IoT data analysis for avalanche forecast

Abstract

In mountainous regions worldwide, avalanches pose a significant hazard to infrastructure and human life. Avalanche identification and prediction have relied on labor-intensive and error-prone human observation and analysis. Artificial intelligence (AI) tools have become a promising means of improving the precision and speed of avalanche detection and prediction in recent years. To address this problem, we are developing an early warning system that utilizes Machine Learning techniques. This article examines the most recent research in AI techniques for avalanche detection and prediction, along with the details of our solution implementation. The system gathers and analyses data from various sensors, including meteorological and auditory sensors, to determine the potential risk of an avalanche. An analysis made on the data collected from the sensors placed in the selected region is also presented. Recent investigations have shown that AI algorithms can identify and forecast avalanches. The article discusses the potential benefits and challenges of implementing an early warning system for avalanches, highlighting the importance of continued research in this area. The research was carried out over 2 years within the MEWS project and is based on the fruitful collaboration between all partners involved in the project. The IoT stations used in the research were located in Norway. The current paper presents an analysis of the data collected in the project and a method for avalanche prediction using Feed Forward Neural Networks.