1. Introduction
Cloud–climate feedback continues to represent one of the greatest uncertainties in climate models (Cubasch et al. 2001). This is regardless of the decade-long research after the works of Cess et al. (1990) and Senior and Mitchell (1993, 1996), which highlighted the sensitivities of cloud feedbacks in general circulation models to their physical parameterizations. Evaluation of cloud and radiation fields in the models against multiple observational datasets is therefore highly desirable (Del Genio et al. 1996; Kiehl et al. 1998b; Klein and Jakob 1999; Norris and Weaver 2001; Webb et al. 2001; Randall et al. 2003). Since cloud feedback cannot be directly observed, the validation of a model's basic cloud climatology becomes a useful test.
The present study attempts to evaluate the basic climatology of clouds and radiation in the National Center for Atmospheric Research (NCAR) Community Atmospheric Model (CAM2; Collins et al. 2003) by using satellite observations of clouds and radiation. This is greatly facilitated by employing an International Satellite Cloud Climatology Project (ISCCP) simulator, as is used in Klein and Jakob (1999) and Webb et al. (2001). We first illustrate several deficiencies of the model cloud fields and point out that these deficiencies paradoxically do not produce major errors in the radiation fields. We then analyze the model biases of the vertical structures and optical properties of clouds and identify causes of the spurious balance of the top-of-the-atmosphere (TOA) radiative forcing fields.
The purpose of this paper is to document the major cloud biases in CAM2 through comparison against observations. We also attempt to identify cloud biases at different altitudes and optical thickness according to different regimes of dynamical circulations in order to offer a close link to the model physical parameterizations.
The paper is organized as follows. Section 2 briefly describes the ISCCP, Earth Radiation Budget Experiment (ERBE) data, and the CAM2 simulations. Section 3 compares the global distribution of the total cloud amount and cloud radiative forcing (CRF) between the data and the model. Major differences are highlighted, along with the associated similarities of the cloud-forcing fields. Section 4 uses three selected regions to analyze the model biases of vertical structures and optical properties of clouds, and the possible causes of model biases. Results are generalized to the globe. The last section contains a summary and discussions.
2. Data and method
a. ISCCP
We use the ISCCP D1 product from 1983 to 2001 to describe the satellite cloud climatology. ISCCP uses the radiance measurements from a visible channel and an infrared channel on up to five geostationary and two polar-orbiting satellites, together with the Television Infrared Observation Satellite (TIROS) Operational Vertical Sounder (TOVS) temperature and moisture soundings as well as surface data as input. The visible channel is used to derive the cloud optical thickness, and the infrared channel is used to derive the cloud-top temperature and thus the pressure (Rossow and Schiffer 1991; Rossow et al. 1996). Typical cloud types are classified according to their top pressure and optical thickness (Rossow et al. 1996).
b. ERBE
The longwave (LW) and shortwave (SW) cloud radiative forcing data at the TOA are from ERBE. We use the combined product of the S-4 data from the Earth Radiation Budget Satellite (ERBS) and National Oceanic and Atmospheric Administration (NOAA) satellites NOAA-9 and NOAA-10. The monthly data span from January 1985 to December 1989 and are gridded at 2.5° × 2.5° resolution. More detailed descriptions of the ERBE data and their error characterization can be found in Barkstrom and Smith (1986) and Harrison et al. (1990).
c. The model
The CAM2 simulation is carried out from 1981 to 2001 with forcing from observed SST. The monthly SST is the blended Hadley Centre Sea Ice and SST (HadISST)/Reynolds product, produced at NCAR (Hack et al. 2002).
CAM2 differs from the earlier version of the community atmospheric model, the community climate model (CCM3; Kiehl et al. 1998a), in several aspects related to clouds. Low-level marine stratus is determined following Klein and Hartmann (1993). Convective anvils associated with cloud fractions are made proportional to the convective detrainment rate above 500 hPa following Rasch and Kristjánsson (1998). Cloud water is predicted using a prognostic scheme that contains the macrophysical formulation of Zhang et al. (2003) and the cloud microphysical package by Rasch and Kristjánsson (1998).
d. The ISCCP simulator
The ISCCP simulator1 used in this study was developed by Drs. Mark Webb and Steve Klein (Klein and Jacob 1999, Webb et al. 2001). Similar diagnostics to derive ISCCP-type clouds can be found in Yu et al. (1996) and Del Genio et al. (1996). The simulator is added as a run-time diagnostic package in CAM2. It diagnoses clouds to emulate the ISCCP algorithm by downscaling the grid-level cloud fraction to top-viewed cloudy and cloud-free subcolumns, along with a maximum–random overlapping in the vertical consistent with the CAM2 radiation codes. Composite subgrid-scale clouds of seven ranges of optical depth are reported for seven pressure ranges of cloud-top height. More details can be found in the appendix of Klein and Jakob (1999). The clouds of the thinnest optical interval are beyond the minimum range detectable by ISCCP. Thus they are not included in the comparison. The remaining 42 cloud types are classified in the same way that ISCCP classifies them.
To show the difference in the total cloud amount because of the use of the ISCCP simulator, we use Fig. 1 to show the total cloud amount from the model output and from the ISCCP simulator. Figure 1a shows the December–January–February (DJF) season and Fig. 1b shows the June–July–August (JJA) season. It is evident that the ISCCP simulator reduces the model total cloud amount primarily in the Tropics. In the midlatitudes, the cloud amount from the two sources agrees well with each other. Reduction through the ISCCP simulator occurs primarily in very high thin clouds over regions of deep convections. This is caused by the restriction of cloud optical thickness larger than 0.1 in the ISCCP algorithm.
