Testing Cloud Radiation Algorithms in Climate Models and in the Atmosphere



Sam Iacobellis, Dana Lane, and Richard Somerville


Introduction

In this presentation we show results from the Single Column Model (SCM) produced using different cloud parameterizations. Our primary goal is to determine how cloud radiative forcing may be affected by the inclusion of cloud liquid water as a prognostic variable. We also examine how sensitive the cloud optical properties are to the parameterization of the effective cloud droplet radius.

Model Description

Three different configurations of the SCM were used in this study. These configurations differed only in the parameterization of cloud amount, cloud droplet radius, and/or cloud optical thickness. These model configurations are summarized below in Table 1.

Cloud Liquid Water Cloud Amount Cloud Optical Thickness Effective Droplet Radius
NOCW None Slingo (1987) McFarlane et al. (1992)
Calculated as fnc(T,P)
Not Applicable
CWTRF Explicit Tiedtke (1993) Water: Slingo (1989)
Ice: Fu (1996)
Calculated as fnc(LWP)
Fixed at 10 microns
CWTRI Explicit Tiedtke (1993) Water: Slingo (1989)
Ice: Fu (1996)
Calculated as fnc(LWP)
Water: Bower et al (1994)
Ice: Suzuki et al (1993)
Calculated as fnc(LWC,T)
Table 1: SCM Configurations.

Observational data collected within the TOGA-COARE Intensive Flux Array (IFA) (Figure 1) was used to force the SCM. During COARE, sounding measurements were collected at the four sites defining the perimeter of the IFA every three hours. Lin and Johnson (1996) used objective analysis to compute the horizontal advective fluxes of heat, moisture and momentum along with vertical profiles of temperature, humidity, and velocity. This data is available via ftp from Colorado State University. We concentrate our study on three time periods when sounding measurements were present from all four locations. These periods (10Nov92-01Dec92, 18Dec92-23Jan93, and 31Jan93-18Feb93) contain a total of 79 days. The SCM is integrated using a timestep of 7.5 minutes. The SST and the surface fluxes of latent and sensible heat are specified from observations. To reduce model drift, the SCM temperature and humidity profiles are relaxed to observed values using a time constant of 24 hours.

Results

The SCM precipitation from all 3 configurations compares very well with both satellite and surface measurements. The SCM precipitation during the second time period (18Dec92-23Jan93) is compared against that measured by the SSMI on board the DSDP satellites in Figure 2. All three configurations of the SCM produce very similar precipitation results. The three large precipitation events during this period were well reproduced by the SCM. Between the large precipitation events, the SCM appears to overestimate the precipitation compared to the SSMI measurements. In Figure 3 the diurnal variation of the mean model precipitation during disturbed periods is compared to measurements gathered by Sui et al (1997). Sui et al (1997) define disturbed conditions as those periods with a mean brightness temperature Tbb of less than 270K and a standard deviation of Tbb greater than 19K. The daily mean has been removed from the curves shown in Figure 3. The diurnal variation from each of the three SCM runs shows a similar structure to the observations. Both the model and observations have a precipiation maximum during the early morning hours, albeit the observed maximum is at 3AM while the model maximum is at 12 Midnight.

The model produced cloud forcing terms averaged over the 79 days of the three time periods are shown in Table 2 along with measured values from the same periods from ISCCP data. The model run NOCW has a longwave cloud forcing (CWLW) and cloud fraction very similar to ISCCP, but the shortwave cloud forcing is 14 W m-2 lower. This implies that the clouds produced by the SCM are not as reflective as those measured by ISCCP. The cloud forcing terms from run CWTRF are very close to those from ISCCP, but the model cloud fraction is higher than observed. One possible explaination is that like NOCW, the model clouds from CWTRF are less reflective than the observed clouds, but now since there are more of them, they also reflect more sunlight producing a higher value of CFSW. Most of these additional clouds would probably be located in the lower troposphere since the CFLW does not change much from the NOCW value. Another possible explaination is that the ISCCP data underestimates the cloud fraction. The radiative results from run CWTRI are very similar to CWTRF except for a decrease in the CFSW of 16 W m-2 and is due to the parameterization of the effective cloud droplet radius. These differences are discussed below.

