A regression-based statistical correction of mesoscale simulations for near-surface wind speed using remotely sensed surface observations

Wind speed is an important meteorological variable for various scientific communities. In this study, numerical mesoscale simulations were performed over the Republic of Korea in 2006, to produce wind information distributed homogeneously with space. Then, an attempt was made to statistically correct the simulated nearsurface wind speed using remotely sensed surface observations. The weak wind season (WWS, from May to October) and strong wind season (SWS, from November to April) were classified on the basis of the annual mean wind speed. Although the spatial features and monthly variation pattern of the near-surface wind speed were reasonably simulated in the Weather Research and Forecasting (WRF) model, the simulations overestimated the observed values. To correct the simulated wind speeds, a regression-based statistical algorithm with different constants and coefficients for WWS and SWS was developed using match-up datasets of wind observations and satellitederived variables (land surface temperature and normalized difference water index). The corrected wind speeds showed reasonable performance for both WWS and SWS with respect to observed values. The monthly variation in the corrected wind speeds over the Republic of Korea also matched better with observations throughout the year, within a monthly bias range of approximately ± 0.2 m s^?1. The proposed algorithm using remotely sensed surface observations may be useful for correcting simulated near-surface wind speeds and improving the accuracy of wind assessments over the Republic of Korea.