1. Introduction
Stratocumulus clouds in the tropics and subtropics are known to be the “climate refrigerators” by reflecting a significant amount of sunlight back to space, and a 5% increase of their coverage would be sufficient to offset the global warming induced by doubling CO2 (Randall et al. 1984; Slingo 1990; Bretherton et al. 2004; Wood 2012), and thus accurate simulation of stratocumulus clouds in climate models is crucial for long-term climate predictions. However, stratocumulus clouds have been poorly simulated by global climate models, and the feedback associated with stratocumulus and shallow cumulus clouds has been demonstrated to be one of the largest sources of uncertainty in climate change projections (Bony and Dufresne 2005; Webb et al. 2006; Medeiros et al. 2008; Soden and Vecchi 2011; Qu et al. 2014). Stratocumulus clouds also play an important role in annual mean and seasonal cycle of the eastern Pacific Ocean and global climate (Ma et al. 1996; Nigam 1997; Yu and Mechoso 1999; Gordon et al. 2000), and in determining the strength and phase of El Niño–Southern Oscillation (Gudgel et al. 2001).
The stratocumulus cloud deck over the southeastern Pacific Ocean (SEP) is the largest and most persistent subtropical stratocumulus deck in the world (Wood et al. 2011; Mechoso et al. 2014). It is formed and maintained by complex interactions among the underlying cold SST, overcapping warm and dry air, and complicated internal physical processes (Fig. 1). The underlying cold SST is formed by intense coastal upwelling driven by alongshore winds associated with the South Pacific subtropical high and the barrier of Andes Cordillera (Garreaud and Muñoz 2005), together with the reduction of solar heating by stratocumulus clouds. The overcapping warm and dry air is driven by the large-scale subsidence associated with the sinking branches of Walker circulation and Hadley circulation, which is enhanced by orographic effects of the Andes (Richter and Mechoso 2006). The internal physical processes involve cloud microphysics, cloud-top radiative cooling, marine boundary layer turbulence, drizzles, aerosols, and regional diurnal circulations (Wood 2012; Bretherton et al. 2004). Convective instability and turbulence in the stratocumulus-topped boundary layer (STBL) are driven mainly by the cloud-top longwave cooling and evaporative cooling, which are partially reduced by shortwave warming and latent heating inside the cloud layer, and the resulting STBL turbulence is enhanced by latent heating in updrafts and cooling in downdrafts. Turbulent eddies and evaporative cooling drives entrainment at the top of the STBL, which tends to deepen the STBL, maintaining it against large-scale subsidence. Drizzle reduces the liquid water path and albedo and can lead to increased mesoscale variability, stratification of the STBL, and in some cases cloud breakup. For a given cloud thickness, polluted clouds tend to produce more numerous and smaller cloud droplets, greater cloud albedo, and drizzle suppression. Feedbacks between radiative cooling, precipitation formation, turbulence, and entrainment help regulate stratocumulus. The stratocumulus cloud cover is well correlated with the lower troposphere stability (LTS) and estimated inversion strength (EIS), both of which are measures of the temperature inversion strength (Slingo 1987; Klein and Hartmann 1993; Norris 1998; Wood and Bretherton 2006). The liquid water path of the SEP stratocumulus clouds is often close to the adiabatic value, and thus determined by cloud thickness (Zuidema et al. 2005, 2012; Bretherton et al. 2004, 2010). The cloud thickness is primarily maintained by a strongly negative cloud–radiation–turbulent–entrainment feedback (Zhu et al. 2005), and the thickness could vary due to changes in turbulent driving, vertical gradient of moisture and moist static energy, large-scale subsidence, and inversion strength (Zhu et al. 2007; Zhang and Bretherton 2008; Caldwell and Bretherton 2009; Brient and Bony 2013; Bretherton et al. 2013).
Schematic depiction of the large-scale forcing and physical processes for a stratocumulus-topped boundary layer. LTS is lower troposphere stability and EIS is estimated inversion strength. Adapted from Bretherton et al. (2004), Wood (2012), and Wood and Bretherton (2006).
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Schematic depiction of the large-scale forcing and physical processes for a stratocumulus-topped boundary layer. LTS is lower troposphere stability and EIS is estimated inversion strength. Adapted from Bretherton et al. (2004), Wood (2012), and Wood and Bretherton (2006).
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Schematic depiction of the large-scale forcing and physical processes for a stratocumulus-topped boundary layer. LTS is lower troposphere stability and EIS is estimated inversion strength. Adapted from Bretherton et al. (2004), Wood (2012), and Wood and Bretherton (2006).
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Because of the complex physical processes and feedbacks described above, the SEP stratocumulus cloud deck has proven particularly difficult for climate models to simulate (Hannay et al. 2009; Wyant et al. 2010). Hannay et al. (2009) evaluated the hindcasts of SEP stratocumulus clouds at 20°S and 85°W during the Eastern Pacific Investigation of Climate (EPIC) cruise of October 2001 by one operational forecast model and three global climate models. They found that the planetary boundary layer (PBL) depth is too shallow in the models when compared to EPIC observations, and improved PBL depths are achieved with more physically realistic PBL schemes. All the models produce a strong diurnal cycle in the liquid water path (LWP), but there are large differences in the amplitude and phase when compared to the EPIC observations. Wyant et al. (2010) evaluated the simulations of SEP stratocumulus by eight regional models, six operational forecast models, and three global climate models. They found that the models have unrealistic spatial patterns of mean cloud cover with only a few models agreeing well with satellite observations. Most models also underestimate the marine boundary layer depth in the eastern part of the stratocumulus region and the diurnal cycle of liquid water path at 20°S and 85°W.
Difficulty in simulating stratocumulus clouds also exists in many climate models participating in the Intergovernmental Panel on Climate Change (IPCC) assessment reports (Clement et al. 2009; Caldwell et al. 2013). Clement et al. (2009) evaluated the low-level clouds in the northeast Pacific simulated by 18 coupled climate models participating in the World Climate Research Programme’s (WCRP’s) phase 3 of the Coupled Model Intercomparison Project (CMIP3). They found that only one of the 18 models simulates the correct sign correlations between cloud cover in the northeast Pacific and the local thermal structure (SST and LTS) and circulation [sea level pressure (SLP) and midtropospheric vertical velocity] associated with a weakening of tropical atmospheric circulation under increased greenhouse gases. Caldwell et al. (2013) examined the stratocumulus clouds in five marine stratocumulus cloud regions simulated by 10 CMIP3 coupled models and found that model cloud fraction compares poorly with observations of mean state, variability, and correlation with estimated inversion strength.
Recently, in preparation for the IPCC Fifth Assessment Report (AR5), phase 5 of the Coupled Model Intercomparison Project (CMIP5) has conducted a comprehensive set of long-term climate simulations for both the twentieth century’s climate and various climate change scenarios for the twenty-first century (Taylor et al. 2012). Since the Fourth Assessment Report (AR4) and CMIP3 model simulations were conducted, modeling groups have made considerable efforts on developing their climate models. As a result, the CMIP5 models on average have higher horizontal resolutions than those used in the CMIP3 models, and have improved subgrid-scale parameterizations that have been newly developed or revised from that used in the CMIP3 version. Some of the CMIP5 models have also evolved from “climate system models” to “Earth system models” that include biogeochemical components and time-varying carbon fluxes between the ocean, atmosphere, and terrestrial biosphere. In collaboration with CMIP5, the Cloud Feedback Model Intercomparison Project (CFMIP; Bony et al. 2011) phase II developed the CFMIP Observation Simulator Package (COSP; Bodas-Salcedo et al. 2011; Pincus et al. 2012), making it possible to directly compare the model simulated clouds with satellite observations. So far, eight CMIP5 models have made their CFMIP simulations available. Therefore, it is of interest to examine the SEP stratocumulus simulated by this new generation of global climate models.
