Hyper-Localized Weather Forecasting System

Abstract

Traditional weather forecasting methods are inconvenient and do not consider detecting weather patterns for remote locations. A hyper-localized Internet of Things (IoT)-based weather forecasting app using artificial intelligence can turn the disadvantage into a strength. This study attempts to overcome the shortcomings of conventional weather forecasting methods by utilizing machine learning models, i.e., Long Short-Term Memory (LSTM) with attention mechanisms, to make precise and timely predictions for a specific location. The system uses real-time information from IoT sensors to enhance the precision of the predictions and detect specific weather patterns. The system includes data acquisition using IoT devices, preprocessing techniques, feature extraction, and model training. The findings confirm the effectiveness of the attention-based LSTM model in predicting various weather parameters, of which temperature and sea level pressure obtained the maximum accuracy.