Analysis of the Impact of Mixing-Length Formulations on Wind and Temperature Predictions in Two TKE PBL Schemes

Recent results from Zhang et al. (2004) show that the turbulent kinetic energy (TKE) predicting Gayno-Seaman (G-S; Shafran et al. 2000) turbulence scheme reproduces well the observed time variations of surface temperature (T_sfc) and the vertical profile of potential temperature. However, it also produces pronounced phase errors in the diurnal cycle of surface wind speed (V_sfc). This motivates us to investigate why V_sfc behaves differently using G-S versus similar TKE-based schemes.

The nonhydrostatic MM5 model is applied on 20-21 May 2002 to compare two planetary boundary layer (PBL) schemes: G-S and the Miller-Yamada-Janjic (Eta; Janjic 1997, 2001) scheme. Results from MM5 experimentation show that the G-S scheme is superior to the Eta scheme for predicting daytime variations of T_sfc and mixed layer height, but it underestimates daytime V_sfc. In particular the G-S scheme shows a distinct phase lag of 4-6 h in the simulation of diurnal variations of V_sfc, which is consistent with the results of Zhang et al. (2004). In addition, the Eta scheme also underestimates afternoon V_sfc and PBL height but produces smaller phase errors in V_sfc. Comparisons of profiles of wind speed, eddy diffusivity and momentum flux between model simulations from the two schemes suggest that different formulations of mixing length may contribute to these differences in V_sfc. We plan to further investigate the effect of the mixing length formulations on the performance of the G-S scheme.