface to 680 hPa and for the whole column.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
2B-GEOPROF-lidar cloud fraction along the transect. The boundary layer height is also shown as estimated by two methods as discussed in section 3b.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
We choose to use the 2B-GEOPROF-lidar product for cloud fraction instead of data from the International Satellite Cloud Climatology Project (ISCCP; Rossow and Schiffer 1991, 1999). This is because the ISCCP data are less sensitive than the CALIPSO lidar, ISCCP data are available only during the day, and cirrus clouds obscure the ISCCP view of low clouds. Because high clouds account for up to 13% of the total cloud cover north of 15°N (Fig. 4), cirrus obscuration could cause significant inaccuracy in the low cloud fraction.
b. Boundary layer height
Establishing the BL height along the transect is a key consideration as it determines the column height used in closing the energy and water budgets. The BL height is needed for all but the surface flux terms: the advective terms, the radiative terms, and the entrainment terms. Errors in the BL height of only a few hundred meters can lead to significant errors in these terms.
We considered obtaining estimates for the BL height from various sources: the gradient in the ERA-Interim relative humidity (RH; see, e.g., von Engeln and Teixeira 2013), the gradient in the CloudSat climatological longwave cooling, the gradient in 2B-GEOPROF-lidar cloud fraction, and COSMIC radio occultation (RO) data (Ao et al. 2012). Restricting our analysis to monthly mean variables allowed us to choose the RO method which does not suffer from the instrumental and model-imposed vertical resolution limitations of the other methods. A notable limitation of the COSMIC RO dataset is that it suffers from poor sampling coverage and has been taken from a data aggregation at the spatial scale of 5°.
Figure 2 shows contours of the RO BL height in the northeastern Pacific region, and Fig. 5 shows the RO BL height along the transect (overlaid on the 2B-GEOPROF-lidar cloud fraction). We have averaged JJA data for the years 2007–11 and filtered the two-dimensional surface with a Gaussian filter with σ = 2, which has a negligible effect on the BL height but ameliorates low-amplitude jitter in the data, which could add unnecessary noise to the gradient.
In Fig. 5, it is apparent that the COSMIC BL height corresponds to the top of the vertical profile of cloud fraction over most of the transect, adding confidence to both of these sets of data. It is also apparent that the estimate using the gradient in ECMWF RH is systematically low. This raises the possibility that the ECMWF model does not establish the BL height correctly in this region, which could cause error in other budget terms that rely on ECMWF data such as horizontal advection.
To estimate the uncertainty in the RO BL height, we compared it to radiosonde profiles from the MAGIC campaign. The method used to determine BL height from the radiosonde data is described in section 3h. Figure 6 shows the comparison.
Comparison of COSMIC BL height estimate to inversion height estimates derived from MAGIC radiosonde profiles, in the JJA mean. The blue circles are estimates from individual radiosonde profiles, the blue line is their binned mean, the red line is the COSMIC estimate along the MAGIC track, the green line is the COSMIC estimate along the GPCI transect, and the black line is the ECMWF gradient in RH estimate along the GPCI transect. The shading shows uncertainty derived from the difference between the COSMIC and MAGIC estimates along the MAGIC transect, as described in the text.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
Along the MAGIC transect, the COSMIC data shows a bias near the coast relative to the MAGIC data, probably due to coastal topography being averaged into the 5° grid. This discrepancy was also observed by von Engeln et al. (2005). We have estimated the one-sided uncertainty due to this bias near the coast, along our portion of the GPCI transect (for which we lack radiosonde data), by comparing it to the MAGIC transect on the basis of the shortest distance to the coast. Specifically, we bias correct the entire COSMIC BL surface, determining the correction at each point on the surface by calculating its distance to the coast and selecting the corresponding bias correction from the MAGIC track. We perform the bias correction to a maximum distance of 1450 km from the coast and smoothly taper into the original surface at this boundary. Performing the correction over the entire surface allows us to include the bias correction in our estimates of the horizontal gradient in BL height. The RMS error between the COSMIC and MAGIC estimates, calculated over the MAGIC track between 22.2° and 31.7°N to exclude the coast bias, is 160 m. We take this as an estimate of the bias and random uncertainty (at the 1σ level) in the BL height far from the coast. The green shading in Fig. 6 combines this RMS uncertainty with the coast bias, adding them in quadrature along our portion of the GPCI transect. We note that the lower BL height uncertainty bound approximates the ECMWF BL height estimate obtained from the gradient in ECMWF RH method; therefore, contributions from BL height–dependent errors lurking in the ECMWF model, if they exist, will be included to some degree in our uncertainty estimates.
We have propagated uncertainty in the BL through the analysis, by recalculating the horizontal advection terms, radiation terms, and entrainment terms at BL height upper and lower uncertainty bounds. In what follows, we present total uncertainty estimates in these terms by adding the RMSE estimates with the propagated BL height uncertainties in quadrature. In the final budgets, however, we add propagated BL height uncertainties first, before adding them in quadrature to the individual RMS uncertainty estimates.
c. Horizontal advection
We now estimate the energy and water deposited into the boundary layer by advection of liquid water potential temperature and total water along the transect using ERA-Interim data, at 0.5° horizontal resolution.
