Optimizing interpolation methods and assessing accuracy of UAS-Echo sounder data for enhanced bathymetric mapping in inland waters

Recent advancements in bathymetric data collection, particularly the miniaturization of sensors and their integration with unoccupied aerial systems (UAS), have revolutionized 3D data acquisition using active remote sensing. This study evaluates the feasibility and accuracy of UAS-mounted echo sounders for bathymetric mapping in shallow inland waterbodies – an environment where traditional bathymetric techniques are often costly and inefficient. To assess data reliability, UAS-echo sounder depth measurements were compared to in-situ reference data across a dataset of 11,439 points, yielding a root mean square error (RMSE) of 10.20 cm and an R-squared value of 72.8%. Field surveys were conducted at Pond D1 at Mississippi State University’s Delta Research and Extension Centre in Stoneville, Mississippi. UAS echo sounder data were collected on 21 August 2024, and the drained-pond LiDAR reference survey was conducted on 9 September 2024. When restricting the analysis to 95.45% of the dataset (within two standard deviations), accuracy improved to an RMSE of 6.59 cm with an R-squared value of 86.6%. Given the sparse nature of single-beam echo sounder data, interpolation is essential for generating continuous bathymetric surfaces. Among 10 interpolation methods tested, the Topo to Raster interpolation method outperformed others, achieving an RMSE of 6.30 cm (R2 = 82.9%) for 5-metre flight line spacing and an RMSE of 7.10 cm (R2 = 81.5%) for 10-metre spacing. Additionally, the study examined how varying sampling intervals along flight lines impact data accuracy. Results indicate that 5-metre and 10-metre sampling intervals offer the best balance between data density and survey efficiency, maintaining RMSE values between 6.6 cm and 8.1 cm with R-squared values consistently above 70%. However, larger intervals (20–30 metres) significantly reduced data accuracy, particularly in complex underwater environments. These findings provide a framework for optimizing UAS-echo sounder surveys for efficient, high-resolution bathymetric mapping in shallow waters, while revealing how interpolation methods behave under data fundamentally distinct from terrestrial DEM assumptions.