Figure 2a shows a time–height cross section of the simulated cloud amount by CAM2 at the Southern Great Plains (SGP) site of the Atmospheric Radiation Measurement Program (ARM) for March 2000. Figure 2b shows the corresponding cloud amount and cloud-top distribution from the ISCCP simulator. As expected, the ISCCP algorithm cannot describe the multilayered structures of clouds. The ISCCP simulator also tends to slightly underestimate the cloud altitudes. These deficiencies, however, are not so serious as to invalidate the general descriptions of clouds, as long as they are properly accounted for in the interpretation of the results.
3. Comparison with ISCCP and ERBE data
Since cloud scene identification is less reliable at high latitudes in both ISCCP and ERBE, we only include results from 60°N to 60°S in the following discussions. All data in this section are seasonal means for the northern winter (DJF) and summer (JJA) seasons during the ERBE period of January 1985 to December 1989.
Figure 3a shows the geographical distribution of the total cloud amount in DJF from ISCCP. Several major features can be clearly identified: the maximum cloud amount in the midlatitude storm tracks of the two hemispheres, particularly over the oceans; the cloud amount maximum associated with the intertropical convergence zone (ITCZ); and the minimum cloud amount in the subtropics of both hemispheres. The northern subtropics corresponds to less clouds than the southern subtropics, reflecting the stronger downward branch of the Hadley circulation in the north in DJF, except over the western oceans where zonally driven subsidence associated with the Walker-type circulation dominates the southern subtropics.
These features of the total cloud distribution are closely mirrored in the cloud forcing from ERBE. Figures 3b and 3c show the longwave and shortwave cloud forcing at the TOA in DJF. They are consistent with each other except off the west coasts of Peru, California, and Namibia; over southeast China; and over the narrow band of the eastern tropical Pacific. In these regions, a substantial cloud amount is associated with large shortwave but little longwave cloud forcing, suggesting low-top stratus or stratocumulus clouds in these regions. These results are consistent with the ground-based observations of low clouds in Klein and Hartmann (1993).
The simulated cloud amount in DJF is shown in Fig. 3d. While the model captures the maximum cloud bands in the Tropics and in the storm tracks and the minimums of clouds in the two subtropics, it differs greatly in the total cloud amount from the data. First, the cloud amount in the two storm tracks of the two hemispheres is substantially less in the model. Second, the cloud minimums in the subtropics of the two hemispheres are exaggerated. Third, in the Tropics, the cloud amount is overestimated over the western Pacific warm pool, the Maritime Continent, and the continental convection center over Africa, but it is underestimated over the Indian Ocean, the Atlantic, and South America.
Figure 4a summarizes these differences in a zonally averaged format. The underestimation of total cloud amount in the model is widespread poleward from about 20°N and 20°S, reflecting the underestimation of clouds in both the storm track cloud maximums and the subtropical cloud minimums. Averaged between 20°–60°N and 20°–60°S, the model total cloud amount is only about 73% of the observed. The consistency of the total cloud amount in the Tropics between the model and the data in the figure is a result of the cancellation of errors of overestimation over the tropical convection centers and underestimation over the rest of the Tropics.
These different cloud distributions, between the model and the data, however, correspond to similar distributions of cloud radiative forcing in the model and in ERBE. The zonally averaged cloud forcing in the model, shown in Figs. 4b and 4c for the longwave and shortwave, respectively, agrees much better with the observations than the cloud amount does. These similarities in the cloud forcing can be even seen in its geographical distributions, as is shown in Figs. 3e and 3f.
These main features are also seen in the comparison of model JJA climatology with the observations (Fig. 5). Averaged from 30° to 60°N and 30° to 60°S, the model simulated 70% of the observed cloud amount in JJA. These differences in cloud amount, however, register very little signature in the zonal distribution of the cloud forcing as shown in Figs. 5b and 5c.
Two questions emerge from the above discussions. One is why the simulated cloud amount differs so much from the observations. The second is why the model produces reasonable cloud-forcing fields at the TOA given the cloud amount biases. These questions are further examined in the next section.
4. Analysis of model biases
We first select three regions to represent the storm tracks, the subtropical dry regions, and the tropical convection centers, in order to analyze the vertical structures of the cloud biases in the model. These three regions are highlighted in Figs. 3a and 3d. We then generalize the discussion to other regions. In this section, all ISCCP data and CAM2 results represent their climatology averaged from 1983 to 2001.
a. In the storm tracks
Figure 6a shows the frequency distribution of DJF clouds in ISCCP, stratified against cloud-top pressure and cloud optical thickness, in the North Pacific storm track of 40°–60°N, 160°–225°E. This is also the region used in Webb et al. (2001), but they used the month of July 1988. There is a maximum frequency center at an optical depth of 2.5 with cloud-top pressure ranging from 500 hPa to near the surface. The vertical pattern of this maximum cloud frequency suggests layered clouds with different tops but comparable thickness. The maximum frequency also extends to around 400 hPa with a larger optical thickness at 10. These are deep clouds. As was also pointed out in Webb et al. (2001), there is a general tendency for higher cloud top to associate with larger optical thickness, an indication of larger liquid and ice water paths associated with deeper clouds.
Figure 6d shows the climatological DJF cloud frequency distribution in the North Pacific storm track from CAM2. Several differences stand out. First, the model overestimated high-top optically thick and thin clouds. Second, the model missed the observed frequency maximum of middle and low clouds at an optical thickness of about 2.5. Third, the model had a spurious frequency maximum of optically thick low clouds. The model exaggerated optically thick clouds of all altitudes.
When integrated over the optical thickness, Fig. 6b shows that cloud amount below 500 hPa in the model is about half of what is observed. Above 500 hPa, the model cloud amount exceeds the observation by about 20%. In ISCCP, the cloud amount decreases with increasing altitude; in the model, the cloud frequency maximum is around 400 hPa. The underestimation of low and middle clouds dominates the total cloud amount bias, with the total cloud cover at 64% in the model versus 85% in the data. The underestimation of the total cloud amount shown in the previous section is therefore already after the partial offset from the excessive spurious high clouds in the model. Figure 6c shows that when integrated over all of the cloud tops, the model underestimated clouds of optical thickness smaller than 20 by about 50% and overestimated clouds of optical thickness larger than 20 by a factor of 2.