To estimate the total advection into the BL, we integrate from the surface to h, the BL top as estimated from COSMIC RO data. Since h does not in general fall exactly at a vertical level boundary, the topmost level only contributes its appropriately scaled amount. To minimize numerical noise arising from the gradient operation, which occurs at a wavelength of approximately twice the horizontal resolution (i.e., 1°), we perform zero-phase low-pass filtering on θl and qt before the gradient operation.
As our convention is that energy deposited into the BL has a positive sign, we take a positive temperature or moisture gradient moving to the east or to the north to have a positive sign. Figure 1 shows the SST and surface winds across the region. From this we can see that, by our choice of signs, both u advection and υ advection of potential temperature should tend to be negative close to sea level. Figure 7 gives the horizontal advection of θl and qt along the transect, summed up to the boundary layer, using JJA data from 2007 to 2010.
Horizontal advection of θl and total water in the BL from ECMWF, NCEP-2, and MERRA reanalysis datasets. The green shading gives the propagated BL height uncertainty estimate, and the red and blue shading give the total uncertainty estimates for these budget terms (BL height and RMS added in quadrature).
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
We have estimated the uncertainty by calculating the root-mean-square difference between advective terms calculated from ECMWF data and a mean of two independent reanalysis datasets: the 2007–10 JJA means of variables from the National Centers for Environmental Prediction (NCEP)–U.S. Department of Energy (DOE) Atmospheric Model Intercomparison Project, version 2 (AMIP-II) reanalysis 2 (NCEP-2) and the 2007–10 JJA means of variables from the Modern Era-Retrospective Analysis for Research and Applications (MERRA). These two comparison datasets were chosen arbitrarily. Because NCEP-2 does not provide liquid water at pressure levels, we calculate q instead of qt; as mentioned above, the two variables are sufficiently similar for use in an uncertainty estimate. We find RMS uncertainty estimates of 5 and 10 W m−2 for advection of θl and qt, respectively. To propagate uncertainty from the COSMIC BL height estimate, we rerun the advection calculations separately with the upper and lower bound of BL height.
d. Surface latent and sensible heat fluxes
We take estimates of surface latent and sensible heat fluxes from ERA-Interim at 0.5° resolution. We take upward surface fluxes to be positive. The ECMWF estimates of the surface heat fluxes are in broad agreement with those in Karlsson and Teixeira (2014), who give annual mean values. Latent heat increases steadily across the transect, as the SST warms and more solar radiation is absorbed by the ocean as the cloud fraction decreases.
To estimate the RMS uncertainties, we compare ECMWF fluxes along the MAGIC track to MAGIC LHF and SHF flux estimates. The MAGIC fluxes were estimate using version 3.0 of the Coupled Ocean–Atmosphere Response Experiment (COARE 3.0) bulk algorithm (Fairall et al. 2003). Our RMS uncertainty estimates are 18 and 10 W m−2 for LHF and SHF, respectively. Figure 8 shows the ECMWF surface LHF and surface SHF along the GPCI transect and along the MAGIC track and surface fluxes estimated from MAGIC data, all in the JJA mean. We note a significant discrepancy between ECMWF and MAGIC SHF, which appears unphysically low; therefore, the SHF uncertainty might be overestimated.
ECMWF surface fluxes along the GPCI transect and MAGIC track and fluxes estimated from MAGIC data. The uncertainty estimates along the GPCI transect are obtained from the RMSE between ECMWF fluxes and MAGIC fluxes along the MAGIC track.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
e. Precipitation
Condensation of water vapor generating rain deposits latent heat into the boundary layer and removes water, so long as the rain reaches the ocean. We define low rain as originating from a low cloud–top height of 3.4 km or lower, approximately 680 hPa (consistent with the definition of low clouds). We remove all precipitation contributed by high-topped synoptic-scale weather which moves through the region from time to time. We note that this filter is not applied to other variables in the analysis. We obtain precipitation rates from the CloudSat level 2C precipitation profile product (2C-RAIN-PROFILE; L’Ecuyer and Stephens 2002; Lebsock and L’Ecuyer 2011), which accounts for evaporation of falling hydrometeors in its retrieval model and as such represents an estimate of surface precipitation (Comstock et al. 2004). After averaging, the data resolution is 4° in longitude and 2° in latitude.
Figure 9 shows the BL heating from rain along the transect in watts per square meter, in the JJA climatological mean. The net BL heating from rain latent heat release is estimated from the surface precipitation amounts via a factor of 28.48 W m−2 (mm day−1)−1, which we calculate using a latent heat for condensation for water at 17°C (the approximate temperature at 850 hPa along the transect) of 2451 J g−1, obtained from an empirical cubic fit of tabulated data. The heating from the mean rain at the surface is assumed to be the integrated heating along the vertical BL column released by condensation. This heat going into the BL is a positive term in the energy budget but a negative term in the water budget. The red line shows the heating from shallow rain that was estimated to reach the surface, while the magenta line shows the total rain estimated to reach the surface, from 2007 to 2010. At least two synoptic storm systems with high rain rates are present in the data.