The purpose of this study is to evaluate the SEP stratocumulus cloud deck simulated by eight CMIP5–CFMIP climate models. The questions to be addressed are these:
How well do the CMIP5–CFMIP models simulate the climatology and variability of SEP stratocumulus clouds?
How well do the CMIP5–CFMIP models simulate the cloud–radiation feedback associated with SEP stratocumulus clouds at seasonal to decadal time scales?
Are the cloud biases in the models, if any, associated with possible biases in large-scale environment?
This paper is organized as follows. The models and validation datasets used in this study are described in section 2. The results are presented in section 3. A summary and discussion are given in section 4.
2. Models and validation datasets
We use 27 years (model years 1979–2005) of the historical simulations from eight global climate models participating in CMIP5 and CFMIP. All models except CAM5 are coupled GCMs; CAM5 is an atmosphere GCM forced by observed SST. CAM5 is the atmospheric component of National Center for Atmospheric Research (NCAR) Community Earth System Model version 1.0 (CESM1.0). CAM5 was used because CFMIP simulations are available only for CAM5 but not for the coupled CESM1.0. Table 1 shows the model names and acronyms, their horizontal and vertical resolutions, and their PBL, convection, and cloud schemes. For each model we use 27 years of monthly mean longwave and shortwave radiative fluxes for both all sky and clear sky, SST, air temperature, and cloud properties from International Satellite Cloud Climatology Project (ISCCP) simulator including total cloud cover, cloud top pressure (Ptop), cloud albedo, and Ptop-cloud optical depth (τ) histogram of cloud cover. The ISCCP simulator converts the model clouds to be comparable to ISCCP observations (Klein and Jakob 1999; Webb et al. 2001).
List of the CMIP5/CFMIP global climate models analyzed in this study.
The observational datasets used for evaluating model simulations include clouds, radiative fluxes, SST, and air temperature (Table 2). For each variable, different datasets are used whenever possible in order to bracket the uncertainties associated with measurement/retrieval/analysis. The cloud radiative forcing (CRF) is defined as the difference between clear sky flux and all sky flux at the top of the atmosphere, with positive sign denoting downward flux (energy gain for the atmosphere–ocean system). We calculated the longwave CRF (LWCRF), shortwave CRF (SWCRF), and net CRF (NETCRF).
Observational datasets used in this study.
3. Results
a. Annual mean and seasonal cycle
Figure 2 shows the map of SST in observations and eight CMIP5–CFMIP models. Note that CAM5 is forced by observed SST. The observed SST isotherms tilt along the northeast–southwest direction between equator and 30°S. While this spatial pattern of SST is determined by a combination of oceanic, atmospheric, and air–sea feedback processes, coastal upwelling and horizontal heat advection of the upper ocean largely contribute to the SST distribution in this region (e.g., Zheng et al. 2010; Shinoda and Lin 2009). In the models, the isotherms generally tilt more toward the east–west direction, especially in the region close to the equator. This result is similar to the diagnosis of CMIP3 models (Zheng et al. 2011). Zheng et al. (2011) demonstrated that errors in coastal upwelling and heat advection by Ekman currents significantly contribute to the warming biases in SST. Hence similar oceanic processes may contribute to the warm SST bias in CMIP5 models, which is beyond the scope of the current study.
Horizontal map of annual mean SST (°C) for (a) observations of ERSST and (b)–(i) the eight models. The box denotes the core stratocumulus region (10°–30°S, 70°–90°W).
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Horizontal map of annual mean SST (°C) for (a) observations of ERSST and (b)–(i) the eight models. The box denotes the core stratocumulus region (10°–30°S, 70°–90°W).
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Horizontal map of annual mean SST (°C) for (a) observations of ERSST and (b)–(i) the eight models. The box denotes the core stratocumulus region (10°–30°S, 70°–90°W).
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 3 shows the horizontal map of total cloud coverage. The observed clouds show a clear local maximum along the coast within 10°–30°S and 270°–290°W. Only two models (HAES and CAM5) produce a similar pattern but with much lower cloud cover. All the other models simulate an overly low cloud cover and some of the models show a noisy spatial distribution (CANE, MIR5, and MPIE).
As in Fig. 2, but for total cloud cover (%).
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 2, but for total cloud cover (%).
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 2, but for total cloud cover (%).
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 4 displays the horizontal map of cloud-top pressure. The observed stratocumulus cloud deck shows a spatially uniform cloud-top pressure of about 700 hPa (Fig. 4a). This is lower than the cloud-top pressure observed from in situ field experiments (Bretherton et al. 2004; Wood et al. 2011), which is caused by the aforementioned underestimate of stratocumulus cloud-top pressure by the ISCCP retrieval method. Clouds in the two IPSL models are mainly high clouds with cloud-top pressure around 300 hPa (Figs. 4d,e). The other six models simulate low or middle clouds with cloud-top pressure between 550 and 700 hPa, which is slightly lower than the observed cloud-top pressure.
As in Fig. 2, but for cloud-top pressure (hPa) and ISCCP instead of ERSST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 2, but for cloud-top pressure (hPa) and ISCCP instead of ERSST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 2, but for cloud-top pressure (hPa) and ISCCP instead of ERSST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 5 displays the vertical structure of clouds represented by the Ptop–τ histogram of annual mean cloud cover averaged over 10°–30°S, 70°–90°W. The observed clouds are dominated by low and middle clouds with cloud-top pressure higher than 560 mb (1 mb = 1 hPa), especially the stratocumulus clouds with cloud-top pressure between 680 and 800 mb and τ between 3.6 and 23. Six of the eight models (CANE, HAES, MIR5, MPIE, MRIC, and CAM5) successfully reproduce the dominance of low and middle clouds, but the cloud covers tend to be smaller than observation. In the two IPSL models, however, the clouds are dominated by high clouds, especially cirrus and cirrostratus clouds. This is interesting since the IPSL models produce the best SST distribution in this region among all the coupled models, and thus the high clouds are generated over the cold SST in the model.
Ptop–τ histogram of annual mean cloud cover (%) averaged over 10°–30°S, 70°–90°W for (a) ISCCP and (b)–(i) the eight models. Cloud cover larger than 9% are shaded in red, those between 5% and 9% in magenta, and those between 1% and 5% in yellow.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Ptop–τ histogram of annual mean cloud cover (%) averaged over 10°–30°S, 70°–90°W for (a) ISCCP and (b)–(i) the eight models. Cloud cover larger than 9% are shaded in red, those between 5% and 9% in magenta, and those between 1% and 5% in yellow.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Ptop–τ histogram of annual mean cloud cover (%) averaged over 10°–30°S, 70°–90°W for (a) ISCCP and (b)–(i) the eight models. Cloud cover larger than 9% are shaded in red, those between 5% and 9% in magenta, and those between 1% and 5% in yellow.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Since the ISCCP clouds in the SEP region are dominated by clouds in the lower troposphere with cloud-top pressure greater than 560 mb, we just define those clouds as lower troposphere clouds, and their cloud coverage is called lower troposphere cloud cover (LTCC). Previous studies have shown that the ISCCP retrieval method underestimates cloud-top pressure in regions with strong temperature inversions such as the stratocumulus regions, which caused some low-level clouds to be mistakenly identified as midlevel clouds (Clement et al. 2009). Previous studies of the northeast Pacific stratocumulus region have noted that surface observations of low-level cloud cover are in better agreement with the sum of ISCCP middle plus low cloud cover than low cloud cover alone (Minnis et al. 1992; Clement et al. 2009).