These multiple model biases in cloud optical thickness and altitudes compensate for each other in the TOA cloud-forcing fields. This is demonstrated in Fig. 7, which shows a breakdown of the contribution to the cloud forcing by different types of clouds in ISCCP and in CAM2. The attribution is calculated using NCAR's column radiation model with CAM2's radiation package, similar to Hartmann et al. (2001).
In ISCCP, high-top clouds of all optical thicknesses contribute appreciably to the longwave cloud forcing at the TOA. Middle-top optically thin and medium clouds contribute comparable fractions owing to their large amount (Fig. 7a). In contrast, Fig. 7b shows that high clouds, predominantly high-top optically thick clouds, mainly contribute to the longwave cloud forcing in the model. The first-order compensation of errors occurs in the model between optically thick high clouds and low-top and middle-top optically thin and medium clouds. Thus, even though the model underestimates cloud amount substantially, its longwave cloud forcing is close to that of the observations.
For the shortwave cloud forcing, Fig. 8c shows that the three leading contributors of ISCCP cloud types are the medium thickness clouds of all altitudes. In the model, however, the leading contributors are the optically thick clouds with high tops (Fig. 7d).
To gain some insight into the large-scale atmospheric conditions in which the model cloud differs from the observations (Bony et al. 1997), the dashed lines in Figs. 8a and 8e show the frequency histogram of the 500-hPa large-scale pressure vertical velocity from the National Centers for Environmental Prediction (NCEP)–NCAR reanalysis and CAM2 for one winter season of 1999/ 2000. Figures 9a and 9b further show the vertical distribution of the large-scale pressure vertical velocity plotted against its value at 500 hPa. It is evident that the frequency distribution of the vertical velocity is remarkably similar between the model and the reanalysis. This is consistent with the findings by Sanders et al. (2000) that the CCM captures the observed storm track activities fairly well.
Figures 8b–d show the frequency of different cloud types binned against the 500-hPa vertical velocity in ISCCP. These are compared with the corresponding CAM2 distributions in Figs. 8f–h. Upward motion of stronger than −100 hPa day−1 is predominantly associated with high-top/optically thick clouds in observations and in CAM2, but in the model these clouds have a much higher frequency and larger optical thickness (solid lines in Fig. 8e vs Fig. 8a). A significant amount of middle and low clouds are also present in regions of rising motions in observations. They are much less common in the model.
In regions of subsidence, observations show a considerable amount of middle- and low-top optically thin and intermediate clouds (Figs. 8c and 8d). These are associated with stratocumulus and shallow cumulus as a result of the temperature contrast between post-cold-frontal air and the relatively warm surface. The model underestimated these clouds in this subsidence regime (Figs. 8g and 8h).
To further understand the above cloud biases, we show in Fig. 9c cloud frequency histograms diagnosed solely from the model stratiform cloud scheme. This is very similar to Fig. 6d. Separate diagnostic calculations of cloud fraction from the convective source as described in Rasch and Kristjánsson (1998) and from the boundary layer cloud parameterization following Klein and Hartmann (1993) confirm small contributions from these sources (not shown).
Since relative humidity is the main control variable in the stratiform cloud scheme, we show the relative humidity distribution from the reanalysis and in CAM2 stratified against the pressure vertical velocity in Figs. 9d and 9e. In the regime of strong upward motion, the relative humidity in CAM2 is actually smaller than in the reanalysis, even though the model diagnosed substantially more optically thick high-top clouds. We therefore speculate that the two possible sources of excessive optically thick clouds in the deep upward motion regime are the parameterization of cloud amount and/ or cloud condensate.
Figures 8d and 8h show that the overestimation of optically thick low clouds in the model is associated with very weak vertical motion. Figure 9e also shows that in this regime the model is actually drier than in the reanalysis. Therefore, possible causes of the model cloud biases should also be from either the cloud amount parameterization or the cloud microphysics.
In the subsidence regime, the model has a relatively drier lower to middle troposphere (Fig. 9e), which could contribute to the deficient cloud amount. Furthermore, the model shallow convection scheme does not generate clouds. The occurrence of shallow and deep convection2 for the month of January is shown in Fig. 10. It is evident that shallow convection is frequent in the model in this region. There is, however, no cloud associated with these convective activities. Therefore, the underestimation of low- and middle-top clouds in the model is likely due to both the drier bias and the lack of convective clouds in the model.
The above features of model cloud biases—the overestimation of optically thick high-top clouds and very low clouds and the deficient medium thick middle and low clouds—are representative in other storm track areas. The generalization of these results is shown using the global climatology of high-top/optically thick clouds for DJF in ISCCP and CAM2 in Figs. 11a and 11b. The middle-top/optically medium and thin clouds from ISCCP and the CAM2 are shown in Figs. 11c– f. The lack of middle-top clouds in the model in the selected storm track region can be seen. The low-top/ optically thick clouds in the data and in CAM2 are shown in Figs. 11g and 11h. It is evident that the same deficiencies are widespread over all storm tracks in both hemispheres.
Norris and Weaver (2001) used daily cloud forcing to infer the model cloud biases in an earlier version of the CAM and reported model biases in its daily distributions. Klein and Jakob (1999) did a composite study of clouds with midlatitude cyclones and showed a lack of middle clouds west of the cold fronts in the ECMWF model. These are consistent with our analysis here. The lack of intermediate optical thickness clouds was also shown in Webb et al. (2001) for the Met office's Hadley Centre model and the European Centre for Medium-Range Weather Forecasts (ECMWF) model, but was less obvious in the Laboratoire de Météorologie Dynamique (LMD) GCM for this region in the month of July 1988.
b. In the warm pool
Figure 12a shows the cloud frequency histogram in DJF from ISCCP for the selected region of 10°N–10°S, 120°–150°E in the warm pool. The maximum cloud frequency is at around 300 hPa with an optical depth of 3.0. There is an extension of the maximum at the same altitude to a higher optical depth.