Condensational heating across the transect (W m−2) for the JJA mean. The red line is CloudSat data from 2007 to 2010, while the cyan line is GPCP data. The red shading gives the RMS error relative to the GPCP data along the transect. The CloudSat total rain along the transect is shown in magenta.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
We once again estimate the uncertainty by calculating the RMS error with a secondary dataset, this time from the Global Precipitation Climatology Project (GPCP) version 2.2 JJA climatology from 1979 to 2010 (Adler et al. 2003). This RMS error is 6 W m−2. However, the condensational heating cannot be less than zero, and we account for this in the uncertainty (red shading in the figure).
f. Radiation
Downwelling SW radiation is absorbed by the BL atmosphere, while upwelling LW radiation is emitted by the clear atmosphere and cloud tops. We estimate the LW and SW contributions to the BL energy budget by using the CloudSat level 2B lidar radiation fluxes and heating rate products (2B-FLXHR-lidar) retrieved from instruments on the A-Train satellites: CloudSat’s Cloud Profiling Radar, CALIPSO’s Cloud–Aerosol Lidar with Orthogonal Polarization, and Aqua’s Moderate Resolution Imaging Spectroradiometer (Henderson et al. 2013). The 2B-FLXHR-lidar SW estimates have been normalized as appropriate to the latitude, day, and local time (LT) of each observation in an attempt to remove the diurnal cycle of solar insolation. We take the mean of retrievals from the ascending and descending nodes (with flyover times of 1330 and 0130 LT, respectively), which smooths out the relatively small LW diurnal cycle. We produce a monthly climatology product at a resolution of 4° in longitude, 2° in latitude, and 250 m in vertical.
We can convert a heating rate Q (in kelvin per day) to power added to a slab of thickness δh = 250 m (in watts per square meter) via the factor ρcpδh/(24 × 3600). Summing the heating contributions from slabs within the boundary layer is equivalent to evaluating ΔBLFSW,LW from Eq. (11). As in the case of horizontal advection, in summing over the BL, the topmost slab only contributes its appropriately scaled amount since the BL height h does not in general fall on a slab boundary. Figure 10 shows the heating rates integrated vertically over the BL along the transect, in the JJA mean from 2007 to 2010.
Radiation heating (LW and SW) in the BL along the transect. The dotted line shows the LW data before low-pass filtering. Green shading shows the propagated BL height uncertainty, while blue and red shading show the total uncertainty for these terms.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
There is no ancillary dataset of atmospheric heating rates for comparison, but Henderson et al. (2013) give RMS estimates between 2B-FLXHR-lidar and the Aqua satellite’s Clouds and the Earth’s Radiant Energy System (CERES) top-of-atmosphere (TOA) outgoing fluxes for various resolutions and time scales. With a mean of 15 months of data in our study, we choose the 2°, 1-yr estimates: 19 W m−2 for SW and 6 W m−2 for LW. While TOA fluxes are fundamentally different from our BL column sum of heating rates, we nonetheless take these RMS errors (on a relative basis) as characteristic uncertainty estimates.
We have added in quadrature the propagated bias uncertainty from the COSMIC BL height estimate. For the LW data, we have also added in quadrature the RMS error between the low-pass-filtered and the raw data. The uncertainty in the LW term is dominated by the propagated BL height uncertainty. The mean difference along the transect between the mean ascending and descending node observations of the LW heating rate is 5 W m−2. The error from taking the ascending/descending average to be representative of the diurnal mean is smaller than this and therefore insignificant relative to the propagated BL height uncertainty.
g. Subsidence
As with horizontal advection, to estimate the uncertainty in the ECMWF-derived subsidence, we compare it to the 2007–10 JJA mean of NCEP-2 data and MERRA data along the transect, at the BL height h. We add in quadrature the propagated bias from the COSMIC BL height estimate. Figure 11 shows a plot of estimates from the reanalyses, in the JJA average from 2007 to 2010. Table 1 gives subsidence estimates and uncertainties in different regimes of the transect.
ECMWF, NCEP-2, and MERRA subsidence rates at h in the JJA climatological mean. Negative values indicate downward motion. The dotted red line shows the ECMWF subsidence before low-pass filtering. Green shading shows the propagated BL height uncertainty, and red shading shows the total uncertainty.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
The JJA climatological entrainment and subsidence rates along the transect from this study (i.e., from the ECMWF-derived subsidence and the COSMIC BL height) and from Wood and Bretherton (2004).
Our uncertainty estimate is high relative to the estimate of 20% for the northeastern Pacific given in Wood and Bretherton (2004). They compare the reanalysis surface divergence rates to satellite observations to estimate the subsidence uncertainty.
h. Cross-inversion changes in θl and qt
To estimate the entrainment of θl and qt, the second to last terms in the budget equations [Eqs. (10) and (11)], we must estimate the values of the variables just above the inversion. We can then estimate Δq and Δθl at points along the transect. This task would be difficult or impossible using reanalysis data because of its low vertical resolution. Fortunately, we can use radiosonde data from the MAGIC campaign to construct a lookup table for mean values of Δq and Δθl.
