Performance evaluation of the WRF model in a tropical region: Wind speed analysis at different sites

Abstract

In this study, the performance of the mesoscale Weather Research and Forecasting (WRF) model is evaluated using combinations of three planetary boundary layers (PBL) (YSU, ACM2, and MYJ) and three land surface model (LSM) schemes  (RUC, Noah and Noah-MP) in order to identify the optimal parameters for the determination of wind speed in a tropical region. The state of Bahia in Brazil is selected as the location for the case study and simulations are performed over a period of eight months between 2015 and 2016 (dry and rainy seasons). The results of the simulations are compared with observational data obtained from three towers equipped with anemometers at heights of 80, 100, 120 and 150 m, strategically placed at each site and evaluated with statistical indices: MB, RMSE, MAGE, IOA, R, Fac2 and standard deviation. Overestimation of wind speed is observed in the simulations, despite similarities between the simulated and observed wind directions. In addition, the accuracies of simulations corresponding to sites that are closer to the ocean are observed to be lower. The most accurate wind speed estimates are obtained corresponding to Mucugê, which is located farthest from the ocean. Finally, analysis of the results obtained from each tower accounting for periods with higher and lower precipitation reveals that the combination of the PBL-YSU scheme with the LSM-RUC scheme yields the best results.