Figure 12d shows the corresponding model climatology from CAM2. The model overestimated optically thick clouds of all altitudes and underestimated middle and low-top/optically medium and thin clouds. These bias features are similar to those for the storm track clouds. Additionally, the model overestimated high thin cirrus in this region.
As seen in the vertical cloud amount integrated over optical thickness in Fig. 12b, the model underestimated cloud amount by about 46% below 310 hPa, but overestimated cloud above 310 hPa. When integrated over all altitudes in Fig. 12c, the model overestimated both optically thick and thin clouds, but underestimated optically medium clouds. The underestimation of middle and low-top/optically intermediate and thin clouds is also shown in Webb et al. (2001) in the Met Office and ECMWF models, but is less severe in the LMD GCM for a larger domain in the tropical warm pool. The overestimation of thin high clouds is also obvious in the three models. The overestimation of high-top optically thick clouds appears, however, to be a distinct feature of CAM2.
The reason for these cloud biases in the Tropics can be further examined by studying Fig. 13. It shows convection occurrences, cloud amount, and relative humidity at a single model grid at 9.8°N, 137.8°E. Relative humidity from a radiosonde sample at the Yap Island (9.48°N, 138.08°E) is also shown in Fig. 13d. There is almost continuous coverage of convection in the model, more than what the observed relative humidity suggests. There is also overestimation of relative humidity in the upper troposphere. Separate calculations of clouds from the model convective and stratiform cloud schemes show that about two-thirds of the model high clouds are from its stratiform scheme, and the rest is from the convective scheme. Overestimations of upper-tropospheric relative humidity and convection occurrence in the model should have both contributed to the excessive high clouds. The latter is a known feature of CAM2 (e.g., Xie and Zhang 2000), and the two might be related to each other. This does not, of course, exclude the possibility of algorithmic bias in the cloud amount parameterization from both sources.
Optically thick high clouds in the model are associated with deep events in the relative humidity field in Fig. 13c, which correspond to deep large-scale-diagnosed upward motion. It is possible that the reason of the overestimation of optically thick high-top clouds is the same as those for storm track clouds in dynamical regimes of upward motion.
The general underestimation of middle and low clouds in the model is related to the lack of clouds from the model deep and shallow convection schemes below 500 hPa. The mean relative humidity of the model in the lower troposphere is actually larger than the reanalysis, except in the very lowest model layer, in which it is drier.
These multiple cloud biases offset each other to minimize the cloud-forcing difference between the model and the observations. The near cancellation of longwave and shortwave cloud forcing is maintained (Kiehl 1994). This compensation of errors is similar to what was shown previously for the storm tracks. The excessive thin high-cloud bias at the analyzed warm pool region is representative of other regions of tropical convections. This is shown in Figs. 11i and 11j.
c. The subtropics
We now examine the comparison in the subtropics by selecting the region of 12.5°–25°N, 150°–210°E. Figure 14 shows the cloud frequency distribution for the DJF season. The observation shows primarily optically thin low and middle clouds, while the model shows optically thick low clouds and very high thin clouds. These biases are similar to what were shown for the midlatitude storm tracks and the deep convective centers, except that the overestimation of optically thick low clouds is more predominant. These biases are also similar to biases shown for the Hawaii trade cumulus region in the Met Office and ECMWF models by Webb et al. (2001).
Figure 15 shows the comparison of different cloud types in ISCCP binned against the large-scale vertical velocity in the NCEP–NCAR reanalysis versus those in CAM2 for the winter season of 1999/2000. In ISCCP (Figs. 15c and 15d), optically thin and medium low clouds dominate in the weak ascent and subsidence regime where the 500-hPa pressure vertical velocity is larger than −100 hPa day−1 which should mostly be associated with trade cumulus. CAM2 (Figs. 15f and 15g), however, failed to produce these clouds.
It is noted that large-scale subsidence in the reanalysis extends to lower levels than in CAM2, and CAM2 seems to have overestimated the thickness of the PBL height (Figs. 16a and 16b). The model tends to overestimate humidity from 800 to 900 hPa in the subsidence regime as a result of the deeper boundary layer (Fig. 16c), even though clouds are underestimated. The model is somewhat drier in other parts of the middle and lower troposphere, which could be partially responsible for the underestimation of middle clouds. The model diagnosed more optically thick clouds in the lowest model layer, which is actually drier than in the reanalysis. Overall, we speculate that the drier lower to middle troposphere and the lack of clouds from the model convection scheme both contribute to the deficient low-to middle-top clouds in the model. The excessive high-top thin cirrus in the model was found to be from the model stratiform cloud scheme. It is possibly related to the cold bias, and thus the large relative humidity, in the model. It is restricted to be above the 150-hPa level, and the exact cause of this cold bias is not clear. Similar to other regions, the model cloud errors compensate for each other to produce the reasonable cloud forcing at the TOA. These are shown in Fig. 17.
d. Global biases
We now generalize the major cloud biases of model cloud structures to all latitudes. Figures 18a and 18b show the difference of the cloud frequency histogram between CAM2 and ISCCP for each zonally averaged 15° latitude bin for the DJF and JJA seasons, respectively. The principal mode of the difference is consistent with what was shown for the four regions discussed above: overestimation of high-top optically thin, high-top optically thick, and low-top optically thick clouds and underestimation of middle and low clouds of intermediate and thin optical thickness. The underestimation of clouds dominates in the midlatitudes and subtropics to register the signature of deficient total cloud at these latitudes, and in the Tropics it almost offsets the overestimation of thin cirrus to produce a reasonable total cloud amount. Compensation of radiative forcing between high and low clouds and between optically thick and intermediate clouds yields the reasonable cloud-forcing simulation in the model. These features apply to both the winter and summer seasons.