Figure 6 displays the horizontal map of LTCC. The observed LTCC has similar pattern as the total cloud cover but with slightly smaller value (Fig. 6a). The two IPSL models show little LTCC (Figs. 6d,e). For the other six models, the spatial pattern of LTCC is similar to that of total cloud cover but with slightly smaller magnitude. All models significantly underestimate the LTCC.
As in Fig. 2, but LTCC (%) and ISCCP instead of ERSST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 2, but LTCC (%) and ISCCP instead of ERSST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 2, but LTCC (%) and ISCCP instead of ERSST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 7 shows the horizontal map of cloud albedo. The observed stratocumulus cloud deck has a spatially uniform cloud albedo of 0.35–0.5 (Fig. 7a). The two IPSL models simulate overly small albedo associated with high clouds (Figs. 7d,e). The other six models display a variety of different and unrealistic cloud albedo, with three models (CANE, MIR5, and MRIC) showing overly thin clouds, while three models (HAES, MPIE, and CAM5) show overly thick clouds.
As in Fig. 2, but for cloud albedo and ISCCP instead of ERSST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 2, but for cloud albedo and ISCCP instead of ERSST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 2, but for cloud albedo and ISCCP instead of ERSST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 8 displays the seasonal cycle of SST, total cloud cover, cloud top pressure, LTCC, cloud albedo, LWCRF, SWCRF, and NETCRF averaged over 10°–30°S, 70°–90°W. The observed SST (Fig. 8a) shows a clear seasonal cycle with maximum temperature in Southern Hemisphere (SH) late summer and fall (February and March), and minimum temperature in SH spring (September). Again, note that CAM5 is forced by observed SST. All the coupled models reproduce quite well the phase of the observed SST seasonal cycle. The two IPSL models also reproduce well the absolute value of SST throughout the year, while the other five coupled models tend to have a warm SST bias, especially for SST minimum in SH spring. The observed total cloud cover is roughly out of phase with SST, with a minimum in March and maximum in October (Fig. 8b). Five models (HAES, IPSL, IPSM, MRIC, and CAM5) capture the observed inverse phase relationship between cloud cover and SST, while the other three models produce an unrealistic in-phase relationship between cloud cover and SST. The observed cloud-top pressure (Fig. 8c) shows a very weak seasonal cycle with higher cloud top associated with warmer SST. Only one model (HAES) simulates the correct phase of seasonal cycle. The seasonal cycle of LTCC is similar to that of total cloud cover for observation and all models except the two IPSL models, which have little LTCC. (Fig. 8d). The observed cloud albedo is roughly out of phase with SST (Fig. 8e). Only two models (HAES and CAM5) reproduce the observed phase of seasonal cycle. The observed LWCRF shows a weak seasonal cycle with the maximum in May–June and the minimum in November (Fig. 8f), which is likely caused by the combined effect of total cloud cover (Fig. 8b), cloud-top pressure (Fig. 8c), and cloud thickness (Fig. 8e). Only two models (MPIE and CAM5) capture the observed phase of seasonal cycle. The two IPSL models produce overly large longwave CRF because of the dominance of high clouds in those models (Figs. 4 and 5). The observed SWCRF displays a significant seasonal cycle with strongest (most negative) value in October–November (Fig. 8g), which is caused by the concurrence of maximum total cloud cover (Fig. 8b) and maximum cloud albedo (Fig. 8e) at that time. Only two models (HAES and CAM5) capture the observed phase, while the other models tend to simulate the most negative SWCRF around January. Because the SWCRF is much larger than the LWCRF, the NETCRF generally follows the SWCRF in both observation and models (Fig. 8h).
Seasonal cycle of (a) SST, (b) total cloud cover, (c) cloud-top pressure, (d) lower troposphere cloud cover, (e) cloud albedo, (f) LWCRF, (g) SWCRF, and (h) NETCRF for observations and the eight models. All data are averaged over 10°–30°S, 70°–90°W.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Seasonal cycle of (a) SST, (b) total cloud cover, (c) cloud-top pressure, (d) lower troposphere cloud cover, (e) cloud albedo, (f) LWCRF, (g) SWCRF, and (h) NETCRF for observations and the eight models. All data are averaged over 10°–30°S, 70°–90°W.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Seasonal cycle of (a) SST, (b) total cloud cover, (c) cloud-top pressure, (d) lower troposphere cloud cover, (e) cloud albedo, (f) LWCRF, (g) SWCRF, and (h) NETCRF for observations and the eight models. All data are averaged over 10°–30°S, 70°–90°W.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
In summary, the state-of-the-art global climate models still have significant difficulty in simulating the SEP stratocumulus clouds. Comparing with observation, the models tend to simulate significantly less cloud cover, higher cloud top, and a variety of unrealistic cloud albedo. The insufficient cloud cover leads to overly weak shortwave CRF and net CRF. All models simulate well the phase of the SST seasonal cycle, but only two models (HAES and CAM5) could capture the phases of seasonal cycle for LTCC, cloud albedo, SWCRF, and NETCRF.
b. Cloud–radiation feedback
The observed seasonal cycle displayed in Fig. 8 is characterized by reduction of cloud cover, cloud albedo, and shortwave CRF associated with local SST warming at the seasonal time scale, leading to a positive cloud feedback.1 Figure 9 displays the vertical structure of cloud cover change associated with local SST warming. All data are averaged over 10°–30°S, 70°–90°W. Linear regression is calculated between the cloud fraction in each Ptop–τ pixel and local SST. For observation, linear regression is calculated for the period between July 1983 and June 2008. In observation, there is a significant decrease of stratocumulus clouds and stratus clouds, but a slight increase of cirrus clouds associated with SST warming (Fig. 9a). Only the HAES, CAM5, and MRIC models capture the decrease of low clouds but the magnitude is too small in the MRIC model (Figs. 9c,h). The two IPSL models simulate a decrease of cirrus and cirrostratus clouds associated with SST warming (Figs. 9d,e). The CANE, MIR5, and MPIE models generally show an increase of low clouds associated with SST warming (Figs. 9b,f,g).
As in Fig. 5, but for the linear regression of the Ptop–τ histogram to SST (% K−1). Values larger than +0.5 are shaded in magenta and those between 1% and 5% in yellow. Values smaller than −0.9 are shaded in blue, those between −0.5 and −0.9 in cyan, and those between −0.1 and −0.5 in green.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 5, but for the linear regression of the Ptop–τ histogram to SST (% K−1). Values larger than +0.5 are shaded in magenta and those between 1% and 5% in yellow. Values smaller than −0.9 are shaded in blue, those between −0.5 and −0.9 in cyan, and those between −0.1 and −0.5 in green.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 5, but for the linear regression of the Ptop–τ histogram to SST (% K−1). Values larger than +0.5 are shaded in magenta and those between 1% and 5% in yellow. Values smaller than −0.9 are shaded in blue, those between −0.5 and −0.9 in cyan, and those between −0.1 and −0.5 in green.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
The linear regression coefficient shown in Fig. 9 includes all the time scales the data could cover, ranging from subseasonal to decadal time scales. Next we check about the physical relationship at each different time scales by calculating the cross-spectrum and regression between cloud/radiation variables and the local SST. The cross-spectrum between two time series was calculated following Jenkins and Watts (1968) and Salby and Hendon (1994). For both time series, a 5% cosine tapper was applied to the edges of the time series to control the effect of side lobes, and the Fourier spectrum was calculated for the resulted time series using discrete Fourier transform. Both spectra were smoothed to a spectral bandwidth of 0.015 month−1. The coherence squared and phase differences between the two smoothed spectra were then calculated. For linear regression, the two raw time series were filtered using the Murakami (1979) filter to different time scales including subannual (3–11 months), annual (11–13 months), biennial (13–36 months), interannual (3–7 yr), and decadal (7–20 yr). Then for each time scale, the lag-0 regression coefficient was calculated between the two filtered time series. Note that the length of data used may not be sufficiently long for analyzing the decadal variability.
