Development of an AI-assisted UAV-based LiDAR and multispectral survey workflow for precision coastal geomorphology mapping and change detection.

Project Abstract

This study presents an integrated AI-assisted UAV-based LiDAR and multispectral workflow engineered to enable high-precision coastal geomorphology mapping and change detection. The research responds to the growing demand for rapid, repeatable, and scalable coastal monitoring tools that can quantify shoreline dynamics, dune migration, cliff retreat, and sediment transport under evolving climate and anthropogenic pressures. A multi-sensor data fusion framework is developed to harness the complementary strengths of LiDAR’s accurate topographic elevation and multispectral imagery’s spectral discrimination for substrate and vegetation mapping. The workflow comprises data acquisition protocols, preprocessing pipelines, and AI-driven analytical modules designed to deliver seamless end-to-end processing from field campaigns to geospatial outputs suitable for decision support. Key methodological innovations include the development of an automated flight planning system that optimizes UAV mission parameters (altitude, overlap, and sensor sequencing) to maximize data quality while minimizing operational time. A robust LiDAR point cloud processing chain integrates ground classification, noise reduction, and precise geo-referencing using RTK/PPP methods to produce high-accuracy bare-earth Digital Elevation Models (DEMs) and DSMs. The multispectral data are atmospherically corrected and radiometrically calibrated, enabling reliable calculation of vegetation indices and coastal material classifications. A novel AI-based fusion strategy leverages deep learning to co-register LiDAR-derived surfaces with spectral features, enabling accurate discrimination of sand, mud, rock, vegetation, and man-made structures even in morphologically complex shorefaces. For change detection, the project introduces a time-aware neural architecture that analyzes multi-temporal DEMs, surface models, and spectral composites to quantify shoreline position shifts, dune morphology changes, and cliff erosion rates with quantified uncertainty. The workflow supports hierarchical change analysis at multiple spatial scales—from centimeter-level transect measurements to grid-based coastal zone summaries—facilitating both local management actions and regional planning. A comprehensive validation framework employs independent ground truth, terrestrial laser scans, and high-resolution orthophotos collected across multiple campaigns to assess accuracy, precision, and repeatability. Sensitivity analyses examine the influence of UAV flight geometry, sensor calibration, and temporal sampling on change metrics. The outputs include detailed coastal habitat and geomorphological maps, high-fidelity DEM time series, change maps with velocity fields, and uncertainty estimates that support risk assessment and shoreline management decisions. The study also explores transferability across different littoral environments and discusses operational considerations, including data management, computational requirements, and workflow automation. Findings demonstrate that the AI-assisted UAV workflow significantly enhances the reliability and efficiency of coastal monitoring, reducing the time from data collection to actionable insights while maintaining rigorous accuracy standards. The research contributes to methodological advances in survey geoinformatics and provides a scalable framework for policymakers, scientists, and engineers engaged in proactive coastal resilience planning.

Project Overview

What This Project Is About

---

A practical exploration of how drones equipped with laser and camera sensors can map coastlines more accurately. The project combines simple data collection with basic computer tools to create clear maps showing how coastlines change over time.

---

---

The Problem It Addresses

---

Coastlines slowly erode or build up due to waves, tides, and storms. Traditional surveys can miss small changes or be tedious to repeat. This project seeks a faster, repeatable way to monitor coastal changes and identify areas at risk.

---

---

Objectives of the Project

---

1. Learn how to plan and conduct UAV flights for coastal surveys.
2. Capture accurate 3D and color data of shorelines using LiDAR and multispectral sensors.
3. Process data to create up-to-date maps and detect changes over time.
4. Apply simple AI methods to help classify coastal features (sand, vegetation, water).
5. Evaluate the accuracy of the resulting maps against reference data.
6. Provide an easy-to-use workflow that others can replicate.

---

---

What You Will Do Step by Step

---

1) Learn basic UAV flight planning and safety. 2) Collect LiDAR and multispectral data over a chosen coast. 3) Preprocess data to remove errors and align datasets. 4) Build simple 3D models and color maps of the coast. 5) Run change-detection analyses to compare recent data with past data. 6) Validate results with ground checks or reference maps. 7) Document the workflow and create user-friendly guidelines. 8) Present findings with clear visuals and a short report.

---

---

Expected Outcome

A repeatable, beginner-friendly workflow that produces coastal maps showing erosion or accretion, with basic change-detection results. The project should yield practical insights for coastal monitoring teams and can inform planning and risk awareness.