CFSW CFLW CLD FRAC
NOCW -78 50 0.72
CWTRF -95 46 0.87
CWTRI -79 45 0.86
ISCCP -92 50 0.73
Table 2: Cloud Forcing Results.

The vertical profile of the mean cloud amount from each model run is shown in Figure 4. The cloud amounts from CWTRF and CWTRI are nearly identical and have significantly more low clouds and slightly more high cirrus clouds than run NOCW. Additionally, run NOCW produces more clouds in the middle troposphere between 900 and 400 mb than either CWTRF or CWTRI. These differences are attributable mostly to the different cloud parameterizations. The peak in low clouds seen in the CWTRF and CWTRI values are due are boundary layer clouds which are explicitly included in the Tiedtke cloud parameterization. The lack of clouds in the 900-400 mb region is due to the absence of cumulus detrainment in this region and may be an artifact of the particular cumulus convection routine that was used. These runs used the cumulus parameterization from CCM3 (Zhang and McFarlane, 1995) which does not allow the detrainment of cloud liquid water in regions where the cumulus mass flux is increasing with height.

The vertical profile of the mean cloud optical thickness from each model run is shown in Figure 5. The optical thicknesses from run NOCW were calculated using the midlayer temperature and pressure and are essentially based on climatology. The inclusion of interactive cloud liquid water (CWTRF and CWTRI) acts to increase the cloud optical thickness of low clouds and to decrease the cloud optical thickess of high cirrus clouds. The differences between runs CWTRF and CWTRI are smaller than the differences noted above but are still significant. Interactively calculating the cloud droplet radius (run CWTRI) acts to increase the optical thickness of low clouds and to decrease the optical thickness of higher clouds. The net effect reduces the mean shortwave cloud forcing and has only a slight effect on the mean longwave cloud forcing.

Figure 6 shows the model mean cloud droplet radius versus height for runs CWTRF and CWTRI. Compared to the constant droplet radius of 10 microns in CWTRF, the mean droplet radius produced by run CWTRI is smaller below 600 mb and larger above 600 mb. These differences are responsible for the differences in cloud optical thicknesses (shown in adjoining plot). These results show that the use of a constant value of 10 microns in not a good approximation for either water or ice clouds in the tropical atmosphere and can lead to potentially significant variations in cloud optical properties.

Conclusions


Due to the uncertainties in both the model we hesitate to use the results of this preliminary study to rate the various parameterizations against each other. We do feel that the results of this study are important in showing the differences in cloud radiative forcing and cloud optical properties that would result from the different types of parameterizations. We can say fairly confidently is that the cloud optical properties and hence the cloud forcing terms are sensitive to both the inclusion of interactive cloud liquid water and to the cloud droplet radius parameterization.

References

Bower, K. N., T. W. Choularton, J. Latham, J. Nelson, M. B. Baker, and J. Jenson, 1994: A parameterization of warm clouds for use in atmospheric general circulation models. J. Atmos. Sci., 51, 2722-2732.

Fu, Q. A., 1996: An accurate parameterization of the solar radiative properties of cirrus clouds for climate models. J. Climate, 9, 2058-2082.

Lin, X., and R. H. Johnson, 1996: Heating, moistening, and rainfall over the western Pacific warm pool during TOGA COARE. J. Atm. Sci., 53, 3367-3383.

McFarlane, N. A., G. J. Boer, J.-P. Blanchet, and M. Lazare, 1992: The Canadian Climate Centre second-generation general circulation model and its equilibrium climate. J. Climate, 5, 1013-1044.

Slingo, A., 1989: A GCM parameterization for the shortwave radiative properties of water clouds. J. Atmos. Sci., 46, 1419-1427.

Slingo, J. M., 1987: The development and verification of a cloud prediction scheme for the ECMWF model. Q. J. R. Meteorol. Soc., 113, 899-927.

Sui, C.-H., K.-M. Lau, Y. N. Takayabu, and D. A. Short, 1997: Diurnal variations in tropical oceanic cumulus convection during TOGA COARE. J. Atmos. Sci., 54, 639-655.

Suzuki, T., M. Tanaka, and T. Nakajima, 1993: The microphysical feedback of cirrus cloud in climate change. J. Meteor. Soc. Japan, 71, 701-713.

Tiedtke, M., 1993: Representation of clouds in large-scale models. Mon. Wea. Rev., 121, 3040-3061.

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.

Back to SCM Research at SIO