Figure 10 shows the cross-spectrum and lag-0 regression between LTCC and SST for observation and the models. For coherence squared, the dashed line denotes the 95% confidence level. Phase difference is plotted in dots for only the frequencies with coherence squared higher than the 95% confidence level. A positive (negative) phase difference means LTCC lags (leads) SST at that time scale. When the coherence squared is higher than the 95% confidence level, the sign of regression coefficient is generally consistent with the phase difference, with a positive (negative) sign when the absolute value of the phase difference is smaller (larger) than 0.25 cycles. The observed LTCC decreases with local SST warming at all time scales (subannual to decadal), with a 4%–6% reduction of total cloud cover for 1°C of local SST warming (Fig. 10a). Only the HAES and CAM5 models capture the observed inverse relationship with nearly realistic magnitude (Figs. 10c,i). The MRIC model and two IPSL models also simulates reduction of LTCC at all time scales, but the magnitude is too small compared with observations, especially for the two IPSL models (Figs. 10d,e,h). The MIR5 model simulates an increase of LTCC associated with SST warming at all time scales (Fig. 10g), while CANE (Fig. 10b) and MPIE (Fig. 10g) show an increase for subannual to biennial time scales but a decrease for interannual to decadal time scales.
Coherence squared (Coh2), phase difference, and lag-0 regression coefficients at different time scales between LTCC and SST for (a) observations, and (b)–(i) the eight models. All data are averaged over 10°–30°S, 70°–90°W. For coherence squared, the dashed line denotes the 95% confidence level. Phase difference is plotted (dots) for only the frequencies with coherence squared >95% confidence level. A positive (negative) phase difference means LTCC lags (leads) SST at that time scale.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Coherence squared (Coh2), phase difference, and lag-0 regression coefficients at different time scales between LTCC and SST for (a) observations, and (b)–(i) the eight models. All data are averaged over 10°–30°S, 70°–90°W. For coherence squared, the dashed line denotes the 95% confidence level. Phase difference is plotted (dots) for only the frequencies with coherence squared >95% confidence level. A positive (negative) phase difference means LTCC lags (leads) SST at that time scale.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Coherence squared (Coh2), phase difference, and lag-0 regression coefficients at different time scales between LTCC and SST for (a) observations, and (b)–(i) the eight models. All data are averaged over 10°–30°S, 70°–90°W. For coherence squared, the dashed line denotes the 95% confidence level. Phase difference is plotted (dots) for only the frequencies with coherence squared >95% confidence level. A positive (negative) phase difference means LTCC lags (leads) SST at that time scale.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 11 shows the cross-spectrum and regression between cloud albedo and SST. The observed albedo decreases significantly with SST warming at all time scales (Fig. 11a), suggesting a thinning of the stratocumulus clouds. Three models (HAES, CAM5, and CANE) capture this cloud thinning but with magnitude smaller than in observations (Figs. 11b,c). The MIR5 model simulates a cloud thickening (Fig. 11f), while the MPIE and MRIC models produce cloud thinning at shorter (from subannual to biennial) time scales but cloud thickening at longer (from ENSO to decadal) time scales (Figs. 11g,h). The two IPSL models show a slight cloud thickening that is not statistically significant (Figs. 11d,e).
The final sign and magnitude of shortwave cloud feedback are shown in Fig. 12 by the cross-spectrum and regression between SWCRF and SST. The SWCRF is determined by both the total cloud cover, which is dominated by LTCC, and the cloud albedo. The observed SWCRF increases with SST warming at all time scales (Fig. 12a), indicating a weakening of the magnitude of SWCRF and thus a positive cloud feedback. This is understandable because the stratocumulus clouds are becoming both fewer (Fig. 10a) and thinner (Fig. 11a) associated with SST warming, and thus reflect less sunlight back to space. Only the HAES and CAM5 models simulate the correct sign of this feedback (Fig. 12c). Three models (CANE, MIR5, and MPIE) generate significant negative feedback, while the other three models (IPSL, IPSM, and MRIC) simulate weak negative or positive feedbacks at different time scales. The cross-spectrum and regression between NETCRF and SST are similar to those for SWCRF since the SWCRF is much larger than the LWCRF in this region (not shown).
In short, the observed cloud feedback at subannual to decadal time scales is characterized by reduction of cloud cover, cloud albedo, and shortwave CRF associated with local SST warming, leading to a positive cloud feedback. Only two of the eight models (HAES and CAM5) capture the observed cloud feedback.
c. Connection to large-scale environment
Both physical schemes and large-scale environment affect the generation and maintenance of clouds in CMIP5–CFMIP models. Therefore, the cloud biases in the models could be caused by 1) biases in the model-simulated large-scale environment and/or 2) misrepresentation of physical processes by model schemes. The feedback between those two can either amplify or damp the initial biases. First we look at the large-scale environment important for the stratocumulus clouds, including the large-scale subsidence, humidity structure, and temperature structure. Figure 13 shows the vertical profile of annual mean vertical pressure velocity averaged over 10°–30°S and 70°–90°W for two new reanalysis datasets [Interim European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-Interim) and Climate Forecast System Reanalysis (CFSR)] and the eight models. Both reanalysis demonstrate the large-scale subsidence throughout the troposphere. The magnitudes are similar in the lower troposphere but have larger differences in the upper troposphere. The models simulate reasonably well the large-scale subsidence, except that the MPIE model shows a weak upward motion near the surface.
Vertical profile of annual mean vertical velocity for two reanalysis datasets (ERA-Interim and CFSR) and the eight models. All data are averaged over 10°–30°S, 70°–90°W.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Vertical profile of annual mean vertical velocity for two reanalysis datasets (ERA-Interim and CFSR) and the eight models. All data are averaged over 10°–30°S, 70°–90°W.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Vertical profile of annual mean vertical velocity for two reanalysis datasets (ERA-Interim and CFSR) and the eight models. All data are averaged over 10°–30°S, 70°–90°W.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 14 displays the vertical profile of linear regression coefficient between vertical velocity and SST averaged over 10°–30°S and 70°–90°W. The two reanalyses consistently show a vertical dipole with an upward motion anomaly in the lower troposphere (below 500 mb) and a downward motion anomaly in the upper troposphere (above 500 mb) associated with local SST warming, which may be associated with the changes of both the Walker circulation and local Hadley circulation. The models generally reproduce the vertical dipole, but the height of the node (the zero cross height) differs greatly among different models, ranging from 600 mb for MIR5 and MRIC to 250 mb for IPSL. Furthermore, the models generally simulate better the magnitude of upward motion anomaly in the lower troposphere, but scatter more for the downward motion in the upper troposphere, where the two reanalyses also show larger differences. The modeling study of Bretherton et al. (2013) shows that less subsidence tends to thicken stratocumulus clouds. Therefore the weakened subsidence is not likely the reason for the observed reduction of cloud albedo associated with SST warming.