5. Discussion and summary
We wish to raise several caveats in the interpretation of the presented results. First, in ISCCP, since a visible channel is used to derive the optical thickness, the ISCCP data used here represents daytime climatology, while the CAM2 results are from 24-h climatology. Diurnal variations of clouds are known to exist. A sensitivity study has been carried out to examine model daytime-only clouds. We have found that the major cloud biases highlighted above are very similar. Second, in ISCCP, partially cloudy pixels affect both the scene identification and the estimations of cloud top and optical thickness. The finite resolution of ISCCP pixels may produce a positive bias in the ISCCP cloud amount and a negative bias in the cloud optical thickness. This would cause the model to underestimate cloud amount and overestimate its optical depth. Given that the original pixel size of ISCCP data is only several kilometers, we believe the reported biases are most likely real. Furthermore, for low clouds, in particular low clouds associated with trade cumulus and stratocumulus in the subtropics where there are boundary layer inversions, ISCCP may have underestimated the cloud amount due to its use of the TOVS temperature profile in detecting cloud tops. Our analysis showed overall underestimation of low clouds in the model in comparison with ISCCP. Therefore, the sign of the model bias should be correct. With respect to the comparison of cloud radiative forcing from the model with ERBE in section 3, we note that potential uncertainties in the ERBE data associated with their scene identification, angular distribution models, and diurnal models could reach up to 8 W m−2 in the shortwave forcing and 4 W m−2 in the longwave forcing (Loeb et al. 2003; N. G. Loeb 2003, personnel communication) in the zonally averaged monthly fields. These, however, should not change the overall conclusion that CAM2 agrees better with satellite measurements in TOA cloud radiative forcing fields than in clouds.
We have shown that CAM2 underestimated cloud amount in the storm tracks and in the subtropics of the two hemispheres. It overestimated total cloud amount in the tropical convective regions. Yet, its simulation of the TOA cloud radiative forcing is close to ERBE measurements. When stratified against cloud-top pressure and optical depth, the model is shown to have the following biases: (i) overestimation of high optically thin clouds, (ii) overestimation of high-top optically thick clouds, (iii) overestimation of low-top optically thick clouds, and (iv) significant underestimation of middle- and low-top optically intermediate and thin clouds. In general, the excessive high clouds and deficient middle-top clouds compensate for errors in the longwave cloud forcing, while excessive optically thick clouds and deficient optically medium clouds compensate for errors in the shortwave cloud forcing.
We have also examined the association of large-scale atmospheric conditions with the cloud biases. The overestimation of high-top optically thick clouds occurs in the regime of large-scale deep upward motion in both the midlatitudes and the Tropics. The overestimation of very thin high clouds in the Tropics is associated with larger relative humidity and excessive convection frequency in the model. The overestimation of low-top optically thick clouds is confined to the lowest model layer, presumably due to the stratiform cloud parameterization layers that should be described by a moist boundary layer cloud parameterization. The general lack of middle- and low-top optically intermediate and thin clouds in the midlatitude storm tracks and in the subtropics of the model is associated with both a drier model atmosphere and the inability of the model convection to produce clouds, while the lack of these clouds in the warm pool is more related to the inability of model convection to produce clouds. The understanding of the exact causes of these model biases, tracing back to specific cloud parameterization components, however, requires more diagnostic studies and sensitivity tests. These will be pursued in the future.
Acknowledgments
The authors wish to thank Drs. Steve Klein at GFDL and Mark Webb at the Met Office for making their ISCCP simulator available to this study, and Drs. Jim Hack and Jeff Kiehl at NCAR for valuable discussions at different stages of this work. We also thank the two anonymous reviewers whose comments have led to improvements to the original paper. The ISCCP data was obtained from NASA Langley Data Center. This research was supported by the ARM program of the U.S. Department of Energy under Grant EFG0298ER62570, by the National Science Foundation under Grant ATM901950, and by NASA, under its TRMM and GPM programs, to the Stony Brook University.
REFERENCES
Barkstrom, B. R., and G. L. Smith, 1986: The Earth Radiation Budget Experiment: Science and implementation. Rev. Geophys, 24 , 379–390.
Bony, S., K. M. Lau, and Y. C. Sud, 1997: Sea surface temperature and large-scale circulation influences on tropical greenhouse effect and cloud radiative forcing. J. Climate, 10 , 2055–2077.
Cess, R. D., and Coauthors, 1990: Intercomparison and interpretation of climate feedback processes in 19 atmospheric general circulation models. J. Geophys. Res, 95 , 16601–16615.
Collins, W., and Coauthors, cited 2003: Description of the NCAR Community Atmospheric Model (CAM2). [Available online at http://www.ccsm.ucar.edu/models/atm-cam/docs/cam2.0/description.pdf.].
Cubasch, U., and Coauthors, 2001: Projection of future climate change. Climate Change 2001: The Scientific Basis, J. T. Houghton et al., Eds., Cambridge University Press, 527–582.
Del Genio, A. D., M-S. Yao, W. Kovari, and K. K-W. Lo, 1996: A prognostic cloud water parameterization for global climate models. J. Climate, 9 , 270–304.
Hack, J. J., 1994: Parameterization of moist convection in the National Center for Atmospheric Research Community Climate Model (CCM2). J. Geophys. Res, 99 , 5551–5568.
Hack, J. J., J. Hurrell, J. Rosinski, and J. Caron, cited 2002: The NCAR CGD annual scientific report 2002. [Available online at http://www.cgd.ucar.edu/asr02/CMS.htm.].
