Mapping topo-bathymetry of transparent tufa lakes using UAV-based photogrammetry and RGB imagery.

Seasonal or interannual precipitation differences lead to changes in the water level of a tufa lake, while the underwater topography affects its local depth. Therefore, topo-bathymetry is important to study and protect the aquatic environment of tufa lakes. However, traditional field-based topo-bathymetric surveying methods (e.g., sounding rod or sonar) inevitably disturb the fragile lake ecosystem. In recent years, the emerging remote sensing technology of unmanned aerial vehicles (UAV) has provided a cost-effective solution for measuring topo-bathymetry without disturbance. In this paper, taking Spark Lake in Jiuzhaigou, China, as an example, we captured red-green-blue (RGB) images using a fixed-wing UAV and produced a digital elevation model (DEM) prior to the Jiuzhaigou Earthquake using Structure-from-Motion (SfM) photogrammetry. The underwater topography of Spark Lake was obtained by refraction correction and water color inversion based on the DEM and orthophoto, respectively. For refraction correction, a water depth correction model based on Snell's Law was used. For water color inversion, general band ratio models were replaced by a band difference model (blue band - green band). The qualities of the resulting DEMs produced by the two methods were evaluated against the topography of the drained Spark Lake after the earthquake, and the corresponding DEMs of difference (DoD) were also analyzed. The coefficient of determination (R2) and root mean square error (RMSE) are 0.88 and 1.32 m for refraction correction, and 0.86 and 1.37 m for water color inversion, respectively. The results demonstrated the feasibility and effectiveness of applying UAV-acquired RGB imagery and the two optical remote sensing methods to topo-bathymetric mapping of transparent tufa lakes. • UAV-acquired RGB images were applicable for testifying topo-bathymetry of tufa lake. • Refraction correction demonstrated a good performance for water depths up to 12 m. • The difference between blue and green bands was the optimal spectral combination. • Water color inversion exhibited a fine performance across the entire lake depth range. [ABSTRACT FROM AUTHOR]