A versatile and widely used tool in atmospheric research and operational forecasting is the Weather Research and Forecasting (WRF) model. The WRF model utilizes advanced algorithms and physics parameterizations to simulate complicated atmospheric phenomena. However, this model relies on a configuration that encompasses spin-up time, different Planetary Boundary Layer (PBL), and Land Surface Model (LSM) schemes to describe the physics near the surface. This work aims to optimize the configuration of the WRF model by evaluating the impact of the spin-up period, PBL, and LSM schemes to obtain the best representation of the wind field to calculate the power production of a wind farm located in the complex terrain of La Rumorosa, Baja California, Mexico. This site was chosen because of the availability of data, the challenging conditions of the terrain complexity, and severe changes in the wind direction. Results indicate that one day spin-up is the best option to predict the wind speed for one month or even one year. Moreover, the best statistical metrics are obtained when the Mellor-Yamada-Nakanishi–Niino Level 2.5 (MYNN2) PBL scheme is coupled with Noah-Multiphysics LSM. The best set-up for the WRF model was used to simulate the performance of a wind turbine in the farm with excellent results.