Real-time Snow Depth Measurement: Development and Construction of a System to Conduct Vehicle- Based Measurements of Snow Depth in Real-Time

Abstract

This thesis presents a ground-based solution for measuring snow depth in real-time, specifically applicable to a vehicle. The primary motivation behind this study is the desire to enhance efficiency in snow-related domains through precise monitoring of snow depth. Typically, ground penetrating radars (GPR) have been used in conjunction with human operators to monitor ice and snow layers. However, this approach incurs costs in terms of both time and performance, leading to potential errors and uncertainties. To address these challenges, this research explores the integration of artificial neural networks with GPR technology, aiming to boost their effectiveness in measuring snow depth. Various sensors, simulation methods, and neural network models were examined to develop high-performance solutions. Through extensive testing, a trained convolutional neural network achieved an accuracy of 1 cm on controlled experiments, showcasing the potential of combining these approaches. By enabling the system to conduct 20 depth measurements per second with this level of precision, a real-time solution for measuring snow depth is achieved. Furthermore, the thesis proposes that incorporating the propagation speed of signals through the measured snow can further enhance the reliability of these monitoring systems. Such enhancements would provide valuable data to various fields that rely on accurate information about snow depth. By improving the overall efficiency and accuracy of snow depth measurements, this research opens up possibilities for advancements in snow-related domains.