Tests of RANS-based PBL schemes against LES, RUC, and tall tower data

Wind energy resource assessments and forecasts are particularly sensitive to the behavior of boundary layer schemes in the numerical weather prediction models due to their influence on the evolution of the convective boundary layer depth during the day and the distribution of momentum within the PBL, particularly in the surface layer up to 200 m AGL. These two processes strongly influence the diurnal wind distribution at the height of the wind turbine as well as the average wind speed, both of which are important factors in the financial feasibility of wind farms.

We tested the YSU and transilient turbulence parameterizations in the MM5 model against large eddy simulations, rapid update cycle model analyses, and tall tower data to determine their suitability with respect to the CBL depth, vertical momentum distribution, diurnal cycle, and average wind speed. Although the two schemes performed very similarly and provided better agreement with the above data sources than did other vertical mixing schemes, the smaller computational expense of the YSU parameterization made it preferable to the transilient scheme. It was found that the average wind speeds occasionally did not agree well with tower data. Analysis of the modeling results revealed temperature bias errors, of large spatial extent, that were consistent with the wind speed errors. MM5 simulations with the NOAH land surface parameterization option were then performed and found to produce better agreement with tower wind speed data compared to simulations with the six-layer soil temperature model.