Improving Monthly Precipitation Forecasting with GRU-LSTM Model in Turkey

Abstract

Protecting lives and property in Turkey requires reliable precipitation forecasts due to the country's susceptibility to extreme weather events like heavy precipitation, flash floods, and typhoons. On the other hand, Predictions enable officials to implement preventive actions and alert the community. Hence, researching forecasting precipitation in Turkey is essential. This research uses two combined deep learning hybrid models of Convolutional Neural Network, Long Short-Term Memory (CNN-LSTM), and Gated Recurrent Units-Long Short-Term Memory (GRU-LSTM) to predict the monthly precipitation of Istanbul and Konya between 2000 to 2023. From the research results, it can be concluded that the GRU-LSTM model is generally better.