摘要
Accurate simulation of variations in planetary boundary layer (PBL) depth is important for weather prediction and climate studies as well as for carbon cycle analysis. The PBL is difficult to represent in global models because of the need to represent strong gradients associated with capping inversions at arbitrary heights everywhere. Bulk parameterizations of boundary layer processes are therefore an attractive solution. We evaluated a bulk PBL parameterization locally by prescribing horizontal advective tendencies from high-frequency regional meteorological analyses, and running the PBL formulation in a single-column version of a climate model. We compared simulated variations in PBL depth to observations of radar reflectivity and vertical profiles of CO2 made at a tall tower in northern Wisconsin during 1999. The model captures many features of the observed diurnal and synoptic variability, but tends to underestimate mid-day maxima in PBL depth. Observed late afternoon collapse of the PBL due to decoupling from an underlying stable surface layer is not simulated. The model underestimates mid-day mixing during calm conditions, suggesting underestimation of buoyancy forcing. Conversely, it overestimates PBL depth under windy conditions, suggesting the parameterization is overly sensitive to shear forcing. Global model simulations cannot be compared to specific dates, but monthly mean diurnal cycles show reasonably good agreement to observations at this site. The simulated PBL in the GCM is generally too shallow at mid-day during the summer months, but is well simulated in spring (when it is deeper than summer) and autumn (when it is shallower than in summer). Seasonal rectifier forcing is slightly underestimated by the model at this site.