As in Fig. 13, but for the linear regression coefficient between vertical velocity and SST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 13, but for the linear regression coefficient between vertical velocity and SST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 13, but for the linear regression coefficient between vertical velocity and SST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 15 shows the vertical profile of annual mean specific humidity. The two reanalyses agree quite well below 850 mb, but have a slight difference between 850 and 700 mb, with CFSR being wetter than ERA-Interim. The MIR5 and CAM5 models simulate the most realistic humidity profile. Two models (HAES and MRIC) are wetter than the reanalyses. The other four models (CANE, IPSL, IPSM, and MPIE) are drier than the reanalyses below 850 mb, but wetter than the reanalyses above 800 mb. Overall, all models except MIR5 underestimate the vertical moisture gradient between the boundary layer and free troposphere.
As in Fig. 13, but for specific humidity.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 13, but for specific humidity.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 13, but for specific humidity.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 16 displays the vertical profile of the (a) regression coefficient and (b) correlation coefficient between specific humidity and SST. The regression coefficients for the two reanalyses show moistening of the whole troposphere associated with local SST warming with magnitude increasing with height within the boundary layer, but decreasing with height above the boundary layer (Fig. 16a). The moisture gradient between surface and free troposphere is enhanced with SST warming. All models simulate moistening throughout the troposphere and a decrease of moistening magnitude above the boundary layer. Five of the eight models (CANE, MIR5, MPIE, MRIC, and CAM5) capture the increase of moistening magnitude with height inside the boundary layer, but the other three models (HAES, IPSL, and IPSM) do not reproduce that well. All models simulate well the increase of moisture gradient between the surface and free troposphere associated with SST warming. The correlation coefficients illustrate some other interesting points (Fig. 16b). The correlation coefficients for the two reanalyses show a nearly perfect correlation inside the boundary layer, decrease quickly with height from the top of boundary layer to 700 mb, and then change slowly with height above 700 mb. The four models with better low cloud simulation (HAES, MRIC, MPIE, and CAM5) capture the observed structure, but the four models with worse low cloud simulation (CANE, IPSL, IPSM, and MIR5) show a too small correlation around 700 mb, suggesting a decoupling of the lower troposphere with the boundary layer.
As in Fig. 13, but for the (a) regression coefficient and (b) correlation coefficient between specific humidity and SST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 13, but for the (a) regression coefficient and (b) correlation coefficient between specific humidity and SST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 13, but for the (a) regression coefficient and (b) correlation coefficient between specific humidity and SST.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 17a shows the vertical profile of annual mean temperature. The two reanalyses consistently display the temperature inversion, which is higher and stronger in ERA-Interim than in CFSR. All models except CAM5 fail to produce the inversion. There is also a warm bias in boundary layer temperature, but a cold bias in lower troposphere temperature in all models except CAM5. This is confirmed by the vertical profile of potential temperature θ (Fig. 17b). All models except CAM5 tend to have a warmer θ than observation in the boundary layer, but a colder θ above 800 mb, suggesting an insufficient LTS in the models. This may contribute to the insufficient LTCC in the models.
As in Fig. 13, but for annual mean (a) temperature, and (b) potential temperature.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 13, but for annual mean (a) temperature, and (b) potential temperature.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 13, but for annual mean (a) temperature, and (b) potential temperature.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 18 displays the vertical profile of the regression coefficient between θ and SST. In observation, there is a warming from surface to 200 mb associated with local SST warming, but the warming magnitude changes with height. The warming rate decreases with height in the boundary layer, but increases with height from above the boundary layer to a maximum around 400 mb. Most importantly, the warming magnitude in the free troposphere is weaker than at the surface, indicating a reduction of LTS associated with SST warming. All models simulate warming between the surface and 200 mb, a decrease of warming rate with height inside the boundary layer, and an increase of warming rate with height with height from above the boundary layer to a maximum in the upper troposphere. However, all models except CAM5 produce a stronger warming than observations above the boundary layer, leading to a weaker reduction of LTS associated with local SST warming.
As in Fig. 14, but for potential temperature.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 14, but for potential temperature.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
As in Fig. 14, but for potential temperature.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
The excessive free troposphere warming in the models is possibly caused by excessive latent heating in the models. This is confirmed in Fig. 19a, which shows the time series of annual mean precipitation, which is a measure of vertically integrated latent heating. In observations, the annual mean precipitation is generally smaller than 0.3 mm day−1, which comes mainly from drizzle from the stratocumulus clouds. All models except CANE produce excessive precipitation. Comparison with Fig. 18 indicates that models with larger (smaller) precipitation tend to have larger (smaller) warming in upper/middle troposphere associated with local SST warming, suggesting that the excessive troposphere warming in the models is likely caused by excessive latent heating in the models. One exception is CANE, which produces weak precipitation but large troposphere warming, suggesting that the troposphere warming in that model is caused by physical processes other than latent heating.
Time series of (a) annual mean precipitation for observations (GPCP) and the eight models and (b) ratio between annual mean convective precipitation and annual mean total precipitation for the eight models except CANE. All data are averaged over 10°–30°S, 70°–90°W.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Time series of (a) annual mean precipitation for observations (GPCP) and the eight models and (b) ratio between annual mean convective precipitation and annual mean total precipitation for the eight models except CANE. All data are averaged over 10°–30°S, 70°–90°W.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Time series of (a) annual mean precipitation for observations (GPCP) and the eight models and (b) ratio between annual mean convective precipitation and annual mean total precipitation for the eight models except CANE. All data are averaged over 10°–30°S, 70°–90°W.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Figure 19b shows the ratio between convective precipitation and total precipitation for the models. Note that convective precipitation output is not available for CANE. Previous observational studies showed that precipitation in the SEP stratocumulus region is dominated by convective precipitation, which accounts for 70%–80% of the total precipitation (Schumacher and Houze 2003). The two models with better stratocumulus cloud simulation (HAES and CAM5) generate convective precipitation fraction that is consistent with observations. The other five models produce overly large stratiform precipitation.
Figure 20 illustrates the scatterplot of LTCC versus LTS. The observation shows a high correlation of 0.80 between ISCCP LTCC and ERA-Interim LTS, with a regression coefficient of 5.42% K−1, which is consistent with the results of previous observational studies (Klein and Hartmann 1993; Norris 1998; Wood and Bretherton 2006). Only one model (HAES) captures the observed correlation and regression coefficients, suggesting that the model physical schemes master well the formation and maintenance of low clouds. Five other models (IPSL, IPSM, MPIE, MRIC, and CAM5) also show high correlation, but with too small regression coefficients, suggesting that the model physical schemes are in the right direction in simulating the low clouds. Two models (CANE and MIR5) display very low correlation, suggesting that the model physical schemes cannot properly represent some key aspect of low cloud formation.
Scatterplot between LTS and LTCC for (a) observations (ERA-Interim LTS vs ISCCP LTCC) and (b)–(i) the eight models. All data are averaged over 10°–30°S, 70°–90°W. The linear regression coefficient (r, K %−1) and correlation coefficient (c) are also shown on the top of each panel.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Scatterplot between LTS and LTCC for (a) observations (ERA-Interim LTS vs ISCCP LTCC) and (b)–(i) the eight models. All data are averaged over 10°–30°S, 70°–90°W. The linear regression coefficient (r, K %−1) and correlation coefficient (c) are also shown on the top of each panel.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
Scatterplot between LTS and LTCC for (a) observations (ERA-Interim LTS vs ISCCP LTCC) and (b)–(i) the eight models. All data are averaged over 10°–30°S, 70°–90°W. The linear regression coefficient (r, K %−1) and correlation coefficient (c) are also shown on the top of each panel.
Citation: Journal of Climate 27, 8; 10.1175/JCLI-D-13-00376.1
To summarize, the cloud and radiation biases in the models are associated with model biases in both large-scale environment and model physics. For the large-scale environment, the models simulate reasonable large-scale subsidence and humidity structure, but show problems in simulating the temperature structure including missing temperature inversion, insufficient LTS, and insufficient reduction of LTS with local SST warming. For model physics, six of the eight models can simulate the increase of low cloud cover associated with larger LTS, but the regression coefficient is generally too small with only one model capturing the observed magnitude.
