Temperature Prediction as a Climate Change Factor in Coastal Area

Temperature, which is considered to be the most important facto of climate change in any region, is always the focus of climate change researchers. One of the smart methods to predict this factor is artificial neural networks. In this research, the temperature variable was predicted as the input of the neural network model. The information used in this research includes the daily temperature of ۲۵ years related to the statistical period from ۱۹۹۹ to ۲۰۲۳, in this research the network was first proposed in its simplest form. The simplicity of the network reduced the number of trainable parameters of the network and increased the training speed of the neural network. The increase of this complexity to the network happened until the network was in overfitting mode, and then the trial and error method and changing the network training parameters were used to prevent overfitting. Finally, a network was selected that included ۳ layers and the number of neurons [۵, ۶, ۵]. The results of this research indicate the acceptable performance of the neural network model in temperature prediction, provided that the number of training parameters is selected.

Keywords:

Temperature , Prediction , Coastal Area