Rainfall rate prediction using recurrent neural network with long short-term memory algorithm: Iraq case study

Rainfall is one of the primary sources of water for many countries in the world. Recently, the problem of variant rainfall rates has emerged in most countries, especially in the Middle East, due to the phenomenon of global warming. Consequently, this phenomenon affected all aspects of human life, especially the agricultural sector. To address this problem, machine learning algorithms were adopted to predict rainfall in Al-Diwaniya city in Iraq. A Recurrent Neural Network (RNN) algorithm based on Long- and Short-Term Memory (LSTM) technology was applied. This technique was implemented by calculating the weight of previous observations or time shift variables in the form of time series based on the simulating neural networks. This network is trained to reach the minimum Mean Square Error (MSE) rate by adjusting the values of the estimated weights for the chosen model structure. The finding of the study showed the prediction values for LSTM are better than the RNN algorithm according to the MSE values that are obtained.