Multi-Level Multi-Sensor Data Fusion in Bathymetric Modeling: A Systematic Review

Abstract

Accurate subaqueous morphological modelling is crucial for sustainable aquatic ecosystem management, yet single-sensor hydrographic surveys remain constrained by environmental limitations. Through a systematic review, this study aims to evaluate the efficacy of multi-level, multi-sensor data fusion approaches (acoustic, optical, LiDAR, and satellite) in bathymetric modelling. A search of five electronic databases (Scopus, Web of Science, IEEE Xplore, ScienceDirect, SpringerLink) up to 14 August 2026 was conducted to identify relevant peer-reviewed studies. From 685 initial records, 62 studies were included for systematic synthesis. The eligible studies’ methodological quality was assessed via a customized six-domain risk-of-bias framework. The results indicated significant improvements in spatial coverage and accuracy at complex land–water transition zones compared to single-source baselines, reducing depth retrieval errors by 18% to over 50% and achieving sub-decimeter accuracies (RMSE < 0.10 m). Feature-level Geospatial Artificial Intelligence (GeoAI) further enhanced nonlinear environmental modelling and benthic habitat classification (accuracies > 98%). This review identified that multisensor fusion effectively enhances bathymetric accuracy, providing a critical foundation for sustainable water-resource management. While standardized uncertainty quantification remains an operational bottleneck, future interventions integrating predictive 4D Digital Twins show immense promise for long-term ecological monitoring, flood-risk mitigation, and climate-resilient infrastructure planning.