Abstract
Weather prediction methods have evolved significantly over the past fifty years, including advances in numerical weather prediction, high-performance computing, mesoscale modeling, assimilation of observations from new sources, and ensemble prediction systems. In recent years, machine learning techniques, including Long Short- Term Memory (LSTM) models, have shown promise in improving the accuracy of weather predictions. In this paper, we aim at investigating the use of LSTM models to improve the accuracy of open weather datasets. Our work was conducted as part of the AUGEIAS research project, which aims at capitalizing on research results in the field of Internet of Things (IoT) and Low Power Wide Area Networks (LPWAN). More specifically, the project target is to create a smart ecosystem, utilizing machine learning techniques that will enable and optimize the use of treated wastewater reuse in agriculture. Our research indicates that, within the context of the AUGEIAS project, single-step LSTM models outperform multistep models in enhancing weather prediction accuracy.