Investigation of Cloud Representation from Stratocumulus to Deep Convection in the Weather Research and Forecasting Model

How to represent clouds remains a top issue in numerical modeling including regional and global weather/climate models, particularly for stratocumulus in the planetary boundary layer (PBL) off the west coasts of California and PBL trade-wind shallow cumulus near Hawaii. Trade-wind shallow convection is an important cloud that represents a transition from the stratocumulus to deep convective clouds. It plays an important role in transporting moist air from PBL to free troposphere, and enhancing the deep convection downstream. Without realistic stratocumulus and shallow convection simulated, the modeled deep convective system tends to be too weak.

In this study, we evaluate two versions of Weather Research and Forecasting Model (WRF), one with Bretherton and Park shallow convective scheme (WRF_SHCU) and the other without (WRF_CU). The parameterization consists of a mass flux scheme with entrainment–detrainment plume model coupled to a 1.5-order turbulence closure model with an entrainment closure for convective boundary layers. The simulations are done for the area over Northern Western Pacific Ocean off coast of California, a covering the transition from the stratocumulus, trade-wind to ITCZ.

Results of simulated cloud properties, such as, cloud ice water content (CIWC), cloud liquid water content (CLWC), radiation, and other macro-physical fields are compared against ECMWF interim analyses as well as CERES, SRB and CloudSat observations. While the dynamical fields in both WRF simulations are similar to ECMWF reanalysis data, the model simulated CIWC/CLWC shows disagreement with ECMWF and CloudSat in both the vertical and horizontal distributions. In general, the models underestimated CIWC/CLWC in ITCZ, and over estimate at low level. Compare with ECMWF interim and WRF, the ice water path (IWP) in ITCZ is under estimated in WRF. IWP in WRF_SHCU is even lower than in WRF_CU, implying weaker deep convection and/or drier PBL. Furthermore, both WRF simulations show unrealistically high cloud cover over the transition zone. Although WRF_SHCU produced slightly less (about 10%) low cloud cover, it is still too much comparing with ECMWF and CloudSat data. This indicates that current shallow convection schemes are not yet suitable for simulating the trade-wind shallow cumulus cloud cover. Deep and mid-level convection are changes, which in turn influence high-level clouds CWC. Generally, a better agreement with the observations can be obtained. For a better understanding of the scheme's impact, more in deep examinations is needed. In this presentation, some results from simulations with and without shallow convection schemes are discussed.