Local wind speed forecasting based on WRF-HDWind coupling

Wind speed forecasts obtained by Numerical Weather Prediction models are limited for fine interpretation in heterogeneous terrain, in which different roughnesses and orographies occur. This limitation is derived from the use of low-resolution and grid-box averaged data. In this paper a dynamical downscaling method is presented to increase the local accuracy of wind speed forecasts. The proposed method divides the wind speed forecasting into two steps. In the first one, the mesoscale model WRF (Weather Research and Forecasting) is used for getting wind speed forecasts at specific points of the study domain. On a second stage, these values are used for feeding the HDWind microscale model. HDWind is a local model that provides both a high-resolution wind field that covers the entire study domain and values of wind speed and direction at very located points. As an example of use of the proposed method, we calculate a high-resolution wind field in an urban-interface area from Badajoz, a South-West Spanish city located near the Portugal border. The results obtained are compared with the values read by a weathervane tower of the Spanish State Meteorological Agency (AEMET) in order to prove that the microscale model improves the forecasts obtained by the mesoscale model. •A method for the accurate estimation of a wind field that meets the local characteristics of the terrain is proposed.•This procedure is based on a dynamic downscaling by coupling the WRF mesoscale model with the HDWind microscale model.•The joint operation of the two models allows to improve the wind estimation provided by the mesoscale model.•The proposed dynamic downscaling supposes a slight increase in simulation time.•This improvement can be useful to improve the estimation of near-surface winds.