Sensitivity Tests Using the Revised PBL Parameterization in the CSU GCM

Recent revisions have been made to the planetary boundary layer (PBL) parameterization in the Colorado State University General Circulation Model (CSU GCM), which have improved the simulation of PBL clouds in the model. In addition, a number of sensitivity tests have recently been completed following some of the recent modifications made to the UCLA atmospheric general circulation model (Li and Arakawa, 1998), which involve improved representation of PBL moist processes.

Among the recent changes to the CSU GCM's PBL parameterization was the inclusion of an iterative scheme to correct the PBL top temperature (TB) and surface temperature in cases where the PBL is partially or fully saturated. Previously, TB was computed from the PBL potential temperature and PBL top pressure assuming a dry adiabatic lapse rate which results in an underestimation of TB. In addition, the "jump" values across the PBL top have been defined more precisely and consistently throughout the code. Finally, consolidation of the model subroutines that compute PBL quantities has been completed, and during this process a number of minor bugs were corrected.

With these changes, a July simulation was performed with the CSU GCM, and the resulting PBL cloud incidence field is shown in the upper left panel of the attached figure (labelled "control"). The distribution of the subtropical PBL clouds off the west coasts of North America, South America and Africa is quite satisfactory, as is the cloud amount in the southern hemisphere "storm track region".

One of the recent modifications to the UCLA AGCM (Li and Arakawa, 1998) involves the specification of the level B+ thermodynamic quantities (T, q, etc.). Previously, both models used the predicted value at the center of the layer above the PBL as the B+ value. However, in the case of water vapor mixing ratio, when the qB+ value is entrained into the PBL, the resulting drying of the PBL may be overestimated since this value is generally less than the value expected at the PBL top. A value for qB+ obtained by downward extrapolation from above should give a more realistic amount of moisture entrainment, since in most cases the water vapor mixing ratio decreases with height above the top of the PBL. Another July simulation was performed with this modification and the resulting PBL cloud incidence field is shown in the upper right panel of the attached figure. Indeed, the extrapolation of the qB+ value increased the amount of PBL clouds, particularly in the storm track region as well as the northern Atlantic.

A second sensitivity test was performed whereby the temperature at level B+ was recomputed in a similar manner (see the lower left panel). This changes the In this experiment, there is a noticeable decrease in the PBL cloud amount in the subtropical regions compared to both the control and extrapolate-qB+ experiments, with an increase seen in the Arctic region.

Finally, another test was performed where the term containing the jump in the longwave radiative flux across the PBL top was removed from the entrainment rate formula. This effectively reduces the entrainment rate in regions with , thus This gave a slightly higher PBL cloud incidence than in the "control" simulation.

The Li and Arakawa step 3 revision whereby a PBL fractional cloudiness is computed based on a formula involving the ratio of the subgrid scale orography to the PBL depth is currently being tested.

References

Li, J.-L. and A. Arakawa, 1998: Improved representation of PBL moist processes in the UCLA general circulation model. Journal of Climate, submitted.