Once the inversion is identified, we find the boundaries of the inversion and use them to estimate the inversion jumps in our variables of interest. Based on visual inspections of soundings, we estimate uncertainties in Δqt and Δθl to be 8% and 10%, respectively. These uncertainties come from occasional apparent irregularities in the radiosonde data, algorithm overestimates or underestimates of maximum values of θ and minimum values of q at the inversion top, and occasional weak “stair step” inversions where the top of the inversion is not clear. We note that 8 of the 212 individual radiosonde BL height estimates (see Fig. 6) appear to be anomalously high. We examined the temperature and specific humidity profiles corresponding to these eight data points. In five of the cases, the BL height algorithm appears to have made a realistic choice; in the other three cases, the algorithm misidentified the inversion or no clear inversion was present. As some other inversions are estimated too low, we decided not to cut these three points. We also calculated errors on the mean from individual measurements, which were 5% for both Δqt and Δθl; we use the larger uncertainty estimates when propagating uncertainties. Figure 12 shows estimates of Δqt and Δθl for soundings launched in JJA during the course of the MAGIC campaign.
The Δqt and Δθl from MAGIC radiosonde profiles for the JJA mean, along the MAGIC transect. Individual points correspond to individual radiosondes.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
We note that, while Caldwell et al. (2005) correct for apparently thicker inversions in θl due to possible condensation on the radiosonde, which then evaporates above the actual inversion, we do not. While we did see this effect in some soundings—where the inversion as traced in the θ sounding was thicker than the inversion as traced in the RH or q soundings—we included this effect in our uncertainty estimate, and there were approximately as many soundings that appeared to overestimate Δθ. We do not believe that this effect is a significant source of uncertainty in our analysis.
Finally, we map the measured inversion jumps along the MAGIC transect to estimated jumps along our portion of the GPCI transect by interpolating the MAGIC values linearly between 15° and 35°N along the GPCI transect. While simple, this mapping is rather crude; based on comparison to interpolating the MAGIC values between 10° and 35°N, we estimate that this mapping contributes an additional 3 W m−2 of uncertainty to the entrainment terms. (This additional uncertainty is included in the error bar in Fig. 14.) Our estimates of cross-inversion jumps Δq and Δθl agree well with those given in Caldwell et al. (2005), who find values in the ranges from −7 to −6 g kg−1 and 9–11 K, respectively, in the southeastern Pacific subtropical basin at 20°S, 85°W.
i. Deviation from an ideal mixed layer
To estimate the final terms in the budget equations, we estimate
4. Results
We are now nearly in a position to combine the budget terms into energy and water budgets along the transect. First, we must estimate entrainment rates.
a. Entrainment rates
In this section, we estimate the climatological entrainment rates along the transect from the ECMWF subsidence data and COSMIC observations of the BL height and the entrainment of θl and qt.

Figure 13 shows the resulting entrainment rates along the GPCI transect. We have low-pass filtered the subsidence along the transect to remove numerical noise. Component sources of uncertainty include the RMS uncertainties from the two ECMWF components, subsidence and the horizontal winds at the BL height h; the COSMIC component, the gradient of h; and propagated BL height uncertainties in
The climatological entrainment rate along the transect for the JJA season. Green shading shows the propagated BL height uncertainty, and blue shading shows the total uncertainty estimate, calculated as described in the text.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
Figure 14 shows the entrainment of θl and qt along the transect, which are terms in the energy and water budgets. The RMS uncertainties are estimated from the quadrature sum of the entrainment RMS uncertainty and the uncertainties in the cross-inversion changes; these are then added in quadrature to the propagated BL height uncertainties.
The climatological entrainment of θl and qt along the transect for the JJA season. Green shading shows the propagated BL height uncertainty, and red or blue shading shows the total uncertainty estimate for the corresponding term, calculated as described in the text.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
In Table 1, we give our estimates of the entrainment rates along the transect and compare them to those in Wood and Bretherton (2004). We note that we do not observe the marked decrease along the transect reported by Wood and Bretherton (2004); two factors contributing to the discrepancy could be the bias in the COSMIC BL height near the coast and the use of the NCEP reanalysis to estimate subsidence in Wood and Bretherton (2004) versus our use of the ECMWF reanalysis. Furthermore, Wood and Bretherton (2004) report climatological results from September–October 2000, as opposed to our 2007–10 JJA climatology. Caldwell et al. (2005) report a mean entrainment value of 4 ± 1 mm s−1 over a 6-day period in the southeastern Pacific at 20°S, 85°W. Faloona et al. (2005) report values of 1.2–7.2 mm s−1, observed during seven research flights of DYCOMS-II in a relatively small region off the coast of San Diego (29°–32.5°N, 120.5°–123°W). Duynkerke et al. (2004) give entrainment rates for LES of stratocumulus, using forcings from the First ISCCP Regional Experiment (FIRE) campaign (Albrecht et al. 1988), of 5.8 and 3.6 mm s−1 for daytime and nighttime, respectively, whereas Stevens et al. (2005) present LES results of between 3.8 and 5.9 mm s−1 using forcings from the DYCOMS-II research flight 01.
b. Observational energy and water budgets along the transect
Figure 15 shows the energy budget from 35° to 15°N along the GPCI transect. This budget is a diurnally averaged climatological JJA mean from 2007 to 2011.
