Influence of land-use misrepresentation on the accuracy of WRF wind estimates: Evaluation of GLCC and CORINE land-use maps in southern Spain

• WRF configured with Corine Land Cover represents more accurately the spatial variability of land use than GLCC.
• The better representation of land use associated with high roughness length values allows more accurate wind estimates.
• The overestimation or underestimation of roughness length contributes to underestimation or overestimation of wind speed.
• Only land use associated with high roughness length values shows an influence on wind direction.
• The influence of land use representation on wind estimates decreases at higher measuring elevation.

In this work, we evaluate the influence of land-use representation accuracy on the reliability of wind speed and direction estimates derived from the Weather Research and Forecasting (WRF) model. To this end, the 100-m spatial resolution Coordination of Information on the Environment (CORINE) land-use dataset was implemented as static geographic data in WRF. Next, a set of one-year long simulations at 1-km spatial resolution was conducted using both the CORINE and Global Land Cover Characterization (GLCC) land-use datasets, the latter the default in WRF. The simulations were conducted for three locations in southern Spain, and were characterized by variable land-use composition and topography. At these locations, wind speed and direction estimates were compared against observations at different measurement elevations. Results showed that the selection of land-use database has a major influence on wind estimate bias. The effect on the wind direction distribution is also significant, whereas that on the standard deviation is much weaker. CORINE provided a more reliable land-use representation than GLCC. Nevertheless, as a consequence of the interpolation procedure used for land use in the domain setup, this representation did not necessarily translate to a superior roughness length, thereby affecting wind speed and direction estimates. This was particularly so for areas of high spatial variability in land-use categories. In such areas, the misrepresentation of land use may result in large wind speed estimation errors.