Harrison, E. F., P. Minnis, B. R. Barkstrom, V. Ramanathan, R. D. Cess, and G. G. Gibson, 1990: Seasonal variation of cloud radiative forcing derived from the Earth Radiation Budget Experiment. J. Geophys. Res, 95 , 18687–18703.
Hartmann, D. L., A. M. Leslie, and Q. Fu, 2001: Tropical convection and the energy balance at the top of the atmosphere. J. Climate, 14 , 4495–4511.
Kiehl, J. T., 1994: On the observed near cancellation between longwave and shortwave cloud forcing in tropical regions. J. Climate, 7 , 559–565.
Kiehl, J. T., J. J. Hack, G. B. Bonan, B. A. Boville, D. L. Williamson, and P. J. Rasch, 1998a: The National Center for Atmospheric Research Community Climate Model: CCM3. J. Climate, 11 , 1131–1150.
Kiehl, J. T., J. J. Hack, and J. W. Hurrell, 1998b: The energy budget of the NCAR Community Climate Model: CCM3. J. Climate, 11 , 1151–1178.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6 , 1587–1606.
Klein, S. A., and C. Jakob, 1999: Validation and sensitivities of frontal clouds simulated by the ECMWF model. Mon. Wea. Rev, 127 , 2514–2531.
Loeb, N. G., K. Loukachine, N. Manalo-Smith, B. Wielicki, and D. Young, 2003: Angular distribution models for top-of-atmosphere radiative flux estimation from the Clouds and the Earth's Radiant Energy System instrument on the Tropical Rainfall Measuring Mission satellite. Part II: Validation. J. Appl. Meteor, 42 , 1748–1769.
Norris, J. R., and C. P. Weaver, 2001: Improved techniques for evaluating GCM cloudiness applied to the NCAR CCM3. J. Climate, 14 , 2540–2550.
Randall, D. A., and Coauthors, 2003: Confronting models with data: The GEWEX cloud systems study. Bull. Amer. Meteor. Soc, 84 , 455–469.
Rasch, P. J., and J. E. Kristjánsson, 1998: A comparison of the CCM3 model climate using diagnosed and predicted condensate parameterizations. J. Climate, 11 , 1587–1614.
Rossow, W. B., and R. A. Schiffer, 1991: ISCCP cloud data products. Bull. Amer. Meteor. Soc, 72 , 2–20.
Rossow, W. B., A. W. Walker, D. Beuschel, and M. Roiter, 1996: International Satellite Cloud Climatology Project (ISSCP) description of new cloud datasets. WMO/TD 737, World Climate Research Programme (ICSU and WMO), 115 pp.
Sanders, F., S. L. Mullen, and D. P. Baumhefner, 2000: Ensemble simulations of explosive cyclogenesis at ranges of 2–5 days. Mon. Wea. Rev, 128 , 2920–2934.
Senior, C. A., and J. F. B. Mitchell, 1993: Carbon dioxide and climate: The impact of cloud parameterization. J. Climate, 6 , 393–418.
Senior, C. A., and J. F. B. Mitchell, 1996: Cloud feedbacks in the unified UKMO GCM. Climate Sensitivity to Radiative Perturbations, Physical Mechanisms and Their Validation, H. Le Treut, Ed., Springer, 191– 202.
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.
Xie, S. C., and M. H. Zhang, 2000: Impact of the convection triggering function on single-column model simulations. J. Geophys. Res, 105 , 14983–14996.
Yu, M., M. Doutriaux, G. Sèze, H. Le Treut, and M. Desbois, 1996: A methodology study of the validation of clouds in GCMs using ISCCP satellite observations. Climate Dyn, 12 , 389–401.
Zhang, G. J., and N. A. McFarlane, 1995: Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model. Atmos.– Ocean, 33 , 407–446.
Zhang, M., W. Lin, C. Bretherton, J. J. Hack, and P. J. Rasch, 2003: A modified formulation of fractional stratiform condensation rate in the NCAR Community Atmospheric Model (CAM2). J. Geophys. Res.,108, 4035, doi:10.1029/2002JD002523.
Zonally averaged total cloud amount in CAM2 from the ISCCP simulator (solid) and from the direct model output (dashed): (a) DJF and (b) JJA
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Zonally averaged total cloud amount in CAM2 from the ISCCP simulator (solid) and from the direct model output (dashed): (a) DJF and (b) JJA
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Zonally averaged total cloud amount in CAM2 from the ISCCP simulator (solid) and from the direct model output (dashed): (a) DJF and (b) JJA
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Time–pressure cross section of the cloud amount from the CAM2 simulation at the ARM SGP for Mar 2000. Dark shading indicates no data below surface. (b) Cloud amount and cloud-top pressure from the ISCCP simulator that corresponds to the cloud distribution in (a)
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Time–pressure cross section of the cloud amount from the CAM2 simulation at the ARM SGP for Mar 2000. Dark shading indicates no data below surface. (b) Cloud amount and cloud-top pressure from the ISCCP simulator that corresponds to the cloud distribution in (a)
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Time–pressure cross section of the cloud amount from the CAM2 simulation at the ARM SGP for Mar 2000. Dark shading indicates no data below surface. (b) Cloud amount and cloud-top pressure from the ISCCP simulator that corresponds to the cloud distribution in (a)
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) DJF ISCCP total cloud amount, (b) DJF LW CRF from ERBE, and (c) DJF SW CRF from ERBE. (d)–(f) Same as in (a)–(c), except for DJF CAM2. Units for CRF are W m−2. [The three regions of NP, WP, and the subtropics are boxed in (a) and (d)]
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) DJF ISCCP total cloud amount, (b) DJF LW CRF from ERBE, and (c) DJF SW CRF from ERBE. (d)–(f) Same as in (a)–(c), except for DJF CAM2. Units for CRF are W m−2. [The three regions of NP, WP, and the subtropics are boxed in (a) and (d)]
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) DJF ISCCP total cloud amount, (b) DJF LW CRF from ERBE, and (c) DJF SW CRF from ERBE. (d)–(f) Same as in (a)–(c), except for DJF CAM2. Units for CRF are W m−2. [The three regions of NP, WP, and the subtropics are boxed in (a) and (d)]
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Zonally averaged DJF total cloud amount in ISCCP (solid) and CAM2 (dashed). (b) Zonally averaged DJF LW CRF from ERBE (solid) and CAM2 (dashed). (c) Same as in (b), except for SW CRF. CRF units are W m−2