The climatological boundary layer energy budget along the transect for the JJA mean. Green shading shows the propagated BL height uncertainty, and gray shading shows the total uncertainty estimate, calculated as described in the text.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
Table 2 summarizes the individual terms in the energy budget for the mean along the transect from 35° to 15°N, and also in increments of 5° latitude. For comparison, values from a previous observational study (Caldwell et al. 2005) are also given. Caldwell et al. (2005) used 6 days of ship data, in October 2001, from a region approximately 1500 km off the coast of Chile (20°S, 85°W). The CloudSat/CALIPSO low cloud mean CF for that location in October 2007 was 0.9. The maximum CF along the transect in our study of 0.8 occurs at 30°N (Fig. 4).
The JJA climatological boundary layer energy budget along the transect from this study. All values are in watts per square meter. Positive values indicate energy added into the BL. Also shown are results from the study of Caldwell et al. (2005) that was located at 20°S, 85°W over 6 days in October 2001. The CloudSat/CALIPSO low cloud mean for that location in October 2007 was 0.9, similar to 30°N along the transect in our study.
Figure 16 shows the water budget along the transect. Table 3 gives the individual terms in the water budget and compares them to values given in Caldwell et al. (2005).
As in Fig. 15, but for the boundary layer water budget.
Citation: Journal of Climate 27, 24; 10.1175/JCLI-D-14-00242.1
The JJA climatological boundary layer water budget along the transect from this study. Also shown are results from Caldwell et al. (2005). All values are in watts per square meter. Positive values indicate water added into the BL.
In both the energy and water budgets, our results between 20° and 30°N are remarkably similar to those given by Caldwell et al. (2005). The primary difference is that the Caldwell et al. (2005) study observes more LW cooling, which is balanced by more warming from entrainment and SW absorption than we observe in our study.
Because we assume that the mean in the time derivative (storage) terms is zero over the 15 summer months of our study, our residuals represent the degree to which the budget is in error (i.e., not equal to zero). This is also the interpretation of the Caldwell et al. (2005) residuals, although they do estimate the time derivative terms from observations. For their energy and water budgets, this term is 0 and −5 W m−2, respectively. For the sake of comparison, we subtract this term from their water budget residual.
Our uncertainty estimates take into account both bias errors and sampling uncertainties. We add propagated biases due to the COSMIC BL height estimate from the various budget terms together (not in quadrature), before adding the RMSE estimates (in quadrature). We note that the uncertainties in Caldwell et al. (2005) are sampling uncertainties only: that is, the standard deviations on the mean values.
5. Discussion and conclusions
Using satellite and surface observations combined with reanalysis data, we have constructed the climatological water and energy budgets of the subtropical marine boundary layer. We have focused on the transition from marine stratocumulus to trade cumulus occurring along the GPCI transect from 35° to 15°N.
We see no sharp transition between these two BL cloud regimes in the JJA climatological mean examined here (which does not exclude the possibility of sharp transitions at any given moment). Instead, we observe a gradual transition. Nonetheless, it is convenient (if somewhat artificial) to map the two major regimes to portions of the transect: the stratocumulus regime is roughly between 25° and 35°N and the cumulus regime is roughly between 15° and 25°N. Figure 4 shows that the cloud fraction increases from the coast to a peak in middle of the stratocumulus regime, and then begins decreasing steadily at about 25°N. Meanwhile, Fig. 5 shows an approximately linear increase in COSMIC BL height from 35° to 15°N. We note that, in this same figure, the cloud-top height exhibits more of a dual regime than a linear BL height climatology. Choosing cloud-top height instead of COSMIC BL height might have caused some of the budget terms to take on more of a dual-regime shape as they evolve along the transect.
The water budget is dominated by an approximately linear increase in evaporation along the transect. This increase is balanced by increases in rain, horizontal advection drying, and entrainment drying of roughly equal magnitudes. The energy budget, on the other hand, shows comparatively subtle differences between the two regimes. Small increases in rain heating and shortwave radiation of 15 and 5 W m−2, respectively, which together account for only about 25% of the total energy budget magnitude, are balanced by a 6 W m−2 decrease in entrainment heating, a 4 W m−2 increase in advective cooling, and 5 W m−2 due to an increasing tendency for the BL to be unmixed. Surprisingly, we do not observe a significant change in the longwave term between regimes, despite a gradual decrease in low cloud cover from 85% to 55% across the transition. The absence of a significant LW cooling gradient along the transect is due to the increase in BL water vapor that balances the decrease in cloudiness.