4. Summary and discussion
This study examines the SEP stratocumulus clouds and associated cloud feedback simulated by eight CMIP5–CFMIP global climate models. Twenty-eight years of historical simulations from each model were analyzed and compared with long-term observations of clouds, radiative fluxes, CRF, SST, and large-scale atmosphere environment. The results show that the state-of-the-art global climate models still have significant difficulty in simulating the SEP stratocumulus clouds and associated cloud feedback. Comparing with observations, the models tend to simulate significantly less cloud cover, higher cloud top, and a variety of unrealistic cloud albedo. The insufficient cloud cover leads to overly weak shortwave CRF and net CRF. Only two of the eight models capture the observed positive cloud feedback at subannual to decadal time scales. The cloud and radiation biases in the models are associated with 1) model biases in large-scale temperature structure including the lack of temperature inversion, insufficient lower troposphere stability (LTS), and insufficient reduction of LTS with local SST warming, and 2) improper model physics, especially insufficient increase of low cloud cover associated with larger LTS.
The simulated LTCC–LTS relationship (Fig. 20) is likely related to PBL, shallow convection, and/or cloud schemes in the models. Among the eight CMIP5–CFMIP models analyzed in this study, the HAES and CAM5 models are the only ones that capture many aspects of the SEP stratocumulus clouds and associated cloud feedback. It is interesting to note that HAES and CAM5 are the only ones among the eight models that use cloud-top radiative cooling to drive boundary layer turbulence (Table 1). Clement et al. (2009) examined the simulation of northeast Pacific stratocumulus clouds by 18 IPCC AR4/CMIP3 models, and found that the Hadley Centre Global Environment Model, version 1 (HadGEM1), which is an early version of HAES and uses similar boundary layer turbulence scheme, was the only model that could reproduce the observed relationships between cloud cover and regional meteorological conditions. Later, Broccoli and Klein (2010) showed that the Geophysical Fluid Dynamics Laboratory Climate Model, version 2.1 (GFDL CM2.1), another model that uses cloud-top radiative cooling to drive boundary layer turbulence, also simulated well the relationship between low cloud cover and LTS.2 These results suggest that using cloud-top radiative cooling to drive boundary layer turbulence may help to improve the simulation of stratocumulus clouds.
As reviewed by Lock et al. (2000) and Bretherton and Park (2009), the PBL schemes in atmospheric GCMs have evolved through three stages: 1) local schemes (e.g., Louis 1979) relating the diffusivity to the local stability, which work well for stable conditions but not for unstable conditions; 2) nonlocal schemes (e.g., Holtslag and Boville 1993) with nonlocal diffusivity whose magnitude is determined by surface forcing, which works well for unstable conditions driven by surface forcing but not for unstable conditions driven by forcing from boundary layer top, such as the stratocumulus-topped boundary layer; and 3) nonlocal schemes with consideration of cloud-top forcing (e.g., van Meijgaard and van Ulden 1998; Lock et al. 2000; Grenier and Bretherton 2001; Bretherton and Park 2009). For example, the HAES model uses the Lock et al. (2000) scheme with two separate K-profiles (diffusion coefficients), one for surface sources of turbulence and one for cloud-top sources, and the entrainment across the top of the boundary layer is coupled to the radiative fluxes and the dynamics through a subgrid inversion diagnosis. The CAM5 model uses the Bretherton and Park (2009) scheme that is formulated using moist thermodynamics with turbulence being affected by all processes that affect the vertical structure of the atmosphere, such as cloud-top radiative cooling. Previous studies showed how these schemes improved cloud-topped boundary layer simulation and reduced climate biases in a single GCM (e.g., Martin et al. 2000; Park and Bretherton 2009). The present study conducted an intercomparison of different models using different type of PBL schemes, and found that the nonlocal schemes with consideration of cloud-top forcing tend to give better simulation of stratocumulus clouds and associated cloud feedback.
In addition to improved model physics, our results suggest that a realistic large-scale temperature structure is important for simulating the SEP stratocumulus clouds and cloud feedback. This is consistent with the conclusions of Caldwell et al. (2013), which used large-scale conditions from GCMs to drive an offline atmospheric mixed-layer model (MLM) that generally simulates better stratocumulus clouds than the GCM cloud parameterizations. They found that replacing the various GCM cloud parameterizations with a single physics package (the MLM) does not reduce intermodel spread in low-cloud feedback because the MLM is more sensitive than the GCMs to existent intermodel variations in large-scale forcing. Nevertheless, there are interactions and feedbacks between the model physics and model-simulated large-scale state, and improved model physics may help improving model-simulated large-scale state and the overall simulation of SEP stratocumulus clouds and associated cloud feedback.
Acknowledgments
We highly appreciate the very insightful reviews from three anonymous reviewers, which significantly improved the manuscript. We acknowledge the World Climate Research Programme’s Working Group on Coupled Modeling, which is responsible for CMIP, and we thank the climate modeling groups (listed in Table 1 of this paper) for producing and making available their model output. For CMIP the U.S. Department of Energy’s Program for Climate Model Diagnosis and Intercomparison provides coordinating support and led development of software infrastructure in partnership with the Global Organization for Earth System Science Portals. J.-L. Lin was supported by the NOAA Climate Program Office Modeling, Analysis, Predictions and Projections (MAPP) Program as part of the CMIP5 Task Force under Grant GC10-400, the NASA MAP Program, and NSF Grant ATM-0745872. T. Shinoda is supported by NOAA CPO MAPP and ESS (GC10-400, NA11OAR4310110) and the ONR/LASP project (Program Element 601153N).
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.
Bodas-Salcedo, A., and Coauthors, 2011: COSP: Satellite simulation software for model assessment. Bull. Amer. Meteor. Soc., 92, 1023–1043, doi:10.1175/2011BAMS2856.1.
Bony, S., and J.-L. Dufresne, 2005: Marine boundary layer clouds at the heart of cloud feedback uncertainties in climate models. Geophys. Res. Lett., 32, L20806, doi:10.1029/2005GL023851.
Bony, S., and Coauthors, 2011: CFMIP: Towards a better evaluation and understanding of clouds and cloud feedbacks in CMIP5 models. CLIVAR Exchanges, No. 56 (2), International CLIVAR Project Office, Southampton, United Kingdom, 20–22.
Bretherton, C. S., and S. Park, 2009: A new moist turbulence parameterization in the Community Atmosphere Model. J. Climate, 22, 3422–3448, doi:10.1175/2008JCLI2556.1.
Bretherton, C. S., and Coauthors, 2004: The EPIC 2001 stratocumulus study. Bull. Amer. Meteor. Soc., 85, 967–977, doi:10.1175/BAMS-85-7-967.
Bretherton, C. S., R. George, R. Wood, G. Allen, D. Leon, and B. Albrecht, 2010: Southeast Pacific stratocumulus clouds, precipitation and boundary layer structure sampled along 20°S during VOCALS-REx. Atmos. Chem. Phys., 10, 10 639–10 654, doi:10.5194/acp- 10-10639-2010.
Bretherton, C. S., P. N. Blossey, and C. R. Jones, 2013: Mechanisms of marine low cloud sensitivity to idealized climate perturbations: A single-LES exploration extending the CGILS cases. J. Adv. Model. Earth Syst., 5, 316–337, doi:10.1002/jame.20019.
Brient, F., and S. Bony, 2013: Interpretation of the positive low-cloud feedback predicted by a climate model under global warming. Climate Dyn .,40, 2415–2431, doi:10.1007/s00382-011-1279-7.
