Use of Contactless Spatial Data Collection Methods for Snow Cover Monitoring: Case Studies from Czech Mountains

Abstract

Accurate documentation of snow cover is critical for hydrological modeling, climate adaptation planning, and risk assessment in mountainous regions. This study presents a comprehensive methodology for snow cover monitoring using UAV-based LiDAR scanning, tailored to the specific environmental and technical constraints of Central European mountain ranges. Field campaigns were conducted across several Czech border mountain locations (Ore Mountains, Giant Mountains, Beskids Mountains), utilizing DJI Matrice 300 RTK equipped with Zenmuse L1 or L2 LiDAR sensors. Due to limitations in deploying traditional ground control points (GCPs) in remote and protected areas, the methodology emphasizes reliance on GNSS RTK corrections and minimal GCP use. The influence of two GNSS reference networks (CZEPOS and TopNet) was evaluated through photogrammetric analysis, revealing systematic elevation biases and spatial autocorrelation, with TopNet yielding slightly better results.

Various point cloud post-processing workflows were tested, including smoothing and noise filtering in DJI Terra, TerraSolid, and CloudCompare. The best visual and statistical results were obtained using a combined approach supplemented by a single foldable GCP. Ground point classification methods were assessed in both snow-free and snow-covered conditions. The most reliable method for snow-free filtering was the Spatix-based algorithm in TerraSolid, while snow-covered scenes required custom multi-criteria filtering in CloudCompare.

Validation was performed using over 4,500 RTK GNSS ground points and manual snow probe measurements. The methodology proved robust despite uncertainties from vegetation interference and manual measurement limits. This study delivers practical guidelines for operational snow cover documentation under constrained field conditions, and proposes improvements for future automation and validation.