In the course of constructing the energy and water budgets, we have also estimated the cloud-top entrainment rate along the transect. Our value, with a mean of 3.5 ± 1.5 mm s−1, agrees with previous observational estimates.
Both the energy and water budgets close to within 3 W m−2 in the mean along the transect, well within the total uncertainty budget obtained from uncertainty estimates in each of the budget terms. In fact, our uncertainty on the budget closure is about an order of magnitude greater than the residuals themselves. While the low residuals are promising, the large disagreement between different datasets, especially reanalysis datasets used to estimate subsidence and horizontal advection, which contribute significantly to our high uncertainties, is troubling and may serve as a caution to investigations depending on a particular reanalysis dataset. Indeed, future regional budget analyses could contribute to a more fundamental understanding of the strengths and weaknesses of the various datasets.
Much of the uncertainty in our analysis is tied to the BL height, especially near the coast. The budget terms contributing the most uncertainty are the entrainment terms, which depend on the subsidence rate at the inversion height. It is challenging to directly measure subsidence, and it is challenging to directly measure entrainment. However, if uncertainty in the BL height could be reduced significantly, an improvement in either entrainment or subsidence measurements would translate into improved knowledge of the other variable. If advective terms can be precisely measured with soundings, for example, it may be possible to construct regional budgets that constrain estimates of subsidence and entrainment to unprecedented precision.
Our budget analysis provides a holistic view of the subtropical marine boundary layer in the climatological mean, which should be useful for validating and improving global models, large-eddy simulations (e.g., Chung et al. 2012; van der Dussen et al. 2013), parameterizations, intercomparisons such as the Cloud Feedback Model Intercomparison Project (CFMIP)/Global Energy and Water Cycle Experiment (GEWEX) Atmospheric System Studies (GASS) Intercomparison of Large-Eddy and Single-Column Models (CGILS; Zhang et al. 2013), and in particular attempts to unify the parameterizations of BL cloud regimes (e.g., Golaz et al. 2002; Suselj et al. 2013).
The dataset developed and assembled in this work has the potential to provide deeper theoretical interpretations of the physical processes behind the transition. In particular, the development of simple and more theoretical models of the SCT transition based on these datasets is a key goal of our research in the medium to long term. However, it should be noted that currently the most successful theoretical models of the physics of subtropical BL clouds are concerned with the two extremes of the transition: mixed-layer models provide a good understanding of stratocumulus boundary layers and mass-flux models of cumulus BLs, but key implicit assumptions of these models like cloud cover being either 100% (for mixed-layer models) or zero (for mass-flux models) make them somewhat unsuitable for deeper investigations of the physics of the SCT transition.
Mixed-layer and mass-flux models of the subtropical BL cannot explain and fully represent the physics of the transition, and in this context the current dataset is more complex and is not really suitable for investigations that only make use of mixed-layer and/or mass-flux models of the BL. In fact, the dataset we developed may well turn out to be ideal for what is currently clearly needed (and lacking), which is new theoretical models of the subtropical BL that fully represent the physics of the transition.
Acknowledgments
The research described in this publication was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. We also acknowledge the NASA MEaSUREs program. Data were obtained from the Atmospheric Radiation Measurement Program (ARM) sponsored by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, Climate and Environmental Sciences Division. Data were obtained from the NASA CloudSat project. NCEP-2 data were provided by the NOAA/OAR/ESRL/PSD, Boulder, Colorado. We thank Chi Ao for providing COSMIC data and Mike Reynolds for providing MAGIC surface flux data. We are grateful for helpful suggestions from three anonymous referees.
REFERENCES
Adler, R. F., and Coauthors, 2003: The Version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 1147–1167, doi:10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.
Albrecht, B. A., D. A. Randall, and S. Nicholls, 1988: Observations of marine stratocumulus clouds during FIRE. Bull. Amer. Meteor. Soc., 69, 618–626, doi:10.1175/1520-0477(1988)069<0618:OOMSCD>2.0.CO;2.
Albrecht, B. A., C. S. Bretherton, D. Johnson, W. H. Scubert, and A. S. Frisch, 1995: The Atlantic Stratocumulus Transition Experiment—ASTEX. Bull. Amer. Meteor. Soc., 76, 889–904, doi:10.1175/1520-0477(1995)076<0889:TASTE>2.0.CO;2.
Ao, C. O., D. E. Waliser, S. K. Chan, J.-L. Li, B. Tian, F. Xie, and A. J. Mannucci, 2012: Planetary boundary layer heights from GPS radio occultation refractivity and humidity profiles. J. Geophys. Res.,117, D16117, doi:10.1029/2012JD017598.
Bretherton, C., and M. C. Wyant, 1997: Moisture transport, lower-tropospheric stability, and decoupling of cloud-topped boundary layers. J. Atmos. Sci., 54, 148–167, doi:10.1175/1520-0469(1997)054<0148:MTLTSA>2.0.CO;2.
Bretherton, C., S. K. Krueger, M. C. Wyant, P. Bechtold, E. Meijgaard, B. Stevens, and J. A. Teixeira, 1999: A GCSS boundary-layer cloud model intercomparison study of the first ASTEX Lagrangian experiment. Bound.-Layer Meteor., 93, 341–380, doi:10.1023/A:1002005429969.