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Zonally averaged DJF total cloud amount in ISCCP (solid) and CAM2 (dashed). (b) Zonally averaged DJF LW CRF from ERBE (solid) and CAM2 (dashed). (c) Same as in (b), except for SW CRF. CRF units are W m−2
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Zonally averaged DJF total cloud amount in ISCCP (solid) and CAM2 (dashed). (b) Zonally averaged DJF LW CRF from ERBE (solid) and CAM2 (dashed). (c) Same as in (b), except for SW CRF. CRF units are W m−2
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Same as in Fig. 4, except for JJA
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Same as in Fig. 4, except for JJA
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Same as in Fig. 4, except for JJA
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) DJF cloud frequency stratified against cloud-top pressure ( pc) and cloud optical depth (τ) for the NP storm track. (b) Cloud frequency integrated over τ, for ISCCP (solid) and CAM2 (dashed). (c) Cloud frequency integrated over pc for ISCCP (solid) and CAM2 (dashed). (d) Same as in (a), except from CAM2. Abscissas in (a) and (d) are optical depths, in (b) and (c) and are cloud fractions. Ordinates in (a), (b), and (d) are cloud-top pressure, and are optical depth-in (c)
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) DJF cloud frequency stratified against cloud-top pressure ( pc) and cloud optical depth (τ) for the NP storm track. (b) Cloud frequency integrated over τ, for ISCCP (solid) and CAM2 (dashed). (c) Cloud frequency integrated over pc for ISCCP (solid) and CAM2 (dashed). (d) Same as in (a), except from CAM2. Abscissas in (a) and (d) are optical depths, in (b) and (c) and are cloud fractions. Ordinates in (a), (b), and (d) are cloud-top pressure, and are optical depth-in (c)
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) DJF cloud frequency stratified against cloud-top pressure ( pc) and cloud optical depth (τ) for the NP storm track. (b) Cloud frequency integrated over τ, for ISCCP (solid) and CAM2 (dashed). (c) Cloud frequency integrated over pc for ISCCP (solid) and CAM2 (dashed). (d) Same as in (a), except from CAM2. Abscissas in (a) and (d) are optical depths, in (b) and (c) and are cloud fractions. Ordinates in (a), (b), and (d) are cloud-top pressure, and are optical depth-in (c)
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Decomposed DJF LW CRF from ISCCP for the NP storm track based on cloud-top height and optical thickness. (b) Same as in (a) except from CAM2. (c), (d) Same as in (a) except for SW CRF and SW CRF, respectively
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Decomposed DJF LW CRF from ISCCP for the NP storm track based on cloud-top height and optical thickness. (b) Same as in (a) except from CAM2. (c), (d) Same as in (a) except for SW CRF and SW CRF, respectively
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Decomposed DJF LW CRF from ISCCP for the NP storm track based on cloud-top height and optical thickness. (b) Same as in (a) except from CAM2. (c), (d) Same as in (a) except for SW CRF and SW CRF, respectively
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Frequency histogram (dashed) of 500-hPa pressure vertical velocity (ω) and total optical thickness (solid) from NCEP–NCAR reanalysis and ISCCP for DJF of 1999/2000. Abscissa is ω in hPa day−1. (b)–(d) Frequency of high, middle, and low clouds of various optical thickness binned against ω from ISCCP. Solid, dashed, and dotted lines represent optically thick, intermediate, and thin clouds, respectively. (e)–(h) Same as in (a)–(d) except from CAM2. The ordinates for frequency of ω are on the right axis in (a) and on the left in (e)
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Frequency histogram (dashed) of 500-hPa pressure vertical velocity (ω) and total optical thickness (solid) from NCEP–NCAR reanalysis and ISCCP for DJF of 1999/2000. Abscissa is ω in hPa day−1. (b)–(d) Frequency of high, middle, and low clouds of various optical thickness binned against ω from ISCCP. Solid, dashed, and dotted lines represent optically thick, intermediate, and thin clouds, respectively. (e)–(h) Same as in (a)–(d) except from CAM2. The ordinates for frequency of ω are on the right axis in (a) and on the left in (e)
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Frequency histogram (dashed) of 500-hPa pressure vertical velocity (ω) and total optical thickness (solid) from NCEP–NCAR reanalysis and ISCCP for DJF of 1999/2000. Abscissa is ω in hPa day−1. (b)–(d) Frequency of high, middle, and low clouds of various optical thickness binned against ω from ISCCP. Solid, dashed, and dotted lines represent optically thick, intermediate, and thin clouds, respectively. (e)–(h) Same as in (a)–(d) except from CAM2. The ordinates for frequency of ω are on the right axis in (a) and on the left in (e)
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Reanalysis pressure vertical velocity as a function of its value at 500-hPa for the NP storm track for DJF of 1999/2000. (b) Same as in (a) except from CAM2. (c) Cloud histogram diagnosed from the CAM2 RH-based stratiform cloud scheme. (d) Reanalysis RH as a function of 500-hPa pressure vertical velocity. (e) Same as in (c) except from CAM2
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Reanalysis pressure vertical velocity as a function of its value at 500-hPa for the NP storm track for DJF of 1999/2000. (b) Same as in (a) except from CAM2. (c) Cloud histogram diagnosed from the CAM2 RH-based stratiform cloud scheme. (d) Reanalysis RH as a function of 500-hPa pressure vertical velocity. (e) Same as in (c) except from CAM2
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Reanalysis pressure vertical velocity as a function of its value at 500-hPa for the NP storm track for DJF of 1999/2000. (b) Same as in (a) except from CAM2. (c) Cloud histogram diagnosed from the CAM2 RH-based stratiform cloud scheme. (d) Reanalysis RH as a function of 500-hPa pressure vertical velocity. (e) Same as in (c) except from CAM2
