Comparative assessment of bathymetric methods using unmanned aerial vehicle (UAV) high-resolution multispectral imaging

Abstract

Unmanned Aerial Vehicles (UAVs) equipped with multispectral sensors offer a promising, cost-effective alternative for high-resolution bathymetric mapping in dynamic coastal environments. This study evaluates the feasibility of adapting two established satellite-derived bathymetric (SDB) methods, the Stumpf (S) and Caballero-Stumpf (CS), to UAV-based multispectral imagery (UDB, UAV-Derived Bathymetry). Four UAVs flights are conducted over Luarca (Northern Spain), capturing data from a port and an adjacent beach with varying turbidity and environmental conditions. Results indicates that the UDB_green model, based on the green band reflectance, consistently outperforms in accuracy the UDB_red model, based on the red band, with median absolute errors (MedAE) ranging from 0.41 to 0.67 m for depths up to 7 m. Conversely, UDB_red exhibits poor performance in these waters. A composite methodology integrating multiple UAV flights is also tested through the first UAV-based implementation of the CS compositing method, originally developed for satellite imagery to address turbidity. However, it does not yield significant accuracy improvements over the traditional S model or single-image results, highlighting the influence of environmental factors and flight-specific parameters on UDB data quality. Given these findings, and considering that UAV platforms offer key operational advantages, such as higher spatial resolution, flexible acquisition timing, and better suitability for small or cloud-prone coastal areas, there is a strong need further validation of UAV-based techniques under varied coastal conditions.