Analysis of morphological changes under controlled environmental conditions using UAV image-based bathymetry in very shallow waters

Abstract. Coastal environments present a challenge to traditional hydroacoustic methods (e.g. echo sounders) primarily due to their very shallow depth. Therefore, fast and effective methods of reconstructing the seafloor morphology are needed. This study aimed to evaluate the accuracy, uncertainty and reliability of the proposed UAV–SfM-based processing with geometric refraction correction for detecting morphological changes in a controlled shallow-water environment. The analysis focused on the impact of flight altitude (12 m and 61 m), water surface modelling and uncertainty propagation on classifying erosion, accumulation and non-change zones. Four UAV data acquisition campaigns were conducted. Total uncertainty was defined as a combination of instrumental errors, photogrammetric model accuracy, and water surface determination accuracy. Based on this, the LoD95 thresholds used for change detection with the M3C2 method were calculated. The obtained LoD95 values confirmed that changes exceeding ~2–3 cm can be reliably detected. The classification consistency between data sets obtained from low and high flight height reached an average overall accuracy of 77%, proving the stability and robustness of the developed process. These results demonstrate the efficiency and repeatability of the proposed workflow, which allows for the quantitative assessment of morphological changes in controlled shallow water conditions, with potential applications in similar natural environments.