Optimizing solar energy forecasting in Senegal: Sensitivity analysis of WRF-solar radiation schemes

Abstract Solar irradiance forecasting is essential in maximizing solar energy utilization and facilitating the shift towards an eco-friendly and sustainable energy future. However, accurate solar irradiance from numerical weather prediction remains challenging. This research evaluates the performance of state-of-the-art Weather Research and Forecasting tailored to the solar energy applications (WRF-Solar) at two operational solar plants in Senegal: Diass and Ten Merina. The experiments include different shortwave radiation schemes (Dudhia and RRTMG) with RRTMG coupled with dynamic aerosols. In addition, the impact of shallow convection on the different experiments is investigated. A total of six simulations is performed and assessed under various sky conditions using hourly GHI measurements for 2020. Results indicate that the RRTMG scheme coupled with aerosols outperforms other simulations, exhibiting a maximum correlation (R) of 0.85, skill score (SS) of 0.17, and the lowest RMSE value (160 W/m 2 ) and MAE (110 W/m 2 ). However, WRF-Solar exhibits poor performance across all experiments (RMSE = 386 W/m², R = 0.55, SS = -1.48) under cloudy skies. The influence of the shallow convection scheme in the WRF-Solar model on GHI estimation was found to be limited under different atmospheric conditions at both sites. These findings offer valuable insights that can enhance solar energy forecasting accuracy and support reliable solar power generation and renewable energy optimization, benefiting energy providers, policymakers, and communities in Senegal.