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Occurrences of deep and shallow convections in CAM2 for Jan 2000. The layers having convection are marked using dots for deep convection and diamonds for shallow convection. Thin lines are formed by continuous dots. The abscissa is the calendar day in Jan
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Occurrences of deep and shallow convections in CAM2 for Jan 2000. The layers having convection are marked using dots for deep convection and diamonds for shallow convection. Thin lines are formed by continuous dots. The abscissa is the calendar day in Jan
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Occurrences of deep and shallow convections in CAM2 for Jan 2000. The layers having convection are marked using dots for deep convection and diamonds for shallow convection. Thin lines are formed by continuous dots. The abscissa is the calendar day in Jan
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a), (b) High-top thick clouds in ISCCP and CAM2; (c), (d) middle-top medium thick clouds in ISCCP and CAM2; (e), (f) middle-top thin clouds in ISCCP and CAM2; (g), (h) low-top thick clouds in ISCCP and CAM2; and (i), ( j) high-top thin clouds in ISCCP and CAM2
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a), (b) High-top thick clouds in ISCCP and CAM2; (c), (d) middle-top medium thick clouds in ISCCP and CAM2; (e), (f) middle-top thin clouds in ISCCP and CAM2; (g), (h) low-top thick clouds in ISCCP and CAM2; and (i), ( j) high-top thin clouds in ISCCP and CAM2
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a), (b) High-top thick clouds in ISCCP and CAM2; (c), (d) middle-top medium thick clouds in ISCCP and CAM2; (e), (f) middle-top thin clouds in ISCCP and CAM2; (g), (h) low-top thick clouds in ISCCP and CAM2; and (i), ( j) high-top thin clouds in ISCCP and CAM2
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Cloud frequency distribution for the WP: 10°N–10°S, 120°–150°E; same as in Fig. 6
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Cloud frequency distribution for the WP: 10°N–10°S, 120°–150°E; same as in Fig. 6
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Cloud frequency distribution for the WP: 10°N–10°S, 120°–150°E; same as in Fig. 6
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Occurrences of deep and shallow convections in WP from CAM2 for DJF of 1999/2000. Symbols are the same as in Fig. 10. (b) Simulated cloud amount corresponding to (a); (c) simulated RH, and (d) RH soundings from Yap Island
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Occurrences of deep and shallow convections in WP from CAM2 for DJF of 1999/2000. Symbols are the same as in Fig. 10. (b) Simulated cloud amount corresponding to (a); (c) simulated RH, and (d) RH soundings from Yap Island
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Occurrences of deep and shallow convections in WP from CAM2 for DJF of 1999/2000. Symbols are the same as in Fig. 10. (b) Simulated cloud amount corresponding to (a); (c) simulated RH, and (d) RH soundings from Yap Island
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Cloud frequency for the subtropics: 12.5°–25°N, 150°E–150°W; same as in Fig. 6
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Cloud frequency for the subtropics: 12.5°–25°N, 150°E–150°W; same as in Fig. 6
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Cloud frequency for the subtropics: 12.5°–25°N, 150°E–150°W; same as in Fig. 6
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
The 500-hPa pressure vertical velocity histogram; same as in Fig. 8, except for the subtropics
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
The 500-hPa pressure vertical velocity histogram; same as in Fig. 8, except for the subtropics
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
The 500-hPa pressure vertical velocity histogram; same as in Fig. 8, except for the subtropics
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Reanalysis pressure vertical velocity as a function of its value at 500-hPa for the subtropics for DJF of 1999/2000. (b) Same as in (a) except from CAM2. (c) Difference of RH between CAM2 and reanalysis as a function of 500-hPa pressure vertical velocity
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Reanalysis pressure vertical velocity as a function of its value at 500-hPa for the subtropics for DJF of 1999/2000. (b) Same as in (a) except from CAM2. (c) Difference of RH between CAM2 and reanalysis as a function of 500-hPa pressure vertical velocity
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
(a) Reanalysis pressure vertical velocity as a function of its value at 500-hPa for the subtropics for DJF of 1999/2000. (b) Same as in (a) except from CAM2. (c) Difference of RH between CAM2 and reanalysis as a function of 500-hPa pressure vertical velocity
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
CRF decomposition for the subtropics; same as in Fig. 7
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
CRF decomposition for the subtropics; same as in Fig. 7
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
CRF decomposition for the subtropics; same as in Fig. 7
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Difference of cloud frequency between CAM2 and ISCCP for different latitude bands during (a) DJF and (b) JJA. The abscissa between any two latitude labels is optical depth increasing from left to right
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Difference of cloud frequency between CAM2 and ISCCP for different latitude bands during (a) DJF and (b) JJA. The abscissa between any two latitude labels is optical depth increasing from left to right
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Difference of cloud frequency between CAM2 and ISCCP for different latitude bands during (a) DJF and (b) JJA. The abscissa between any two latitude labels is optical depth increasing from left to right
Citation: Journal of Climate 17, 17; 10.1175/1520-0442(2004)017<3302:EOCATR>2.0.CO;2
Version 2.2.1.1 is used in this study.
Here, shallow convection refers to convection from the Hack (1994) scheme and deep convection refers to penetrative convection from the Zhang and McFarlane (1995) scheme.