Caldwell, P., C. S. Bretherton, and R. Wood, 2005: Mixed-layer budget analysis of the diurnal cycle of entrainment in southeast Pacific stratocumulus. J. Atmos. Sci.,62, 3775–3791, doi:10.1175/JAS3561.1.
Chen, T., W. B. Rossow, and Y. Zhang, 2000: Radiative effects of cloud-type variations. J. Climate, 13, 264–286, doi:10.1175/1520-0442(2000)013<0264:REOCTV>2.0.CO;2.
Chung, D., and J. Teixeira, 2012: A simple model for stratocumulus to shallow cumulus cloud transitions. J. Climate, 25, 2547–2554, doi:10.1175/JCLI-D-11-00105.1.
Chung, D., G. Matheou, and J. A. Teixeira, 2012: Steady-state large-eddy simulations to study the stratocumulus to shallow-cumulus cloud transition. J. Atmos. Sci., 69, 3264–3276, doi:10.1175/JAS-D-11-0256.1.
Comstock, K. K., R. Wood, S. E. Yuter, and C. S. Bretherton, 2004: Reflectivity and rain rate in and below drizzling stratocumulus. Quart. J. Roy. Meteor. Soc., 130, 2891–2918, doi:10.1256/qj.03.187.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc.,137, 553–597, doi:10.1002/qj.828.
Duynkerke, P. G., and Coauthors, 2004: Observations and numerical simulations of the diurnal cycle of the EUROCS stratocumulus case. Quart. J. Roy. Meteor. Soc.,130, 3269–3296, doi:10.1256/qj.03.139.
Fairall, C. W., E. F. Bradley, J. E. Hare, A. A. Grachev, and J. B. Edson, 2003: Bulk parameterization of air–sea fluxes: Updates and verification for the COARE algorithm. J. Climate, 16, 571–591, doi:10.1175/1520-0442(2003)016<0571:BPOASF>2.0.CO;2.
Faloona, I., and Coauthors, 2005: Observations of entrainment in eastern Pacific marine stratocumulus using three conserved scalars. J. Atmos. Sci., 62, 3268–3285, doi:10.1175/JAS3541.1.
Golaz, J.-C., V. E. Larson, and W. R. Cotton, 2002: A PDF-based model for boundary layer clouds. Part II: Model results. J. Atmos. Sci.,59, 3552–3571, doi:10.1175/1520-0469(2002)059〈3552:APBMFB〉2.0.CO;2.
Henderson, D. S., T. L’Ecuyer, G. Stephens, P. Partain, and M. Sekiguchi, 2013: A multisensor perspective on the radiative impacts of clouds and aerosols. J. Appl. Meteor. Climatol., 52, 853–871, doi:10.1175/JAMC-D-12-025.1.
Karlsson, J., and J. Teixeira, 2014: A simple model of the northeast Pacific stratocumulus to cumulus transition based on the climatological surface energy budget. J. Climate, 27, 4111–4121, doi:10.1175/JCLI-D-13-00534.1.
Kawai, H., and J. A. Teixeira, 2010: Probability density functions of liquid water path and cloud amount of marine boundary layer clouds: Geographical and seasonal variations and controlling meteorological factors. J. Climate, 23, 2079–2092, doi:10.1175/2009JCLI3070.1.
Klein, S. A., and D. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate,6, 1587–1606, doi:10.1175/1520-0442(1993)006<1587:TSCOLS>2.0.CO;2.
Lebsock, M. D., and T. S. L’Ecuyer, 2011: The retrieval of warm rain from CloudSat. J. Geophys. Res.,116, D20209, doi:10.1029/2011JD016076.
L’Ecuyer, T. S., and G. L. Stephens, 2002: An estimation-based precipitation retrieval algorithm for attenuating radars. J. Appl. Meteor., 41, 272–285, doi:10.1175/1520-0450(2002)041<0272:AEBPRA>2.0.CO;2.
Mace, G. G., Q. Zhang, M. Vaughan, R. Marchand, G. Stephens, C. Trepte, and D. Winker, 2009: A description of hydrometeor layer occurrence statistics derived from the first year of merged Cloudsat and CALIPSO data. J. Geophy. Res.,114, D00A26, doi:10.1029/2007JD009755.
Randall, D. A., J. J. A. Coakley, C. W. Fairall, R. A. Kropfli, and D. H. Lenschow, 1984: Outlook for research on subtropical marine stratiform clouds. Bull. Amer. Meteor. Soc., 65, 1290–1301, doi:10.1175/1520-0477(1984)065<1290:OFROSM>2.0.CO;2.
Rossow, W. B., and R. A. Schiffer, 1991: ISCCP cloud data products. Bull. Amer. Meteor. Soc., 72, 2–20, doi:10.1175/1520-0477(1991)072<0002:ICDP>2.0.CO;2.
Rossow, W. B., and R. A. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc., 80, 2261–2287, doi:10.1175/1520-0477(1999)080<2261:AIUCFI>2.0.CO;2.
