IMPROVING SNOW WATER EQUIVALENT RETRIEVAL WITH HIGH-RESOLUTION UAV AND GPR REMOTE SENSING SYSTEMS

Date

2026-3-5

Author

Yıldız, Semih

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

131
views

0
downloads

Cite This

Quantification of snow water equivalent (SWE) is essential for hydrological modelling, snowmelt forecasting, and climate research, yet sparse ground observations and low-resolution satellites fail to capture its spatial variability. This study develops and validates an integrated framework for high-resolution estimation and spatial extrapolation of SWE by combining unmanned aerial vehicle (UAV) photogrammetry, ground-penetrating radar (GPR), and machine learning (ML) techniques. The study was conducted in the Karanlıkdere Catchment in the Ilgaz Mountains, northern Türkiye. The study was carried out in two stages. In the first stage, UAV and GPR observations were jointly used to derive local-scale SWE estimates with high spatial accuracy. A Phantom-4 Pro UAV and a 450 MHz GPR mounted on a snowmobile were used to acquire snow depth (SD) and radar two-way travel time (TWT) data. SD was obtained by subtracting a snow-free digital surface model (DSM) from a snow-covered DSM. TWT and SD were used to estimate electromagnetic velocity and derive snow density using empirical dielectric–density relationships. The resulting SWE estimates (mean 384 mm) showed a root mean square error (RMSE) of 63 mm compared with manual snow-tube measurements. In the second stage, these SWE estimates were extrapolated to the catchment using Sentinel-1 backscatter and terrain attributes within a ML framework. Among five algorithms tested, the CatBoost model performed best (RMSE = 32 mm, R² = 0.78). The results demonstrate that integrating UAV, GPR, and ML provides an effective framework for sptaillay continuous high-resolution SWE mapping in complex mountainous terrain.

Subject Keywords

Digital Elevation Model, Ground Penetrating Radar, Machine Learning, Snow Water Equivalent, UAV Photogrammetry

URI

https://hdl.handle.net/11511/119050

Collections

Graduate School of Natural and Applied Sciences, Thesis