P68 Verification of the WRF topographic parameterization for surface wind speed bias over the Iberian Peninsula

Lorente-Plazas, R., University of Murcia, Spain, P. A. Jiménez, National Center for Atmospheric Research (NCAR) and CIEMAT, Spain, J. Dudhia, NCAR, J. P. Montávez, and P. Jimenez Guerrero, University of Murcia, Spain

The Weather Research and Forecasting (WRF) model introduces a bias for the surface wind speed when we compare to observational records. Several studies have pointed out that the WRF tends to overestimate the wind speed independently of the WRF version, PBL scheme, world region, horizontal or vertical resolution, the input data... Less studies pay attention is the underestimation on hills and mountains. To overcome this drawback, a new parameterization has been developed by Jimenez and Dudhia (2012). This new scheme parametrizes the subgrid scale orographic effects on momentum. The aim of this work is testing the improvements of this parameterization over the Iberian Peninsula. For this task, two simulations have been performed, without (Reference simulation) and with (New simulation) the topographic effects activated. The simulations encompass 2005 with 10km spatial resolution, using WRF 3.4 version modified to obtain the wind speed at 2m. These simulations are validated by comparison with 400 weather stations with wind speed records at 2m.

The preliminary results show a systematic overestimation for the surface wind speed in the Reference simulation, more noticeable at the coastline. When the topographic scheme is activated, there is a remarkable bias reduction with an increase of the number of stations where the wind is underestimated. This underestimation is more remarkable in the windiest months, mostly in the plain areas of the IP. The analysis of the diurnal cycle shows a good agreement with the observations during the night, but there is an underestimation during the daytime. These results suggest that the influence of the sub-grid drag under unstable conditions brakes the wind speed more than the observations. In order to investigate this underestimation two analysis methods are proposed. First, we activate the topographic effects only during the stable conditions and switch off this scheme in the unstable conditions. Second, we study this wind bias variation with the structure of turbulence by assessing the correlation with convective velocity.