Sandu, I., and B. Stevens, 2011: On the factors modulating the stratocumulus to cumulus transitions. J. Atmos. Sci., 68, 1865–1881, doi:10.1175/2011JAS3614.1.
Sandu, I., B. Stevens, and R. Pincus, 2010: On the transitions in marine boundary layer cloudiness. Atmos. Chem. Phys., 10, 2377–2391, doi:10.5194/acp-10-2377-2010.
Slingo, A., 1990: Sensitivity of the earth’s radiation budget to changes in low clouds. Nature, 343, 49–51, doi:10.1038/343049a0.
Stephens, G. L., 2005: Cloud feedbacks in the climate system: A critical review. J. Climate,18, 237–273, doi:10.1175/JCLI-3243.1.
Stephens, G. L., and Coauthors, 2012: An update on Earth’s energy balance in light of the latest global observations. Nat. Geosci., 5, 691–696, doi:10.1038/ngeo1580.
Stevens, B., and Coauthors, 2003: Dynamics and chemistry of marine stratocumulus–DYCOMS-II. Bull. Amer. Meteor. Soc.,84, 579–593, doi:10.1175/BAMS-84-5-579.
Stevens, B., and Coauthors, 2005: Evaluation of large-eddy simulations via observations of nocturnal marine stratocumulus. Mon. Wea. Rev., 133, 1443–1462, doi:10.1175/MWR2930.1.
Suselj, K., J. Teixeira, and D. Chung, 2013: A unified model for moist convective boundary layers based on a stochastic eddy-diffusivity/mass-flux parameterization. J. Atmos. Sci., 70, 1929–1953, doi:10.1175/JAS-D-12-0106.1.
Teixeira, J., and Coauthors, 2011: Tropical and subtropical cloud transitions in weather and climate prediction models: The GCSS/WGNE Pacific Cross-Section Intercomparison (GPCI). J. Climate, 24, 5223–5256, doi:10.1175/2011JCLI3672.1.
Trenberth, K. E., L. Smith, T. Qian, A. Dai, and J. Fasullo, 2007: Estimates of the global water budget and its annual cycle using observational and model data. J. Hydrometeor., 8, 758–769, doi:10.1175/JHM600.1.
Trenberth, K. E., J. T. Fasullo, and J. Kiehl, 2009: Earth’s global energy budget. Bull. Amer. Meteor. Soc., 90, 311–323, doi:10.1175/2008BAMS2634.1.
Vaisala Oyj, 2013: Humidity conversion formulas. Vaisala Tech. Rep. B210973EN-F, 17 pp. [Available online at http://www.vaisala.com/Vaisala%20Documents/Application%20notes/Humidity_Conversion_Formulas_B210973EN-F.pdf.]
van der Dussen, J. J., and Coauthors, 2013: The GASS/EUCLIPSE model intercomparison of the stratocumulus transition as observed during ASTEX: LES results. J. Adv. Model. Earth Syst.,5, 483–499, doi:10.1002/jame.20033.
von Engeln, A., and J. Teixeira, 2013: A planetary boundary layer height climatology derived from ECMWF reanalysis data. J. Climate, 26, 6575–6590, doi:10.1175/JCLI-D-12-00385.1.
von Engeln, A., J. Teixeira, J. Wickert, and S. A. Buehler, 2005: Using champ radio occultation data to determine the top altitude of the planetary boundary layer. Geophys. Res. Lett.,32, L06815, doi:10.1029/2004GL022168.
Wong, S., E. J. Fetzer, B. H. Kahn, B. Tian, B. H. Lambrigtsen, and H. Ye, 2011: Closing the global water vapor budget with AIRS water vapor, MERRA reanalysis, TRMM and GPCP precipitation, and GSSTF surface evaporation. J. Climate, 24, 6307–6321, doi:10.1175/2011JCLI4154.1.
Wood, R., 2012: Stratocumulus clouds. Mon. Wea. Rev., 140, 2373–2423, doi:10.1175/MWR-D-11-00121.1.
Wood, R., and C. Bretherton, 2004: Boundary layer depth, entrainment, and decoupling in the cloud-capped subtropical and tropical marine boundary layer. J. Climate,17, 3576–3588, doi:10.1175/1520-0442(2004)017〈3576:BLDEAD〉2.0.CO;2.
Wyant, M. C., C. 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, doi:10.1175/1520-0469(1997)054<0168:NSAACM>2.0.CO;2.
Zhang, M., and Coauthors, 2013: CGILS: Results from the first phase of an international project to understand the physical mechanisms of low cloud feedbacks in single column models. J. Adv. Model. Earth Syst.,5, 826–842, doi:10.1002/2013MS000246.
Zhu, P., and Coauthors, 2005: Intercomparison and interpretation of single-column model simulations of a nocturnal stratocumulus-topped marine boundary layer. Mon. Wea. Rev., 133, 2741–2758, doi:10.1175/MWR2997.1.
Data are available at the ARM data archive (http://www.archive.arm.gov).