Fitting and Time Series Neural Networks Modelling to Forecast Weather and Wind Speed

Considering 40 years of data collected from 1981 to 2020, a method for weather and wind speed specific daily forecasting was proposed. It combines fitting and time series neural networks to predict weather and wind speed in Jarab, located in the northeastern of Al-Baha region, Saudi Arabia. The developed forecasting models are based on historical data and were used to predict the weather and wind speed up to 2040. A graphical user interface was built in-house and added to the application. The application shows that the combined fitting and time series neural networks can accurately predict weather and wind speed. The metrics of the proposed model have minimum correlation coefficient values of 0.934 and 0.969 for fitting and time series neural networks, respectively. The overall correlation coefficients of the predicted and actual values are 0.9991, 0.9949, 0.9973, 0.9407, 0.9381, and 0.9335 for temperature, relative humidity, pressure, and wind speed at 2, 10, and 50 m above the ground, respectively.