Broccoli, A. J., and S. A. Klein, 2010: Comment on “Observational and model evidence for positive low-level cloud feedback.” Science,329, 277, doi:10.1126/science.1186796.
Caldwell, P., and C. S. Bretherton, 2009: Response of a subtropical stratocumulus-capped mixed layer to climate and aerosol changes. J. Climate, 22, 20–38, doi:10.1175/2008JCLI1967.1.
Caldwell, P., Y. Zhang, and S. A. Klein, 2013: CMIP3 subtropical stratocumulus cloud feedback interpreted through a mixed-layer model. J. Climate, 26, 1607–1625, doi:10.1175/JCLI-D-12-00188.1.
Chikira, M., and M. Sugiyama, 2010: A cumulus parameterization with state-dependent entrainment rate. Part I: Description and sensitivity to temperature and humidity profiles. J. Atmos. Sci., 67, 2171–2193, doi:10.1175/2010JAS3316.1.
Clement, A., R. Burgman, and J. Norris, 2009: Observational and model evidence for positive low-level cloud feedback. Science, 325, 460–464, doi:10.1126/science.1171255.
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.
Derbyshire, S. H., A. V. Maidens, S. F. Milton, R. A. Stratton, and M. R. Willett, 2011: Adaptive detrainment in a convective parameterization. Quart. J. Roy. Meteor. Soc., 137, 1856–1871, doi:10.1002/qj.875.
Garreaud, R. D., and R. C. Muñoz, 2005: The low-level jet off the west coast of subtropical South America: Structure and variability. Mon. Wea. Rev., 133, 2246–2261, doi:10.1175/MWR2972.1.
Gordon, C. T., A. Rosati, and R. Gudgel, 2000: Tropical sensitivity of a coupled model to specified ISCCP low clouds. J. Climate, 13, 2239–2260, doi:10.1175/1520-0442(2000)013<2239:TSOACM>2.0.CO;2.
Gregory, D., and P. R. Rowntree, 1990: A mass flux convection scheme with representation of cloud ensemble characteristics and stability-dependent closure. Mon. Wea. Rev.,118, 1483–1506, doi:10.1175/1520-0493(1990)118<1483:AMFCSW>2.0.CO;2.
Grenier, H., and C. S. Bretherton, 2001: A moist PBL parameterization for large-scale models and its application to subtropical cloud-topped marine boundary layers. Mon. Wea. Rev.,129, 357–377, doi:10.1175/1520-0493(2001)129<0357:AMPPFL>2.0.CO;2.
Gudgel, R. G., A. Rosati, and C. T. Gordon, 2001: The sensitivity of a coupled atmospheric–oceanic GCM to prescribed low-level clouds over the ocean and tropical landmasses. Mon. Wea. Rev., 129, 2103–2115, doi:10.1175/1520-0493(2001)129<2103:TSOACA>2.0.CO;2.
Hannay, C., D. L. Williamson, J. J. Hack, J. T. Kiehl, J. G. Olson, S. A. Klein, C. S. Bretherton, and M. Köhler, 2009: Evaluation of forecasted southeast Pacific stratocumulus in the NCAR, GFDL and ECMWF models. J. Climate, 22, 2871–2889, doi:10.1175/2008JCLI2479.1.
Holtslag, A. A. M., and B. A. Boville, 1993: Local versus nonlocal boundary layer diffusion in a global climate model. J. Climate,6, 1825–1842, doi:10.1175/1520-0442(1993)006<1825:LVNBLD>2.0.CO;2.
Hourdin, F., and Coauthors, 2006: The LMDZ4 general circulation model: Climate performance and sensitivity to parametrized physics with emphasis on tropical convection. Climate Dyn.,27, 787–813, doi:10.1007/s00382-006-0158-0.
Jenkins, G. M., and D. G. Watts, 1968: Spectral Analysis and Its Applications. Holden-Day, 525 pp.
Klein, S. A., and D. L. 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.
Klein, S. A., and C. Jakob, 1999: Validation and sensitivities of frontal clouds simulated by the ECMWF model. Mon. Wea. Rev., 127, 2514–2531, doi:10.1175/1520-0493(1999)127<2514:VASOFC>2.0.CO;2.
Lock, A. P., A. R. Brown, M. R. Bush, G. M. Martin, and R. N. B. Smith, 2000: A new boundary layer mixing scheme. Part I: Scheme description and single-column model tests. Mon. Wea. Rev.,128, 3187–3199, doi:10.1175/1520-0493(2000)128<3187:ANBLMS>2.0.CO;2.
Loeb, N. G., B. A. Wielicki, D. R. Doelling, G. L. Smith, D. F. Keyes, S. Kato, N. Manalo-Smith, and T. Wong, 2009: Towards optimal closure of the earth’s top-of-atmosphere radiation budget. J. Climate, 22, 748–766, doi:10.1175/2008JCLI2637.1.
Louis, J.-F., 1979: A parametric model of vertical eddy fluxes in the atmosphere. Bound.-Layer Meteor.,17, 187–202, doi:10.1007/BF00117978.
Ma, C.-C., C. R. Mechoso, A. W. Robertson, and A. Arakawa, 1996: Peruvian stratus clouds and the tropical Pacific circulation: A coupled ocean–atmosphere GCM study. J. Climate, 9, 1635–1645, doi:10.1175/1520-0442(1996)009<1635:PSCATT>2.0.CO;2.
Martin, G. M., M. R. Bush, A. R. Brown, A. P. Lock, and R. N. B. Smith, 2000: A new boundary layer mixing scheme. Part II: Tests in climate and mesoscale models. Mon. Wea. Rev., 128, 3200–3217, doi:10.1175/1520-0493(2000)128<3200:ANBLMS>2.0.CO;2.
Mechoso, C., and Coauthors, 2014: Ocean–cloud–atmosphere–land interactions in the southeast Pacific: The VOCALS program. Bull. Amer. Meteor. Soc., in press.
Medeiros, B., B. Stevens, I. M. Held, M. Zhao, D. L. Williamson, J. G. Olson, and C. S. Bretherton, 2008: Aquaplanets, climate sensitivity, and low clouds. J. Climate, 21, 4974–4991, doi:10.1175/2008JCLI1995.1.
Minnis, P., P. W. Heck, D. F. Young, C. W. Fairall, and J. B. Snider, 1992: Stratocumulus cloud properties derived from simultaneous satellite and island-based instrumentation during FIRE. J. Appl. Meteor., 31, 317–339, doi:10.1175/1520-0450(1992)031<0317:SCPDFS>2.0.CO;2.
Murakami, M., 1979: Large-scale aspects of deep convective activity over the GATE area. Mon. Wea. Rev., 107, 994–1013, doi:10.1175/1520-0493(1979)107<0994:LSAODC>2.0.CO;2.
Neale, R. B., and Coauthors, 2012: Description of the NCAR Community Atmosphere Model (CAM 5.0). NCAR Tech. Rep. NCAR/TN-486+STR, 274 pp. [Available online at http://www.cesm.ucar.edu/models/cesm1.0/cam/docs/description/cam5_desc.pdf.]
Nigam, S., 1997: The annual warm to cold phase transition in the eastern equatorial Pacific: Diagnosis of the role of stratus cloud-top cooling. J. Climate, 10, 2447–2467, doi:10.1175/1520-0442(1997)010<2447:TAWTCP>2.0.CO;2.
Nordeng, T. E., 1994: Extended versions of the convective parameterization scheme at ECMWF and their impact on the mean and transient activity of the model in the tropics. ECMWF Tech. Rep. 206, 41 pp.
