Evaluation of the short and medium-term forecast quality of global solar irradiance from GFS-MOS and WRF-Solar models for the northeast region of Brazil

Abstract

The prediction of global horizontal irradiance (GHI) is crucial due to the rapid growth of photovoltaic energy, a sustainable and increasingly important source in the global energy landscape. The Northeast region of Brazil has exceptional potential for solar energy generation, thanks to its high levels of solar radiation throughout the year. This underscores the need for accurate irradiance forecasts to optimize the use and efficiency of photovoltaic systems in the region. This study aims to predict GHI over short- and medium-term horizons using the Global Forecast System (GFS), the Weather Research and Forecasting-Solar (WRF-Solar) model, and a hybrid method called GFS-MOS for the Northeast of Brazil (NEB). The GFS-MOS combines GFS forecasts with the Model Output Statistics (MOS) algorithm, which uses observational data for statistical refinement. The results demonstrate that GFS-MOS outperforms WRF-Solar in most aspects, particularly in simulating low GHI values, where GFS-MOS demonstrated superior accuracy. GHI forecasts were validated against data from 137 meteorological stations of the National Institute of Meteorology (INMET) for the period from 2020 to 2022. Additionally, the GFS-MOS method further improved the accuracy of GHI predictions. The performance of the GFS, the GFS-MOS, and the WRF-Solar was compared, with the GFS-MOS demonstrating the best overall performance. It is concluded that the application of statistical refinement techniques, such as GFS-MOS, can significantly improve solar irradiance forecasting.