Day-Ahead Solar Irradiance Forecasting in a Tropical Environment

Abstract

Day-ahead solar irradiance forecasting is carried out using data from a tropical environment, Singapore. The performance of the weather research and forecasting (WRF) model is evaluated. We explore various combinations of physics configuration setups in the WRF model and propose a setup for the tropical regions. The WRF model is benchmarked using persistence and two seasonal time series models, namely, the exponential smoothing (ETS) and seasonal autoregressive integrated moving average (SARIMA) models. It is shown that the WRF model outperforms the SARIMA model and achieves accuracies comparable with persistence and ETS models. Persistence, ETS, and WRF models have relative root mean square errors (rRMSE) of about 55–57%. Furthermore, we find that by combining the forecasting outputs of WRF and ETS models, errors can be reduced to 49%.