Snow Depth Classification using MultiSensory Ubiquitous Platform and Machine Learning

In the drastic period of climate change the continuous data monitoring of snow characteristic is required. The immensely impact of snow on hydro production, water resource management and its inhabitants, drive to the need for the importance of snow information such as its extent, dynamics and water it holds at global and local scale. At present, there are various approaches such as traditional ground-based approach, optical satellite imaging and the radio, which are available for snow monitoring at global scale. However, the use of these approaches incurs from large labor and high monitoring cost. Since, the advance in sensor technologies and Internet of Things (Iot), provides an appealing possibility to develop a framework for monitoring snow parameters at enormously low cost. In this study. we implemented two machine learning classifiers model based on the input acquired from the low-cost wearable sensor platform. The results of Random forest classifier showed the accuracy of 88.8%, indicate a promising alternative in snow depth measurements with in-situ validation, when data or wireless sensor network are not available or affordable.

Nasim Sofeem, Oussalah Mourad, Haghighi Ali Torabi, Klove Bjron

A4 Article in conference proceedings

Proceedings of the FRUCT’25, Helsinki, Finland, 5-8 November 2019

Nasim, S., Oussalah, M., Haghighi, A. T., Klove, B., Snow depth classification using MultiSensory ubiquitous platform and machine learning, Proceedings of the FRUCT’25, Helsinki, Finland, 5-8 November 2019, ISSN: 2305-7254, p. 546-551