Artificial neural networks approach for forecasting of monthly relative humidity in Sivas, Turkey
Abstract
Relative humidity is a crucial parameter for various agricultural and engineering
applications and atmospheric dynamics; hence its accurate and reliable estimation is essential.
This study aims to predict monthly relative humidity by means of the artificial neural networks
(ANNs) method using neighbouring data in Sivas Province, Turkey. Nineteen years (2000-
2018) monthly mean relative humidity data of five measurement stations was used for ANN
analysis. The prediction accuracy of the ANN models was evaluated with the coefficient of determination
(R2), mean absolute error (MAE), mean absolute percentage error (MAPE) and root
mean squared error (RMSE). Contour plot maps were also generated for visual comparison.
R2, MAE, MAPE and RMSE values ranged between 0.952-0.965, 1.916-2.586, 3.422-4.974
and 2.472-3.391, respectively. The results showed that the ANN method provided satisfactory
predictions for relative humidity.