Abstract: The simulation effect of WRF model on the low-level wind field is closely related to the choice of parameterization schemes for the boundary layer and land surface processes. To find an optimal wind field simulation scheme suitable for a certain coastal site under different climatic backgrounds, 9 sets of experiments were designed by combining several commonly used boundary layer schemes (YSU, ACM2, MYNN 2.5) and land surface process parameterization schemes (RUC, Noah, Pleim-Xiu) in the model. The wind field in four representative months, namely, January (winter), April (spring), July (summer) and October (autumn) of 2016, was simulated, and compared with the wind observation data of 10, 30, 50, 70 and 100 m tower layer at the site. The results show that: 1) All 9 schemes can simulate the daily variation trend of wind speed and the simulated value is slightly high. For the wind speed in January, April, July and October, the root mean square error is respectively 2.61-2.95 m/s, 2.47-3.13 m/s, 2.78-3.45 m/s and 3.34-3.77 m/s, while the average absolute error is 2.27-2.64 m/s, 2.04-2.63 m/s, 2.21-2.92 m/s, 3.09-3.52 m/s respectively. 2) The simulated wind rose diagrams are in good agreement with the actual measurements. 3) The optimal wind speed simulation combination is MYNN 2.5 + Pleim-Xiu in January, April and October, and YSU + Pleim-Xiu in July. 4) There is little difference in the wind direction simulation between the schemes. 5) Pleim-Xiu in the land surface process parameterization schemes is more appropriate for the wind field simulation of this site. In the case, the accuracy of simulation results can be improved by choosing reasonable boundary layer schemes.

Key words: WRF model, Low level wind field, Boundary layer, Land surface process, Parameterization