We have used the PSU/NCAR numerical model (MM5) to investigate the effect of different boundary layer schemes on numerical simulations of the low-level jet (LLJ) over the central U.S. and subsequent convective precipitation. Three different boundary layer parameterizations, which were Blackadar's local scheme, Troen & Mahrt's non-local scheme, and Burk & Thompson's second-order closure scheme, were used to study the impact of boundary layer schemes on the formation of LLJ. The initial data used in this study came from the NCEP/NCAR reanalysis data for two events during the CASES experiment, namely May 6th and May 17th, 1997. These two events are used because our analysis shows that they exemplify the two main types of LLJs, namely LLJs that primarily reflect synoptic forcing as compared with those that have closer relationship with boundary layer processes.
The preliminary results show that the choice of boundary layer scheme has a significant effect on the formation of the predicted LLJ. The three boundary layer schemes produced substantial differences in the structure of the horizontal wind within the wind speed maximum core of the LLJ. Furthermore, these differences influenced the location of convection associated with the LLJ. These results indicate that boundary-layer schemes thus have both a "direct" effect on convection by their influence on the thermodynamic state of the atmosphere, and an "indirect" effect by their influence on mesoscale dynamical processes that determine convergence patterns and moisture fluxes.
It is noteworthy that while the schemes differed, none was clearly superior to the rest when compared to observed low-level jet development and precipitation patterns. That is, each scheme had its strengths and weaknesses for certain aspects of the problem or certain geographical regions. The results point especially toward the challenge of improved parameterizations of nocturnal boundary layer turbulence in environments that are both strongly stratified and strongly sheared