Exploring the Efficacy of Artificial Neural Networks in Climate Prediction: Societal and Environmental Implications in Tungurahua, Ecuador

This study explores the efficacy of artificial neural networks in predicting climate variables, specifically temperature, in Tungurahua Province, Ecuador. Utilizing climate data from the Mula Corral meteorological station, three types of neural networks—Long Short-Term Memory (LSTM), bidirectional LSTM, and Gated Recurrent Unit (GRU)—were evaluated. The data set, spanning from April 1, 2013, to April 1, 2024, included measurements of humidity, wind, precipitation, and temperature. Missing data were addressed using linear interpolation to ensure continuity. The LSTM model demonstrated superior performance, achieving a Root Mean Square Error (RMSE) of 0.71, making it the optimal choice for these data. Artificial Intelligence (AI) based temperature predictions offer significant benefits for agricultural planning, enabling farmers to optimize planting and harvesting times to enhance crop yields, including potatoes, corn, and beans, which are vital for the region’s food security. Moreover, these predictions facilitate the anticipation and mitigation of adverse impacts from extreme weather events, safeguarding public health and food security. Accurate long-term climate predictions also aid in efficient water resource management, contributing to economic and environmental savings. This research underscores the transformative potential of AI in climate prediction, providing a robust foundation for informed decision-making in agriculture, resource planning, and risk management. Furthermore, the neural network approach can be adapted to other regions with similar climatic conditions, opening avenues for future research and applications in meteorology and climate change.