Registration Details    
Govt. of India Trust : E/11049/Rajkot
Income tax of India 80G : AADTK8161HF20221
Income tax of India 12A : AADTK8161HE20217

Volume 1 – Issue 1 – 2020

Original Research Article

A Long-Term Prediction Of Global Solar Radiation Over Nigeria Using The Nonlinear Autoregressive Neural Network

Olusola Samuel Ojo*, Babatunde Adeyemi

Department of Physics, P.M.B. 704, The Federal University of Technology, Akure, (NIGERIA)

PAGE NO: 9-17


In this study, surface data of minimum and maximum temperature, relative humidity, and wind speed were used as input variables to create, trainand validate the network in which global solar radiation serves as a target. These surface data were obtained from the archives of the Era-interim database of the European center for Medium-Range weather forecast (ECMWF) ata resolution of 0.25o x 0.25o. The dataspan36 years (1980-2015) for the four climatic regions of Nigeria the viz Sahel, Guinea Savannah, Derived Savannah, and Coastal regions. The research aims to evaluate the predictive ability of the Nonlinear Autoregressive Neural Network with Exogenous Input (NARX) model compared with the Multivariate Linear Regression (MLR) model. The efficiency of the two models was compared using the statistical metrics such as correlation coefficient (CC), coefficient of determination (CD), index of agreement (IA), mean bias errors (MBE), root mean square errors (RMSE) and t-statistics (T-test). Analyses showthat in the Sahel region, for instance, the NARX and MLR model have values of 0.876 and 0.450 for CC, 0.767 and 0.240 for CD, 0.813 and 0.763 for IA, 0.081 and 0.846 for MBE, 14.192 and 17.234 for RMSE and 0.4230 and 2.135 for t-test respectively. The statistical metrics for other regions followed similar patterns. Therefore, it can be inferred from these metrics thatthe NARX model gives a better prediction of global solar radiation than the traditional common MLR models in all the zones in Nigeria.