Impact of MBES Measurement Noise on Seabed Model Accuracy: A Comparative Analysis of Interpolation and Smoothing Techniques

This study investigates the accuracy of seabed digital terrain models (DTMs) generated from multibeam echo sounder (MBES) data, focusing on mitigating the impact of random measurement noise. Despite the precision of modern MBES systems, random errors inherent in the data can distort generated models, particularly when accurate interpolation methods are used, as they tend to replicate this noise. Many existing studies overlook this issue or use validation methods that do not account for such noise, leading to unreliable assessments of model accuracy. The use of a dedicated virtual sea survey simulator enables the generation of realistic test datasets with controlled noise and a known reference surface, which in turn allows for reliable accuracy assessment of the generated models, including those created using smoothing or noise-reduction techniques. The presented approach enables a rigorous and repeatable assessment of how different algorithms — and their optimised parameters — mitigate measurement noise and improve model accuracy. A total of 23 variants of widely used methods, including moving average (MA), inverse distance weighting (IDW), local polynomial interpolation (LPI), Kriging, and various smoothing filters, were tested on both synthetic and real-world surfaces. The results indicate that specific interpolation strategies, especially MA and IDW with tailored parameter settings, combined with appropriately configured smoothing filters, can significantly reduce the influence of random errors by up to 75% — thereby enhancing model accuracy. The findings support a refined approach to seabed modelling that prioritises faithful shape reproduction, noise mitigation, and input data averaging. This leads to more reliable DTMs that comply with the International Hydrographic Organisation (IHO) standards and benefit applications in chart production, marine environmental protection, and underwater construction.