Norris, J. R., 1998: 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, doi:10.1175/1520-0442(1998)011<0369:LCTOTO>2.0.CO;2.
Park, S., and C. S. Bretherton, 2009: The University of Washington shallow convection scheme and moist turbulence schemes and their impact on climate simulations with the Community Atmosphere Model. J. Climate,22, 3449–3469, doi:10.1175/2008JCLI2557.1.
Pincus, R., S. Platnick, S. A. Ackerman, R. S. Hemler, and R. J. P. Hofmann, 2012: Reconciling simulated and observed views of clouds: MODIS, ISCCP, and the limits of instrument simulators. J. Climate, 25, 4699–4720, doi:10.1175/JCLI-D-11-00267.1.
Qu, X., A. Hall, S. A. Klein, and P. M. Caldwell, 2014: On the spread of changes in marine low cloud cover in climate model simulations of the 21st century. Climate Dyn., doi:10.1007/s00382-013-1945-z, in press.
Randall, D. A., J. A. Coakley, C. W. Fairall, R. A. Knopfli, and D. H. Lenschow, 1984: Outlook for research on marine subtropical stratocumulus clouds. Bull. Amer. Meteor. Soc., 65, 1290–1301, doi:10.1175/1520-0477(1984)065<1290:OFROSM>2.0.CO;2.
Richter, I., and C. R. Mechoso, 2006: Orographic influences on the subtropical stratocumulus. J. Atmos. Sci., 63, 2585–2601, doi:10.1175/JAS3756.1.
Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 1015–1057, doi:10.1175/2010BAMS3001.1.
Salby, M. L., and H. H. Hendon, 1994: Intraseasonal behavior of winds, temperature, and motion in the tropics. J. Atmos. Sci., 51, 2207–2224, doi:10.1175/1520-0469(1994)051<2207:IBOCTA>2.0.CO;2.
Schumacher, C., and R. A. Houze, 2003: The TRMM Precipitation Radar’s view of shallow, isolated rain. J. Appl. Meteor., 42, 1519–1524, doi:10.1175/1520-0450(2003)042<1519:TTPRVO>2.0.CO;2.
Shinoda, T., and J.-L. Lin, 2009: Interannual variability of the upper ocean in the southeast Pacific stratus cloud region. J. Climate, 22, 5072–5088, doi:10.1175/2009JCLI2696.1.
Slingo, A., 1987: The development and verification of a cloud prediction scheme for the ECMWF model. Quart. J. Roy. Meteor. Soc., 113, 899–927, doi:10.1002/qj.49711347710.
Slingo, A., 1990: Sensitivity of the earth’s radiation budget to changes is low clouds. Nature, 343, 49–51, doi:10.1038/343049a0.
Smith, T. M., and R. W. Reynolds, 2004: Improved extended reconstruction of SST (1854–1997). J. Climate,17, 2466–2477, doi:10.1175/1520-0442(2004)017<2466:IEROS>2.0.CO;2.
Soden, B. J., and G. A. Vecchi, 2011: The vertical distribution of cloud feedback in coupled ocean–atmosphere models. Geophys. Res. Lett., 38, L12704, doi:10.1029/2011GL047632.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498, doi:10.1175/BAMS-D-11-00094.1.
Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 1779–1800, doi:10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2.
van Meijgaard, E., and A. P. van Ulden, 1998: A first-order mixing–condensation scheme for nocturnal stratocumulus. Atmos. Res.,45, 253–273, doi:10.1016/S0169-8095(97)00080-X.
von Salzen, K., N. A. McFarlane, and M. Lazare, 2005: The role of shallow convection in the water and energy cycles of the atmosphere. Climate Dyn., 25, 671–688, doi:10.1007/s00382-005-0051-2.
Webb, M., C. Senior, S. Bony, and J.-J. Morcrette, 2001: Combining ERBE and ISCCP data to assess clouds in the Hadley Centre, ECMWF and LMD atmospheric climate models. Climate Dyn., 17, 905–922, doi:10.1007/s003820100157.
Webb, M., and Coauthors, 2006: On the contribution of local feedback mechanisms to the range of climate sensitivity in two GCM ensembles. Climate Dyn., 27, 17–38, doi:10.1007/s00382-006-0111-2.
Wood, R., 2012: Stratocumulus clouds. Mon. Wea. Rev., 140, 2373–2423, doi:10.1175/MWR-D-11-00121.1.
Wood, R., and C. S. Bretherton, 2006: On the relationship between stratiform low cloud cover and lower-tropospheric stability. J. Climate, 19, 6425–6432, doi:10.1175/JCLI3988.1.
Wood, R., and Coauthors, 2011: The VAMOS Ocean–Cloud–Atmosphere–Land Study Regional Experiment (VOCALS-REx): Goals, platforms, and field operations. Atmos. Chem. Phys., 11, 627–654, doi:10.5194/acp-11-627-2011.
Wyant, M. C., and Coauthors, 2010: The PreVOCA experiment: Modeling the lower troposphere in the southeast Pacific. Atmos. Chem. Phys., 10, 4757–4774, doi:10.5194/acp-10-4757-2010.
Yu, J.-Y., and C. R. Mechoso, 1999: A discussion on the errors in the surface heat fluxes simulated by a coupled GCM. J. Climate, 12, 416–426, doi:10.1175/1520-0442(1999)012<0416:ADOTEI>2.0.CO;2.
Yukimoto, S., and Coauthors, 2011: Meteorological Research Institute–Earth System Model version 1 (MRI-ESM1): Model description. MRI Tech. Rep. 64, 83 pp.
Zhang, M., and C. S. Bretherton, 2008: Mechanisms of low cloud climate feedback in idealized single-column simulations with the Community Atmospheric Model, version 3 (CAM3). J. Climate, 21, 4859–4878, doi:10.1175/2008JCLI2237.1.
Zheng, Y., T. Shinoda, G. N. Kiladis, J.-L. Lin, E. J. Metzger, H. E. Hurlburt, and B. S. Giese, 2010: Upper-ocean processes under the stratus cloud deck in the southeast Pacific Ocean. J. Phys. Oceanogr., 40, 103–120, doi:10.1175/2009JPO4213.1.
Zheng, Y., T. Shinoda, J.-L. Lin, and G. N. Kiladis, 2011: Sea surface temperature biases under the stratus cloud deck in the southeast Pacific Ocean in 19 IPCC AR4 coupled general circulation models. J. Climate, 24, 4139–4164, doi:10.1175/2011JCLI4172.1.
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.
Zhu, P., J. Hack, J. Kiehl, and C. S. Bretherton, 2007: Climate sensitivity of tropical and subtropical marine low clouds to ENSO and global warming due to doubling CO2. J. Geophys. Res.,112, D17108, doi:10.1029/2006JD008174.
Zuidema, P., E. R. Westwater, C. Fairall, and D. Hazen, 2005: Ship-based liquid water path estimates in marine stratocumulus. J. Geophys. Res., 110, D20206, doi:10.1029/2005JD005833.
Zuidema, P., D. Leon, A. Pazmany and M. Cadeddu, 2012: Aircraft millimeter-wave passive sensing of cloud liquid water and water vapor during VOCALS-REx. Atmos. Chem. Phys., 12, 355–369, doi:10.5194/acp-12-355-2012.
Here we discuss the feedbacks or interactions among clouds, radiation, and local SST at seasonal to decadal time scales, which are at different time scales and spatial scales from the cloud feedbacks associated with doubling CO2.
The Met Office models, GFDL models, and CAM5 happen to be the only models within the CMIP3 and CMIP5 models that use cloud-top radiative cooling to drive boundary layer turbulence (Qu et al. 2014).